© Bachar Cheaib, 2020
Étude de l’évolution contemporaine de systèmes microbiens environnementaux et hôtes associés dans
un contexte d’écotoxicologie
Thèse
Bachar Cheaib
Doctorat en biologie
Philosophiæ doctor (Ph. D.)
Québec, Canada
Étude de l’évolution
contemporaine de systèmes
microbiens environnementaux et
hôtes associés dans un contexte
d’écotoxicologie
Thèse
Bachar Cheaib
Doctorat en biologie
Philosophiae doctor (Ph.D.)
Sous la direction de :
Nicolas Derome, directeur de recherche
ii
Résumé
Les microbes ou micro-organismes sont les producteurs primaires des
services écosystémiques pour les cycles biogéochimiques de la terre et les
systèmes biologiques. Les xénobiotiques marquent une nouvelle ère
anthropogénique « l’anthropocène », et ils représentent une source de
sélection artificielle de la structure et de la composition de la biodiversité
microbienne. Par conséquent, les perturbations anthropogéniques sont
néfastes pour les systèmes microbiens et induisent des changements
adaptatifs ou des dommages dans leurs répertoires génotypiques.
L’assemblage des communautés microbiennes durant la résistance et la
résilience est gouverné par des processus éco-évolutifs.
Ce travail découle de l’intersection transdisciplinaire de l’écotoxicologie,
l’écologie microbienne, la métagénomique et la bioinformatique. L’objectif de
ce travail consiste à étudier les signatures adaptatives de la résistance et de
la résilience microbienne selon deux modèles. Le premier est
environnemental (E) composé d’un bassin versant lacustre contaminé par des
métaux lourds. Le deuxième modèle est hôte-associé (HA), constitué d’un
système expérimental d’exposition de la Perchaude (Perca flavescens) au
chlorure de cadmium selon deux régimes constant et graduel.
Trois nouveautés résument les travaux de cette thèse de doctorat.
Premièrement, le phénomène de découplage taxon-fonction a été démontré
pour la première fois, dans le système E sous un gradient sélectif de pollution,
et au sein du microbiote cutané dans le système HA durant sa période de
résilience.
Deuxièmement, des altérations significatives de la diversité taxonomiques et
fonctionnelles mettent en évidence des signatures adaptatives du résistome
et de l’érosion des fonctions métaboliques dans le système E. Quant au
système HA, le stress métallique a augmenté la prévalence significative de
souches pathogènes et des opportunistes avec une dysbiose cutanée de la
perchaude accompagnée par une réduction de sa capacité de résistance à
une colonisation bactérienne massive.
iii
Troisièmement, la modélisation de l’assemblage bactérien de microbiote du
système HA montre des rôles confondus de l’ontogenèse et de la force de
sélection durant la période de résistance. La persistance des effets à long
terme de la sélection durant le stade de résilience a été expliquée par une
augmentation inattendue de la bioaccumulation du cadmium dans les tissus
hépatiques de l’hôte.
En conclusion, nos travaux montrent que l’adaptation des répertoires
métagénomiques peut être décelée par le phénomène de redondance
fonctionnelle observée à l’échelle de découplage taxon-fonction, ce qui reflète
potentiellement une stratégie adaptative par transfert horizontal de gènes
partagés entre les communautés microbiennes environnementales sous
perturbation graduelle.
Dans le système HA, l’assemblage de microbiote montre un gradient de
processus neutres et non neutres. Enfin, la dérive taxonomique serait une
force écologique non négligeable plus importante dans le système
environnemental que dans le système intestinal durant et après la
perturbation.
iv
Abstract
Microbes or microorganisms are the primary producers of ecosystem services
for biogeochemical cycles of the earth and biological systems. Xenobiotics
mark a new anthropogenic era, "the Anthropocene," and they represent a
source of artificial selection of the structure and composition of microbial
biodiversity. As a result, anthropogenic disturbances are detrimental to
microbial systems and induce adaptive changes or damage in their
metagenomic repertories. During resistance and recovery, the ecological
processes governing the assembly of microbial communities cannot be
dissociated from those of microbial evolution.
This work stems from the transdisciplinary intersection of ecotoxicology,
microbial ecology, metagenomics and bioinformatics. The main goal is to
understand the adaptive signatures of microbial resistance and resilience in
two models. The first is environmental (E) composed of a lake-bound
watershed contaminated by heavy metals. The second model is host-
associated (HA), consisting of an experimental system of perch (Perca
flavescens) intoxicated with cadmium using two steady and gradual regimes.
Three novelties summarize the work of this doctoral thesis. Firstly, the
phenomenon of taxon-function decoupling has been demonstrated for the
first time, in the E system under selective pollution gradient, and second,
within the cutaneous microbiota in the HA system during its recovery stage.
Third, the microbiota assembly modelling in the HA system suggested mixed
effects of ontogenesis, and selective pressure during the period of resistance
and recovery. The increase in cadmium bioaccumulation in liver tissues of
perch can argue the persistence of the long-term effects of selection during
the recovery stage.
In conclusion, our work showed that the adaptation of microbial metagenomic
repertories could be revealed through functional and taxonomic redundancy
patterns observed at the scale of taxon-function decoupling. The gap between
functional and taxonomic diversity reflects an adaptive strategy by horizontal
gene transfer among environmental communities microbial under gradual
disruption.
v
In the HA system, the microbiota assembly shows a gradient of neutral and
non-neutral processes. Finally, the taxonomic drift is a significant ecological
force, more effective in the environmental system than in the intestinal
system during and after the disruption.
vi
Table des matières
Résumé ................................................................................................................ ii
Abstract .............................................................................................................. iv
Liste des tableaux .......................................................................................... x
Liste des Schémas ......................................................................................... xi
Liste des figures ............................................................................................ xii
Liste des abréviations ................................................................................ xiii
Remerciements .............................................................................................. xvi
Introduction Générale .................................................................................. 1
L’anthropocène: des effets anthropiques sur la biodiversité et
l’assemblage des communautés microbiennes ......................................... 2
Les questions et les chapitres de cette thèse ............................................ 4
Chapitre 1 – Méthodes et concepts de base en écologie
microbienne ...................................................................................................... 8
1.1 Les microbes sont ubiquitaires et l’environnement les filtre ... 9 1.2 Accès à la biodiversité microbienne, de la boite de Pétri au
métagénome ....................................................................................................................... 9 1.3 L’avènement du séquençage de nouvelle génération ................. 10 1.4 Les approches de la métagénomique .................................................... 12
1.5 L’approche d’amplicons basée sur un gène marqueur
universel .............................................................................................................................. 12 1.6 L’approche métagénomique globale basée sur le séquençage
de l’ADN total. .................................................................................................................. 14 1.7 Approche de l’opération taxonomique universelle (UTO) ........ 14 1.8 L’approche des variants des séquences d’amplicons (VSA).... 16 1.9 Les mesures écologiques et phylogénétiques de la diversité. 17 1.9.1 Les mesures de la diversité alpha ...................................................... 17 1.9.2 Les mesures de la diversité Beta ......................................................... 19 1.9.2.1 Les distances écologiques ..................................................................... 19 1.9.2.2 Les distances phylogénétiques .......................................................... 20 1.10 Réseaux d’interactions microbiennes ............................................... 20 1.11 Les quatre forces évolutives et processus écologiques ......... 22 1.11.1 La mutation ..................................................................................................... 22 1.11.2 La sélection ..................................................................................................... 24 1.11.3 La migration .................................................................................................... 26 1.11.4 La dérive génétique .................................................................................... 27 1.11.5 Interactions des forces évolutives .................................................... 28 1.11.6 De l’évolution à l’écologie ...................................................................... 30 1.11.6.1 L’adaptation Locale ................................................................................ 30 1.11.6.2 L’adaptation rapide ................................................................................ 31 1.12 Processus écologiques et modèles ..................................................... 32
vii
1.13 Introduction à la modélisation mathématique en écologie et
évolution microbienne................................................................................................. 33
1.13.1 Préambule historique de la modélisation ...................................... 34 1.13.2 Aperçu sur les modèles d’assemblage microbiens .................. 37 1.13.2.1 Les modèles métaboliques ................................................................. 37 1.13.3 Les modèles kinétiques ............................................................................ 38 1.13.4 Les modèles spatiaux ................................................................................ 38 1.13.4.1 Modèles basés à l’échelle individuelle ........................................ 39 1.13.4.2 Modèles basés à l’échelle populationnelle ............................... 39 1.13.4.3 Modèles basés à l’échelle de communauté .............................. 40 1.13.5 Les modèles neutres en écologie microbienne .......................... 40 1.13.6 Le modèle neutre de Sloan .................................................................... 42 1.13.7 Neutralisme et déterminisme dans l’assemblage du
microbiote de l’hôte. .................................................................................................... 45 1.13.8 Les modèles éco-évolutifs ...................................................................... 46 1.14 Récapitulatif ..................................................................................................... 48
Chapitre 2: Taxon-function decoupling as an adaptive
signature of lake microbial metacommunities under a chronic
polymetallic pollution gradient .............................................................. 50
2.1 Résumé ................................................................................................................... 51 2.2 Abstract .................................................................................................................. 52 2.3 Introduction ......................................................................................................... 53 2.4 MATERIALS AND METHODS ......................................................................... 57 2.4.1 Lake characteristics and locations ..................................................... 57 2.4.2 Metallic and chemical gradient surveys .......................................... 58 2.4.3 Water sampling .............................................................................................. 58
2.4.4 DNA extraction and metagenome sequencing ............................ 58 2.4.5 Bioinformatic and statistical analysis ............................................... 58 2.5 Results ..................................................................................................................... 61 2.5.1 Decoupling taxon-function ...................................................................... 61 2.5.2 Detangled taxonomic structure and function diversity ......... 62
2.5.3 Canonical correlations of taxon and function .............................. 62 2.5.4 Taxonomic variation signatures........................................................... 63 2.5.5 Role of trace metals in taxonomic variation signatures ........ 64 2.5.6 Function variation signatures ............................................................... 64 2.5.7 Role of trace metals in function variation signatures ............. 66 2.6 Discussion ............................................................................................................. 67 2.6.1 Decoupling taxon-function as a signature of adaptive
strategies ............................................................................................................................ 67 2.6.2 Taxonomic adaptive signatures ........................................................... 69 2.6.3 Functional adaptive signatures ............................................................ 71 2.7 Conclusions ........................................................................................................... 73 2.8 Figures ..................................................................................................................... 74 2.9 Supplementary figures .................................................................................. 82 2.10 Supplementary Material ........................................................................... 94
Chapitre 3: From networks to models: The Yellow Perch
(Perca flavescens) microbiome assembly under metal toxicity
.............................................................................................................................. 96
3.1 Resumé ................................................................................................................... 97
viii
3.2 Abstract .................................................................................................................. 98 3.3 Introduction ......................................................................................................... 99 3.4 Materials and methods ................................................................................ 102 3.4.1 Fish rearing. ................................................................................................... 102 3.4.2 Exposure regimes to cadmium. .......................................................... 102 3.4.3 Host-microbiota and water sampling. ............................................ 102 3.4.4 Metal concentration in water and fish liver. ............................... 103
3.4.5 DNA extraction to Illumina Miseq sequencing. ......................... 103 3.4.6 Analysis of 16S rDNA amplicons. ...................................................... 104 3.4.7 Correlational networks. .......................................................................... 104 3.4.8 Metacommunity assembly modelling. ............................................ 105 3.5 Results ................................................................................................................... 105 3.5.1 Metal concentrations in water and host livers .......................... 105 3.5.2 Mixed Effects of time and treatment on metacommunity
alpha diversity ............................................................................................................... 106 3.5.3 An important effect of time on the taxonomic composition
of metacommunities ................................................................................................... 106 3.5.4 Community-level phylogenetic divergence ................................. 107 3.5.5 Correlational metacommunity networks ...................................... 107 3.5.5.1 Substantial role of rare taxa in the metacommunity
network connectivity ................................................................................................. 107 3.5.5.2 Reduced network connectivity in gut communities under
cadmium stress ............................................................................................................. 107 3.5.5.3 Negative correlations in Skin Mucous Community
networks suggest dysbiosis ................................................................................... 108 3.5.5.4 Fragmentation of water microbial community networks . 108 3.5.6 Stochasticity in water community assembly and
determinism in that of host microbiota .......................................................... 109
3.6 Discussion ........................................................................................................... 109 3.6.1 Phylogenetic divergence at the community-level revealed
the impact of Cadmium exposure. ..................................................................... 110 3.6.2 Gradual disconnection of abundant taxa from the main gut
interacting network. ................................................................................................... 111 3.6.3 Rare OTUs play a pivotal role in community assembly. ....... 112 3.7 Conclusions ......................................................................................................... 112 3.8 Perspective ......................................................................................................... 113 3.9 Tables ..................................................................................................................... 114 3.10 Figures .............................................................................................................. 119 3.11 Supplementary Figures ........................................................................... 125 3.12 Supplementary Material ......................................................................... 131
Chapitre 4: Community recovery dynamics in yellow perch
microbiome after gradual and constant metallic perturbations
............................................................................................................................ 134
4.1 Resumé ................................................................................................................. 135 4.2 Abstract ................................................................................................................ 136 4.3 Introduction ....................................................................................................... 137 4.4 Methods ................................................................................................................ 139 4.4.1 Fish rearing .................................................................................................... 139 4.4.2 Exposure regimes to cadmium ........................................................... 140
4.4.3 Recovery after the exposure to Cadmium.................................... 140
ix
4.4.4 Host-microbiota and water sampling ............................................. 140 4.4.5 Metal concentration in water and fish liver ................................ 141 4.4.6 DNA extraction, libraries preparation and 16S amplicons
sequencing ....................................................................................................................... 141 4.4.7 Bioinformatics and biostatistics analyses .................................... 141 4.4.7.1 Reads preprocessing and OTUs clustering ................................ 141 4.4.7.2 Post-OTUs analysis, networks and function prediction. ... 143 4.5 Neutral and deterministic models to asses the recovery of
community assembly. ................................................................................................ 144 4.6 Results ................................................................................................................... 144 4.6.1 Cadmium concentration bioaccumulation in the fish liver
during recovery time .................................................................................................. 144 4.6.2 Genotypic signatures of community recovery ........................... 145 4.6.3 Microbial taxonomic composition change during recovery 146 4.6.4 Correlational networks of host and water microbiome ....... 148 4.6.5 Recovery of microbial functional diversity a time T5. .......... 149 4.6.6 The role of neutral and deterministic processes in the
recovery of host microbiota ................................................................................... 149 4.7 Discussion ........................................................................................................... 150 4.8 Conclusions ......................................................................................................... 154
4.9 Tables ..................................................................................................................... 155 4.10 Figures .............................................................................................................. 158 4.11 Supplementary figures ............................................................................ 166 4.12 Supplementary Material ......................................................................... 168
Discussion et conclusions générales.................................................. 170
Bibliographie ................................................................................................. 178
x
Liste des tableaux
Table 3. 1 Statistics’ summary of Cadmium concentration variation over
time and treatments in water tanks and fish livers. ............................. 114
Table 3. 2 Statistical summary of alpha-diversity changes over time and
treatments. .................................................................................. 115
Table 3. 3 Phylogenetic divergence at the community level. ................. 117
Table 4. 1 Statistics of Cd concentrations in water and fish liver over time
and treatments. ............................................................................ 155
Table 4. 2 Phylogenetic divergence in host and water microbiomes. ...... 156
xi
Liste des Schémas
Schéma 1. 1 Approches métagenomiques de l’analyse de microbiote. ...... 15
Schéma 1. 2. Boite noire de l’écologie des communautés. ...................... 33
Schéma 1. 3. Espace des modèles éco-évolutifs. .................................. 48
xii
Liste des figures
Figure 2. 1 Bioinformatics analysis pipeline. ......................................... 74
Figure 2. 2 Composition of metacommunities based on the ORF approach. 75
Figure 2. 3 Function abundance classification based on ORF approach. ..... 76
Figure 2. 4 Polymetallic resistance genes (PMRG) abundance correlation
with trace metals. ........................................................................... 78
Figure 2. 5 Decoupling of taxon and function between metacommunities
based on the subsampled reads approach. .......................................... 80
Figure 2. 6 Coupling of taxon and function between metacommunities based
on the subsampled reads approach .................................................... 81
Figure 3. 1 Linear variations of alpha-diversity over time and between
treatments explained by the linear mixed model in water and host-microbial
communities ................................................................................ 119
Figure 3. 2 Phylogenetic divergence at the community level ................. 120
Figure 3. 3 Correlational co-abundance networks of gut microbial
community. .................................................................................. 121
Figure 3. 4 Correlational co-abundance networks of skin microbial
community. .................................................................................. 122
Figure 3. 5 Correlational co-abundance networks of water microbial
community. .................................................................................. 123
Figure 3. 6 Bar plots of neutral OTUs change at community and meta-
community levels. ......................................................................... 124
Figure 4. 1 Schematic illustration of the perch microbiome recovery
experiment. ................................................................................. 158
Figure 4. 2 Alpha-diversity dynamics in the water and perch microbiome.
.................................................................................................. 158
Figure 4. 3 Taxonomic composition dynamics of host communities ........ 159
Figure 4. 4 Heatmaps of differential abundance among host and water
communities ................................................................................ 159
Figure 4. 5 Function diversity dynamics in host and water microbiome. .. 160
Figure 4. 6 Recovery dynamics of the networks of host communities. .... 161
Figure 4. 7 Recovery dynamics of the network of water communities ..... 162
Figure 4. 8 Centrality plots of host microbiome networks ..................... 163
Figure 4. 9 Percentage of neutral OTUs over time and treatment. ......... 164
Figure 4. 10 Demographic variation of metacommunity neutrality across
water and host microbiome ............................................................. 165
xiii
Liste des abréviations
16S rDNA: 16S Ribosomal DNA;
ACE: Abundance-based coverage estimator;
Al: Aluminium;
AMD: Acid Mine Drainage;
ANOVA: Analysis of variance;
BAR-mc: Arnoux Bay, medium contaminated;
BEB: back extraction buffer;
BH: Benjamini-Hochberg correction test;
CC : Cadmium Constant Concentration;
Cd : Cadmium;
CdCl2: Cadmium Salts;
CPAUL: Comités de protection des animaux de l’université Laval ;
Ctrl: regime of negative Control;
Cu: Cooper;
CV: Cadmium Variable Concentration;
DAS-lc: Dasserat Lake, low contaminated;
FDR: False Discovery Rate;
Fe: Iron;
HGT: Horizontal Gene Transfer;
LAR-hc: Arnoux Lake, high contaminated;
Lead: Pb;
Mn: Manganese;
MRPP: Multiple Response Permutation Procedure;
xiv
NGS: Next generation sequencing;
NLS: non-linear least squares model;
NMDS: non-metric Multi-Dimensional Scaling;
OPA-nc: Opasatica Lake, not contaminated;
ORF: Open Reading Frame;
OTU: Operational taxonomic Unit;
PCA: Principal Component Analysis;
PCG: Polymetallic Contamination Gradient;
PCR: Polymerase Chain Reaction;
PERMANOVA: Permutational analysis of variance;
PMRG: Polymetallic resistance genes;
ppb: parts per billion;
rCCA: Regularized canonical correlation analysis;
RDP: Ribosomal Database Project;
TUR-hc: Turcotte Lake, high contaminated;
UTO : Unité Taxonomique Opérationnelle;
Zinc : Zn;
xv
À Hadi, Amar et Farah
À toute ma famille
À la révolution de 17 Octobre
xvi
Remerciements
En premier, je présente mes profonds et sincères remerciements à mon
directeur Pr. Nicolas Derome de m’avoir donné l’opportunité de continuer
dans la recherche scientifique, et la chance de saisir ma première opportunité
québécoise. Merci de m’avoir accueilli dans ton laboratoire comme un
assistant de recherche en premier temps (Décembre 2013-Mai 2014) et
comme un étudiant en doctorat en deuxième temps (Juin 2014-Octobre
2018) ! Merci de m’avoir accordé toute ta confiance pour mener mon projet
avec toute liberté ! Merci de m’avoir donné avec toute générosité tous les
matériels et les consommables nécessaires pour mener mes travaux dans ton
laboratoire ! Merci de m’avoir donné toute cette belle chance.
J’aimerais aussi remercier les membres de mon comité d’encadrement : Pr.
Connie Lovejoy, Pr. Louis Bernatchez et Pr. Jacques Corbeil. Merci pour votre
confiance aux différentes étapes de ce doctorat et pour vos suggestions.
Je voudrais remercier de tout mon cœur Dr. Martin LLewellyn de m’avoir
accordé également toute sa confiance pour réaliser des collaborations
scientifiques sur ses projets durant son post-doctorat dans notre laboratoire.
Merci Martin pour ton soutien tous les niveaux et merci de m’avoir accueilli
dans ton laboratoire comme un chercheur postdoctoral à l’université de
Glasgow, même avant de soutenir ma thèse.
Merci infiniment à Dr. Mohamed Alburaki de m’avoir soutenu durant cette
thèse. Merci Mohamed de m’avoir donné l’opportunité de collaborer sur ton
projet et de m’avoir transmis ta passion pour les abeilles, et de me faire
connaitre leur diversité génétique en Syrie et au Liban.
Grand merci à tous mes proches collaborateurs, Hamza Seghouani, Sarah El-
Khoury, François-Etienne Sylvain, Pierre-Luc Mercier et Dr. Umer Ijaz.
Je tiens à remercier particulièrement Jeff Gauthier d’avoir lu mon texte en
Français. Merci pour toutes les discussions infinies et intéressantes sur les
aspects différents de la recherche scientifique. Tu es un collègue agréable,
altruiste, et toujours prêt pour aider tous les collègues au laboratoire.
xvii
Merci spécial à Amina Abed et Vani Mohit d’avoir corrigé minutieusement
plusieurs textes en Français à plusieurs occasions durant mon parcours. Merci
à toi Amina pour tes conseils concernant les protocoles de l’extraction d’ADN.
Merci à Dr. Anne Dalziel, Dr. Amanda Xuereb, Dr. Ciara Keating, Eleanor
Lindsay, et Matt Bywater d’avoir m’aidé dans les corrections de mes
manuscrits en anglais.
Merci à tous les anciens et les nouveaux membres du laboratoire Derome,
pour leur qualité humaine et leur aide durant des longues journées
d’échantillonnage au LARSA. Merci à Émie, Sidki, Laurence, Katherine,
Camille et Sara-Jane.
Merci aux professionnels de recherche du Laboratoire Bernatchez, Alysse et
Cecilia pour leur dépannage de matériels durant les longues semaines de
grève.
Je souhaiterais remercier spécialement Dr. Michel Lavoie pour ses conseils
d’or concernant la manipulation de Cadmium. Merci Michel, tu m’avais ouvert
les yeux sur des aspects que j’ignorais en Chimie environnementale.
Je tiens à remercier les personnels du LARSA particulièrement Jean-
Christophe Therrien, et je n’oublie pas la plateforme de séquençage de l’IBIS
particulièrement Dr. Brian Boyle.
À mes collègues co-fondateurs du Club Bioinformatique, Jeff Gauthier, Dr.
Anthony Vincent et Éric Normandeau, je vous exprime tous mes sincères
remerciements.
Je souhaiterais remercier Mario Boutin une personne qui a laissé une
empreinte humaine et un vide éternel à l’IBIS, et aux services de la Laverie.
Mario était toujours plein de joie et d’amour pour ses collègues, que ton âme
repose en paix.
À tous(tes) mes ami(e)s Nabil, Carlos, Aref, Yazan, Hayan, Mohamed, Fawzi,
Aoun, Rabih, Imad, Roba, Sarah, Émilie, Hector sans oublier personne, je
souhaiterais vous remercier pour votre encouragement et pour tous les bons
moments. Merci de m’avoir soutenu sur place et à distance durant toutes les
étapes de ma vie canadienne.
xviii
À ma mère, mon père, mes frères et sœurs, aucun mot ne suffit de vous
exprimer ma gratitude, sans vos encouragements, je ne serais jamais arrivé
à finir ma thèse.
xix
Avant-propos
Cette thèse de doctorat en biologie et bioinformatique comporte cinq
chapitres incluant une introduction et une conclusion générale. Mes travaux
sont constitués de trois articles de recherche, identifiés chapitres 2, 3 et 4
dans la table des matières. Les trois chapitres sont rédigés en anglais car le
premier fut publié et les deux derniers demeurent en révision dans des revues
scientifiques internationales utilisant la langue anglaise. Le thème global de
cette thèse s’articule sur l’axe de l’évolution contemporaine de deux modèles
systèmes microbiens : environnemental et hôte-associé dans un contexte
d’écotoxicologie.
Le 1er chapitre est une revue bibliographique des principes de bases en
écologie microbienne. Le deuxième chapitre fait l’objet d’une publication
scientifique dans le Journal « Frontiers in Microbiology » et présente les
signatures adaptatives détectées au sein d’un système microbien
environnemental sous un gradient de pression sélective, induite par une
exposition chronique aux métaux traces au cours de plus de 60 ans
d’exploitation minière.
Le 3ème et le 4ème chapitres présentent les empreintes de l’évolution
expérimentale d’un système microbien hôte-associé dans un contexte
d’intoxication métallique artificielle selon deux régimes de sélection, constant
et graduel. Le chapitre 3 se focalise sur les signatures de la résistance de
microbiote de l’hôte en fonction de l’intensité du polluant, tandis que chapitre
4 décrit la résilience de la structure de microbiote en fonction du gradient de
stress métallique. Les deux chapitres sont sous révisions depuis quelques
mois, dans les deux revues « Nature ISME » et « Microbiome »
respectivement.
Pour chacun des chapitres de ma thèse, j’ai formulé les hypothèses
scientifiques et la théorie des questions adressées avec l’appui de mon
directeur Nicolas Derome. Pour le deuxième chapitre la prise d’échantillons
sur le terrain a été assurée par Pierre-Luc Mercier le troisième co-auteur de
la publication. Le deuxième cosignataire, Malo Le Blouch a contribué
également dans l’analyse descriptive des données de séquençage. Pour les
chapitres 3 et 4, j’ai principalement contribué à la conception, la planification
xx
et à la conduite des expériences avec l’aide de mon directeur Nicolas Derome
et mon collaborateur Hamza Seghouani. La prise des échantillons et la
dissection des poissons ont été réalisée grâce à l’aide de tous les membres
du laboratoire. Après l’échantillonnage, j’ai effectué la filtration de l’eau,
l’extraction de l’ADN, les préparations des librairies, les analyses bio-
informatiques et biostatistiques des résultats et la rédaction des articles.
Dans le chapitre 3, les deux co-auteurs Katherine Vandal-Lenghan et Pierre-
Luc Mercier ont contribué à l’extraction de l’ADN de certains échantillons et
la quantification des librairies de séquençage.
Les coauteurs internationaux de l’Université de Glasgow, Martin LLewellyn et
Umer Ijaz ont contribué dans l’amélioration de l’analyse des résultats et dans
la qualité de la rédaction des manuscrits de chapitre 3 et 4.
Les détails des articles publiés ou soumis se trouvent ci-dessous :
Article I. Taxon-function decoupling as an adaptive signature of lake
microbial metacommunities under a chronic polymetallic pollution gradient
Auteurs : Bachar Cheaib, Malo Le Boulch, Pierre-Luc Mercier, et Nicolas
Derome
Publié le 3 Mai 2018 dans le journal Frontiers in Microbiology (Front Microbiol.
2018 ; 9 : 869.).
Article II. From networks to models: the Yellow Perch (Perca flavescens)
microbiome assembly under metal toxicity
Auteurs : Bachar Cheaib, Hamza Seghouani, Martin Llewellyn, Katherine
Vandal-Lenghan, Pierre-Luc Mercier, and Nicolas Derome.
Soumis le 05 Mars 2019 dans journal Nature ISME, rejeté uniquement par
l’éditeur (accepté par l’arbitre) le 30 septembre 2019 avec l’option de
resoumettre une version courte (Reference ISMEJ-19-00341A)
Article III. Community recovery dynamics in yellow perch microbiome after
gradual and constant metallic perturbations
Auteurs : Bachar Cheaib, Hamza Seghouani, Umer Zeeshan Ijaz, and Nicolas
Derome.
xxi
Cet article a été publié dans le journal Microbiome. Microbiome 8, 14 (2020)
https://doi.org/10.1186/s40168-020-0789-0
En parallèle, je me suis intéressé à l’évolution d’une métalloenzyme le
carbonique anhydrase. C’est une famille protéique ubiquitaire capable de
substituer les ions métalliques pour assurer la photosynthèse chez les
phytoplanctons. J’ai rédigé un manuscrit (Juin 2017) sur ce sujet sous forme
d’un chapitre supplémentaire sous la direction de Connie Lovejoy. Ce travail
serait bientôt envoyé à une revue scientifique afin de le publier.
D’un autre côté, les travaux de recherche de la thèse ont été communiqué
sous forme de présentations et affiches à des journées scientifiques
départementales et des organisations scientifiques québécoises et
canadiennes, et à des conférences internationales. J’ai personnellement
conçu et communiqué chacune de ces présentations et des affiches. Mes
participations en personne sont citées ci-dessous :
Conférences locales à l’Université Laval, Québec, Canada
Plusieurs éditions de la journée d’étudiante (2014, 2015, 2016 et 2018) de
l’Institut de Biologie Intégrative et des Systèmes (IBIS)
Deux éditions de colloque annuel de département de Biologie (2015,2018)
avec un prix de distinction (Edition 2015) de la fondation Richard Bernard.
Conférences et réunions annuelles canadiennes
Plusieurs éditions (2014, 2015, 2017) de la Réunion annuelle des Ressources
Aquatiques Québec (RAQ), Québec.
Colloque annuel conjoint RÉAQ-EcoBIM (ÉcoBIM 2015), INRS, Québec.
Conférence Genomes to / aux Biomes (2014), 1ère réunion conjointe de la
Société canadienne d'écologie et d'évolution (SCEE), de la Société canadienne
de zoologie (CSZ) et de la Société canadienne des limnologie (SCL), 25-29
mai, Montréal.
BISP 2016. Congrès BiSP (Bactériologie intégrative : Symbiose &
Pathogenèse), third edition, Université Laval Québec, Canada.
Conférences internationales
xxii
Le 17ème Symposium de la Société Internationale de l'Écologie Microbienne
(ISME17), du 12 - 17 Août 2018, Leipzig, Allemagne. Ma participation été
financé en partie par les Fonds général pour les études supérieures (FGES)
du Conseil de recherches en sciences naturelles et en génie (CRSNG).
Le 16ème Symposium de la Société Internationale de l'Écologie Microbienne
(ISME16), du 21 au 26 août 2016, Montréal, Canada.
GLBIO-CCBC 2016, Great Lakes Bioinformatics et Conférence Canadienne sur
la Biologie Computationnelle, du 16 au 19 mai 2016, Toronto, Canada.
Mes recherches dans le laboratoire de Nicolas Derome, ne se restreignent pas
aux chapitres présentés dans ce document. J’ai contribué en tant que co-
auteur aux analyses bioinformatiques et biostatistiques de quatre
publications sur des problématiques qui rentrent dans l’intérêt général de
cette thèse
La première porte sur l’impact de l’acidité de l’eau sur la résilience de
microbiote d’un poisson amazonien le Tambaqui publiée dans le journal
Scientifc Reports
Sylvain, F.-É., Cheaib, B., Llewellyn, M., Gabriel Correia, T., Barros
Fagundes, D., Luis Val, A., Derome, N., 2016. pH drop impacts differentially
skin and gut microbiota of the Amazonian fish tambaqui (Colossoma
macropomum). Sci. Rep. 6, 32032.
La deuxième porte sur la causalité entre pesticides, expression des gènes et
maladies infectieuses des abeilles (varroa) publié dans Journal of economic
entomology.
Alburaki, M. Cheaib B, et al. Agricultural Landscape and Pesticide Effects on
Honey Bees (Hymenoptera: Apidae) Biological Traits. Journal of economic
entomology 110, 835–847;2017.
La troisième propose un candidat probiotique pour réduire l’effet du
parasitisme sur la mortalité des abeilles, publié dans Frontiers in Ecology and
Evolution
El Khoury S, Rousseau A, Lecoeur A, Cheaib B, Bouslama S, Mercier PL,
Demey V, Castex M, Giovenazzo P, Derome N. Deleterious Interaction
xxiii
Between Honeybees (Apis mellifera) and its Microsporidian Intracellular
Parasite Nosema ceranae Was Mitigated by Administrating Either Endogenous
or Allochthonous Gut Microbiota Strains. Frontiers in Ecology and Evolution.
2018 May 23; 6:58. Doi: 10.3389/fevo.2018.00058.
La quatrième révèle une sélection diversifiante des protéases de surface de
Trypanosoma cruzi GP63 parmi les patients atteints de la maladie de Chagas
chronique et congénitale. Elle est publiée dans le journal PLOS Neglected
Tropical Diseases
Llewellyn MS, Messenger LA, Luquetti AO, Garcia AL, Torrico F, Tavares SBN,
Cheaib B, Derome N et al. 2015. Deep sequencing of the Trypanosoma cruzi
GP63 surface proteases reveals diversity and diversifying selection among
chronic and congenital Chagas disease patients. PLoS NTD.
Depuis le début de la dernière année de cette thèse, j’entame un stage post-
doctoral avec mes collaborateurs, William Sloan, Martin LLewellyn et Umer
Ijaz à l’école ingénieure de l’Université de Glasgow. Mes recherches portent
sur la modélisation écologique de la dynamique de microbiote des
salmonidés. Mes recherches se concentrent sur la quantification du rôle relatif
des processus neutres et non-neutres dans la colonisation du tube digestif
des juvéniles des salmonidés dans la nature et en aquaculture. Mes résultats
pour l’instant sont publiés dans « le journal of AEM Applied and Environmental
Microbiology »
Durant cette thèse, j’ai transformé mes difficultés financières en opportunités
pour développer un gout pour l’enseignement en assurant mes fonctions
d’auxiliaire d’enseignement de biostatistique sous forme des travaux dirigés
avec Frederic Maps. Durant mon post-doc, j’ai également saisi l’opportunité
d’assister des travaux dirigés sur les bases de programmation (Python) avec
mon collaborateur Umer Ijaz à l’université de Glasgow.
Sur le plan social et scientifique général, j’ai initié l’idée du Club
Bioinformatique de l’IBIS que j’ai co-fondé plus tard avec mes chers collègues
Jeff Gauthier, Antony Vincent, et Eric Normandeau de plusieurs laboratoires
de l’IBIS. Cette expérience m’a offert la chance de partager et d’améliorer
mes connaissances et d’acquérir des nouvelles problématiques récentes en
bioinformatique biostatistique et en biologie évolutive
1
Introduction Générale
L’anthropocène, soit l’époque géologique contemporaine, est caractérisée par
un impact colossal de l’activité humaine sur la biosphère (Balter 2013; Larson
et al. 2014; Waters et al. 2016; Tucker et al. 2018). De l’industrie chimique,
plastique et métallurgique, aux pesticides et insecticides, à la pollution des
eaux, et la perturbation des cycles biogéochimiques de la terre, la liste est
longue et ne cesse de s’allonger avec des centaines d’altérations irréversibles
des écosystèmes. Les effets anthropiques sur les écosystèmes terrestres,
fluviaux et marins et les systèmes biologiques sont devenus alors tout à la
fois néfastes et étendus à l’échelle mondiale. Ainsi, le progrès technologique
d’Homo sapiens (industriel, numérique, communication, énergétique,
transport, nucléaire, militaire, etc.), sa croissance démographique et la
prolongation de son espérance de vie ne pourraient plus continuer au
détriment des ressources de la biodiversité et de l’équilibre de nos
écosystèmes (Crutzen and Stoermer 2000; Pelletier and Coltman 2018). Au-
delà des enjeux économiques, politiques de la réalité et en dépit de l’absence
des règlementions éthiques efficaces par les nations unies, les chercheurs ne
cessent de saisir l’opportunité pour soulever des questions fondamentales
concernant l’accélération artificielle de l’évolution génotypique des systèmes
biologiques sous l’effet de la pression sélective induite par les effets
anthropiques. À court et long terme, les études de l’évaluation de l’impact
des xénobiotiques sur la biodiversité tout au long de la chaine trophique
démontrent des perturbations même à l’échelle la plus fine du vivant, les
micro-organismes ou les microbes. Les xénobiotiques sont des produits
chimiques trouvés mais non produits par les organismes ou l'environnement.
Certains produits chimiques naturels (endobiotiques) deviennent des
xénobiotiques lorsqu'ils sont présents dans l'environnement à des
concentrations excessives. Le terme « xeno » dans « xénobiotiques » vient
du mot grec « xenos » qui signifie un invité, ami ou étranger (Soucek 2011).
Sans métabolisme, de nombreux xénobiotiques atteindraient des
concentrations toxiques (Croom 2012).
2
L’anthropocène: des effets anthropiques sur la biodiversité et
l’assemblage des communautés microbiennes
Il est connu que les microbes (Bactéries, Archées, Eucaryotes unicellulaires,
Virus, etc.) sont ubiquitaires, vivant en communautés, en biofilms,
planctoniques ou associées avec d’autres forme de vie. Les communautés
microbiennes planctoniques se trouvent dans l’eau, le sol et l’air, et les
symbiotiques se sont associées aux Métazoaires et aux Plantes, et aux autres
microbes unicellulaires. Elles sont les productrices primaires des services
écosystémiques du sol, et de l’eau, ou de l’hôte et constituent la partie
majeure de la biodiversité sur terre.
Les communautés microbiennes impliquées dans ces services s’avèrent très
impactés dans leurs répertoires génotypiques et phénotypiques par les
xénobiotiques. Citons par exemple les antibiotiques, les métaux toxiques, les
plastifiants chimiques, les biocides comme les pesticides, insecticides, et
désinfectants. Ces polluants chimiques non dégradables sont souvent des
xéno-estrogènes synthétiques qui interfèrent avec les récepteurs des
systèmes biologiques comme des perturbateurs endocriniens. Dans ce
contexte de perturbation, des centaines d’exemples peuvent être cités, nous
citons ici quelques-uns.
L’exemple de traitement par des antibiotiques est un excellent argument qui
témoigne de l’adaptation rapide des microbes et l’accélération de la cadence
de l’évolution microbienne. Les études montrent que les souches bactériennes
traitées par des antibiotiques (Vincent et al. 2019) acquièrent de la résistance
par conjugaison, un mécanisme de transfert horizontal parmi d’autres
(transformation, transduction) bien connus chez les microbes. Cette
résistance prépondérante aux antibiotiques est acquise horizontalement et
avant de se transmettre verticalement d’une génération bactérienne à l’autre
par division clonale (Holmes et al. 2016; von Wintersdorff et al. 2016; Cesare,
Eckert, and Corno 2016). Les biocides (triclosan, toluène, méthyl-mercure,
proflavine, etc.) sont des polluants de classes chimiques différentes qui
conduisent aussi à l’évolution du résistome bactérien (l’ensemble de gènes
de résistance bactériens) par des mutations génomiques (substitutions,
délétions, insertions) ou par l’acquisition de nouveaux gènes par transfert
3
horizontal. Peu importe les mécanismes, si les mutations génétiques sont
fixées dans les populations à travers les générations, elles peuvent selon la
stabilité de l’environnement conférer des avantages phénotypiques qui
contribuent à des capacités adaptatives comme la dégradation des
xénobiotiques. D’autre part, les néonicotinoïdes (Clothianidine,
thiaméthoxame ou imidaclopride) affectent la physiologie, le comportement
des pollinisateurs, et par conséquent la santé de l’Homme et de notre
environnement (Raine 2018; Crall et al. 2018; Alburaki et al. 2016; Doublet
et al. 2015; van der Sluijs et al. 2013; Di Prisco et al. 2013). Par exemple,
les pesticides et les insecticides contiennent des molécules chimiques qui
perturbent non seulement les capacités cognitives (Zhang and Nieh 2015;
Siviter et al. 2018) et hygiéniques des abeilles (Boutin et al. 2015; E. Zhang
and Nieh 2015; Siviter et al. 2018), mais aussi leur microbiote (Raymann,
Shaffer, and Moran 2017; Motta, Raymann, and Moran 2018), ce qui
augmente leur susceptibilité aux pathogènes et de ce fait perturbe l’équilibre
immunité-symbionte/microbiote.
Le dernier exemple concerne les métaux lourds, ayant une masse atomique
élevée, certains laissent de traces toxiques (Cd, Ni, Cu, Co, Al etc.) qui
altèrent la biodiversité microbienne dans le sol, les milieux aquatiques et les
systèmes biologiques (Koschorreck 2008; Huang, Kuang, and Shu 2016;
Hudson-Edwards and Dold 2015). Les études publiées par notre laboratoire
ont montré une corrélation entre les concentrations des métaux avec la
variation la composition des communautés microbiennes (Laplante and
Derome 2011; Laplante, Boutin, and Derome 2013). Par exemple, les métaux
lourds déversés par le drainage minier acide, en particulier le cadmium,
augmente l’acidité de l’eau, et induit des changements dans la composition
taxonomique des communautés bactériennes favorisant les
Alphaprotéobactéries.
Les effets des métaux traces toxiques sur l’environnement (Nordstrom 2011;
Bejan and Bunce 2015; Lavoie, Fortin, and Campbell 2012; Wu et al. 2016;
X. Zeng, Chen, and Zhuang 2015) et les êtres vivants (Vymazal 1987;
Giguère et al. 2004; Lacroix and Hontela 2004) sont connus, mais très peu
étudiés dans le cas des symbiontes microbiens associées aux systèmes
biologiques aquatiques, par exemple les Poissons (S. Zhang et al. 2015;
4
Bridges et al. 2018)
En général, les xénobiotiques perturbent la diversité, la structure et les
fonctions des communautés microbiennes, leur assemblage ainsi que leurs
interactions, selon des processus moléculaires évolutifs et adaptatifs
méconnus.
Les études de l’assemblage du microbiote de l’Homme sous antibiothérapie
(Costello et al. 2012) ou du microbiote de poisson euryhalin lors de
l’acclimatation à la salinité (V. T. Schmidt et al. 2015), étaient principalement
centrées sur des processus déterministes, avec peu de preuves d’une
colonisation stochastique. De plus, le rôle des processus écologiques et
évolutifs dans la résilience de la structure des communautés microbiennes
après perturbation reste à déchiffrer. Théoriquement, la nature de ces
processus peut varier entre neutre (stochastique) (Sloan et al. 2006;
Jayathilake et al. 2017) et sélective (déterministe) (Stegen et al. 2012; Q.
Zeng et al. 2017). Dans le contexte de la résilience du microbiote de l’hôte
ou environnemental, peu d’études ont été dédiée à cette question. Ces
derniers ont révélé que ce sont les processus déterministes qui induisaient la
dynamique de la succession bactérienne étudiée dans un contexte du sol
perturbé soit par un gradient d’épuisement des éléments nutritifs (Song et
al. 2015), soit par un choc thermique (Jurburg et al. 2017), ou par une
réhydratation pluviale(Placella, Brodie, and Firestone 2012)
Néanmoins, il reste encore beaucoup à faire pour comprendre les mécanismes
de l’assemblage des communautés microbiennes résilientes chez diverses
espèces d’organismes hôtes sous un contexte de perturbation et de résilience.
Les questions et les chapitres de cette thèse
Dans cette thèse, les impacts des effets anthropiques sur la diversité, la
structure, la fonction et la composition des communautés microbiennes dans
deux écosystèmes ; environnemental, et en association avec un hôte seront
discutés. En admettant que les connaissances actuelles sur l’évolution
microbienne à partir des microbes cultivables ne soient pas représentatives
de celles de la majorité inconnue, les approches de séquençage de nouvelle
génération « Next generation sequencing » (NGS) et le progrès de la bio-
5
informatique (banques des séquences, réseaux, algorithmiques), ainsi les
modèles récents en écologie et évolution microbienne nous ouvrent alors des
voies interdisciplinaires et intégratives prometteuses pour trouver des
réponses à nos questions. Ainsi dans cette étude, nous nous sommes
intéressés à trois problématiques :
1) La dynamique de la structure et de la composition des
communautés microbiennes, et leurs fonctions écosystémiques
2) L’évolution de leurs interactions et leur assemblage dans un
contexte anthropogénique
3) La résilience de leur structure et assemblage et leurs interactions
avec leur écosystème après perturbation (environnement et hôte).
Le premier chapitre est une revue bibliographique qui résume les méthodes
et les concepts de base en écologie microbienne.
Le deuxième chapitre étaye l’hypothèse que les perturbations chroniques
des communautés microbiennes lacustres par des métaux traces
entraineraient un découplage entre la diversité taxonomique et fonctionnelle.
Pour vérifier cette hypothèse, des communautés microbiennes ont été
échantillonnées dans cinq lacs exposés à un gradient de contamination
polymétallique (PCG), appartenant à un même bassin versant, situé à
proximité d’une mine de cuivre historiquement active pendant plus de
soixante ans. Avec une approche métagénomique, bio-informatique, nous
avons alors caractérisé le niveau d’intégrité des fonctions représentant les
services écosystémiques.
Le troisième chapitre étaye l’hypothèse qu’un processus de gradient
sélectif gouvernerait l'assemblage du microbiote d’un organisme soumis à des
perturbations graduelles et constantes.
Pour évaluer cette hypothèse, une approche d’évolution expérimentale à
court terme a été utilisée (six mois), laquelle, compte tenu du temps de
génération des Bactéries (20 à 30 minutes chez E. coli) équivaut à plus de
8000 générations. Plus précisément, nous avons testé expérimentalement au
laboratoire comment l’exposition prolongée constante ou graduelle au
6
cadmium (Cd) de 1200 individus des perchaudes (Perca flavescens) avait
affecté l’assemblage de leurs microbiotes par séquençage d’amplicons du
gène de l’ARNr 16S. Nous avons constaté qu’un gradient de sélection induit
par un gradient de concentration d’un métal toxique a perturbé non
seulement la physiologie de l’hôte, mais également le recrutement et
l’assemblage de son microbiote.
Le quatrième chapitre traite de la dynamique de la résilience du microbiote
de la perchaude après l’arrêt de l’exposition à des quantités sous-létales de
cadmium. Le terme « résilience » est employé pour décrire le changement
qui se produit lorsqu'une communauté retourne à un autre état stable après
perturbation. Le rétablissement des communautés microbiennes dépend du
type, de la durée, de l’intensité et du gradient de perturbation.
Nous focalisons sur les modèles car les rôles des processus écologiques et
évolutifs dans la résistance et la résilience de la structure du microbiote des
hôtes restent à documenter. Théoriquement, et telle que mentionné ci-
dessus, la nature de ces processus varie entre neutralisme (stochastique)
(Stephen P. Hubbell 2006; Sloan et al. 2006) et sélection (déterministe)
(Webb et al. 2002; Chase 2003). Ces derniers opèrent soit par filtrage
environnemental et exclusion compétitive (Cadotte et al. 2010; Stegen et al.
2012). Le peu d’études disponibles a révélé que ce sont des processus
déterministes qui régissent la dynamique de la succession bactérienne durant
le rétablissement des communautés microbiennes des sols. Dans notre étude
expérimentale (chapitre 3 et 4), nous avons évalué la contribution relative
des processus neutres et non-neutres à la résistance et à la résilience de
l’assemblage du microbiote de la perchaude à la suite d’un gradient
expérimental d’exposition métallique. Étant donné que les juvéniles de
perchaude peuvent tolérer des doses non létales de cadmium sans subir de
dommages physiologiques importants (Giguère et al. 2006; Campbell et al.
2005a), notre modèle hôte-microbiote est optimal pour étudier la capacité de
résilience du microbiote après une perturbation liée à l'exposition au
cadmium. Premièrement, nous avons quantifié avec des méthodes
appropriées les niveaux de métaux traces dans des échantillons de foie et
d’eau. Ensuite, nous avons comparé l’état de résilience de la structure et de
la fonction de la communauté dans l’eau et du microbiote de l’hôte entre des
7
régimes constants et variables d’exposition au cadmium. Afin de démêler
l'effet du xénobiotique du développement de l'hôte (Sylvain and Derome
2017; Burns et al. 2016a) sur l'ontogenèse, l'assemblage du microbiote a
également été évalué dans des conditions stables en tant que régime témoin.
Deuxièmement, nous discutons des changements globaux de la diversité
taxonomique et fonctionnelle ainsi que la prévalence des agents pathogènes.
Nous évaluons la contribution relative des processus neutres et non-neutres
dans la résilience du microbiote de perchaude de deux régimes d’exposition
métallique constant et graduel.
8
Chapitre 1 – Méthodes et concepts de
base en écologie microbienne
Ce chapitre est une bibliographie des méthodes et concepts de base utiles
pour assurer la bonne compréhension des trois chapitres de recherche de
cette thèse.
9
1.1 Les microbes sont ubiquitaires et l’environnement les filtre
Que leur mode de vie soit libre ou associé à un hôte (microbiote) les micro-
organismes constituent le générateur fonctionnel des métabolites primaires
et secondaires de tout écosystème. Leur importance est donc centrale et
primordiale dans les cycles biogéochimiques de la terre, la qualité des eaux,
en agriculture (fixation de l’azote, biostimulation de la croissance par
solubilisation d’éléments minéraux, biocontrôle des phytopathogènes), en
médecine (la médecine personnalisée pour les diabétiques, la médecine
régénératrice après transplantation des cellules souches, le traitement des
maladies infectieuses par des probiotiques), en industrie agroalimentaire et
en sécurité alimentaire, entre autres. Les nouvelles découvertes en
microbiologie, en écologie et en évolution microbienne ont, clairement, mis
en évidence l’implication de l’environnement à sélectionner ses communautés
microbiennes. Un simple exemple très répandu, les nodules racinaires des
plantes recrutent les Bactéries bénéfiques fixatrices d’azote (Brill 1975; Dos
Santos et al. 2012; Yang et al. 2017). Le système immunitaire de l’Homme
joue le rôle du filtre qui retient les bactéries bénéfiques responsables de la
maintenance de son homéostasie (Belkaid and Hand 2014; Belkaid and
Harrison 2017; Belkaid and Segre 2014). Ainsi, le microbiote contrôle et
régule de nombreuses fonctions vitales de l’hôte telles que l’immunité et
l’assimilation des nutriments chez les modèles animaux en génétique et en
médecine (Rawls, Samuel, and Gordon 2004; Wong and Rawls 2012; Heys et
al. 2018; Gould et al. 2018; Wong et al. 2015).
1.2 Accès à la biodiversité microbienne, de la boite de Pétri au
métagénome
Les microbes cultivables au laboratoire ne représentent qu’une infime
minorité (< 5%) de la biodiversité microbienne (Donachie and Begg 1970;
Konopka 1984; Amann, Ludwig, and Schleifer 1995; Connon and Giovannoni
2002; Nichols et al. 2008; Staley and Konopka 1985). Jusqu’à récemment,
ceci représentait une contrainte majeure pour caractériser la composition
génétique de l’ensemble des communautés microbiennes dans un échantillon
d’un environnement donné. Après deux décennies du séquençage du premier
génome bactérien (Haemophilus influenzae par Fleischmann et al (1995)
10
(Fleischmann et al. 1995), l’avènement des technologies de séquençage de
nouvelle génération Next generation sequencing » (NGS) a favorisé l’accès
aux ressources génétiques microbiennes de différents écosystèmes (sol, eau,
hôte-symbionte) d’une manière exhaustive. L’ensemble du répertoire
génétique d’une communauté microbienne est appelé métagénome. Il faut
toutefois garder à l’esprit que les découvertes fascinantes basées sur la
minorité cultivable, sans le recours aux approches métagénomiques, ont
permis aux biologistes de construire des banques des données généralistes
(GenBank, KEGG. UNIPROT etc.) , spécialistes et expertes (SEED, IMG/M,
BRENDA, etc.) à l’échelle des fonctions, des domaines, des protéines, de
gènes, de génomes et même des populations (Benson et al. 2013; Kanehisa
et al. 2014; “UniProt: The Universal Protein Knowledgebase” 2017;
Schomburg et al. 2004; I.-M. A. Chen et al. 2019). Ces connaissances
acquises contribuent aujourd’hui à donner des réponses claires aux questions
centrales portant sur les bases génétiques de la diversité microbienne, de
leurs capacités adaptatives et de leurs potentiels évolutifs dans un contexte
anthropique, ou naturel.
1.3 L’avènement du séquençage de nouvelle génération
Le NGS (next-generation sequencing) désigne une nouvelle série de
technologies qui a révolutionné le séquençage des acides nucléiques et
aminés. Le rendement du séquençage a augmenté de manière exponentielle
alors que le prix ne cesse de diminuer rendant ces technologies très rentables.
Les technologies NGS ont prouvé leur efficacité, rapidité et surtout leur
exactitude (Goodwin, McPherson, and McCombie 2016). Plusieurs
méthodologies de NGS ont remarquablement influencé nos connaissances
dans presque tous les domaines de la biologie moderne, depuis l’évolution
moléculaire jusqu’au diagnostic des maladies contagieuses et infectieuses et
en passant par les thérapies des maladies génétiques, le champ d’application
est très large. Depuis la découverte de la structure de l’ADN en 1953, à
l’élucidation du code génétique, cela nous a pris plus d’un quart de siècle pour
développer la méthode Sanger, une première génération de séquençage de
l’ADN, basée sur une synthèse chimique connue sous le nom « séquençage
par synthèse » ; un seul acide nucléique est déterminé à la fois, en fonction
de la longueur des séquences d’ADN d’un gène ou d’un génome. Une décennie
11
plus tard, le séquençage Sanger a été automatisé par « Applied Biosystems »
en 1987, traçant depuis, la voie vers le séquençage du premier génome
bactérien (Haemophilus influenzae) en 1995 (Fleischmann et al. 1995), puis
la levure (premier génome eucaryote) en 1996 (Goffeau et al. 1996), et le
génome de l’Homme en 2001 (Lander et al. 2001). Ce dernier a coûté 150
millions de dollars (https://www.genome.gov/about-genomics/fact-
sheets/Sequencing-Human-Genome-cost) au consortium international du
projet « Human genome » et une longue décennie de cartographie physique
et génétique pour compléter le génome. À la fin du projet en 2001, le
séquençage de Sanger avait atteint ses limites théoriques et techniques. Par
la suite, afin d’accélérer le séquençage, augmenter le rendement et
notamment réduire le coût et le temps, une nouvelle ère des méthodes
s’ouvra en 2005 avec la technologie 454.
La démocratisation du NGS avait par conséquent réduit drastiquement le coût
du séquençage du génome humain de 2.7 milliards ou millions? aux environs
1 000 dollars.
Les méthodes NGS se divise en deux grandes catégories : séquençage par
ligation (« SOLiD », « Complete Genomics ») et séquençage par synthèse
(Illumina, 454, Ion Torrent). La plupart des technologies se base donc sur la
dernière, et se classifie en deux méthodes, le séquençage par clôture cyclique
(Illumina), et le séquençage par addition successive des nucléotides (454,
Ion Torrent). En seconde et en troisième générations le séquençage par
synthèse produit respectivement des courtes et des longues lectures
(Goodwin, McPherson, and McCombie 2016). Les technologies de courtes
séquences (Illumina, Qiagen, 454, Ion Torrent) se caractérisent par les coûts
les plus bas, et produisent des séquences de haute précision, utiles pour la
détection des variants génétiques à l’échelle populationnelle. Cependant, les
technologies de longues séquences (PacBio et ONT) assurent des longues
séquences afin d’assembler des génomes complets ou pour d’autres
applications comme le séquençage des isoformes. Ainsi, chaque technologie
présente ses avantages et ses limites en termes de précision, de fiabilité, de
longueur des séquences et d’erreurs de séquençage (Goodwin, McPherson,
and McCombie 2016; Kumar, Cowley, and Davis 2019; Vincent et al. 2017).
12
Par conséquent, les méthodes NGS procurent un accès aux ressources
génétiques de la biodiversité microbienne. Depuis plus qu’une décennie, elles
ont ouvert le chemin aux approches omiques « Omics », dont la
métagénomique, pour identifier les micro-organismes et caractériser leur
contenu génétique à l’échelle phénotypique, fonctionnelle et métabolique.
1.4 Les approches de la métagénomique
Tout d’abord un métagénome c’est l’ensemble des matériels génétiques des
microorganismes séquencés dans un échantillon d’un environnement donné
(sol, air, eau, animal hôte, plante hôte).
Les méthodes métagénomiques sont souvent subdivisées en deux approches.
La première consiste en l'amplification ou séquençage d’un seul gène
marqueur conservé universellement chez tous les micro-organismes, par
exemple le gène de la sous-unité 16S de l’ARN ribosomique chez les
Bactéries, de la sous-unité 18S chez les Eucaryotes microbiens (zooplanctons
et phytoplanctons). Alors que la deuxième approche consiste en un
séquençage de l'ensemble du contenu génétique d’une communauté
microbienne « Whole-Genome-Sequencing » (WGS), ce qui procure un
aperçu global du répertoire fonctionnel et du contenu génétique du
métagénome en question.
1.5 L’approche d’amplicons basée sur un gène marqueur universel
Cette approche consiste à amplifier un gène marqueur à partir d’un
échantillon d’ADN. Une amplification « PCR » est réalisée pour quantifier
l'abondance relative des variants du marqueur en question au sein de la
communauté microbienne présente dans un échantillon donné. Le gène
ribosomique 16S chez les Procaryotes et le gène 18S chez les Eucaryotes sont
donc des marqueurs universels couramment utilisés en phylogénie
moléculaire et plus généralement pour détecter les changements de
composition des communautés microbiennes.
Le gène de l’ARN ribosomique 16S a été proposé comme un marqueur de
phylogénie pour la première fois par Carl Woese (Woese 1987) . La taille de
ce gène fait environ 1500 paires de bases, il possède des propriétés qui
répondent aux critères d’un candidat d’un marqueur génétique universel pour
différencier les différentes lignées bactériennes. Le gène 16S est ubiquitaire,
13
très conservé entre les différentes lignées bactériennes et évolue lentement
(Gillespie et al. 2006; Pei et al. 2010). Il est composé de de neuf domaines
qui se subdivisent entre deux types de séquences conservées et variables.
Les domaines conservés évoluent très lentement avec peu de mutations
fixées. Les régions variables sont ciblées par des amorces universelles car
elles permettent généralement distinguer entre les différents genres
bactériens (Baker, Smith, and Cowan 2003; Guo et al. 2013; Kembel et al.
2012; Caporaso et al. 2011), même si certaines souches peuvent être autant
plus divergentes au sein d’un même genre que d’entre genres différents
(Johansen et al. 2017; Acinas et al. 2004; Janda and Abbott 2007).
L’amplification de la séquence entière du marqueur n’est pas obligatoirement
nécessaire, cependant une région hypervariable serait ciblée comme un code-
barre génétique. Après amplification de ce dernier, le produit de PCR est
purifié et enfin séquencé. Ensuite, les séquences lues des amplicons sont
analysées par des méthodes bio-informatiques, qui se résument par les
étapes suivantes :
Le contrôle de qualité consiste sur l'élimination de courtes séquences, la
filtration des homopolymères et des séquences ayant un score de qualité
moyen à très bas.
La correction du nombre de copies s’il s’agit du marqueur (16S/18S), et
filtration des chimères produites par le biais d'hybridation non spécifique des
amorces durant l'amplification PCR.
La classification des séquences filtrées en unités de base de diversité tout en
se basant sur un seuil de similarité nucléotidique (97% à 99%) arbitraire
communément utilisé en écologie microbienne, et à la fois contesté (Edgar
2018). Au sein d’une même unité de diversité obtenue avec une similarité de
97%, les études documentent une grande diversité des espèces, par exemple
au sein du genre Bacillus (Maughan and Van der Auwera 2011; Connor et al.
2010), d’où la contestation de l’approche de seuil de similarité. Les unités
taxonomiques sont alors annotées taxonomiquement et comparées
phylogénétiquement afin de comprendre la structure et la composition des
communautés microbiennes en question (Figure 1).
14
1.6 L’approche métagénomique globale basée sur le séquençage de
l’ADN total.
Pour cette approche, l’échantillon d’ADN est fragmenté aléatoirement par
fragmentation mécanique. Les fragments obtenus sont de différentes tailles,
ils sont clonés aléatoirement dans des vecteurs afin de construire une banque
de librairies à petits ou larges inserts. Tout dépend de l’intérêt de l’étude, les
librairies peuvent être criblées afin de cibler un gène d'intérêt biomarqueur
de l’écosystème en question, ou bien séquencées dans son intégralité afin de
permettre la caractérisation du contenu génétique et taxonomique du
métagénome (Schéma 1.1). Après le séquençage à haut débit, l’assemblage
des séquences en « contigs », suffisamment longs permet de prédire le
répertoire des gènes dans l’échantillon. Selon la complexité de la diversité de
l’environnement étudié, les assembleurs peuvent permettre la reconstitution
de génomes presque complets à partir d’un métagénome (Albertsen et al.
2013; Parks et al. 2015; Iverson et al. 2012; Albertsen et al. 2013; Mehrshad
et al. 2016). L’analyse bio-informatique des méta-communautés post-
assemblage se résume en trois étapes principales :
Prédire des cadres de lecture ouverts (« Open Reading Frame » ORF) des
gènes.
Assigner des affiliations taxonomiques de gènes prédits.
Annoter les fonctions de gènes en spécifiant les ontologies de leurs fonctions.
L’abondance de chaque gène est donc déterminée par la fréquence de son
occurrence et la profondeur des séquences lues qui ont servi à l’assemblage
de son contig prédit.
1.7 Approche de l’opération taxonomique universelle (UTO)
15
Schéma 1. 1 Approches métagenomiques de l’analyse de microbiote.
Édité de Morgan XC, Huttenhower C (2012) PLoS Comput Biol 8(12): e1002808.doi:10.1371/journal.pcbi.1002808
Une étape cruciale de toutes les études en écologie microbienne est le
regroupement des séquences des communautés de micro-organismes en
groupes (clusters en anglais) phylogénétiquement proches (97% de
similarité) appelés OTU (« operational taxonomic unit ») (Schéma 1.1).
Cependant, cette approche a été remise en question à plusieurs reprises, car
le concept OTU ne prend pas en compte la théorie de spéciation chez les
Bactéries (Preheim et al. 2013; T. S. B. Schmidt, Rodrigues, and Mering
2014). Des approches alternatives ont été proposé récemment pour
minimiser le biais introduit par le seuil d’identité.
16
La notion de l’OTU représente donc ici l’unité de base de diversité même si le
concept derrière la définition de l’espèce demeure problématique et un sujet
de débat actif (Doolittle and Zhaxybayeva 2009).
1.8 L’approche des variants des séquences d’amplicons (VSA)
Pour pallier le problème du seuil d’identité (97%) arbitraire de l’approche
UTO, il a été proposé récemment que les taxons devraient être définis avec
une approche d’identité exacte des séquences des gènes marqueurs
(Callahan et al. 2016). Cette approche alternative à l’UTO est basée sur la
détermination des taxons par des variants des séquences exacts (VSEs), et
connue sous le terme de variants des séquences d’amplicons (VSA)s « ASVs :
amplicons sequence variants » (Callahan, McMurdie, and Holmes 2017) ou
UTO-à-rayon zéro (zUTOs), tel que proposé par Edgar (2016).
Les VSAs de novo sont déduits d'un processus d’apprentissage automatique
dans lequel les séquences biologiques sont distinguées des erreurs sur la
base, d’un modèle d’erreur via un algorithme de débruitage « Denoising »
DADA2 (Callahan et al. 2016) et d’une fonction (f) de transition de probabilité
des erreurs. Le modèle d’erreur assume que les séquences répétées ont plus
de chances d'être observées que celles contenant des erreurs.
Par conséquent, l'inférence des VSAs de chaque échantillon ne peut pas être
effectuée indépendamment de chaque lecture considérée comme la plus
petite sous unité de données. Selon Callahan et al. (2017), contrairement aux
unités UTOs de novo, les VSAs sont des étiquettes cohérentes car ils
représentent une réalité biologique. Ces unités peuvent être déduites et
comparés avec différentes études ou différents échantillons. Cette approche
n’augmente pas seulement la résolution taxonomique, mais elle simplifie
également les comparaisons entre les études en éliminant la nécessité de
réannoter les taxons lorsque les ensembles de données sont fusionnés
(Glassman and Martiny 2018). Grâce à ces avantages, il y a eu une
augmentation du nombre de pipelines bioinformatiques cherchant à utiliser
les VSEs et à minimiser le biais d’inférence de la diversité (Callahan,
McMurdie, and Holmes 2017; Edgar 2016; Amir et al. 2017). En outre, les
auteurs de cette approche ont déclaré que les VSEs devraient remplacer les
17
OTUs (Callahan, McMurdie, and Holmes 2017). Cependant, avant l'adoption
de toute nouvelle approche, il est vraiment nécessaire de quantifier la
manière dont celle-ci se compare à un grand nombre de recherches
antérieures (Glassman and Martiny 2018). De plus, les classifications d’UTOs
restent biologiquement utiles pour comparer la diversité de grands ensembles
de données ou pour identifier des clades partageant des traits communs
(Delgado-Baquerizo et al. 2018). Parallèlement, une approche nommée Z-
UTOs a été proposée pour prendre en considération les variations intra-
taxons, elle consiste en la construction d’UTOs avec un seuil de 100% de
similarité de séquences (Edgar 2018).
Indépendamment de leur l’exactitude, les VSAs Z-OTUs et UTOs restent des
unités moléculaires de base approximatives de la diversité d’une
communauté. Comme les deux approches sont limitées quant au contexte
fonctionnel, nous aurons besoin de définir une l’unité moléculaire
fonctionnelle de la diversité.
1.9 Les mesures écologiques et phylogénétiques de la diversité
Les matrices d’abondance et d’occurrence des UTO permettent de calculer les
indices de la diversité intra-échantillons (Alpha) et inter-échantillons (Beta).
1.9.1 Les mesures de la diversité alpha
La diversité des UTO au sein d'un échantillon donné est connu sous le terme
diversité Alpha. On distingue deux types des mesures Alpha ; la richesse
spécifique (nombre des UTOs, « chao-1 index » et « ACE ») et l’équitabilité
de la distribution d’abondance des UTO (Shannon et Simpson). La
comparaison des distances écologiques et phylogénétiques inter-
communautés permet de mettre la loupe sur la divergence et la convergence
de leurs structure, fonction et diversité dans des contextes divers (pollution,
traitement, spatio-temporel).
L’indice de la richesse spécifique
Le moyen le plus simple de mesurer la diversité c'est d’énumérer les UTO
présents dans l'échantillon, appelé la richesse spécifique. La richesse ne
considère pas la structure de la communauté et elle ne détecte pas
l'abondance différentielle entre les UTO.
18
D’autres estimateurs de la diversité alpha « chao- index » (Chao 1984) et
« ACE» (Abundance-based coverage estimator of species richness) (Chao and
Lee 1992; Chao and Yang 1993) sont aussi utilisés.
Supposons que SObs soit le nombre des UTOs abondants plus que deux fois
dans un échantillon donné et a soit le nombre inconnu des taxons présents
mais qui ne sont pas observés dans la communauté, la richesse serait :
La richesse sans correction pour les taxons non observés serait
R=SObs
Dans le cas de la richesse avec correction pour les taxons non observés, on
utilise l’indice de choa1 ou l’indice :
L’indice de chao1 :
𝑆𝐶ℎ𝑎𝑜1 = 𝑠obs +𝐹1(𝐹1−1)
2(𝐹2+1)
F1 et F2 sont le nombre de singletons et de doublons, respectivement, et Sobs
est le nombre d'espèces observées
L’indice ACE :
𝑆𝐴𝐶𝐸 = 𝑠abund +𝑠rare
𝐶ACE+
𝐹1
𝐶ACE𝛾𝐴𝐶𝐸
2
Soit Sabund and Srare sont respectivement le nombre d’UTO abondantes et rares,
CACE est l’estimateur de la couverture d’abondance de l’échantillon, F1 est la
fréquence des singletons et 𝛾𝐴𝐶𝐸2 est le coefficient estimé de la variation pour
les OTUs rares.
Le coefficient de variation estimé serait défini par la formule suivante :
𝜸𝑨𝑪𝑬𝟐 = 𝒎𝒂𝒙[
𝑺𝐫𝐚𝐫𝐞
𝑪𝐀𝐂𝐄
∑ 𝒊(𝒊−𝟏)𝑭𝒊𝟏𝟎𝒊=𝟏
(𝑵𝐫𝐚𝐫𝐞)(𝑵𝐫𝐚𝐫𝐞−𝟏)− 𝟏, 𝟎]
Les indices basés sur l’abondance
Il y a plusieurs indices qui peuvent décrire la structure d’une communauté.
Les deux mesures les plus communément utilisées sont Shannon (Lemos et
19
al. 2011; Magurran and McGill 2011) et Simpson (Simpson 1949; Lemos et
al. 2011) . Le dernier ajoute plus de poids sur l'abondance des UTO.
L’indice de Shannon (H)
𝐻 = − ∑ (𝑝𝑖 ln 𝑝𝑖)𝑆𝑖=1
Où s est le nombre total d’UTO et pi est la proportion de la communauté
représentée par l’UTO i.
L’indice de Simpson (D)
𝐷 =1
∑ 𝑝𝑖2𝑠
𝑖=1
Où s est le nombre total des taxons et pi est la proportion de la communauté
représentée par l’UTO i.
Ces indices ne sont pas linéaires, ce qui signifie qu’un échantillon avec un
indice Simpson de 0.8 n'est pas deux fois plus diversifié qu’un échantillon
avec indice Simpson de 0,4. Un moyen plus représentatif de la diversité Alpha
contourne ce problème par le calcul de la diversité effective tel que proposé
par (Jost 2006; 2007).
1.9.2 Les mesures de la diversité Beta
Le changement de la diversité d’un d’échantillon à l’autre est connu sous le
nom de la diversité Beta. La comparaison de cette diversité implique des
calculs matriciels des distances entre les échantillons, basés sur leurs profils
taxonomique ou génétiques (UTO). Il existe plusieurs façons de calculer les
distances tout en se basant sur la similarité des profils d'abondance et la
proximité génétique des UTO. On distingue les distances écologiques des
distances phylogénétiques.
1.9.2.1 Les distances écologiques
Les distances de Jaccard (Jaccard 1912) et Bray-Curtis (Bray and Curtis
1957) sont les mesures les plus communément utilisées. L’indice de Jaccard
mesure les similitudes entre deux ou plusieurs groupes. L'indice de Bray-
Curtis (BC) est utilisé pour quantifier la dissimilarité compositionnelle entre
deux sites écologiques différents. Les abondances sur chaque site sont
exprimées en proportion. La dissimilitude de Bray-Curtis est comprise entre
20
0 et 1, 0 signifiant que les deux sites ont la même composition (c'est-à-dire
qu'ils partagent toutes les espèces) et 1, que les deux sites ne partagent
aucune espèce.
1.9.2.2 Les distances phylogénétiques
Les distances UniFrac (C. Lozupone and Knight 2005; C. A. Lozupone and
Knight 2015), prennent en compte la distance génétique des UTO dans
chaque échantillon. La distance Unifrac pondérée ajoute l'information de
l'abondance relative de chaque UTO à chaque distance génétique. La méthode
construit un arbre phylogénétique de tous les UTO, puis il calcule le ratio des
longueurs de branches d’UTO uniques versus celles d’UTO communs.
La version non pondérée de l’Unifrac transforme l’abondance en occurrence.
Les distances pondérées et non pondérées sont sensibles à la présence des
plus rares et des plus dominants. Une version plus équilibrée a été proposée,
dénommée distance Unifrac généralisée (J. Chen et al. 2012). La variante
Gunifrac prend en compte le paramètre d’équitabilité de l’abondance relative
des UTO dans le processus de pondération.
Les distances phylogénétiques ne peuvent pas décrire les corrélations ni les
co-occurrences des UTO. Pour ce faire les réseaux sont des outils importants
pour visualiser les interactions entre les communautés microbiennes.
1.10 Réseaux d’interactions microbiennes
Les réseaux, largement appliquées en génomique, sont des outils
extrêmement puissants afin d’explorer les relations entre les entités
biologiques, par exemple, la similarité entre deux gènes, les interactions
entre deux protéines, ou en écologie microbienne par la co-occurrence des
taxons (Weiss et al. 2016). Les corrélations de OTU, basées sur l’abondance
ou l’occurrence décrivent bien des liens cohésifs au sein d’une méta-
communauté, mais pas nécessairement leur causalité. En revanche, les
propriétés topologiques des réseaux peuvent prédire les processus
écologiques conducteurs de la dynamique spatiale et temporelle des
interactions microbiennes.
En considérant la simple formulation de Lotka-Volterra des interactions entre
deux espèces, les différents types d’association comme le mutualisme,
21
commensalisme ou le parasitisme ont été aussi élucidés (Lidicker 1979). Il
est important de citer qu’il existe au moins six types d'interactions
qualitativement distinctes.
Le mutualisme concerne les interactions qui sont bénéfiques pour les deux
partenaires d’interactions.
2) Le commensalisme désigne les interactions qui sont bénéfiques pour l'un
et sans effet pour l'autre.
3) Le parasitisme définit les interactions qui sont bénéfiques pour l'un et
néfastes pour l'autre.
4) Le neutralisme se produit quand les interactions n’ont pas d'effet sur les
deux partenaires.
5) L’amensalisme englobe les interactions qui ont des effets négatifs sur l’un
et pas d’effets sur l’autre.
6) La compétition concerne les interactions qui sont néfastes pour les deux
partenaires.
Du point de vue écologique, l’orientation de la direction des arrêtes dans les
réseaux d’interactions indique si une espèce affecte d’autres espèces. Par
exemple, cela voudrait dire, que les réseaux sans arrêtes orientés ne peuvent
pas décrire où distinguer qualitativement la nature des interactions entre OTU
à part les corrélations positives et négatives des abondances. À l'heure
actuelle, il est difficile d’inférer des réseaux dirigés sans l’acquisition des
séries des données temporelles car la modélisation mathématique de la
dynamique de communautés peut nous renseigner sur les types de ces
interactions.
Les fonctions métaboliques sont souvent plus informatives que les réseaux
de co-occurrence. Étant donné que les fonctions métaboliques des microbes
peuvent être différentes ou redondantes, elles peuvent indiquer une
complémentarité ou une redondance métabolique et ainsi déterminer si les
arêtes sont susceptibles de représenter un mutualisme ou une compétition.
Des outils basés sur des modèles métaboliques peuvent fournir de telles
prédictions. Par exemple, le calcul de la consommation et la sécrétion de
métabolites permettent de fournir des indices de complémentarité ou de
22
compétition (Cao et al. 2015). En revanche, d’autres méthodes se focalisent
sur l’analyse du bilan du flux métabolique pour calculer les taux de croissance
des espèces en monoculture et en co-culture. Ces outils permettent aux
chercheurs de prédire les interactions écologiques entre les espèces.
Cependant, les modèles métaboliques nécessitant la séquence complète du
génome, sont souvent limités par une annotation médiocre et peuvent être
inexacts sans vérification manuelle de l’annotation par des experts (Henry et
al. 2010). Par conséquent, leur utilisation est limitée à un petit nombre
d'espèces bien étudiées (Faust et al. 2018).
1.11 Les quatre forces évolutives et processus écologiques
1.11.1 La mutation
La variation génétique est une condition préalable aux changements évolutifs.
En l'absence d'une telle variation, aucun changement ne peut être réalisée.
La variation génétique est essentiellement générée par mutation. Il est donc
clair que la mutation est une force évolutive majeure et moteur de l'évolution
(Hershberg 2015; M. Lynch 2016). Lorsqu’une mutation se produit chez un
individu, sa fréquence augmentera au sein de la population ou disparaîtra de
la population. La fixation des mutations à travers les générations dépend
d’une combinaison avec la force de sélection naturelle, de la migration ainsi
que la dérive génétique.
Chez les bactéries, les gènes correspondent à 85–90% du génome et les
régions intergéniques sont petites et englobent de sites régulateurs (Mira,
Ochman, and Moran 2001). Les gènes ont peu ou pas d'introns et sont
généralement organisés en unités polycistroniques (opérons). Les mutations
ne se restreignent pas aux substitutions nucléotidiques synonymes et non-
synonymes. Rappelons qu’une substitution est un remplacement d’une seule
base nucléotide d’un codon au sein un gène. Si la protéine codée par ce gène
était modifiée, la substitution est synonyme (souvent troisième position du
codon), sinon la substitution est non-synonyme (deuxième position du
codon). Il existe plusieurs types de mutations telles que les insertions et de
délétions et les mécanismes de transferts horizontaux connus chez les
procaryotes ; la conjugaison la transduction et la transformation. Les
insertions et les délétions de matériel génétique ont généralement des effets
23
délétères sur l'expression des gènes (Price, Arkin, and Alm 2006). Les
mécanismes du transfert horizontal représentent une source d'hérédité
horizontale qui permet un flux des gènes transdomaines en éliminant les
barrières génétiques même entre des lignées évolutives distantes (Dagan,
Artzy-Randrup, and Martin 2008; Aminov 2011; Zhaxybayeva, Lapierre, and
Gogarten 2004). Le transfert horizontal est essentiel pour l’adaptation des
cellules aux nouvelles conditions écologiques, y compris les environnements
cliniques et ceux contenant des antibiotiques (Niehus et al. 2015). Les
mécanismes des transferts horizontaux chez les Bactéries sont fréquents et
se produisent naturellement à l’échelle des secondes (Kejnovsky and Trifonov
2016; Andam, Williams, and Gogarten 2010; Syvanen 2012). De l’autre coté,
les Bactéries n'échappent pas aux remaniements connus en génomique,
comme les phénomènes de duplication, de transposition et d'intégration qui
se produisent aussi spontanément et dépendamment du contexte évolutif.
Donc, la duplication des gènes ou des opérons peut engendrer une source de
variation importante dans les génomes bactériens. La transposition et les
transferts horizontaux de gènes notamment la conjugaison plasmidique sont
des mécanismes de transport génétique qui facilitent soit l’insertion des
éléments génétiques mobiles (mobilome) soit le flux de gènes accessoires
adaptatifs (Dabbs and Sole 1988; Pérez Carrascal et al. 2016; Ana Segura
2014; Gillings 2013; Dalia et al. 2015). L’intégration via des intégrons
(cassettes des gènes de résistance aux antimicrobiens ou de virulence) est
également connue chez les Bactéries (Frost et al. 2005; Starikova et al. 2012;
Oliveira-Pinto et al. 2016). De même, la recombinaison par site homologue
est largement connue et étudiée (Castillo-Ramírez et al. 2011). En somme,
les sources des mutations chez les Bactéries sont très variées, ce qui suggère
la mutation comme une force majeure génératrice permanente de la diversité
génétique moléculaire.
La mutation est souvent étudiée en assumant que certains types des
mutations d'ADN (par exemple les substitutions synonymes) évoluent
entièrement d'une manière neutre. Les premières études d’évolution
expérimentales des Bactéries (exposition des Bactéries aux virus) ont montré
pour la première fois que les mutations se sont produites avant la sélection
(Luria and Delbrück 1943).L’expérience de Salvador Luira et Max Delbrück
24
fut montré que les mutations surviennent de manière aléatoire spontanée et
ne sont pas induites par le force de sélection naturelle. Plusieurs méthodes
pour éliminer les effets de la sélection naturelle durant les études
expérimentales de mutation ont été discutées (Hershberg 2015), par exemple
les expériences d’accumulation des mutations (Elena and Lenski 2003; Lind
and Andersson 2008; Brockhurst, Colegrave, and Rozen 2011).
1.11.2 La sélection
La sélection naturelle est le concept central de la théorie de l'évolution de C.
Darwin. Cette force évolutive a révolutionnée nos idées sur la façon dont les
êtres vivants ont évolué depuis leur ancêtre commun. Elle affirme que les
variations spontanées étaient une caractéristique commune de toutes les
formes de vie(Moxon 2011). Elle explique comment les populations peuvent
évoluer de telle manière qu'elles deviennent mieux adaptées à leur
environnement au fil du temps et au cours des générations. Les variants
spontanés qui sont mieux adaptés augmentent en fréquence grâce à la
sélection naturelle parce que leur progéniture hérite de ces caractéristiques
avantageuses pour survivre.
Les effets de la sélection sur la fréquence des gènes au cours des générations,
ont été largement étudiés par les généticiens des populations (Dykhuizen
1990). Les causes et la façon dont la variation génétique dans un
environnement crée des différences sélectives (Gac et al. 2012; Wielgoss et
al. 2013; Good et al. 2017; Maddamsetti, Lenski, and Barrick 2015). La
compréhension du phénotype (par exemple les facteurs de la croissance
bactérienne, les voies métaboliques) et la définition des composants
importants de l'environnement comme les facteurs biotiques et abiotiques
deux défis majeurs dans les études expérimentales de la sélection.
La sélection naturelle peut fonctionner de différentes manières en distinguant
trois modes principaux: la sélection négative ou prufiante, la sélection
positive et la sélection balancée (Shapiro 2014). La sélection négative (parfois
appelée purifiante) est la tendance des individus inaptes à se reproduire
moins et donc à être éliminés de la population. La sélection positive favorise
25
la survie et la reproduction de variants avantageux et concurrentiels par
rapport au reste de la population. Lors d'un balayage sélectif, des variants
sélectionnés positivement remplacent des variants non sélectionnés. La
sélection diversifiante pourrait être considérée comme des balayages sélectifs
incomplets mais favorables dans une population (Shapiro 2014). Le balayage
sélectif est le processus par lequel un gène adaptatif ou un allèle
(éventuellement un variant spécifiant une niche) se répand dans une
population par recombinaison plus rapidement que par expansion clonale. Par
conséquent, la variante adaptative est présente dans plus d'un clone et la
diversité n'est pas purgée à l'échelle du génome (Shapiro 2014). La sélection
diversifiante est en cas particulier fréquence-dépendante négative, lorsqu’un
individu avec un phénotype considéré comme "rare" par rapport aux autres
individus pourra gagner en survie ou reproduction grâce à cette rareté. C'est-
à-dire qu'un individu sera favorisé et aura un gain plus important en fitness
si son phénotype est moins fréquent que les autres individus de la population.
C'est une des pressions de sélection les plus fortes, cependant les exemples
dans la nature sont assez rares, tout du moins difficilement observables
directement. La fréquence-dépendante négative a été observée chez les
gènes comportements rares des drosophiles (Fitzpatrick et al. 2007) et le
gènes de coloration des males chez les poissons guppies (Olendorf et al.
2006)
Dans un autre exemple, une mutation particulière est favorisée par la
sélection positive lorsqu'elle est à faible fréquence, mais elle devient délétère
à haute fréquence. La mutation ne balaye jamais toute la population, mais
fluctue autour d'une fréquence intermédiaire (Shapiro 2014). En fonction de
la délimitation des frontières entre les populations, il peut être difficile de
distinguer la sélection diversifiante et la sélection positive chez les micro-
organismes (Shapiro 2014).
Deux autres types de sélection sont aussi connues, la sélection par « auto-
stop » génétique et la sélection purifiante. Lorsqu'une mutation avantageuse
serait fixée dans une population par sélection positive, une autre mutation
neutre ou délétère physiquement liée, peut faire l'auto-stop avec la fixation
de la première. Cet effet est connu sous le terme de la sélection liée (linked
26
selection) par auto-stop génétique (Birky and Walsh 1988; Smith and Haigh
1974).
Les mutations non synonymes sont beaucoup plus nombreuses et ont plus
des effets plus importants sur le fitness que les mutations synonymes (Nei
and Gojobori 1986). Vu que ces effets sur le fitness sont généralement
négatifs, ces mutations sont progressivement éliminées des populations par
sélection purifiante conduisant à de faibles ratios de taux de substitution non
synonymes (dN) sur synonymes (dS) (dN/dS <1) (Kuo, Moran, and Ochman
2009). Toutefois, les substitutions non synonymes peuvent être également
affectées par la sélection naturelle, par exemple en raison du biais lié à
l'usage de codons (Gouy and Gautier 1982; Sharp et al. 2005), en particulier
chez les bactéries à croissance rapide (Vieira-Silva and Rocha 2010). Les
régions intergéniques correspondent rarement à plus de 20% du génome et
sont sujettes à une sélection purifiante à des niveaux intermédiaires entre
ceux des substitutions non synonymes et synonymes (Thorpe et al. 2017).
Globalement, cela signifie que la plupart des mutations du génome bactérien
sont délétères (Rocha 2018).
1.11.3 La migration
La migration n’est pas uniquement une force évolutive mais aussi une force
physique. La forme, la taille et la motilité des cellules microbiennes
apparaissent comme des facteurs cellulaires intrinsèques déterminants dans
le comportement rhéologique des cultures bactériennes vivantes au cours du
processus de croissance. Vu que le fonctionnement d’un écosystème dépend
des interactions sociales et de l'auto-organisation spatiale, ces derniers
dépendent de la motilité des cellules bactériennes (Ebrahimi, Schwartzman,
and Cordero 2019). Avec la migration, le taux et l'impact des transferts
horizontaux peuvent augmenter considérablement pour des locus sous
sélection positive. Sans migration le transfert horizontal devient rare (Niehus
et al. 2015).
La force de migration est très importante car elle nous permet de comprendre
la dynamique spatiotemporelle des communautés microbiennes symbiontes,
27
planctoniques ou biofilms. En écologie microbienne la migration est une force
majeure, mais nous comprenons très peu de choses sur la structure spatiale
et la dynamique temporelle de microbiote chez l'Homme ou n'importe quel
système biologique. Des travaux récents sur le microbiote du poisson zèbre
montrent que l'influence de la structure spatiale sur la dynamique des
populations bactériennes ne peut être révélée que par imagerie directe
(Jemielita et al. 2014; Hammers et al. 2015; Wiles et al. 2016). Comment les
populations microbiennes s'établissent, croissent, fluctuent et persistent ;
sont des questions d’actualités que nous élaborons dans la section des
modèles mathématiques.
1.11.4 La dérive génétique
Cette force est la base de la théorie neutre de l'évolution moléculaire et de la
génétique des populations. La théorie neutre affirme que la grande majorité
des changements évolutifs au niveau moléculaire sont causés par la fixation
aléatoire de mutations neutres, due à la perte aléatoire de certains génotypes
rares dans des populations finies (M. Kimura 1991). La théorie affirme
également que la majeure partie de la variabilité génétique au sein de
l'espèce (telle que le polymorphisme des protéines et de l'ADN) est
sélectivement neutre ou presque, et qu'elle est maintenue dans l'espèce par
l'équilibre entre mutation et extinction aléatoire. Cette théorie suppose que
seule une minorité des mutations de novo ont soit des effets suffisamment
délétères sur le fitness qu'elles ont peu de chances de se fixer dans la
population, soit sont soumises à une sélection positive suffisamment forte
pour contrer les effets de la dérive et se fixer. Il existe toutefois des mutations
soumises à une sélection faible, qui peuvent être fixées lorsque les allèles
alternatifs sont éliminés par dérive génétique (Jensen et al. 2019).
L’importance de la force de la dérive génétique pour dicter la trajectoire
évolutive microbienne à l’échelle de la complexité d’un génome entier a été
largement débattu (Michael Lynch 2006; Charlesworth and Barton 2004;
Daubin and Moran 2004; Michael Lynch and Conery 2003; Hershberg, Tang,
and Petrov 2007). La dérive génétique est considéré comme une force qui
contribue à la réduction de taille de génomes bactériens (Kuo, Moran, and
28
Ochman 2009), par exemple chez les endosymbiontes des insectes (Alleman,
Hertweck, and Kambhampati 2018) , mais pas chez les bactéries symbiontes
marines (Bobay and Ochman 2017).
Contrairement aux Eucaryotes, pour lesquels la densité génique est très
variable (Il n’y a pas de relation entre la taille du génome et le nombre de
gènes) (Ryan Gregory 2002; Michael Lynch and Conery 2003), la taille du
génome des bactéries est étroitement liée au nombre de gènes (Mira,
Ochman, and Moran 2001; Giovannoni et al. 2005). Par conséquent, les
forces évolutives et plus particulièrement les mutations agissent
principalement au niveau des gènes codants ou de leur séquence de
régulation et ont donc des effets importants sur l'architecture globale des
génomes bactériens (Kuo, Moran, and Ochman 2009). En raison des
mutations de novo (délétions, insertions) chez les génomes bactériens
(Andersson and Andersson 2001; Mira, Ochman, and Moran 2001; Nilsson et
al. 2005; Hershberg, Tang, and Petrov 2007), les gènes seront inactivés et
perdus s'ils ne sont pas conservés par sélection.
À l’extrême, les gènes essentiels doivent, par définition, être préservés,
tandis que ceux qui n'offrent aucun effet bénéfique perdront leur fonction
avec le temps. Cependant, la plupart des gènes se situent quelque part entre
ces extrêmes et l'étendue de la dérive génétique déterminera le nombre de
gènes conservés (Ochman and Davalos 2006; Bobay and Ochman 2017;
Khachane, Timmis, and Martins dos Santos 2007).
1.11.5 Interactions des forces évolutives
Il est généralement admis que la sélection naturelle favorise la diminution
des taux de mutation, vu que les mutations sont souvent délétères (Motoo
Kimura 1967; Drake 1991; Dawson 1998; Michael Lynch 2010). Drake (1991)
et collaborateurs postulaient que la réduction de taux de mutation
impliquerait certains coûts physiologiques et par conséquent que la sélection
naturelle entraînerait à la fois la réduction des taux de mutation et la
diminution progressive de cette réduction (Drake 1991). En revanche,
Michael Lynch a suggéré un modèle alternatif selon lequel le taux de mutation
29
n'est pas fixé par sélection naturelle sur le coût physiologique, mais plutôt
par la dérive génétique (Michael Lynch 2010).
Au fur et à mesure que les taux de mutations par site nucléotidique
diminuent, la sélection devient plus faible pour réduire encore les taux de
mutation, jusqu'à atteindre un point où la sélection n'est plus assez puissante
pour contrecarrer l'action de la dérive génétique (Hershberg 2015; Michael
Lynch 2010)
Conformément au modèle de Lynch, les taux de mutation par base étaient
inversement corrélés à la taille effective de la population (Ne) chez les
Procaryotes et les Eucaryotes (Michael Lynch 2010; Sung et al. 2012). Étant
donné que le « Ne » est inversement lié à la dérive, on peut donc affirmer
que les taux de mutation deviennent plus élevés à mesure que la force de la
dérive par rapport à la sélection devient plus forte.
Lynch a ensuite raffiné son modèle de « barrière de dérive » en montrant que
la régression des taux de mutation par rapport à Ne est élevée pour les
Procaryotes par rapport aux Eucaryotes (M. Lynch 2016). La capacité de la
sélection naturelle à raffiner un phénotype est finalement limitée par le bruit
créé par la dérive génétique qui est elle-même limitée par la taille finie de la
population et des effets stochastiques de mutations associées (M. Lynch
2016). Cette découverte suggère que, pour un Ne donné, la sélection peut
être moins efficace pour réduire les taux de mutation chez les Procaryotes.
Pour expliquer ce phénomène, Lynch a suggéré que la magnitude de la
sélection visant à réduire les taux de mutation ne soit pas simplement
fonction du taux de mutation par base, mais également du potentiel de
mutation délétère du génome.
Les communautés microbiennes possèdent des caractéristiques bien
différentes des Métazoaires, car elles évoluent principalement par mutation
(Hershberg 2015; M. Lynch 2016). Ainsi, les mécanismes des transfert
horizontaux chez les Bactéries sont fréquents et se produisent à l’échelle des
secondes(Kejnovsky and Trifonov 2016; Andam, Williams, and Gogarten
2010; Syvanen 2012). Ceci suggère un rôle dominant des forces
30
diversifiantes de l’évolution moléculaire, la mutation et la migration, cette
dernière facilite l’adaptation avec un flux de gènes par transfert horizontal.
1.11.6 De l’évolution à l’écologie
1.11.6.1 L’adaptation Locale
L'adaptation locale est le résultat d'une sélection diversifiante en fonction des
conditions locales, qui peut structurer des populations et des communautés
microbiennes. Formellement, la théorie de l’adaptation locale est souvent
appliquée à des métapopulations c.-à-d. groupes de populations qui
interagissent les unes avec les autres par la migration (S. A. Kraemer and
Boynton 2017). Une population dans une métapopulation est plus adaptée
localement lorsqu'elle est plus sélectionnée par son environnement d'origine,
avant que les populations d'autres habitats soient présents dans cet
environnement (S. A. Kraemer and Boynton 2017; Kawecki and Ebert 2004).
L'habitat ici fait référence aux conditions environnementales dans lesquelles
un génotype existe.
Les recherches expérimentales sur l'adaptation locale chez les microbes ont
porté sur des génotypes individuels en tant qu'unités d'adaptation, car la
caractérisation des populations microbiennes dans la nature reste difficile.
Chaque génotype est traité comme une population et un cadre d’adaptation
locale à la métapopulation est utilisé. À l'opposé, quelques études ont défini
des communautés entières comme unités de divergence (S. A. Kraemer and
Boynton 2017; Hoostal, Bidart-Bouzat, and Bouzat 2008; Z. I. Johnson et al.
2006). La sélection, le flux de gènes par migration, la recombinaison, et la
spéciation microbienne sont des processus principaux mis en œuvre durant
l’adaptation locale (S. A. Kraemer and Boynton 2017).
L'adaptation locale dans la nature a été largement étudiée dans les systèmes
macrobiens (Fraser et al. 2011; Anderson et al. 2010), mais des difficultés
techniques rendent difficiles les études de terrain sur l'adaptation
microbienne locale. Contrairement aux plantes et aux animaux, de nombreux
microbes sont faciles à manipuler en laboratoire, mais leur dynamique reste
difficile à étudier sur le terrain. Par conséquent, la plupart des informations
31
sur l'adaptation microbienne locale proviennent de l'évolution expérimentale
de microbes cultivables en laboratoire (S. A. Kraemer and Boynton 2017).
Deux principaux défis doivent être pris en compte lorsqu'on étudie
l'adaptation locale. Premièrement, des données historiques sont nécessaires
pour identifier les environnements, les échelles spatiales, les populations
microbiennes et les traits avantageux pour une sélection divergente.
Deuxièmement, les microbiologistes ont besoin d'outils pour mener des
expériences de transplantations réciproques rigoureuses. Seules les
expériences de transplantations réciproques fournissent les données de
fitness qui sont nécessaires pour confirmer l'adaptation locale. Ces deux défis
sont surmontables (S. A. Kraemer and Boynton 2017).
Les observations à l’échelle des séquences du génome entier sont
particulièrement utiles pour identifier les pressions sélectives potentielles
conduisant à une adaptation locale (Shapiro and Polz 2014). Ces études en
majorité analysent le génome d'individus appartenant à des populations
différentes afin de détecter la présence de régions génomiques divergentes.
Les statistiques de divergence utiles incluent les mesures FST et la divergence
nucléotidique (Ellison et al. 2011); Des études d'association à l'échelle du
génome peuvent également être utilisées pour détecter la base génétique des
phénotypes d'intérêt (Bush and Moore 2012; Shapiro et al. 2012; Dutilh et
al. 2014). Lorsque les données de séquence du génome entier sont
disponibles, la divergence peut être comparée à cette échelle alors que le
long du génome entier; lorsque seulement quelques gènes peuvent être
séquencés, les locus sélectionnés peuvent être comparés à des locus neutres
ou à des prédictions neutres basées sur un modèle (Sokurenko et al. 2004).
1.11.6.2 L’adaptation rapide
La dynamique active et cryptique, ainsi que l’évolution rapide des
communautés microbiennes peut certainement masquer les interactions
trophiques (Yoshida et al. 2007). Ce que nous appelons « dynamique
cryptique », est un phénomène dans lequel la force ou même l’existence d’un
lien trophique prédateur-proie est masquée par les cycles de dynamique
32
évolutive de la proie. Ce phénomène a été observé dans les chemostats des
rotifères-algues et bactériophages. Les observations expérimentales
confirment que l'évolution des proies peut altérer considérablement la
dynamique prédateur–proie et que, par conséquent, toute tentative de
comprendre les oscillations de la population dans la nature ne peut pas
négliger les effets potentiels de l'évolution rapide en cours (Yoshida et al.
2003).
Plusieurs études ont observé que l'évolution par sélection naturelle peut
fonctionner à la même échelle temporelle que la dynamique écologique, plus
précisément, les recherches éco-évolutives récentes sur la manière dont
l'adaptation rapide influence l'écologie et vice versa. L’évolution par des
forces non adaptatives peut se produire également rapidement, avec des
conséquences écologiques mais, pour tenir compte de l'ensemble des
interactions écologie-évolution (éco-évo), il faut aborder explicitement les
processus au niveau de la génétique et la démographie des populations
(Lowe, Kovach, and Allendorf 2017).
1.12 Processus écologiques et modèles
Par analogie aux quatre forces évolutives principales décrites en génétique
des populations : la mutation, la sélection, la dérive génétique, et le flux de
gènes, en écologie microbienne, quatre processus de base contribuant à
l'assemblage d'une communauté microbienne sont considérés (Vellend 2010;
Nemergut et al. 2013): la diversification, la dispersion, la sélection et la dérive
(Schéma 1.2).
33
Schéma 1. 2. Boite noire de l’écologie des communautés.
Édité de Mark Vellend 2010. Quarterly Review of Biology 85(2):183-206.
Les différentes combinaisons spatio-temporelles de ces processus, pourraient
être importantes dans la structuration des communautés microbiennes.
Les modèles en évolution moléculaire et génétique des populations consistent
donc à estimer les taux d'évolution, l'indice de différenciation du
polymorphisme génétique et la variation dans le temps des fréquences
alléliques alors que les modèles d'évolution en écologie microbienne se basent
principalement sur l’information de l'abondance, l’occurrence, et la distance
phylogénétique pour estimer la fréquence, la richesse des UTOs et leur taux
de migration (Vellend 2010).
1.13 Introduction à la modélisation mathématique en écologie et
évolution microbienne
Un modèle mathématique est défini comme une équation, ou un ensemble
d’équations, qui essayent d’expliquer des exemples de réalité d’une manière
simplifiée, en utilisant uniquement les propriétés les plus pertinentes d'un
système (Perez-Rodriguez and Valero 2013). Ensuite, une théorie scientifique
est fondée lorsqu'une explication mécanistique est donnée pour l’ensemble
de phénomènes observés.
34
L’occurrence et l’abondance des communautés microbiennes, ainsi que leurs
interactions sont de propriétés dynamiques qui changent continuellement
dans le temps (Zaccaria, Dedrick, and Momeni 2017). Il est donc nécessaire
de comprendre leur adaptation aux fluctuations des conditions
environnementales et leur évolution sous l’effet des perturbations
(changements de l'environnement, administration des antibiotiques, ou
l’invasion des pathogènes, xénobiotiques etc.) de ces dernières. La
modélisation mathématique peut alors prédire la dynamique de ces
communautés microbiennes perturbées, et la question majeure qui se pose
ici, est comment les modèles permettent-ils de prédire la dynamique
d'évolution des populations microbiennes ?
La modélisation d’une population consiste à prédire sa taille (N) au cours des
générations en utilisant des équations différentielles en fonction de temps, de
la disponibilité des ressources et des taux de croissance exponentiel (K).
1.13.1 Préambule historique de la modélisation
Initialement, l’idée de taux de croissance exponentielle a été exploré par
Thomas Robert Malthus (1766-1834) par dans un essai intitulé « Principe de
population » (Principle of Population, Six éditions, 1798 à 1826). Puis, le
mathématicien Benjamin Gompertz (1779-1865) avait suggéré les premières
équations différentielles en élucidant l’hypothèse : le taux de croissance (K)
décline exponentiellement avec le temps indépendamment la densité de
population.
En parallèle, le mathématicien Pierre-François Verhulst (1804-1849) fut
intéressé par la théorie des nombres pour décrire dynamique des populations.
Il fut publié dans correspondance mathématique et physique en 1838 :
"Notice sur la loi que la population poursuit dans son accroissement". En fait,
Il proposa l’équation logistique « Logistic equation » comme une alternative
plus réaliste à la loi de Malthus. Il fut assumé que le taux de croissance K est
une fonction de la disponibilité des nutriments. Dès lors, il publia en 1845 un
livre intitulé "Recherches mathématiques sur la loi d'accroissement de la
population" dans les Nouveaux Mémoires de l'Académie Royale des Sciences
et Belles-Lettres de Bruxelles.
35
Quelques décennies plus tard, Alfred James Lotka (1880-1949),
mathématicien, physicien chimiste et statisticien, célèbre pour ses travaux
sur la dynamique des populations et l’énergétique, publia en 1925 un livre
intitulé ‘Les éléments mathématiques pour la biologie’ "Elements of
Mathematical Biology". Lotka fut proposé que la sélection naturelle soit, à la
base, une lutte entre les organismes pour obtenir l’énergie disponible; les
organismes qui survivent et prospèrent sont ceux qui capturent et utilisent
l'énergie à une capacité plus efficace et élevée que ceux de leurs concurrents.
Alors, son contemporain, le mathématicien Vito Volterra (1860-1940), connu
pour ses contributions à la mathématique de biologie et aux équations
intégrales, publia "Leçons sur la théorie mathématique de la lutte pour la vie",
Paris, 1931. Lotka et Volterra ont tous les deux (indépendamment) proposé
des modèles d’écologie théorique décrivant les populations concurrentes et
les systèmes prédateurs-proies (Kingsland 2015; Guerraggio and Paoloni
2013) Ils ont suggéré des équations d'évolution des états d’équilibre stables
décrites avec des équations différentielles de la variation des proies et des
prédateurs dans le temps.
Quelques années plus tard, le biologiste russe, Georgyi Frantsevitch Gauze
(1910-1986) a travaillé sur la structure des populations de microorganismes
en culture, et en 1932 il publia « le principe de l’exclusion compétitive ». Son
travail expérimental était basé sur deux espèces des levures en culture,
formalisé théoriquement par le modèle Lotka & Volterra. Deux ans plus tard,
il a publié le livre "La lutte pour l'existence" (The struggle for existence,
1934). En parallèle, Kermack et McKendrick développèrent les premiers
modèles mathématiques épidémiologiques (Kermack, McKendrick, and
Walker 1927).
Comme la théorie de l’exclusion compétitive de Gauze était fondé uniquement
sur deux microorganismes, Robert McCredie May (1936 -) s’intéressait alors
aux interactions de N espèces et à la dynamique des populations animales et
aux relations entre complexité et stabilité dans les communautés naturelles.
Il a réalisé des avancées majeures en biologie des populations grâce à
l'application de méthodes mathématiques (Sugihara and May 1990). Il a
développé le cadre théorique de l'écologie des communautés (stabilité et
complexité dans des écosystèmes modèles). Il a également proposé des
36
outils de modélisation mathématiques à l’épidémiologie et à l'étude de la
biodiversité (May 1976).
Quant au développement des modèles mathématiques de l’évolution,
n’oublions pas la contribution cruciale des travaux de Karl Pearson et ses
collègues qui ont aussi mis au point un cadre mathématique pour la théorie
de l’évolution (Pearson and Henrici 1896; Gillham 2015) et les contributions
de généticien Sewall Wright et du mathématicien Ronald A. Fisher, au
développement théorique de la génétique des populations (Crow 2010).
Rappelons que l'un des principaux objectifs de la génétique des populations
est de comprendre l'effet de diverses forces évolutives, telles que la dérive,
la sélection, la mutation et la migration, sur la dynamique de fréquence des
allèles au sein d'une population. Le modèle nul de Wright-Fisher décrit
l'échantillonnage d'allèles dans une population sans sélection, sans mutation,
sans migration, sans chevauchement des temps de générations et sans
reproduction aléatoire. En réalité, les populations naturelles n'adhèrent pas à
ces prémisses, mais le modèle de dérive proposé par Wright-Fisher fournit un
outil pour étudier comment l'introduction de forces évolutives peut rendre
plus complexe un modèle relativement simple.
Ces efforts, ainsi que de nombreux autres travaux, ont contribué aux
développements théoriques de l’écologie et de l’évolution (Shou et al. 2015;
Haller 2014) qui sont récemment appliqués depuis peu aux systèmes
microbiens. Néanmoins, le développement de nouveaux modèles
mathématiques en biologie est souvent traité avec scepticisme (Zaccaria,
Dedrick, and Momeni 2017).
Ce scepticisme est en partie attribué à une incertitude sur l'utilité d'un
nouveau cadre de modélisation. Le cadre prend-il en compte les aspects
cruciaux du système biologique? Répond-il aux questions importantes
auxquelles sont confrontés les chercheurs sur le terrain ? Le modèle est-il
assez simple pour inspirer un aperçu des processus importants ? Le modèle
est-il général ou spécifique aux détails et nuances d'un phénomène biologique
particulier ? Ces questions se posent naturellement lors des études des
communautés microbiennes (Zaccaria, Dedrick, and Momeni 2017), et
reflètent les compromis intrinsèques de chaque cadre de modélisation ainsi
que discuté ci-dessous :
37
Les principes de base de la modélisation ont été présentés (Baldwin and
Donovan 1998; Curnow 1985) en classant les modèles comme suit: (1)
dynamique ou statique; (2) déterministe ou stochastique; et (3)
mécanistique ou empirique.
1.13.2 Aperçu sur les modèles d’assemblage microbiens
Cette section introduit une classification des modèles mathématiques
employés et expérimentalement testés en microbiologie et en écologie
microbienne. Il s’agit donc d’un aperçu de certains modèles qui ont bien
fonctionné expérimentalement.
1.13.2.1 Les modèles métaboliques
Le fondement d’un modèle métabolique se base une approche des réseaux
interconnectés des réactions métaboliques potentielles qui décrit les
interactions des produits finaux des gènes ou métabolites. Les données de
séquençage d’un génome complet permettent aux chercheurs de prédire le
potentiel métabolique de l'organisme (Raman and Chandra 2009). Chaque
fois que des données génomiques sont disponibles pour une nouvelle espèce
séquencée, l’annotation des voies métaboliques peut définir un réseau
métabolique complet, éventuellement affiné par l’addition de données
biochimiques de la littérature (Feist et al. 2009). La résolution du modèle peut
aller d’une voie métabolique unique (Fuhrer, Fischer, and Sauer 2005) au
métabolisme primaire entier (Almaas et al. 2004). Les données
stœchiométriques applicables aux réactions ou produits des gènes
précédemment établis sont ensuite assimilés au modèle. Les matrices des
coefficients stœchiométriques résultants rapportent les taux de flux des
réactions enzymatiques et les dérivés temporels des concentrations
métaboliques (Raman and Chandra 2009). Ce type de modèle peut ensuite
être utilisé pour l'analyse de la balance de flux (FBA) ce qui permet aux
chercheurs de corréler un génotype à son phénotype (dans une cellule
individuelle ou dans une communauté) par la dérivation des flux métaboliques
(Song et al. 2014; Hanemaaijer et al. 2015; T. Zhang 2017).
38
1.13.3 Les modèles kinétiques
Les descriptions initiales de communautés microbiennes complexes utilisaient
des approches « coarse-grained » (limitées aux intrants et aux rentrants,
sans mécanismes intermédiaires). La modélisation à gros-grain applique des
paramètres empiriques pour décrire la fonction cinétique de base de la
dynamique d’une communauté (Widder et al. 2016). En général, les modèles
cinétiques décrivent la croissance de cultures bactériennes à travers
l’utilisation des équations différentielles incorporant la concentration du
substrat limitant et les taux de croissance (ou d'absorption) correspondant à
cette concentration (Kessick 1974). Les équations de Monod et de Michaelis-
Menten sont deux équations cinétiques couramment utilisées, exprimant
respectivement la croissance cellulaire et l'absorption de substrat, basées sur
un seul substrat limitant la croissance et l'absorption catalysée par une
enzyme (Oh and Martin 2007; Hanemaaijer et al. 2015). Les équations
différentielles sont utiles pour prédire la vitesse d’une réaction enzymatique
lorsque les substrats sont abondants et que les concentrations du produit final
sont constantes. Afin de rectifier l’irréversibilité des équations de Michaelis-
Menten, Hoh et ses collaborateurs ont conçu un modèle cinétique qui prend
en compte les facteurs limitants, la croissance et la théorie thermodynamique
de l’équilibre d’une réaction (Hoh and Cord-Ruwisch 1996).
1.13.4 Les modèles spatiaux
Les modèles métaboliques et cinétiques décrivent les facteurs principaux qui
conduisent la dynamique d’une communauté microbienne (taux de
croissance, absorption du substrat et production de métabolites). En
comparaison, la dimension spatiale n’a pas été considérée attentivement, la
raison principale était de garder les modèles simples. Cependant, il existe de
nombreux systèmes microbiens pour lesquels la dispersion est essentielle
pour définir la dynamique au sein d'une communauté (Hellweger et al. 2016).
Les difficultés de modélisation de structures spatiales particulièrement les
biofilms, proviennent de la partition de la diversité microbienne en gradients
locaux, de la complexité métabolique de rétention des ressources. Elles
proviennent aussi de la synergie de coopération, ainsi que la compétition
continue pour la reproduction ou régénération dans le temps, et enfin les
39
biofilms comme des forteresses de résistance ayant des structures en
agrégats complexes (Flemming et al. 2016). Ajoutons en outre les écueils
méthodologiques liés à la paramétrisation et temps de calibrage et des calculs
des modèles
1.13.4.1 Modèles basés à l’échelle individuelle
Les premiers efforts pour développer un modèle de biofilm microbien ont été
axés sur l'équilibre de la croissance (Klapper and Dockery 2010). Ces types
de modèles étaient initialement unidimensionnels et incorporaient des
équations de réaction-diffusion pour des nutriments et d'autres composés
produits par les cellules (Klapper and Dockery 2010; Wanner and Gujer
1986). Au fil du temps, les modèles ont augmenté en dimension (2D et 3D)
et ont utilisé la modélisation basée à l’échelle individuelle (IBM) (Ferrer, Prats,
and López 2008; Volker Grimm and Railsback 2005) pour décrire de manière
plus précise le comportement hétérogène couramment observé dans un
biofilm (Kreft et al. 2001; Klapper and Dockery 2010). Bien que la croissance
reste l’objectif principal de nombreux modèles de biofilms, d'autres facteurs
tels que la détection du quorum (Klapper and Dockery 2010; Chopp et al.
2002) et la mécanique des biofilms (Körstgens et al. 2001; Klapper and
Dockery 2010) ont également été représentés.
1.13.4.2 Modèles basés à l’échelle populationnelle
Les efforts de modélisation au niveau de la population ont été résumés d’une
manière plus détaillée dans une revue récente (Hellweger et al. 2016). Ici,
nous résumons les traits saillants de modèles des populations. La
modélisation de population consiste en l’une des deux approches suivantes :
ascendante ou descendante. Dans les approches ascendantes, le niveau
inférieur est décrit afin de prédire le résultat au niveau supérieur.
Dans les approches ascendantes, le niveau inférieur est décrit afin de prédire
le résultat au niveau supérieur. Par exemple, un modèle à l’échelle
individuelle (IBM) peut caractériser un système microbien en utilisant des
interactions / caractérisations individuelles (Railsback and Grimm 2011;
DeAngelis and Mooij 2005); ces individus peuvent être des cellules uniques,
des espèces ou des groupes de microbes dans un contexte spatial et / ou
40
temporel particulier. L’information au niveau de la population apparaît comme
un sous-produit naturel de la description du model kinétique IBM (Hellweger
et al. 2016). Les IBM sont intrinsèquement plus complexes et spécifiques à
chaque cas, mais offrent des prédictions très descriptives et conviennent
mieux à la modélisation de l'hétérogénéité.
À l’inverse, les approches descendantes, telles que l’utilisation de modèles au
niveau de la population (PLM), décrivent les changements au niveau de la
population. Contrairement aux IBM, le temps et l’espace sont souvent
considérés comme continus. Les PLMs peuvent être basés sur des équations
différentielles ordinaires (ODE) ou des équations différentielles partielles
(PDE), en fonction des exigences de structure spatiale du modèle (Edelstein-
Keshet 2005; Gurney and Nisbet 1998).
1.13.4.3 Modèles basés à l’échelle de communauté
Ces modèles considèrent la communauté comme un super organisme, et ils
prédisent des réactions métaboliques en prenant en compte les interactions
entre les gènes (Khandelwal et al. 2013). Ils consistent en l’analyse spatiale
et temporelle de flux de balance métabolique, en utilisant la discrétisation
spatiale et la simulation dynamique et tout en adaptant une solution linéaire
basée sur des reconstructions métaboliques à l’échelle du génome (Henson
2015). Actuellement COMETS est le modèle le plus connu (Harcombe et al.
2014). Harcombe et ses co-auteurs en utilisant des cocultures de E. coli and
S. Enterica ont montré que le modèle dépend de la biomasse d’une souche
(S), de métabolite extracellulaire (M) et du taux d’absorption (T) de M par S.
1.13.5 Les modèles neutres en écologie microbienne
Les mécanismes de la coexistence des espèces et la maintenance de la
biodiversité dans les communautés écologiques ont longtemps été un thème
au cœur de la recherche fondamentale de l'écologie des communautés, dans
laquelle la théorie déterministe des niches et la théorie neutre stochastique
sont bien reconnues comme étant les deux plus influentes.
La théorie de niche stipule que les espèces coexistant dans une communauté
doivent avoir des niches différentes, et que les espèces avec les mêmes
41
exigences de niche ne pourraient pas coexister longtemps (Matthews and
Whittaker 2014)
Bien que la théorie de niche ait été soutenue par de nombreuses études de
terrain et de laboratoire, elle rencontre des difficultés pour expliquer les
mécanismes de coexistence d’espèces par exemple dans les forêts tropicales.
Hubbell (1997, 2001); Wills et al. (1997) ont introduit la théorie neutre de la
biodiversité qui fournissait des perspectives alternatives de la coexistence des
espèces (Stephen P. Hubbell 2001; S. P. Hubbell 1997; Wills et al. 1997). La
théorie neutre de la biodiversité de Hubbell est une théorie basée sur la
dynamique stochastique à l’échelle individuelle qui suppose des équivalences
entre les espèces en interaction, et formulée par une limite de dispersion et
par un modèle de distribution de l’échantillonnage (Etienne 2005; Alonso,
Etienne, and McKane 2006; Rosindell, Hubbell, and Etienne 2011; Rosindell
et al. 2012). Un modèle de distribution permet de tester statistiquement la
théorie neutre avec les données d’abondance des espèces obtenues en
macro-écologie (plantes et animaux) ou des données de séquençage
métagénomique (écologie microbienne).
Compte tenu des caractéristiques uniques de la distribution des données
métagénomiques de l'abondance des OTUs microbiennes, Sloan (Sloan et al.
2006; 2007) avait proposé un modèle neutre alternatif mettant l’accent sur
la neutralité au niveau de l’individu dans les communautés microbiennes.
Contrairement aux théories neutres traditionnelles qui ont été calibrées en
utilisant « une description presque complète de la distribution de l’abondance
des taxons pour la communauté », le modèle de Sloan peut se calibrer avec
des échantillons à petite taille recueillis par des approches moléculaires avec
peu d’espèces et il permet une différence de compétitivité entre les espèces
microbiennes en utilisant un modèle de distribution continue (Sloan et al.
2006; 2007) plutôt qu’un modèle discret utilisé par Hubbell.
La théorie neutre offre donc un outil quantitatif puissant pour identifier les
forces qui influencent l’assemblage des communautés microbiennes dans les
systèmes hôtes-associés. Les modèles neutres permettent de comprendre les
mécanismes qui maintiennent la diversité microbienne et l’impact des
perturbations sur l’assemblage des communautés.
Par exemple, le test de la théorie neutre peut aider à distinguer entre les
forces qui contrôlent la composition et la diversité du microbiote de l’hôte.
42
Est-ce que les facteurs déterministes de l'hôte tel que l'immunité ou plutôt
les forces stochastiques de la naissance, de la mort et de la migration seraient
à l’œuvre?
Si c’est le premier cas, cela suggère que la communauté est formée par la
partition de différentes niches, occupées par des OTUs ayant des fonctions
différentes et la diversité observée avec son hétérogénéité est déterminée
par les facteurs environnementaux déterministes qui délimitent différentes
niches.
Si c’est le deuxième cas, cela suggère que la communauté est essentiellement
composée en grande partie d’un mélange aléatoire d'espèces écologiquement
équivalentes, et que la diversité observée avec son hétérogénéité serait le
fruit des force stochastiques de naissance, de mort et de migration des
différentes espèces.
1.13.6 Le modèle neutre de Sloan
Le modèle de neutre de l’écologie microbienne (Sloan et al. 2006) dérive de
la théorie neutre de la diversité de Hubbell (Stephen P. Hubbell 2001). Le
modèle de Sloan visait à résoudre le problème d’inférer les distributions de
l’abondance des taxons des communautés microbiennes à partir des petits
échantillons métagénomiques. Il suppose qu’une communauté locale
(destination) est saturée avec un nombre total NT d’individus. Dans la
communauté locale, un individu, peut subir soit la mort, soit l’immigration
d’une communauté lointaine (source) qui se produit à un taux δ
indépendamment de l'espèce.
Un immigrant d'une communauté source, avec une probabilité m,
remplacerait immédiatement la personne décédée ou un membre localement
né avec une probabilité de remplacement 1-m.
Par conséquent, la communauté de destination est assemblée ou
réassemblée à travers un cycle continu d'immigration, de reproduction et de
mort. En outre, en supposant que les décès soient répartis uniformément
dans le temps, alors un seul décès est attendu pendant une période de temps
1/δ.
43
Pendant ce temps-là, la nième espèce, dont l’abondance absolue initiale était
de Ni, augmenterait d'un point, resterait le même ou diminuerait d'un point
avec la probabilité spécifiée par les trois expressions suivantes,
respectivement.
𝑃𝑟 (𝑁𝑖 + 1𝑁𝑖
⁄ ) = (𝑁𝑇 − 𝑁𝑖
𝑁𝑇) [ 𝑚𝑝𝑖 − (1 − 𝑚) (
𝑁𝑖
𝑁𝑇 − 1)] (1)
𝑃𝑟 (𝑁𝑖
𝑁𝑖⁄ ) = (
𝑁𝑖
𝑁𝑇) [ 𝑚𝑝𝑖 + (1 − 𝑚) (
𝑁𝑖 − 1
𝑁𝑇 − 1)]
+ (𝑁𝑇 − 𝑁𝑖
𝑁𝑇) [ 𝑚(1 − 𝑝𝑖) + (1 − 𝑚) (
𝑁𝑇 − 𝑁𝑖 − 1
𝑁𝑇)] (2)
𝑃𝑟 (𝑁𝑖 − 1𝑁𝑖
⁄ ) = (𝑁𝑖
𝑁𝑇) [ 𝑚(1 − 𝑝𝑖) + (1 − 𝑚) (
𝑁𝑇 − 𝑁𝑖
𝑁𝑇 − 1)] (3)
Soit xi la fréquence d’occurrence de la nième espèce (OTU) dans la
communauté de destination, c’est-à-dire xi =n/N, où n est le nombre
d'échantillons de la communauté locale où l’espèce i est présent et N est le
nombre total d'échantillons de la communauté locale (Burns et al. 2016a).
Soit pi est la fréquence d'occurrence de la nième espèce dans la communauté
source, c’est-à-dire la contrepartie de xi dans la communauté destination.
Sloan et al. (2006) a montré que xi devrait suivre la distribution bêta
suivante:
𝑥𝑖 ~ 𝐵𝑒𝑡𝑎[ 𝑁𝑇𝑚𝑝𝑖 , 𝑁𝑇𝑚(1 − 𝑝𝑖)] (4)
Plus spécifiquement,
φi(xi ; NT, m) = cxiNTmpi−1 (1 − xi)
NTm(1−pi)−1 , (5)
44
𝑐 =Γ(NTm)
Γ[𝑁𝑇𝑚(1 − 𝑝𝑖)]Γ(𝑁𝑇𝑚𝑝𝑖), (6)
où Ni et NT sont respectivement le nombre total d'individus d'espèce i et le
nombre total de d’individus (de toutes les espèces) dans les échantillons de
la communauté locale, m est la fréquence de migration, et φi représente la
fonction de densité de probabilité.
D'après Burns et al. (2016), le processus pour tester le modèle neutre de
Sloan et al. (2006) peut être résumé en trois étapes (W. Li et al. 2018) :
(i) Calculez pi et xi, avec pi et xi, on peut s’ajuster à la distribution
bêta (équations 2 et 3) et obtenir l’estimation de m.
(ii) Calculer la fonction φi, la fréquence d’occurrence théorique de
l’espèce i dans tous les échantillons de la communauté de
destination basés sur m et la distribution bêta.
(iii) Déterminez si le xi observé de l'espèce i se situe dans un intervalle
de confiance théorique prévu par le modèle neutre et obtenez une
liste des espèces neutres dont xi satisfait la prédiction du modèle
neutre.
Contrairement au modèle neutre de Hubbell, Il n’utilise pas un test statistique
p-valeur permettant de tester la neutralité à l’échelle de la communauté avec
le modèle de Sloan (Sloan et al. 2006; 2007), à part de discerner le
pourcentage d’espèces neutres ou non neutres (W. Li et al. 2018).
Évidemment, il n’est pas facile de définir à quel niveau arbitraire le
pourcentage d’espèces neutres peut être désigné comme majoritaire pour
que toute la communauté soit considérée neutre, comme c’est le cas pour le
modèle de Hubbell. Une autre mesure importante qui peut être utilisée pour
juger la pertinence de l’ajustement avec le modèle de Sloan est le coefficient
de détermination (R2). Nous utilisons un seuil subjectif de R2 = 0,5 pour
accepter le test du modèle Sloan.
Enfin, tester les modèles neutres est un défi en soit, avec un environnement
véritablement multisite ou compartimenté (permettant de calculer des taux
de migration variables entre différentes communautés locales) jusqu’à que
Harris et al. (2017) aient développé une méthode d’apprentissage
automatique basée sur la distribution de Dirichlet qui consiste à calculer une
45
fonction bêta multinomiale. Cependant, le temps de calcul de cette méthode
est très long, et proportionnel à la taille des données.
1.13.7 Neutralisme et déterminisme dans l’assemblage du
microbiote de l’hôte.
Les changements dans la composition et la diversité des espèces peuvent
avoir un prépondérant stochastique important (P. L. Chesson and Warner
1981). Telle que mentionnée, c’est seulement lorsque Hubbell (2001) a
importé la théorie neutre de la génétique des populations à l'écologie que la
dérive a été incorporée dans la théorie de neutralité comme un facteur étant
bien plus qu’un « bruit », et pouvant avoir un fort effet.
Les approches de modélisation mathématique évaluent la contribution
relative des processus stochastiques et déterministes dans de l’assemblage
des communautés microbiennes. L’une de ces approches consiste à appliquer
des modèles de communautés neutres (MCN) (Sloan et al. 2006). La théorie
neutre stipule que le dynamique des UTO varie au hasard, par accroissement
démographique, et des évènements de migration d’une communauté source
vers une communauté cible (Sloan et al. 2006; Burns et al. 2016a).
En revanche, la théorie des processus déterministes « non neutre » prédit
que les conditions environnementales (par exemple intra-hôte) et les
interactions interspécifiques, y compris l’exclusion compétitive, déterminent
l’abondance des espèces microbiennes (Chase and Leibold 2003). En raison
de leur grande pertinence, les MCN ont amélioré notre compréhension de
l’assemblage de communautés microbiennes et ont prédit avec succès la
dynamique de leurs structures (Woodcock et al. 2007; Jayathilake et al.
2017; Venkataraman et al. 2015; Ofiţeru et al. 2010). L’approche MCN la plus
robuste est celle de Sloan et al. (2006), utilisant la théorie de la neutralité de
Hubbell qui est calibrée et capable de reproduire des motifs sur des
échantillons de tailles multiples (Woodcock et al. 2007). Cependant, malgré
les avantages évidents de MCN, ils ne sont pas sans controverses, certains
auteurs affirmant qu'ils ne peuvent pas expliquer toute la variance réelle
entre les communautés microbiennes associées aux organismes hôtes
(McGill, Maurer, and Weiser 2006).
46
Des processus à la fois neutres et non neutres ont été mis en évidence comme
moteurs de l'assemblage de métacommunautés microbiennes dans de
nombreuses études de microbiotes de Vertébrés (Jeraldo et al. 2012). Par
exemple, des processus neutres jouent un rôle majeur dans l’ontogenèse des
communautés microbiennes lors du développement de l'hôte et au cours de
son cycle de vie dans différents modèles de Vertébrés et de Plantes (Jeraldo
et al. 2012; Maignien et al. 2014; McCafferty et al. 2013). À l'inverse, dans
le microbiote intestinal humain, le filtrage sélectif imposé par hôte, plutôt que
l'assortiment de souches, dominait les règles d'assemblage microbien,
suggérant ainsi que les processus non neutres en étaient les principaux
facteurs (Levy and Borenstein 2013; 2014).
D’autre part, des études sur l’assemblage de microbiote du poisson zèbre
(Danio rerio) avaient montré la contribution deux types de processus neutres
et non-neutres (Burns et al. 2016a). Chez trois autres espèces de poissons,
Les processus déterministes sont mis en œuvre durant l'assemblage du
microbiote intestinal au cours des premiers stades de développement et avant
de se réduire progressivement (Yan et al. 2016). Par conséquent, l'interaction
entre les processus évolutifs et leurs impacts sur l'ontogenèse du microbiote
associé à l’hôte (néanmoins chez les Vertébrés) reste ambiguë. Dans
l’ensemble, de nombreuses études sur les écosystèmes associés à l’hôte,
portant soit sur l’ontogenèse sous des conditions neutres, soit sur des
comparaisons cas-témoins, révèlent des ambiguïtés quant à l’influence
relative des processus neutres et non neutres sur l’ontogenèse du microbiote.
1.13.8 Les modèles éco-évolutifs
Les communautés microbiennes naturelles, et même celles sous conditions
expérimentales bien contrôlées sont capables de se diversifier et de s’adapter
via une dynamique combinant les processus d’écologie et d’évolution.
Les modèles théoriques de base (Volterra 1928; D. Tilman 1982) supposent
que les dynamiques écologiques et évolutives ont tendance à diminuer les
espèces co-existantes par exclusion compétitive ou par la sélection des plus
aptes. La contradiction apparente entre observations et théorie attribue un
paradoxe à la biodiversité époustouflante des communautés microbiennes
(Hutchinson 1961; U. Sommer and Worm 2002). Ceci a donc suscité un long
débat en cours sur le mécanisme sous-jacents à l'émergence et à la stabilité
47
de la diversité dans les communautés d'organismes compétitifs (Doebeli and
Ispolatov 2010; U. Sommer and Worm 2002; P. Chesson 2000)
Pour identifier les mécanismes candidats expliquant la différence entre
génération et maintien de la diversité, les modèles théoriques de base en
écologie et évolution ont été étendus par de nombreuses caractéristiques,
incluant la structure spatiale, l’hétérogénéité spatiale et temporelle, les
mesures topologiques des réseaux d’interactions, la prédéfinition des limites
spatiale des niches, les taux de mutation versus sélection, et le compromis
entre le taux de croissance et le rendement métabolique et énergétique.
Cependant, on ignore encore quels paramètres sont essentiels pour expliquer
la biodiversité microbienne, car cette diversité est observée également dans
des conditions stables et homogènes.
Jusqu'à présent, les modèles de dynamique éco-évolutive développés
rentrent dans trois grandes catégories (Farahpour et al. 2018):
1- Modèles dans l'espace génotypique, comme la génétique des
populations (Ewens 2004) et les modèles quasi-espèces (Nowak 2006).
2- Modèles dans l’espace phénotypique, comme modèles de la dynamique
adaptative (Doebeli 2011) et de web world (Drossel, Higgs, and
Mckane 2001).
3- Modèles dans l'espace d'interaction, comme les modèles Lotka-Volterra
(Coyte, Schluter, and Foster 2015; Ginzburg, Akçakaya, and Kim 1988)
et des réseaux évolutifs (Coyte, Schluter, and Foster 2015; Mathiesen
et al. 2011; Allesina and Levine 2011).
Chacune de ces catégories de modèles possède des avantages et des limites
et met l’accent sur un aspect particulier. Cependant, dans la nature, ces
aspects sont bruités par des réactions éco-évolutives mettant en lien le
génotype, le phénotype et le niveau d'interaction
Dans un système biologique en évolution, les mutations peuvent provoquer
des variations phénotypiques si elles sont associées à de nouveaux traits
phénotypiques dans l'espace phénotypique (Soyer 2012). Ces variations ont
un impact écologique uniquement si elles affectent les interactions biotiques
ou abiotiques des espèces; sinon, elles sont neutres sur le plan écologique.
Les variations adaptatives du réseau d'interaction qui en résultent modifient
la composition des espèces par le biais de la dynamique des populations.
48
Enfin, la sélection fréquence-dépendante parfois favorise les espèces ayant
des stratégies adaptatives à leur nouvel environnement.
Nous avons donc un lien entre les réseaux d’interactions et les dynamiques
éco-évolutives, et donc nous n’avons pas besoin de suivre tous les
changements évolutifs au niveau génétique ou phénotypique si nous nous
intéressons à la dynamique macro-éco-évolutive, et aux interactions
affectées par cette dernière (Schéma 1.3). L'évolution peut être considérée
comme une exploration temporelle de l'espace d'interaction, et la
modélisation à ce niveau peut nous aider à étudier comment les réseaux
d'interactions compétitives complexes évoluent et influencent la diversité
(Farahpour et al. 2018).
Schéma 1. 3. Espace des modèles éco-évolutifs.
1.14 Récapitulatif
Dans ce chapitre, nous avons discuté en premier les principes et les concepts
de base en écologie microbienne. Le progrès des nouvelles générations du
séquençage, qui en est à sa deuxième décennie, ne peut que promettre de
gagner encore plus de précision dans la caractérisation de la biodiversité
microbienne via les approches de métagénomique. L’approche des amplicons
basée sur un gène marqueur universel comme le 16S (chez les Bactéries et
Archées) ou le 18S (chez les microbes eucaryotes) permet quantitativement
et qualitativement de caractériser la diversité microbienne dans un
échantillon donné d’un environnement donné. L’approche métagénomique
globale basée sur le séquençage de l’ADN total permet de détecter le contenu
génétique et les fonctions moléculaires d’une communauté donnée. Les deux
approches permettent de détecter et d’énumérer les unités de base de la
49
biodiversité grâce au concept de l’unité taxonomique opérationnelle (UTO)
connu sous le terme anglophone OTU « Operational taxonomic Unit ».
Ensuite, les mesures de la diversité Alpha permettent de caractériser et
d’estimer la richesse dans un échantillon donné. Tandis que les distances
écologiques et phylogénétiques de la diversité Beta permettent de comparer
la divergence entre les communautés basées sur les différentielles de
l’abondance et les distances génétiques des UTO. En deuxième partie de ce
chapitre nous avons discuté les types des interactions hôte-symbionte et la
notion de l’hologénome inspirée du concept de l’unité
évolutive « l’holobionte ». En outre, nous avons discuté le concept des forces
évolutives commun en évolution moléculaire et génétique des populations
emprunté récemment à l’écologie microbienne. Il permet de distinguer entre
les forces diversifiantes (migration, mutation) de celles qui sont réductives
de la biodiversité génétique microbienne (sélection et dérive). D’ailleurs, le
rappel de notions de base en évolution microbienne permet d’assurer une
bonne compréhension de la nature déterministe versus neutre de
l’assemblage des communautés microbiennes planctoniques, en biofilms, ou
associées à l’hôte.
En résumé, toute étude en écologie microbienne hormis le contexte
écotoxicologique, requiert les nouvelles technologies de pointe en biologie
moderne afin d’accéder aux ressources génétiques de la biodiversité
microbienne. Les méthodes de bio-informatiques et biostatistiques comme
celles des réseaux et des modèles d’assemblage permettent efficacement de
caractériser et de quantifier la dynamique des répertoires génétiques au sein
des communautés microbiennes.
50
Chapitre 2: Taxon-function decoupling as an
adaptive signature of lake microbial
metacommunities under a chronic polymetallic
pollution gradient
Bachar Cheaib, Malo Le Boulch, Pierre-Luc Mercier, et Nicolas Derome
Front Microbiol. 2018; 9:869
Front. Microbiol., 03 May 2018 https://doi.org/10.3389/fmicb.2018.00869
51
Taxon-function decoupling as an adaptive signature of lake microbial
metacommunities under a chronic polymetallic pollution gradient
Bachar Cheaib, Malo Le Boulch, Pierre-Luc Mercier, et Nicolas Derome
2.1 Résumé
L'exposition des micro-organismes aux facteurs anthropogéniques de stress
peut entraîner une dégradation de la diversité microbienne. Une telle
adaptation peut laisser des empreintes moléculaires dans leurs répertoires
taxonomique et fonctionnel. Ici, nous avons évalué la diversité taxonomique
et fonctionnelle d'un système de méta-communautés microbiennes lacustres
tout au long d'un gradient de pollution polymétallique subi par un drainage
d’acide minier (DAM) durant plus que 60 ans d'exposition chronique. Par une
approche de séquençage métagénomique, nos résultats mettent en évidence
trois types de signatures adaptatives. Premièrement, une signature du
découplage taxon-fonction a été détectée dans les communautés
microbiennes de lacs à contamination forte et intermédiaire. Deuxièmement,
des changements parallèles dans la composition taxonomique ont été
détectés dans de lacs pollués versus non pollués. Troisièmement, les
variations dans l’abondance des modules fonctionnels furent suggérées pour
expliquer une détérioration progressive des services écosystémiques (par
exemple, la photosynthèse) et une érosion graduelle du métabolisme
secondaire dans les lacs pollués. Dans l'ensemble, les changements dans
l'abondance de taxons et fonctions et, ainsi que les gènes de résistance
polymétalliques tels que copA, copB, czcA, cadR, cusA, étaient corrélés avec
la teneur en métaux traces (principalement le cadmium) et l'acidité
52
2.2 Abstract
Adaptation of microbial communities to anthropogenic stressors can lead to
reductions in microbial diversity and disequilibrium of ecosystem services.
Such adaptation can change the molecular signatures of communities with
differences in taxonomic and functional composition. Understanding the
relationship between taxonomic and functional variation remains a critical
issue in microbial ecology. Here, we assessed the taxonomic and functional
diversity of a lake metacommunity system along a polymetallic pollution
gradient caused by 60 years of chronic exposure to acid mine drainage (AMD).
Our results highlight three adaptive signatures. First, a signature of taxon -
function decoupling was detected in the microbial communities of moderately
and highly polluted lakes. Second, parallel shifts in taxonomic composition
occurred between polluted and unpolluted lakes. Third, variation in the
abundance of functional modules suggested a gradual deterioration of
ecosystem services (i.e. photosynthesis) and secondary metabolism in highly
polluted lakes. Overall, changes in the abundance of taxa, function, and more
importantly the polymetallic resistance genes such as copA, copB, czcA,
cadR,cusA, were correlated with trace metal content (mainly Cadmium) and
acidity. Our findings highlight the impact of polymetallic pollution gradient at
the lowest trophic levels.
53
2.3 Introduction
Micro-organisms represent a significant portion of global biodiversity and are
the engine driving Earth’s biogeochemical cycles and primary production
(Falkowski, Fenchel, and Delong 2008; Green, Bohannan, and Whitaker
2008). Ecosystem services provided by microbes ensure optimal
environmental conditions for all multicellular life forms (Robinson, Bohannan,
and Young 2010). For decades, the implications of taxon-function
relationships in microbial communities have been debated by researchers
(Doolittle and Zhaxybayeva 2009; Bissett et al. 2013; A. C. Martiny,
Treseder, and Pusch 2013; Louca, Parfrey, and Doebeli 2016; Morrissey et
al. 2016). On one hand, researchers showed that even very closely related
taxa exhibited contrasting metabolic and ecological functions (e.g. distinct
growth rates and metabolic substrate utilization profiles), indicating a gap
between taxon phylogeny and the functional repertoires of some bacterial
genera (Jaspers and Overmann 2004; Maharjan et al. 2006; Doolittle and
Zhaxybayeva 2009). These studies employed molecular taxonomic profiling,
either by sequencing SSU (small subunit ribosomal ribonucleic acid) 16S rRNA
(Doolittle and Zhaxybayeva 2009; Jaspers and Overmann 2004) or specific
housekeeping genes (Maharjan et al. 2006). On the other hand, studies
focused on microbial molecular evolution and ecology reported a significant
relationship between phylogenetic groups or taxonomic composition at
different hierarchical levels (i.e. Phylum and Class) with ecological and
functional traits (Webb et al. 2002; Ward et al. 2006; J. B. H. Martiny et al.
2006; Gupta and Lorenzini 2007; Allison and Martiny 2008; Philippot et al.
2010; Gravel et al. 2011; A. C. Martiny, Treseder, and Pusch 2013). The
majority of these genomic studies have been limited to correlating traits with
taxa abundance variation. Additional evidence at the community level is
needed to predict the interplay of evolutionary processes (horizontal gene
transfer, gene loss, selective pressure) and ecological processes (spatial
dispersal limits, biotic interactions, neutral biogeography) drive
metacommunity composition and functional repertoires in complex ecological
contexts.
With advances in sequencing technologies, metagenomic approaches have
the potential to advance our understanding of both the taxonomic and
54
functional composition of complex microbial communities. In this respect,
metagenomic studies have revealed significant coupling between taxonomic
composition or phylogenetic lineages and ecological traits (Bouvier and del
Giorgio, 2002; Philippot et al., 2010) or functional gene repertoires (Debroas
et al. 2009; Goldfarb et al. 2011; Muegge et al. 2011; Fierer, Leff, et al.
2012; Bryant et al. 2012; Langille et al. 2013; A. C. Martiny, Treseder, and
Pusch 2013; Mayali et al. 2014; Forsberg et al. 2014; Vanwonterghem et al.
2014; Morrissey et al. 2016; Larkin and Martiny 2017). For example, in
natural lake communities, associations are reported between taxon
abundance and function (Debroas et al. 2009), and in soil communities from
multiple environments, with chemical substrate variation (Goldfarb et al.
2011) and functional attributes (Fierer, Leff, et al. 2012). Most of these
studies have been conducted in relatively unperturbed environments, and on
microbial communities facing moderate to low selective pressure.
Other microbial community studies, mostly based on 16S rRNA gene analysis,
and rarely complemented by whole metagenome shotgun sequencing,
revealed either partial or marked decoupling between taxonomic composition
and ecological traits (Lima-Mendez et al. 2015) or functional gene repertoires.
Patterns of complete to partial decoupling are often found in natural
environmental conditions (Hooper et al. 2008; Hooper, Mavromatis, and
Kyrpides 2009; Burke et al. 2011; Raes et al. 2011; Smillie et al. 2011;
Barberán et al. 2012; Louca, Parfrey, and Doebeli 2016). This taxon-function
decoupling has rarely been discussed in extreme environments such as acid
mine drainages (AMD) (J. Kuang et al. 2016). These findings highlight the
need to further investigate environments where initial conditions have been
perturbed by xenobiotic factors(Bowen et al. 2011).
The occurrence of taxon-function decoupling has been reported in other
metagenomic studies as functional redundancy between phylogenetically
distant taxa (Burke et al. 2011; Green, Bohannan, and Whitaker 2008; Stokes
and Gillings 2011) and divergent microenvironments (Hooper et al. 2008;
Hooper, Mavromatis, and Kyrpides 2009). To summarize, taxonomic and
functional features could be useful in assessing adaptive response of microbial
metacommunities in disturbed ecosystems. One study, to our knowledge, has
focused on the outcome of microbial taxon-function relationships under
55
selection gradients, indicating possible linkages between the structure and
functioning of soil microbial communities (Fierer, Lauber, et al. 2012). Thus,
it remains uncertain whether taxon-function decoupling is an adaptive
response to a gradual selective pressure. Xenobiotic stressors like antibiotics,
chemical and metallic pollutants erode microbial biodiversity (Parnell et al.
2009), which is predicted to impair or erode ecosystem services (Sandifer
and Sutton-Grier 2014). Therefore, the characterization of taxon-function
decoupling patterns will enhance our understanding of the robustness of
microbial functional networks that ensure key ecosystem services. Here, the
complex connections of microbial biodiversity and ecosystem services (Miki,
Yokokawa, and Matsui 2014) were addressed at the molecular level by
comparing variation in the taxonomic composition and molecular functions of
microbial communities.
We hypothesized that a stress gradient, specifically a polymetallic pollution
gradient over a relatively long evolutionary time scale in terms of bacterial
generation time, would result in adaptive signatures in taxonomic
composition and functional repertoires. Specifically, we predicted that stress
gradients would gradually induce selection for microbial metacommunities
with functional repertoires and a taxonomic composition capable of living in
this harsh environment. To test our hypothesis, we targeted lakes polluted
by a polymetallic gradient of acid mine waters. Heavy metals can originate
either from natural sources such as volcanic activity or anthropogenically by
mines tailings, an important source of AMD. Acidity gradients recorded in lake
waters surrounded by natural volcanic activity (e.g. Indonesian crater lake
Kawah Ijen, Argentinian volcanic lake in Patagonia), have significant effects
on the microbial community composition and biodiversity (Wendt-Potthoff
and Koschorreck 2002; Löhr et al. 2006). AMD is created by the exposure of
sulphidic minerals to air and water forming soluble sulphates (Almeida et al.
2008). Ferrous minerals become oxidized in contact with water producing
ferric ions and H2 (D. B. Johnson and Hallberg 2003; Edwards and Bazylinski
2008). Leached ions into streams generate acidic water by lowering the pH
(< 3). Consequently, other metal ions such as Zn, Hg, Ni, Cr, Cd, Cu, Mn, Al,
As and Pb appear in AMD waters at high concentrations. There are limited
descriptions of microbial diversity in AMD in the literature, especially in
56
impacted environments with high zinc and cadmium concentrations (Almeida
et al. 2008). In AMD polluted surface water, Almeida et al. (2008) showed
that bacterial diversity in Sepetiba Bay, Brazil, which is much higher than
archaeal diversity, was dominated by Proteobacteria, Actinobacteria,
Cyanobacteria and had a high abundance of unclassified bacteria (unknown
strains). Similar composition (dominance of Proteobacteria) was observed
over 59 microbial communities from physically and geochemically diverse
AMD sites across Southeast China (J.-L. Kuang et al. 2013). Kuang and al.
(2013) revealed that pH gradient is a major factor explaining community
differences between AMD communities regardless of the long-distance
isolation and the distinct substrate types. Likewise, the investigation of the
microbial diversity of an extremely acidic, metal-rich water lake (Lake Robule,
Bor, Serbia) revealed low diversity dominated by Proteobacteria strains
(Stankovic et al. 2014). Similar community composition was observed in
bacterioplankton communities exposed to cadmium in coastal water
microcosms (K. Wang et al. 2015). Similar to surface waters, Hemme et al., (2010,
2016) highlighted that chronic exposure to high concentrations of heavy
metals (~ 50 years) in groundwater caused a massive decrease in
biodiversity, characterized by a high abundance of Proteobacteria, as well as
a significant loss in allelic and metabolic diversity. More importantly, Hemme
et al., (2016) pointed to the importance of Horizontal Gene Transfer (HGT)
during the evolution of groundwater microbial communities in response to
heavy metal exposure. However, very few studies were carried out on water
polluted across a polymetallic gradient (J.-L. Kuang et al. 2013; Desoeuvre,
Casiot, and Héry 2016). One of those studies reported the impact of an
extreme poly-metallic gradient (including arsenic) on the diversity and
distribution of arsenic-related genes in river waters (Desoeuvre, Casiot, and
Héry 2016). Other studies on AMD polluted freshwater sediments (Sánchez-
Andrea et al. 2011; Jackson et al. 2015; Ni et al. 2016; Jie et al. 2016)
showed the dominance of Proteobacteria in microbial communities as well as
community specialization. In lake sediments exposed to AMD gradients, the
effects of different metals on specific microbes and microbial activities were
correlated with their respective chemical properties. All these studies used
16S rRNA gene analysis, except for one, which used deep coverage data from
shotgun metagenome sequencing (Hemme et al. 2016).
57
In our study, we used a shotgun metagenomic sequencing approach to
characterize the functional and taxonomic diversity of bacterioplankton from
five lakes within a catchment that was historically exposed to a polymetallic
contamination gradient (PCG) for over sixty years. As the PCG was previously
correlated with taxon abundance variation (Laplante, Boutin, and Derome
2013; Laplante and Derome 2011), taxon-function decoupling was expected
to occur in the most polluted lakes and be absent in less polluted or unpolluted
lakes. Our first objective was to assess the taxonomic and functional
signatures of bacterioplankton adaptation to PCG. Secondly, we aimed to
provide insight into the interplay of biodiversity and ecosystem services under
a stress gradient by analyzing taxon-function variation.
2.4 MATERIALS AND METHODS
2.4.1 Lake characteristics and locations
Over the last sixty years, the Rouyn-Noranda (Western Quebec, Canada)
mining sites have dumped acid mine drainage (AMD) with heavy polymetallic
traces (Laplante and Derome 2011) into surrounding lakes. We targeted five
lakes in this area (Supplementary Figure S2.1). Among them, three have
common surface water interconnected along the same hydrologic basin:
Arnoux Lake (LAR-hc; highly polluted), Arnoux Bay (BAR-mc; medium levels
of pollution), and Dasserat Lake (DAS-lc; the least polluted). The water
polluted by AMD spreads from Arnoux to Dasserat Lake generating a
polymetallic gradient over 20 km. Around 30 km to the south side of this
natural system of connected lakes, Opasatica Lake (OPA-nc), which is a
landlocked unpolluted site, was sampled and considered as an unpolluted
negative control, and ca. 40 km to the north side, Turcotte Lake (TUR-hc),
another landlocked site was selected as a highly polluted lake. Longitude and
latitude coordinates are given in Supplementary file 1. The abandoned mine
site is a source of tailings and eroded mine waste into the Arnoux River, which
drains west to Arnoux Lake, Arnoux Bay, and then Dasserat Lake. These lakes
are irregular in shape and the bathymetry reflects the relief of the underlying
bedrock. The immediate surrounding area consists of hilly terrain, volcanic
rocks, ultramafic rocks, mafic intrusions, granitic rocks, and early and middle
Precambrian sediments (Alpay 2016).
58
2.4.2 Metallic and chemical gradient surveys
pH and Polymetallic concentration (Al, Cd, Cu, Fe, Mn, Pb, Zn) in the studied
lakes was measured in June 2010, a year prior to the present study, using
ICP VISTA Varian-axial mass spectrometer as described in Laplante and
Derome (2011). Trace metal profiles showed a polymetallic gradient in the
three interconnected lakes (Supplementary Figure S2.1). For each lake, we
measured temperature (OPA-nc: 12°C; DAS-lc: 10°C; BAR-mc: 9.9°C; LAR-
hc: 11.5°C; TUR-hc: 9.5°C). Dissolved organic carbon (DOC) were
determined in each sample using a total organic carbon (TOC) analyzer
(Shimadzu) following the non-purgeable organic carbon (NPOC) method
(Laplante and Derome 2011).
2.4.3 Water sampling
Sampling was carried out in September 2011 by collecting 6 L of water per
lake at a depth of 60 cm below the surface. Water samples were sequentially
filtered (3 filters per sample), first through a 47-mm polycarbonate filter with
3-micron pore size, followed by a 0.22 μm nitrocellulose membrane filter
(Advantec) using peristaltic filtration (Masterflex L/S Pump System with Easy-
Load II Pump Head; Cole-Parmer, Vernon Hills, IL, USA). Duplicates of the
0.22 μm filter were placed into cryotubes at -80 °C.
2.4.4 DNA extraction and metagenome sequencing
Filter duplicates were pooled, then genomic DNA was extracted as described
by (Laplante, Boutin, and Derome 2013). Library preparation (TruSeq DNA
Illumina) of paired-end reads (2 × 100 bp read length) was performed by the
McGill University/Genome Quebec Innovation Center for whole metagenomic
shotgun sequencing using a HiSeq™ 2000 Sequencing System. A total of 30
Gbps were obtained and the sequencing data summary is shown in
Supplementary file 2. The sequence files are available from the Sequence
Read Archive (http://www.ncbi.nlm.nih.gov/sra), BioProject ID:
PRJNA449990.
2.4.5 Bioinformatic and statistical analysis
Reads-based approach (Figure 2.1). To first discard methodological biases
including sequencing artifacts, we pre-processed data for quality filtering,
59
chimeric sequences, homopolymers and short reads (cutoff: 50 bp) using the
Nesoni Clip tool (https://github.com/Victorian-Bioinformatics-Consortium)
version 0.133. Overall, the quality of forward reads (R1) was better than
reverse reads (R2). This difference is related to sequencing quality decrease
over the length of reads, in addition to the loss of enzymatic specificity
overtime in the paired-end platform technology. Base calling quality was
selected at a Phred or Q score of 33 (Supplementary file 2). FLASH software
v1.2.11 (Magoč and Salzberg 2011) was used with default parameters (10
to 65 bp overlapping window) to merge paired-end reads.
As a second step, following the selection of good quality reads for all
five metagenomic samples, a sequence similarity search was performed
against the SEED database (Version: May 2015) (Overbeek et al. 2014) using
Diamond v0.7.9.58 (Buchfink, Xie, and Huson 2015). The taxonomic content
of each sample was assigned using the Lowest Common Ancestor (LCA)
method (Huson et al. 2007). Functional abundance was estimated using the
SUPERFOCUS software (Silva et al. 2015) with Diamond (1e-12 as p-value, 60
identity as threshold, 30 base pairs as minimum alignment length). To cope
with missing biological replicates and unequal read numbers across all five
lake samples (varying from 37 to 80 million paired-end reads before
filtration), a read subsampling approach without replacement was used
instead of rarefying or simulating reads from complete genomes. Accordingly,
each metagenomic sample was subsampled 12 times with an equal number
of reads (1 million reads). The uniformity in terms of number of subsampled
reads from all samples were largely respected as in previous studies using
simulated metagenomes (Garcia-Etxebarria, Garcia-Garcerà, and Calafell
2014; Mavromatis et al. 2007). The 60 generated metagenomic pseudo-
replicates of equal size were submitted to our custom pipeline of taxonomic
and functional abundance annotations.
Thirdly, to measure alpha and beta diversity based on feature
abundances, we employed the OTU concept of taxonomic units (Schloss et al.
2009). Considering each feature (genus, function) as an OTU, alpha and beta
diversity were computed using Mothur software (Schloss et al. 2009). UniFrac
distances based on shared and unshared features were computed for each
compared pair of samples. To inspect how environmental factors impact
60
metacommunity composition, subsampled sites were first plotted based on
their feature abundances with non-metric multidimensional scaling (NMDS)
using Bray-Curtis distance between samples. Hence, the OPA-nc sample was
used as control reference to compute the differential abundance of genera.
Then, the computed distance matrix was clustered with Ward’s method based
on minimum variance. Clusters of genus abundance were distinguished with
different colors on the NMDS plot. Next, mixed metals metadata were
projected on NMDS axes by fitting a regression model. The significance of the
“regression coefficient” of the model was computed using a random
permutation test (1000 iterations). Then the regression coefficient between
the randomized response and the fitted values from the model was computed.
The NMDS model was run using the VEGAN package (Oksanen et al. 2016) in
the R statistical environment (R Foundation for Statistical Computing 2008).
To test for a correlation between taxonomical and functional composition, a
Canonical Correlation Analysis using the CCA (González et al. 2008) and
mixOmics (Rohart et al. 2017) packages in R were applied. With CCA, the
function-taxon cross-correlation was computed by maximizing the linear
combinations between the two matrix vectors. Then a regularization function
of CCA from mixOmics was used to deal with the high number of features
(genus, function) compared to the low number of samples (60 subsamples).
Regularization parameters (lambda 1 and lambda 2) were determined
through a standard cross-validation (CV) procedure on a two-dimensional
surface. The optimal value for lambda was obtained by searching for the
largest CV-score on the 2D surface that requires intensive computing time to
converge for the optimal cross-validation value. Choice of canonical
dimensions and graphical representation of features and samples were
performed with mixOmics package.
ORFs-based approach (Figure1). To improve annotation accuracy in terms of
length and coverage, an Open Reading Frame (ORF) prediction approach was
used after de novo assembly. Collinear metagenomic reads belong to the
same genetic unit were merged into same contigs. Firstly, de novo assemblies
of raw reads were performed using the RAY Meta (Boisvert et al. 2012)
assembler. Secondly, to explore contig features and gene contents, contigs
were submitted to the MG-RAST webserver (Glass et al. 2010) and ORF
prediction was conducted using the FragGeneScan tool (Rho, Tang, and Ye
61
2010). Afterwards, contigs were annotated with the BLAT tool implemented
in MG-RAST against the SEED database using stringent filtering parameters
(1e-12 as p-value, 85% identity as threshold, 50 base pairs as minimum
alignment length). Statistical summaries of annotated contigs are available
in Supplementary file 2. Customized microbial annotations from the MG-RAST
webserver were improved using the RESTful API tool (Wilke et al. 2015). An
additional similarity research step based on BLASTx (parameters; identity
threshold of 85%, e-value of 10-12 and minimum alignment length of 50 base
pairs) (Camacho et al. 2009) was performed on contigs against the BacMet
database (Pal et al. 2014) for annotating all polymetallic resistance genes
(hereafter termed PMRGs). After annotations, contig coverage information
determined by the Ray Meta assembler was added to normalize abundance
information. Then, both abundance matrices of taxon and function coverage
(both normalized and non-normalized) were analyzed with the STAMP
software using a differential proportion comparisons test (Parks and Beiko
2010). In a second additional workflow analysis, the ORFs were locally
annotated with Diamond as described above in the “Reads-based approach”
section. BLAT and Diamond provided similar annotation results. To measure
alpha and beta diversity within and between communities, abundance
matrices were adapted for the Mothur software. At the third step, metabolic
abundance was analyzed using MG-RAST metabolite annotations. The
metabolic differential abundance was surveyed using iPATH (Yamada et al.
2011); this tool offers the visualization of shared and specific pathways
between pairs of samples.
2.5 Results
2.5.1 Decoupling taxon-function
To investigate the impact of the polymetallic selection gradient on lake
metacommunity composition, we measured the pattern of decoupling
between taxon and function along the contamination gradient of the five
lakes. We hypothesized that taxon-function decoupling pattern is an adaptive
response of lake metacommunities. To detect this pattern, we performed two
independent analyses: i) taxonomic structure versus functional diversity and
ii) canonical correlation of taxon and function.
62
2.5.2 Detangled taxonomic structure and function diversity
According to alpha-diversity analysis, the highest value of community
richness (chao index) at the genus level was recorded in BAR-mc (OPA-nc:
126.8, DAS-lc: 115.07, BAR-mc: 212.6, LAR-hc: 121.66, TUR-hc: 68). In
contrast to richness, community evenness (Shannon index) was lowest in
TUR-hc (0.116), intermediate in OPA-nc (2.24), gradually decreasing along
the metallic gradient from DAS-lc (2.85) < BAR-mc (2.58) < LAR-hc (2.47).
However, community evenness of functions (OPA-nc: 2.36, DAS-lc: 2.32,
BAR-mc: 2.92, LAR-hc: 2.62, TUR-hc: 2.87) was higher in BAR-mc, LAR-hc
and TUR-hc then OPA-nc and DAS-lc. Then, beta-diversity analysis at the
genus level (Figure 2.2f) revealed two patterns of structural convergence: (i)
between the two independent lake communities, namely the unpolluted
control (OPA-nc) and the low polluted lake (DAS-lc); (ii) between the
interconnected BAR-mc-LAR-hc and the polluted control TUR-hc
communities. Concerning functional diversity distribution, beta-diversity of all
subsystems (Figure 2.3b) revealed two convergent patterns: (i) between the
polluted control (TUR-hc), the highly-polluted gradient lake (LAR-hc), and the
medium-polluted lake (BAR-mc); (ii) between the independent lake
communities, namely the low-polluted DAS-lc and the negative control (OPA-
nc) communities.
2.5.3 Canonical correlations of taxon and function
Regularized canonical correlation analysis (rCCA) of function (subsystems
level 1, 2, 3) with taxon was assessed using a maximal cross-validation
criterion (see Materials and Methods). To detect linear combinations between
function and taxon we separately performed the same rCCA analysis for OPA-
nc / DAS-lc, and then for BAR-mc / LAR-hc / TUR-hc. For OPA-nc / DAS-lc
(Figure 2.6 a-d), we found a maximum variance of only 1% explained by the
first axis computed from the taxon covariance matrix, and 1% explained by
the first canonical correlation principal component computed from the
function subsystems level 1 (results not shown) covariance matrix, even
when we tested the canonical model at the most accurate hierarchical
functional resolution (subsystems level 3). Supported by a high cross-
validation score (0.975), this result suggested a strong coupling between
taxon and function. Conversely, in BAR-mc / LAR-hc / TUR-hc (Figure 2.5 a-
63
d) we found the first axis explained 25% of the variance computed from the
taxon covariance matrix, and 3% (Subsystem level 1) to 6% (subsystems
level 3) explained by the first canonical axis computed from the functional
covariance matrix. This result, supported with a high cross-validation score
(0.99), revealed a weak correlation between taxon and function, thus
suggesting a strong taxon-function decoupling. Using the first two canonical
axes, in BAR-mc, LAR-hc and TUR-hc, a clear separation was observed
between taxon and function (Figure 2.5 c), while the axes are superimposed
in OPA-nc and DAS-lc (Figure 2.6 c).
2.5.4 Taxonomic variation signatures
The metacommunity composition analysis emphasized three major patterns
marked by abundance shifts within and between Proteobacteria,
Cyanobacteria and Actinobacteria phylum (Figure 2.2a). In the first pattern,
Proteobacteria mostly dominated by Betaproteobacteria (Supplementary file
3) reached a higher relative abundance in the highly-polluted (hc) lakes TUR-
hc (99%) and LAR-hc (35%) compared to the moderately-polluted (mc) lake
BAR-mc (20%), the least-polluted (lc) lake DAS-lc (19%), and the unpolluted
(nc) lake OPA-nc (27%). At the genus level, Polynucleobacter, unclassified
Burkholderia and Burkholderia were the most dominant within polluted lakes
TUR-hc, LAR-hc, and BAR-mc, respectively, while Polaromonas was the most
dominant in DAS-lc and OPA-nc (Supplementary file 3). In the second
pattern, Actinobacteria were the most dominant phylum (Supplementary file
3) in less polluted lakes [OPA-nc (53%) and DAS-lc (62%)] and their relative
abundance gradually decreased in more polluted lakes [BAR-mc (33%), LAR-
hc (10%) and completely disappeared in TUR-hc], mainly for the five most
abundant genera: Streptomyces, Frankia, Mycobacterium, Kribbella, and
Nocardioides (Supplementary file 3). In the third pattern, Cyanobacteria
(Supplementary file 3) were abundant in OPA-nc (15.4%) and BAR-mc
(42%), and much less frequent in LAR-hc (4.4%), DAS-lc (0.2 %) and TUR-
hc (< 0.01 %). At genus level, Synechococcus was most dominant,
accounting for 98% and 92% of Cyanobacteria genera in OPA-nc and DAS-lc,
respectively. In contrast, distinct Cyanobacteria genera were dominant in
polluted lakes: the filamentous Anabaena in BAR-mc, unclassified
64
Cyanobacteria in LAR-hc, and both the diazotrophic Cyanothece, and the
filamentous Anabaena in TUR-hc.
Lake metacommunity abundance shifts were further documented using
bootstrapped hierarchical classification and PCA. At the genus level, both
methods showed similar pattern of clustering with high statistical support
(bootstrap values above 75%; more than 95% of explained variation by the
first two PCA components), with BAR-mc, LAR-hc, and TUR-hc grouped
separately from OPA-nc, DAS-lc (Figure 2.2b, c, d).
2.5.5 Role of trace metals in taxonomic variation signatures
NMDS analysis based on ORFs (Figure 2.2e) revealed interesting relationships
(significant R-squared indicating regression model’s goodness of fit) between
taxonomic abundance and different factors such as pH, DOC and trace metals
(mainly Cadmium). OPA-nc and DAS-lc were significantly correlated with DOC
and pH axes, while all other sites exposed to polymetallic gradient (BAR-mc,
LAR-hc, and TUR-hc) were significantly correlated with trace metals axes
(Figure 2.2e). To further analyze the link between abundance differences at
different taxonomic ranks and the trace metal gradient, the same NMDS
analysis was performed using the ORFs approach. The abundance of
Proteobacteria (Supplementary Figure S2.3a), Actinobacteria
(Supplementary Figure S2.3b) and Cyanobacteria (Supplementary Figure
S2.3c) were studied separately. NMDS analyses of abundance shifts at the
genus level revealed significant correlations with different metal axes, pH and
DOC. The shifts in composition within lake metacommunities were not
explained by the same factors. For example, variation in the abundance of
Proteobacteria among lakes was mainly explained by Cd, pH, Mn, Alu, while
Cd and Fe explained variation in the abundance of Actinobacteria, and Alu
and Mn were the main factors explaining variation in the abundance of
Cyanobacteria among lake metacommunities.
2.5.6 Function variation signatures
Our results showed 6801 annotated functions from all communities
distributed into 988 subsystems in level 3, 192 subsystems in level 2, and 28
subsystems in level 1 (see sheet 2 in Supplementary Figure S2.5). At the first
level (see Supplementary Figure S2.5 and file 5), our results of cross-
65
metagenomes comparison suggested that the relative abundance of
“Photosynthesis”, “Cofactors, Vitamins, Prosthetic Groups, Pigments” and
“Respiration” subsystems was significantly highest in OPA-nc while the
“Stress response” was the lowest in this lake. However, Subsystems of “RNA
metabolism”, and mobile elements (Phages, prophages, plasmids and
transposable elements) showed the highest abundance in BAR-mc, followed
by LAR-hc and TUR-hc, and low abundance in OPA-nc and DAS-lc.
Furthermore, the relative abundance of the “carbohydrates” subsystem
decreased gradually in all lakes except from DAS-lc to OPA-nc
(Supplementary file 5). Interestingly, among the 28 subsystems (Level 1),
four subsystems “Nitrogen metabolism”, “Cell cycle and division”, “Sulfur
metabolism”, “and “Motility and Chemotaxis” decreased gradually along the
contamination gradient. In addition, three subsystems (Phosphorus and
Potassium metabolism, Membrane transport) were absent in LAR-hc and
showed specific profiles of low abundance (Supplementary file 5) varying
between 0.2% to 3.8% in BAR-mc and TUR-hc. For multiple subsystems in
Level 1 (n= 12), no gradual abundance variation was observed. However, at
a deeper resolution, many important functions related to metals transport
and resistance from the “Virulence defense and disease”, “Membrane
transport” and “Iron acquisition and metabolism” subsystems showed few
gradual (i.e. Cobalt-Zinc-Cadmium resistance) abundance profiles and high
specific abundance per lake (Supplementary Figure S2.7). At the functional
level, variation abundance was detectable within all subsystems where three
profiles of abundance variation were observed from OPA-nc to TUR-hc: i)
profile 1 (FP1) represents gradual function abundance decrease (106
functions) along the contamination gradient (Supplementary file 5 and Figure
S2.10), ii) profile 2 (FP2) represents gradual function abundance increase
(123 functions) along the contamination gradient Supplementary file 5 and
Figure S2.10, and iii) profile 3 (FP3) represents specific functional abundance
(Supplementary file 5 and Figure S2.11) in control negative OPA-lc (167
functions), or in polluted lakes (225 functions). These functional profiles were
not necessarily observed in one subsystem, but rather multiple profiles were
detectable within one subsystem (Supplementary file 5). For example, under
the “Virulence, Disease and Defense” subsystem, we observed all these
profiles with functions related to metal resistance FP2 (i.e Cobalt-zinc-
66
cadmium CzcA protein, Cation efflux system protein CusA), and FP1 (i.e
Magnesium and cobalt efflux protein CorC), and FP3 (i.e Copper homeostasis)
OPA-nc (see Virulence subsystem in Supplementary file 5). However,
functions related to mobile genes and HGT agents (Supplementary Figure
S2.8) were significantly more abundant in polluted lakes (e.g. Gene transfer
agent proteins, conjugative transfer proteins, DNA repair, CRISPR associated
proteins, integrons). Classification of functional abundance (subsystem levels
1, 2, 3) identified two independent clusters. The first cluster grouped BAR-
mc, LAR-hc and TUR-hc, and the second grouped DAS-lc and OPA-nc
(Supplementary Figure S2.5 and S6). Similar topologies were obtained using
both approaches: ORF (Supplementary Figure S2.4b) and reads subsampling
(Supplementary Figure S2.4c, d). PCA analysis based on the ORF approach
produced the same results, where at least 71% of variance was explained on
the first PC for all subsystem function levels. We only presented a PCA plot
for subsystems abundance in level 1, where more than 82% of variation in
functional abundance was explained by the first component (Figure 2.3a). At
the metabolic level, analysis of enzymes abundance profiles cross-
metagenomes showed different topology which was a dichotomy between
OPA-nc and all others pollution gradient lakes (See Supplementary file 7 and
Supplementary Figure S2.9).
2.5.7 Role of trace metals in function variation signatures
NMDS analysis of functional abundance highlighted two main patterns of
correlation (significant R-squared indicating regression model’s goodness of
fit) with metadata (Figure 2.3c). First, BAR-mc, LAR-hc and TUR-hc were
correlated with Cadmium axis (p-value ≤ 0.05). Second, OPA-nc and DAS-lc
were correlated with pH axis (p-value ≤ 0.05). The same analysis performed
on the subsystems in level 2 (192 functional modules) suggested a significant
contribution of all studied factors (results not shown). At the finest functional
level, lakes ordination based on the NMDS of polymetallic resistance genes
(PMRG)s abundance showed a fit with the cadmium concentration gradient
(Figure 2.4), where DAS-lc was ordinated near BAR-mc and LAR-hc. In NMDS
analysis of PMRGs located on chromosomes (Figure 2.4a), only Cadmium
played a significant role in explaining abundance variation. Similarly, the
67
NMDS analysis of PMRGs located on plasmids provided the same classification
profile even though they do not fit significantly with any metal traces (Figure
2.4b).
2.6 Discussion
2.6.1 Decoupling taxon-function as a signature of adaptive
strategies
Comparing the compositional signatures of taxon and function, we observed
that relative differences in taxon abundance could only partially predict the
impact of metallic toxicity on metacommunity structure (see section “Role of
trace metals in taxonomic variation signatures”). By considering the
signatures of functional abundance of the subsystems explained by pH and
Cadmium in polluted lakes, we could more accurately predict the impact of
metallic contaminants on ecosystem services of lake metacommunities. In
this respect, the contamination gradient explained much variation in
community function structure and provided a powerful way to further assess
the relationship between the distribution of functional abundance and
selective pressure, which may increase gradually with the expelled AMD flow
over time. The impact of the selection gradient on lake metacommunity
composition was tested through two independent analyses, first using
diversity measures, and second by detecting taxon-function decoupling
patterns. Alpha taxonomic diversity suggest a switch in BAR-mc, while the
gradual decrease in evenness based both taxon and function in OPA-nc:
(2.2t;2.3f) , DAS-lc (2.8t; 2.3 f), BAR-mc (2.5t;2.9f), LAR-hc (2.4t;2.6f), TUR-
hc (0.1t;2.8f) could be a potential consequence of composition homogeneity
in community type (e.g. Proteobacteria in TUR-hc). Indeed, this observation
may be related to the low complexity in AMD communities previously
documented for the same lake system (Laplante, Boutin, and Derome 2013;
Laplante and Derome 2011), and for other AMD metacommunities (Allen and
Banfield 2005; Huang, Kuang, and Shu 2016).
The rCCA analysis allowed for the detection of significant spatial correlation
between taxon and function in OPA-nc / DAS-lc, reflecting a coupling between
taxon and function. In these unpolluted lakes, as mentioned above, NMDS
analysis showed that environmental factors (Cadmium, pH and DOC)
68
explained variation in the overall taxonomic and functional composition. At
high resolution (subsystems level 2, 3) NMDS showed a slight difference
between OPA-nc and DAS-lc, but we cannot unequivocally associate these
variations to trace metal ratios. We may have missed other explanatory
environmental and chemical variables (i.e. NO2, NO3, SO4, PO4), or the
potential variation resulting from neutral ecological process, drift or random
reproduction as observed in wastewater habitats (Ofiţeru et al. 2010). Such
coupling is not necessarily absolute but partial, owing to the presence of some
differentiated sub-communities performing the same ecosystem services. In
pristine natural conditions (without stressful anthropogenic inputs), coupling
between taxon and function was observed in freshwater lakes (Langenheder
et al., 2005; (Debroas et al. 2009), and decoupling was observed in oceanic
bacterial communities from contrasting environments (Louca, Parfrey, and
Doebeli 2016).
Overall, in the present study, we found that functional variation between
polluted and unpolluted lakes was better explained by environmental factors
than taxonomic variation between and within functional groups. Concerning
the three lake communities facing exposure to a polymetallic gradient (BAR-
mc / LAR-hc / TUR-hc), the explained variance between taxon (25%) and
function (6%) strongly suggests a decoupling between taxa and functions.
The shared functions in these three polluted lakes reflect a convergent
pattern, which in turn could be interpreted as a predictive signature of the
ecosystem service’s impairment associated with acid mine lake water. This
conclusion is further supported by the NMDS results, where the distribution
of polluted lakes fitted closely to Cadmium. In addition to rCCA, when
comparing tree topologies of structure and function (Figure 2.2f; 3c), we
detected additional patterns of taxon-function decoupling, like the
polymetallic contamination gradient. Such an approach offers interesting
insights into the adaptive strategies used by metacommunities facing long-
term exposure to polymetallic pollution. Often interpreted as an indicator of
HGT in natural communities (Burke et al. 2011; Green, Bohannan, and
Whitaker 2008; Louca et al. 2016; Louca, Parfrey, and Doebeli 2016; Ram et
al. 2005) and AMD communities (Navarro, von Bernath, and Jerez 2013;
Devarajan et al. 2015; L. Chen et al. 2016; Hemme et al. 2016), taxon-
69
function decoupling may provide evidence for selective pressure on microbial
communities (e.g. exerted by metallic exposure). Indeed, as mentioned
above, multiple proteins playing a role in HGT, such as cassettes of integrons
and transposable elements, were present in polluted lakes, and absent in an
unpolluted lake (OPA-nc). We observed more than 14 mobile PMRGs located
on plasmids, and only two PMRGs on both plasmids and chromosomes. The
plasmid location of these PMRGs indicates that bacterial conjugation may be
a vector for HGT. Interestingly, a heatmap of abundance clustering from
chromosomal and plasmid PMRGs (Figure 2.not shown) produced a similar
topology of functional profiles (i.e. OPA-nc; DAS-lc-BAR-mc; LAR-hc-TUR-
hc).
Evolutionarily speaking, such taxon-function decoupling patterns are
expected to be signature of adaptation within communities between closely,
but also distantly, related bacterial strains. Consequently, community
composition in BAR-mc, LAR-hc or TUR-hc may have independently evolved
via HGT events of resistance and regulatory genes. According to functional
abundance results, the potential occurrence of HGT is higher in LAR-hc and
TUR-hc compared to BAR-mc, which is closer to DAS-lc and OPA-nc in terms
of functional distribution. A subset of adaptive beneficial transferred genes is
expected to reach fixation (Lind et al. 2010), but the long term metallic
contamination may have funneled the “metal resistance gene pool” into
different evolutionary trajectories due to the mounting selective pressure.
2.6.2 Taxonomic adaptive signatures
In this study, the overall taxonomic variation suggests three salient patterns
of abundance distribution. First, a “composition gradient” pattern constituted
three shifts in taxonomic structure: i) high abundance of Proteobacteria in
polluted sites (TUR-hc, LAR-hc; BAR-mc), ii) high abundance of
Actinobacteria in unpolluted sites (OPA-nc, DAS-lc), iii) intermediate levels of
Cyanobacteria in all sites, with Nostocales being abundant in polluted lakes
and Chroococcales abundant in unpolluted lakes (Supplementary file 3).
Second, a “community type” pattern suggests that the overall
metacommunity exhibited compositional shifts along the five lakes from wide
(phylum) to narrow (genus) taxonomic levels. Third, a “taxonomic
convergence" pattern highlights parallel changes of community taxonomic
70
structure, thus confirming previous results based on semi-quantitative and
quantitative studies (Laplante and Derome 2011; Laplante, Boutin, and
Derome 2013).
To further reinforce the taxonomic composition analysis, we examined genera
abundance and ORF distributions. Similar ratios of ORFs/Taxa were observed
in the five studied metagenomes (Supplementary Figure S2.2). The number
of annotated ORFs in all metagenomes was comparable. Furthermore,
random subsampling analysis without replacement produced similar results
(slightly different in topology) compared to the ORFs approach, with
remarkable clustering fidelity of subsampled replicates from each
metagenome (supplementary Figure S2.4a). Here, the subsampling approach
revealed consistency in the molecular signal of each lake. We acknowledge
that the subsampling approach used in our analysis cannot replace real
biological replications, but it is rather an indicator of the metagenomic data
robustness to the metacommunity structure.
To understand the sources of variation in contributing to the three major
shifts of relative abundance in community type, combined NMDS and
correlational analyses were performed for each pattern of taxonomic
variation. First, the Proteobacteria genera distribution of eight predefined
clusters (Supplementary file 4) showed that abundance variation between
communities was mainly explained by synergistic interactions of Cd, pH, Mn,
and Alu (Supplementary Figure S2.3a). According to previous studies,
Proteobacteria were among the most abundant phyla in acid mine water
(Laplante, Boutin, and Derome 2013; Streten-Joyce et al. 2013) and in
freshwater lake sediments polluted by “heavy metals” (Ni et al. 2016).
Second, in contrast to Proteobacteria, our results divided Actinobacteria into
four genera abundance clusters (Supplementary file 4) constrained by two
main and opposite explanatory factors, Cd and Fe (Supplementary Figure
S2.3b). In fact, the most abundant Actinobacteria genera (Streptomyces,
Frankia, Mycobacterium), which varied between polluted and unpolluted
lakes, fall in the same abundance cluster (see Actinobacteria in
Supplementary file 4). Indeed, some Actinobacteria (e.g. Streptomyces)
strains are known to have different metal-resistance profiles (Álvarez,
Catalano, and Amoroso 2013). Interestingly, strains like Mycobacterium were
71
able to transport and uptake Cd (Dimkpa et al. 2009). On the other hand,
Cyanobacteria abundance showed different patterns of abundance in polluted
and unpolluted lakes (Supplementary file 4; Supplementary Figure S2.3c)
suggesting that Chroococcales (Cyanothece, Microcystis, Synechocystis,
Thermosynechococcus) and Synechococcales (Synechococcus,
Prochlorococcus) are much more affected by trace metals compared to the
Nostocales (Anabaena, Aphanizomenon, Cylindrospermopsis,
Dolichospermum, Nodularia, Nostoc, Raphidiopsis). Although Cd was not
identified here as a significant explanatory factor, diverse strains of
Nostocales were documented to have the capacity to adsorb Cadmium
(Pokrovsky et al. 2008) and trace metals (Mota et al. 2015). Interestingly,
the sudden break of Nostocales lineages (Supplementary file 3) between the
connected lakes DAS-lc, BAR-mc and LAR-hc is potentially related to
resistance thresholds to trace metals, as higher levels become toxic to
Synechococcus (Ludwig et al. 2015). Furthermore, the high relative
abundance of Chroococcales and Cyanobacteria in OPA-nc and DAS-lc is
potentially related to their role in photosynthesis and DOC mineralization
(Bittar et al. 2015). Overall, our results show that metallic toxicity impacts
metacommunity structure and provides a partial explanation for the relative
shifts in abundance found in the lakes we studied. The dominance of
Proteobacteria in over polluted communities confirms the result previously
observed in the same lake system (Laplante, Boutin, and Derome 2013), and
from various acid mine waters in the world (Almeida et al. 2008; Hemme et al. 2010;
J.-L. Kuang et al. 2013; Stankovic et al. 2014; K. Wang et al. 2015).
2.6.3 Functional adaptive signatures
At the general level (subsystems level 1), only four subsystems showed
gradual variation. At the function level, our results suggest deterioration in
ecosystem services along the contamination gradient, as relative abundance
of functional modules in 18 subsystems such as “Carbohydrates”,
“Photosynthesis”, “Cell division and cycle”, “DNA metabolism” and
“Respiration” decrease gradually. However, under “Virulence defense and
disease” and “Membrane transport” subsystems (level 1), many important
metals transport and resistance functions (i.e. Cobalt-Zinc-Cadmium
resistance) increased between OPA-nc and other lakes (Supplementary
72
Figure S2.7). These profiles of gradual changes were less observable at the
general level (subsystems level 1), and more detectable at the functional
level resolution of many subsystems. The gradual decrease and increase in
relative abundance proportions was clearly observed at le the lowest
molecular function (i.e. Photosynthesis functions) along the polymetallic
gradient. Overall, variation in the functional composition of metacommunities
suggests convergence between BAR-mc/LAR-hc and TUR-hc, two
geographically distant and independent lakes affected by independent AMD
sources.
In contrast to the community classification based on taxonomic composition,
BAR-mc is functionally closer to LAR-hc-TUR-hc than OPA-DAS. NMDS of
functional composition, community hierarchical clustering, and PCA analysis
all find the same classification results. Cadmium and pH were the main factors
explaining functional composition variability among lakes. However,
independent analysis performed on both PRMGs and enzymatic functions
abundance showed that DAS-lc fitted within the polluted lakes (BAR-mc-LAR-
hc-TUR-hc) instead of OPA-nc. PMRGs located on plasmids (Figure 2.4b) were
differentiated from those located on chromosomes (Figure 2.4a) since
plasmid genes are known to house more adaptive genes acquired via bacterial
conjugation (Li and Zhang 2015). Only two experimentally confirmed genes
(copA and actP) were found in both plasmids and chromosomes. CopA is
involved in silver/copper export and homeostasis (Banci et al. 2003; Behlau
et al. 2011; Cha and Cooksey 1991; Outten et al. 2001). Acetate Permease
(ActP) controls copper homeostasis in rhizobium preventing low pH-induced
copper toxicity (Reeve et al. 2002). NMDS analysis based on Chromosomal
PMRG abundance revealed that Cadmium plays a significant role (p-value ≤
0.05) in shaping the differential abundance of these genes. Alternatively,
analysis of plasmid PMRGs did not highlight any significant fit with metal axes
(Figure 2.4b), owing to the low number of annotated PMRGs on plasmids.
Using OPA-nc as an unpolluted reference in our comparative framework,
differential metabolic abundance variation revealed an erosion of biosynthesis
pathways along the contamination gradient (results of compared pathways
not shown). Eroded metabolic functions were associated to degradation of
aromatic compounds, amino acid biosynthesis, and carbohydrates, thus
73
leading to the loss of major bacterial mediated ecosystem services. As
bacterial communities experienced a consistent metallic stress over sixty
years of mining activities, many functions associated with ecosystem services
likely became energetically too expensive to be maintained. Such a selective
environment may have led to community specialization. Community
specialization has recently been demonstrated in soil AMD communities
(Volant et al. 2014) and natural freshwater communities (Pérez, Rofner, and
Sommaruga 2015; Pernthaler 2013; Salcher 2013). In summary, the two
main elements (or factors) that explained the majority of the functional
variation between polluted vs unpolluted communities were pH and Cadmium
concentration. Nonetheless, other metal trace gradients offered partial
explanations for functional variation.
2.7 Conclusions
In this study, we examined adaptive signatures within natural lacustrine
microbial communities living under a gradient of selective pressure induced
by trace metal contamination from over 60 years of mining. Using a
metagenomic approach based on whole genome shotgun sequencing, we
identified a convergence in both taxonomic and function responses, thus
providing evidence for genotypic signatures of adaptive evolution. Strong
selective pressure may drive overall taxon-function decoupling, which may
reflect the occurrence of gene loss and Horizontal Gene Transfer (HGT)
induced by AMD gradient, or the result of strong selection exerted on existing
strains possessing the necessary resistance genetic background. This study
remains a preliminary assessment of decoupling phenomenon and further
studies are eventually needed to understand in a deeper manner the nature
of convergence between unpolluted environments versus polluted
environments in a context of stress gradient. At the taxonomic scale,
metacommunity composition showed marked relative abundance shifts of
major phyla, but was much more marked at the genus level, suggesting a
“community type” adaptation to the metallic gradient within each ecological
niche. At the function scale, we observed the erosion of metabolic pathways
along the metallic gradient despite the higher abundance of functional
categories like stress response, regulation, protein metabolism, and metallic
74
resistance in polluted lakes compared to unpolluted lakes. Investigating the
relationship of both taxonomic and functional signatures, we detected a
decoupling pattern between taxon and function in polluted lakes as an
indicator of adaptation potentially via horizontal gene transfer. These results
suggest, for the first time, a decoupling pattern of taxon-function within
natural communities adapted to a gradient of polymetallic contamination.
This decoupling pattern highlights the gap between microbial biodiversity and
ecosystem services in polluted environments.
2.8 Figures
Figure 2. 1 Bioinformatics analysis pipeline.
Two approaches were developed for this work. With the ORF-based approach,
de novo assembly was performed on raw reads data using Ray Meta software.
Then, the predicted ORFs (Open Reading Frames) were annotated using
Diamond similarity research tool against SEED, which is a curated database.
With the reads-based approach, merged (with FLASH algorithm) and filtered
75
reads (with Nesoni) (length ~ 200 pb in average) were annotated using the
same tool as the ORF-based approach, Diamond algorithm and SEED
database. For both approaches, we used the lowest common ancestor (LCA)
algorithm in the taxonomic assignment and the subsystems hierarchy in
function classification. Diversity measures were computed using mothur
software.
Figure 2. 2 Composition of metacommunities based on the ORF approach.
(a) Metacommunity composition (y-axis) is shown in stacked bars for each
lake metagenome (x-axis). Only phyla with relative abundance (RA) greater
than 1% are shown. (b) Hierarchical clustering of samples based on genus
RA using Ward’s method and Bray-Curtis dissimilarity distance, bootstrap AU
(Approximately Unbiased) p-value and BP (Bootstrap Probability) values are
76
shown on the nodes. (c, d) Principal Component Analysis (PCA) of samples
based on genus RA with different annotation parameters of alignment length
cutoffs (50 pb in c and 30 bp in d) and identity threshold (85% in c and 60
% in d). (e) NMDS of genera abundance fitted to trace metals was performed
with Bray-Curtis distance, three dimensions were a priori defined for distance
rank ordination and stress value was below 0.05. Cadmium (Cd), Manganese
(Mn), and pH significantly fitted with NMDS axes are highlighted in red. NMDS
loadings (NMDS1, NMDS2), and P-value of correlation r2 of trace metals were
reported in Supplementary file 6. Each small dot represents the ordinated
genus, while each large point represents the lake communities’ samples using
a circle for OPA-nc in blue and the control TUR-hc in black, and the connected
lakes are illustrated with squares (LAR-hc in red, BAR-mc in orange and DAS-
lc in yellow). Genus plot coordinates, clusters and dot labels are shown in
Supplementary file 4. (f) Tree based Unifrac distance computed with mothur
is indicated by branch lengths. All these results were obtained using the ORF
based approach with 85% identity threshold, e-value of 10-12, minimum
alignment length of 50 base pairs, and the lowest common ancestor (LCA)
algorithm for taxonomic assignment. OPA-nc (Opasatica Lake) is the negative
control; DAS-lc (Dasserat Lake) is low polluted; BAR-mc (Arnoux Bay) is
medium polluted; LAR-hc (Arnoux Lake) is highly polluted, and TUR-hc
(Turcotte Lake) is the positive control of contamination.
Figure 2. 3 Function abundance classification based on ORF approach.
77
Composition analysis of
metacommunity functions
based on relative abundance
(RA) of subsystems using
principal component analysis
(PCA) (a), tree based Unifrac
distance (b) and NMDS (c)
identified the same pattern.
(a) PCA figure was obtained
from STAMP software (Parks
et al. 2014). (b) NMDS (three a priori predefined dimensions projected into
two dimensions, stress value <0.05, Bray-Curtis distance) axes of all
annotated subsystems level 1 fit significantly with Cadmium (Cd), and pH
using the ORF approach. In d, each small dot represents a subsystem, while
the large dot does represent the lake metagenome indicated with a circle for
the negative control lake (OPA-nc) in blue and the positive control lake (TUR-
hc) in black. The connected lakes are illustrated with squares (LAR-hc in red,
BAR-mc in orange and DAS-lc in yellow). NMDS loadings (NMDS1, NMDS2),
and P-value of correlation r2 of trace metals were reported in Supplementary
file 6. Subsystems plot coordinates, clusters and dot labels are resumed in
Supplementary file 4. (c) Tree based Unifrac distance computed with mothur
is indicated by branch lengths In the ORF based approach the following
parameters were strictly respected; 85% identity threshold, e-value of 10-12,
minimum alignment length of 50 base pairs, and the lowest common ancestor
(LCA) algorithm for taxonomic assignment. OPA-nc (Opasatica Lake) is the
negative control; DAS-lc (Dasserat Lake) is low polluted; BAR-mc (Arnoux
Bay) is medium polluted; LAR-hc (Arnoux Lake) is highly polluted, and TUR-
hc (Turcotte Lake) is the positive control of contamination.
78
Figure 2. 4 Polymetallic resistance genes (PMRG) abundance correlation with trace metals.
79
(a) For PMRG on chromosomes (72 genes), Cadmium (Cd) was significantly
correlated with NMDS axes and it was the main explanatory factor of
abundance variation of these genes between metacommunities. (b) NMDS
axes based on relative abundance of PMRG located on plasmids (27 genes)
do not significantly fit with any trace metal arrows. This NMDS analysis was
performed with Bray-Curtis distance, three dimensions were a priori defined
for distance rank ordination and stress value was below 0.05. NMDS loadings
(NMDS1, NMDS2), and P-value of correlated trace metals are reported in
Supplementary file 6. Each small dot represented an individual PMRG, while
each large point represents the lake communities’ samples using circles for
OPA-nc in blue and the control TUR-hc in black, and the connected lakes were
illustrated with squares, LAR-hc in red, BAR-mc in orange and DAS-lc in
yellow. Thresholds of 75% of identity, minimum alignment length of 50 base
pairs and e-value of 10-12 parameters were strictly respected. PMRG were
annotated by performing Blastn of ORFs against BacMet database using
Diamond software.
80
Figure 2. 5 Decoupling of taxon and function between metacommunities based on the subsampled reads approach.
(a-d) Regularized canonical correlation analysis (rCCA) performed on BAR-
mc, LAR-mc and TUR-hc. (e-h) rCCA performed on OPAnc and DAShc. (a)
rCCA of taxon (genus relative abundance) showed 25% of explained variance
on the first canonical component. (b) rCCA of function (subsystems level3
relative abundance) showed only 6% of explained variance on the first
canonical component. (c) The canonical cross-correlation of taxon-function
identified a decoupling pattern. (d) Cross-validation score converged to a
maximum value of 0.99 when regularization parameters λ1 and λ2 were both
fixed at 0.375.
81
Figure 2. 6 Coupling of taxon and function between metacommunities based on the subsampled reads approach
(a) rCCA of taxon (genus relative abundance) showed 1% of explained
variance on first canonical component. (b) rCCA of function (subsystems
level3 relative abundance) also showed 1% of explained variance on first
canonical component. (c) The canonical cross-correlation of taxon-function
identified a coupling pattern between taxon and function. (d) Cross-validation
scores converged to a maximum value of 0.975 when regularization
parameters λ1 and λ2 were both fixed at 0.0925. rCCA was applied using
mixOmics and CCA package in R.
82
2.9 Supplementary figures
Supplementary figure S2.1. Geographical localisation of the
sampling sites.
This figure shows the geographical localisation (a) of the sampling sites
located in Ryoun-Noranda (West Quebec, Canada) visited in June 2011.
Latitute and longitude coordinates of sampling sites are 48.25005489 and -
79.40574646 in Opasatica lake (OPA-nc); 48.07601448 and -79.3082428 in
Dasserat lake (DAS-lc); 48.24090959 and -79.35012817 in Arnoux Bay
(BAR-mc); 48.25051211 and -79.333992 in Arnoux lake (LAR-hc);
48.30474963 and -79.07742262 in Turcotte lake (TUR-hc). This map was
produced using Arc GIS Esri ® Arc Map™ 10.1 under academic licence
certification. (b) Trace metals concentrations measured in the five sampled
lakes one year before this study (Laplante and Derome 2011). The x-axis
represents the log ratio of trace metal concentrations (mg/l) and the y-axis
represents detection limit in each lake. The metallic gradient showed that
83
Cadmium was under the detection limit in OPA-nc (negative control), at the
detection limit in DAS-lc (low contamination), three times more than the
detection limit in BAR-mc (medium contamination), LAR-hc (high
contamination), and TUR-hc (positive control). Contamination gradient
classification refers to the Cadmium log ratio across the five lakes.
Supplementary figure S2.2. Classification of lake metacommunities
based ORF approach at genus and phylum levels.
(a) Distribution of ORF and annotated genus in the five metagenomes. This
figure showed that not only the number of predicted ORFs (Supplementary
file 2), comparable between metagenomes but also the genus count
(Supplementary file 3). (b) Hierarchical clustering of samples using Ward’s
method and Bray-Curtis dissimilarity distance, bootstrap AU (Approximately
Unbiased) p-value and BP (Bootstrap Probability) value are shown on nodes.
(c, d) principal component analysis (PCA) of samples based on genus relative
abundance (RA) assigned with coverage (c) and without coverage (d)
normalization. Metacommunity clustering based on genus abundance is
84
different at phylum level where BAR-mc was closer to OPA-nc and DAS-lc. (e)
PCA analysis of samples based on function RA with different annotation
parameters of alignment length cutoff (30 bp) and identity threshold (60 %).
(f) Distribution of filtered ORFs on different alignment length cutoffs. For the
ORF based approach (a,b,c,d), the 85% identity threshold, e-value of 10-12
and minimum alignment length of 50 base pairs parameters were selected in
filtering annotations , and the LCA (Lowest Common ancestor) algorithm was
used to assign taxonomy.
Supplementary figure S2.3. Composition of metacommunities based
on the ORFs approach.
NMDS (with Bray-Curtis distance) of genera abundance for major abundant
phyla fitted to trace metals for Proteobacteria (a), Actinobacteria (b),
Cyanobacteria (c). the water pH and trace metals which correlated
significantly with NMDS axes were highlighted in red. Each small point in
figures a, b, and c represented the genus abundance, while each big point
does represent the lake metacommunities samples using circle shape for
OPA-nc in blue and the control TUR-hc in black, and the connected lakes were
illustrated with square shape, LAR-hc in red, BAR-mc in orange and DAS-lc in
yellow. NMDS loadings (NMDS1, NMDS2), and P-value of correlation r2 of
85
trace metals were reported in Supplementary file 6. Genus plot coordinates,
clusters and dot labels are resumed in Supplementary file 4.
Supplementary figure S2.4. Hierarchical clustering of taxon and
function.
(a) Hierarchical clustering of artificial replicates based on genus abundance
using the subsampled reads approach. (b) Hierarchical clustering of samples
based on abundance of subsystem level 1 using the ORF approach. (c, d)
Hierarchical clustering of subsampled replicates based on subsystems level 1
and 3 using the reads approach. Hierarchical clustering was performed using
Ward’s method and Bray-Curtis dissimilarity distance; bootstrap AU
(Approximately Unbiased) p-value and BP (Bootstrap Probability) value are
shown on the nodes.
86
Supplementary figure S2.5. Heatmap of subsystems in level 1
This heatmap represents metagenomes classification based on subsystems in
level 1 (See supplementary file 5). Dendrogram’s topology identified two
clusters. The first cluster grouped BAR-mc, LAR-hc and TUR-hc, and the
second grouped DAS-lc and OPA-nc. The hierarchical clustering of relative
abundance proportions of subsystems, and of samples was performed using
Ward’s method and Bray-Curtis dissimilarity distance. The ORF approach was
used with identity threshold of 85%, e-value of 10-12 and minimum
alignment length of 50 base pairs parameters. Vegan package and heatmap
() function in R were used to produce this figure.
87
Supplementary figure S2.6. Heatmap of subsystems in all levels.
Subsystems relative abundance were clustered cross-metagenomes in
different levels, level2 (981 modules), level3 (192 modules) and function
level (6801 functions) (See supplementary file 5). The same topology was
observed in level 1(See supplementary figure S5) and in all levels. The
hierarchical clustering of relative abundance proportions of subsystems, and
of samples was performed using Ward’s method and Bray-Curtis dissimilarity
distance. The ORF approach was used with identity threshold of 60%, e-value
88
of 10-12 and minimum alignment length of 50 base pairs parameters. Vegan
package and heatmap () function in R were used to produce this figure.
Supplementary figure S2.7. Heatmap of multiple subsystems
abundant in function level.
This heatmap represents cross-metagenomes, the common and most
abundant functions (>2%) in 22 subsystems (See supplementary file 5).
Functions of polymetallic resistance (Cation efflux system protein CusA and
89
Cobalt−zinc−cadmium resistance protein CzcA) showed a profile of gradual
abundance increase along the pollution gradient. The hierarchical clustering
of relative abundance proportions of functions, and of samples was performed
using Ward’s method and Bray-Curtis dissimilarity distance. The ORF
approach was used with identity threshold of 60%, e-value of 10-12 and
minimum alignment length of 50 base pairs parameters. Vegan package and
heatmap () function in R were used to produce this figure.
90
Supplementary figure S2.8. Subsystem of “Phages, prophages,
plasmids, and transposable elements” cross-metagenomes.
91
Under this subsystem multiple relevant functions (level 3) related to mobile
elements and transfer vectors (Gene transfer agents, transposons,
prophages, conjugative plasmids, integrons) were shared between DAS-lc,
BAR-mc, LAR-hc, TUR-hc and depleted in OPA-nc. However, each
metagenome contains specific profile of mobile elements functions such like
agents of gene transfers and conjugative elements in TUR-hc. The
hierarchical clustering of relative abundance proportions of this subsystem
modules, and of samples was performed using Ward’s method and Bray-
Curtis dissimilarity distance. The ORF approach was used with identity
threshold of 60%, e-value of 10-12 and minimum alignment length of 50 base
pairs parameters. Vegan package and heatmap () function in R were used to
produce this figure.
Supplementary figure S2.9. Metabolic abundance cross-
metagenomes.
This heatmap represents 1842 annotated enzymes (See EC
number in supplementary file 7) in all samples. The
hierarchical clustering of relative abundance proportions of
enzymes, and of samples was performed using Ward’s
method and Bray-Curtis dissimilarity distance. The
dendrogram shows dichotomy between OPA-nc metagenome
and all others. The ORF approach was used with identity
threshold of 60%, e-value of 10-12 and minimum alignment
length of 50 base pairs parameters. Vegan package and
heatmap () function in R were used to produce this figure.
92
Supplementary figure S2.10. Gradual variation of functions cross-
metagenomes.
Two heatmaps represent gradual function abundance FP1 (106 functions) and
FP2 (123 functions) along the contamination gradient. The hierarchical
clustering of relative abundance proportions of functions was performed using
Ward’s method and Bray-Curtis dissimilarity distance. The ORF approach was
used with identity threshold of 60%, e-value of 10-12 and minimum
alignment length of 50 base pairs parameters. Vegan package and heatmap
() function in R were used to produce this figure.
93
Supplementary figure S2.11. Specific variation of functions cross-
metagenomes.
94
Two heatmaps represent specific function abundance FP3-OPA-nc (167
functions) and FP3 specific to pollution gradient (225 functions). The
hierarchical clustering of relative abundance proportions of functions was
performed using Ward’s method and Bray-Curtis dissimilarity distance. The
ORF approach was used with identity threshold of 60%, e-value of 10-12 and
minimum alignment length of 50 base pairs parameters. Vegan package and
heatmap () function in R were used to produce this figure.
2.10 Supplementary Material
The Supplementary Material for this article can be found online at this link:
https://www.frontiersin.org/articles/10.3389/fmicb.2018.00869/full#supple
mentary-material
Supplementary file 1. This file contains two tables. The first table
resumed the geographical coordinates of sampled sites. The second table
presented abiotic parameters measured for each sampled site one year
before this study (Laplante and Derome 2011).
Supplementary file 2. This file summarized statistics of reads, contigs, and
ORFs MG-RAST annotations per lake metagenome.
Supplementary file 3. This file summarized in one table the relative
abundance of major taxa at phylum, class, and genus levels.
Supplementary file 4. This file resumed details of NMDS plots. NMDS
loadings, abundance clusters and points labels of all genus (Table1),
Proteobacteria (Table 2), Actinobacteria (Table 3), Cyanobacteria (Table 4),
then of all subsystems (Table 5) were reported for assuming a better
understanding of NMDS figures and supplementary figures.
Supplementary file 5. This file reported subsystems annotations and data
analysis; All subsystems data (dataset output of STAMP software) in table 1,
subsystems level 1 relative abundance (RA) in table 2, list of all annotated
subsystems in table 3, function profiles classification of RA proportions in
table 4, different function profiles (FP) (table 5, 6, 7, 8), resume of FP
occurrence in subsystems level 1 (table 9), summary of most abundant
functions (table 10) and summary of the subsystem “virulence, disease and
defense” (table 11).
95
Supplementary file 6. This file summarized details of NMDS correlation
analysis with metadata. Each table resumed NMDS loadings (NMDS1,
NMDS2), P-value of correlated trace metals of taxa (Table1), subsystems
level 1 (Table 2) and PMRGs.
Supplementary file 7. This file resumed diversity measures based relative
abundance of taxa (genus) and function (subsystems) in table 1 and all
annotated enzymes by their EC-number in table 2.
96
Chapitre 3: From networks to models: The
Yellow Perch (Perca flavescens) microbiome
assembly under metal toxicity
Bachar Cheaib , Hamza Seghouani, Martin Llewellyn, Katherine Vandal-
Lenghan, Pierre-Luc Mercier , and Nicolas Derome
Submitted to the Animal microbiome journal
97
From networks to models: The Yellow Perch (Perca flavescens)
microbiome assembly under metal toxicity
3.1 Resumé
Des processus stochastiques et déterministes gouvernent la structure et
l'assemblage des symbiontes bactériennes. Ils sont très peu étudiés dans un
système hôte-microbiote, particulièrement dans un contexte de perturbation.
Pour évaluer le rôle de ces processus dans la structuration et la composition
du microbiote, deux types d’habitats microbiens, libres (eau) et symbiotiques
(cutané et intestinal) associés à la Perchaude (Perca flavescens), ont été
exposés à deux régimes de sélection, constant et graduel, induits par des
doses non létales de cadmium. Les deux régimes de sélection constant et
graduel montrent une augmentation significative de la diversité alpha et des
corrélations négatives dans les réseaux, ce qui reflète une dysbiose à la fois
dans le microbiote cutané et intestinal. Le modèle non linéaire de moindres
carrés (NLS) suggère une dérive taxonomique ce qui reflète une succession
stochastique durant l’assemblage des communautés d'eau sans cadmium,
alors que l'assemblage de microbiote de l’hôte exposé au cadmium évolue
sous pression de sélection.
98
3.2 Abstract
Stochastic and deterministic processes underpin host-associated microbiota
structure and assembly. To assess the impact of such processes on microbiota
composition, two types of microbial habitats, free-living (water), and host-
associated (skin and gut), were experimentally exposed to either a constant
or gradual selection regime exerted by two sublethal cadmium chloride
dosages. Yellow Perch (Perca flavescens) was used as a vertebrate
ecotoxicological model. Using 16S rRNA gene metabarcoding, quantitative
metrics of the three metacommunities were characterized along development
and across experimental conditions. Both constant and gradual selection
regimes drove a significant alpha diversity increase both in skin and gut
microbiota. Pervasive negative correlations between taxa in both selection
regimes in skin, in addition to the taxonomic convergence with the
environmental bacterial community, suggest a loss of colonization resistance
resulting in the dysbiosis of host-associated microbiota. Furthermore,
network connectivity under stress was exclusively maintained by rare OTUs,
while abundant OTUs were mainly composed of opportunistic invaders such
as Mycoplasma and other genera related to fish pathogens. Finally, non-linear
least squares models (NLS) suggested that stochasticity mainly drove
taxonomic drift in cadmium-free water communities, whereas host-
microbiota assembly evolved in a deterministic (non-neutral) manner. Our
findings enhance our understanding of microbiota community assembly under
anthropogenic selective pressure.
99
3.3 Introduction
Microorganisms drive the biogeochemical cycles of the earth and contribute
towards homeostasis, immunity, physiology, behaviour (K. V.-A. Johnson and
Foster 2018; Stephens et al. 2016) and development (Stephens et al. 2016;
Sylvain and Derome 2017) across a wide range of metazoan hosts (Sherrill-
Mix et al. 2018). Host-microbiota symbioses involve complex and dynamic
associations between obligate and facultative symbionts (Theis et al. 2016).
Disentangling the characteristics of microbial interactions within communities
improves our comprehension of metacommunity assembly (Faust and Raes
2012). Ecological processes (i.e., dispersal, selection and ecological drift)
shape these interactions and govern the assembly rules of the ecological
communities (Nemergut et al. 2013; Vellend 2010). The impact of ecological
processes on community assembly is a long-term debate in macroecology.
Stochastic neutral theory suggests ecological interactions have no impact on
species abundances (macroscopic organisms). In such cases, local
communities are randomly connected to a single metacommunity through
differing rates of migration, death and birth (Bell 2001; Hubbell Stephen P.
2005; Stephen P. Hubbell 2006). In contrast, niche-related deterministic
theory considers that environmental conditions and interspecific interactions,
including competitive exclusion, determine species’ abundance (Chase and
Leibold 2003). In microbial ecology, the advent of culture-independent
approaches such as high-throughput 16S rDNA metabarcoding paved the way
for the conceptual framework of the Operational Taxonomic Unit (OTU) as
unit of microbial diversity. Such advancements have opened new
mathematical (Harris et al. 2017; Jayathilake et al. 2017; Sloan et al. 2006;
Stegen et al. 2012; Q. Zeng et al. 2015; 2017) and network-based (Faust
and Raes 2012; Foster, Krone, and Forney 2008; Taxis et al. 2015) models
for predicting ecological interactions between microbial communities. These
models helped in constructing hypotheses on types of processes driving
microbiomes assemblies over evolutionary time.
Models for quantifying the neutral (Harris et al. 2017; Burns et al. 2016a;
Ofiţeru et al. 2010; Sloan et al. 2006) and deterministic (Morrison-Whittle
and Goddard 2015; O’Dwyer, Kembel, and Sharpton 2015; Stegen et al.
2012; Yeh et al. 2015) processes in different types of microbial ecosystems
100
continue to provide new comprehensive insights regarding the forces
governing microbiome assembly. Both neutral and non-neutral processes
have been evidenced as drivers of the microbial metacommunity assembly in
many vertebrate microbiomes (Jeraldo et al. 2012), as well as within
rainwater microcosms (Langenheder and Székely 2011). For instance, neutral
processes were identified to play a major role during development of host-
associated microbial communities in different domesticated vertebrate and
plant models (Jeraldo et al. 2012; Maignien et al. 2014; McCafferty et al.
2013). Conversely, in the human gut microbiome, network modelling has
revealed that host-filtering, rather than strains-assortment, dominates the
microbial assembly rules, thus suggesting that non-neutral processes are the
key drivers (Levy and Borenstein 2013; 2014).
Studies within the same host species have found contradictory results: in
zebrafish, neutral processes generated substantial variation in gut microbiota
composition across individual hosts, but non-neutral processes (i.e. microbe-
microbe interactions, dispersal, or host-filtering) increased along host
development (Burns et al. 2016a). In contrast, in three other fish species,
deterministic processes shaping gut microbiota assembly were mainly at play
during the first developmental stages before gradually reducing (Yan et al.
2016). Therefore, the interplay between the underlying evolutionary
processes governing microbiota ontogeny in vertebrates is still poorly
understood. Overall, many studies of host-associated ecosystems, focusing
either on ontogeny under neutral conditions or on case-control comparisons,
reveal ambiguities regarding the relative influence of neutral and non-neutral
processes on microbiota ontogeny.
When focusing on case-control surveys, the influence of a given selective
pressure on microbiota composition is more salient. For instance, from the
human gut and oral microbiome under antibiotic therapy (Costello et al.
2012) to the euryhaline fish microbiome during salinity acclimation (V. T.
Schmidt et al. 2015), metacommunity assembly was mostly driven by
deterministic processes, with little evidence for stochastic colonisation.
Nonetheless, much remain to be done for fully understanding the mechanisms
of microbial assembly across a diverse range of host species.
101
In the present study, we first predict that a selection gradient induced by a
concentration gradient of a toxic metal will not only disrupt the host
physiology (Pyle, Rajotte, and Couture 2005a), but will also overwhelm the
recruitment and the assembly of symbiont consortia. Second, we predict that
the host will lose its filtering capacity to recruit the appropriate symbionts,
which in turn, will translate into increased colonisation of opportunistic
strains, as predicted by the colonisation resistance model (Costello et al.
2012). Third, from the interaction networks of OTUs, we expect that the rare
taxa may play an essential role in the structure of metacommunity assembly
(Banerjee, Schlaeppi, and Heijden 2018; Fierer 2017; Jousset et al. 2017; M.
D. J. Lynch and Neufeld 2015; Pester et al. 2010). Finally, microbiota diversity
within late host development stages may decrease due to the increasing host-
filtering capacity that occurs along developmental and life stages (Llewellyn
et al. 2016; Stephens et al. 2016; Sylvain and Derome 2017).
To this end, we measured the effect of directional selection along the
developmental stages of the host organism -juvenile Yellow Perch(Perca
flavescens) were exposed to two selection regimes: a constant (9 ppb) and a
gradual (0.8 to 9 ppb) exposure to sublethal doses of cadmium chloride
(CdCl2), over 90 days. The taxonomic compositional dynamic of two types of
microbial habitats, free-living (water), and host-associated (skin and gut),
were characterised throughout the young developmental stages of the host
using a 16S SSU rRNA gene metabarcoding approach. Being able to cope with
polymetallic gradients generated by acid mine drainages (AMD), the Yellow
Perch is a well-established ecotoxicology vertebrate model: many studies
have measured the impact of heavy metals on their adaptive genetic diversity
(Bélanger-Deschênes et al. 2013), innate immune system (Dautremepuits et
al. 2009), metabolism (Couture and Rajender Kumar 2003), development
(Azizishirazi et al. 2014), parasitism (Marcogliese et al. 2005; 2010; Ryman,
Walleghem, and Blanchfield 2008) and transcriptional activity (Bougas et al.
2013b). As these host functions are closely related to gut microbiota
composition, the Yellow Perch is a convenient model to test the effects of
metal exposure on a vertebrate host-microbiota system. The impact of
cadmium on the genotype of microbial communities in AMD water have been
documented (Cheaib et al. 2018), but not within the host-microbiota system.
102
Using sublethal doses of cadmium as a selection pressure, we aimed at
disantangling the neutral and non-neutral processes shaping the microbiota
assembly, without triggering significant physiological damage and causing
host death.
3.4 Materials and methods
3.4.1 Fish rearing.
After an acclimation period of one month in a 1500 L container, 1200 juvenile
Yellow Perch were reared in 24 tanks (50 fish per tank) of 36 L, each of which
had an independent filtering system circuit. The fish juveniles were fed daily
with the same food from the beginning to the end of the experiment. A second
acclimation period of two weeks was carried out before the start of cadmium
exposure (Supplementary file 1).
3.4.2 Exposure regimes to cadmium.
Control (Control) and Cd treated tanks were randomly distributed in the
aquarium facility. The experiment was designed for two cadmium exposure
regimes (8 tanks per regime), and one negative control regime (8 tanks). In
treated tanks, fish were exposed to cadmium chloride (CdCl2) provided by
Sigma-Aldrich (> 99.9% purity). The cadmium was dissolved in water. For
the regime of cadmium constant concentration (CC), the cadmium chloride
was initially added at 0.8ppb (parts per billion), before gradually increasing
the concentration every five days to reach a maximal concentration at the
end of the first month (T1). This maximal concentration was maintained two
months until the end of treatment (third month, T3). For the regime of
cadmium variable concentration (CV), the compound was initially added at
0.6 ppb, before the concentration was increased every five days to reach the
maximal concentration at the end of treatment (third month, T3). The
maximal CdCl2 concentration was empirically set at 9 g/L as it is the highest
CdCl2 concentration tolerated by Yellow Perch in contaminated Canadian lakes
(Couture, Rajotte, and Pyle 2008).
3.4.3 Host-microbiota and water sampling.
A total of 432 mucosa host microbiota samples were collected for this study,
216 (3 times x 3 regimes x 8 tanks x 3 replicates) skin mucus swabs and 216
103
(3 times x 3 regimes x 8 tanks x 3 replicates) gut tract samples
(Supplementary file2). Water samples were stored in sterile bottles
(Nalgene), 2 liters per tank were filtered using a polycarbonate membrane of
0.22 μm. In total, 144 filters (3 times x 3 regimes x 8 tanks x 2 replicates)
were conserved in 2mL sterile micro-centrifuge tube and directly stored at -
80 C.
3.4.4 Metal concentration in water and fish liver.
Concentrations of metal traces (Cd, Cu and Zn) within the water and liver
were determined with the ICPMS (Ionization Coupled Mass spectrometry)
technology at the Department of Chemistry, Laval University for T0 and T1,
then at INRS (Institut National de la Recherche Scientifique), Quebec, for T1-
T3. Before ICPMS analysis of cadmium ions in water, the CdCl2 in water
samples was fixed by adding 4% of nitric acid. This analysis was performed
every week until the end of the CdCl2 exposure regimes. After lyophilisation,
liver samples were digested with purified nitric acid and kept at room
temperature for five days. The liver acid digestion protocol was adapted from
Pierron et al. (2009). For further details, see the Supplementary file 2. The
metal concentrations were analysed using two-way analysis of variance
(ANOVA) of two independent factors: time and treatment. The interactions
between time and treatment factors were analysed using Tukey’s test and
Wilcoxon rank test depending on the data (metal concentration), which was
assessed for normality with the Shapiro test.
3.4.5 DNA extraction to Illumina Miseq sequencing.
DNA was extracted from all skin mucus and water samples using the Qiagen
DNeasy blood and tissue kit (Supplementary file 2). For all intestine samples,
after an RNA extraction for a transcriptomic project, the DNA was extracted
from TRIzol organic phase using BEB (back extraction buffer) and PCI
(phenol/chloroform/isoamylalcohol 25:24:1) solution (Supp. File 3). The 16S
ribosomal DNA was amplified via PCR using universal primers specific to the
V3-V4 hyper-variable region of the rDNA 16S gene (Werner et al. 2012). The
purified product of first-round PCR was used as a template for the library
preparation by performing second-round PCR. Final amplified DNA was
verified by electrophoresis on 2% agarose gel, and finally, DNA concentration
104
of the product was quantified by florescence using Quant-iT™ PicoGreen™
dsDNA Assay Kit (Thermo Fischer Scientific) (Supplementary file 2).
3.4.6 Analysis of 16S rDNA amplicons.
Sequence analysis was performed with our bioinformatic pipeline as described
previously (Llewellyn et al. 2016; Sylvain et al. 2016). The computed
diversity variation distribution across time and per treatment was plotted and
analysed for significance with linear mixed models considering tank as
random effect and cadmium as fixed effect using the lme4 package in R
(model M1= lmer(diversity.index (Evenness or Richness) ~
Cd.Water*Cd.Liver+Time+(1|Tank)). The linear mixed model (M1) was
compared to a multiple linear regression model M2 [lm(diversity.index ~
Cd.Water*Cd.Liver+Time] without random effects using a simple ANOVA
[anova(M1, M2)]. The analysis of the diversity structure and composition of
metacommunities was performed using the Rhea package (Lagkouvardos et
al., 2017). Briefly, the significance of alpha-diversity indexes (richness and
evenness) and beta-diversity (phylogenetic distance) differences between
experimental groups was assessed using rank statistics tests (Kruskal-
Wallis/Wilcoxon). Beta-diversity was measured using generalised UniFrac (J.
Chen et al. 2012), which considers both dominant and rare OTUs. Calculated
p-values of pairwise comparisons in alpha and beta-diversity were validated
for significance using the B-H (Benjamini-Hochberg) multiple correction tests
(Benjamini and Hochberg 1995) for avoiding the Type I errors (false
positives). Additional details on this analysis were documented in
Supplementary file 2.
3.4.7 Correlational networks.
The Spearman correlation coefficient corrected with false discovery rate FDR
multiple correction test was used to compute the OTU co-abundance
correlation. This coefficient was recently demonstrated as a robust approach
in terms of sensitivity and precision of correlation detection (Weiss et al.
2016). Significant positive and negative correlations were selected with B-H
test. Strong positive (Corr > 0.6), negative (Corr < -0.6) and significant
correlations (p-value FDR < 0.05) were used to build OTU networks.
Cytoscape software (Shannon et al. 2003) was used to perform network
105
visualization and analysis. The number of components indicated the level of
fragmentation (number of subnetworks) in the community. Each node size in
the network was proportional to the average OTU' relative abundance in all
samples, and the edge length was inversely proportional to the significant
coefficient of correlation between two nodes.
3.4.8 Metacommunity assembly modelling.
To investigate the role of neutral processes in community assembly, we fit
the distribution of OTUs to a neutral model of microbial assembly (Sloan et
al. 2006) using a non-linear least squares approach and beta distributions,
which has recently been implemented by others (Burns et al. 2016a). The
neutral model compares the frequency of OTU occurrence to their abundance
in the metacommunity by estimating a parameter (m), which describes the
migration rate. The estimated migration rate (m) is the probability that a
random loss (death or emmigration) of an OTU in a local community is
replaced by dispersal from the metacommunity source (Burns et al. 2016a).
The temporal comparisons (T0-T1; T1-T3) of predicted versus observed OTU
frequencies from the neutral model were used to highlight the percentage of
OTUs fitting the model with a confidence interval of 95%. The goodness of fit
to the neutral model was assessed using R-square as the coefficient of
determination. R-squared = 0.5 was the goodness of fit threshold to Sloan’s
neutral model (Sloan et al. 2006).
3.5 Results
3.5.1 Metal concentrations in water and host livers
The ANOVA demonstrated significant changes in the concentration of metals
in both water and fish liver at a specific time point and over time (Supp. Table
3.1). The Cd concentration had significantly changed between experimental
groups in the fish liver at T3, and in water at T1 and T3. According to ANOVA,
the interaction of time and Cd treatment was significant in both water (p-
value <0.05) and in fish liver (p-value <0.05). At time T3, the total Cd
concentration in water was maximal in CC and CV and significantly different
between control and each Cd regime but was not significantly different
between CC and CV. In the fish liver, the difference in Cd concentration at T3
was significant between all treatments (CC, CV or Control) (Table 3.1).
106
3.5.2 Mixed Effects of time and treatment on metacommunity
alpha diversity
Diversity indices (Table 3.2) in gut microbial community, skin microbial
community and water microbial community measured per treatment (CC, CV
& Control) and across time showed notable trends. Over time and treatment,
the alpha diversity (richness or evenness) varied significantly (Table 3.2-a,
Supp Figure 3.1, Supp. Table 3.2). In the gut microbial community, richness
significantly changed over time between all groups (Control, CC and CV)
(Table 3.2A). In the skin microbial community, significant changes between
groups were detected only at T1 (Table 3.2B, Supp Figure 3.1, Supp. Table
3.2). Statistical comparison using both diversity indices for evenness
(Shannon.effective) and richness (Chao1) demonstrated the importance of
time as a driver in microbial community alpha diversity, rather than treatment
(Figure 3.1).
3.5.3 An important effect of time on the taxonomic composition of
metacommunities
Hierarchical OTU classification at the genus and phyla levels revealed a
significant differential abundance of different taxonomic groups in both water
and host communities according to different conditions (treatment, time
point). These taxonomic differences were significant in Bray-Curtis-based
clustering (Supp. Figures 2.2 & 3.3). OTUs classified as Mycoplasma showed
increasing abundance across all treatments (CC, CV & control). Overall,
fluctuations of richness and evenness measured in water and host-associated
communities were explained, at least partly, by time and treatments.
Although time exerted an effect on richness and evenness in all communities,
the treatment showed a significant effect only on the evenness in skin
mucosal communities. Interaction between both factors (i.e. time and
treatment) revealed a significant effect on richness in all communities, and
on evenness in skin mucosal communities.
107
3.5.4 Community-level phylogenetic divergence
Phylogenetic distances based Gunifrac between OTUs were compared using
PERMANOVA and a multivariate homogeneity test for group dispersions
(variances). Among all host and water communities, time had a significant
effect on community divergence (p<0.005). By T1, significant differences
(p<0.005) among treatments were only observed in microbial skin
communities (Table 3.3, Figure 3.2); however, by T3, both variable (CV) and
constant (CC) cadmium exposure had driven differences in both skin and gut
communities compared to the controls. Among water microbial communities,
phylogenetic distance between treatments was significantly different at all
time points (Table 3.3; Supp. Figure 3.4).
3.5.5 Correlational metacommunity networks
3.5.5.1 Substantial role of rare taxa in the metacommunity network
connectivity
The percentage of OTUs involved in significant correlations was higher in
water community networks compared to those of the skin and gut
communities (Supp. Table 3.3). Within gut communities, abundant OTUs
played peripheral roles in network structure, demonstrating low overall
connectivity (Figure 3.3). In contrast, most connections in each network
existed among rare OTUs – especially in skin communities early in both
cadmium exposure regimes (T1 of CC and CV, Figure 3.4).
3.5.5.2 Reduced network connectivity in gut communities under
cadmium stress
Exposure to cadmium had a significant impact on the network connectivity
and integrity in the gut microbial communities. By T3, significant correlations
involved 78%, 69% and 63% of OTUs from Control, CV, and CC treatments
respectively (Supp. Table 3.3). In the Control group, most abundant OTUs
were connected to a central hub (Figure 3.3). In contrast, in the Cd-treated
groups, abundant OTUs were gradually disconnected from the main network
in small independent hubs or sub-networks. These abundant OTUs were
mainly composed of Mycoplasmas and other genera among Firmicutes like
Bacillus (>6% in CC and CV).
108
3.5.5.3 Negative correlations in Skin Mucous Community networks
suggest dysbiosis
In skin microbial community networks, at T1, network connectivity
represented by average neighbors (AN) was lowest in the Control network
(Control AN: 4) compared to cadmium treatment networks (CVAN: 10.7; CCAN:
24.6) (Supp. Table 3.4). Higher connectivity observed in CV and CC relative
to the Control treatment was manifest in the higher percentage of significant
negative correlations (red edges in networks of Figure 3.4) observed in those
groups (CV neg. corr.: 6.9%; CC neg. corr.: 6.3%) compared to control group
(Control neg. corr.: 2.2%). A significant increase in the abundance of Tenericutes
(Mycoplasma) and Proteobacteria nodes (OTUs) explained many of negative
interactions in CV (71 nodes) and CC (112 nodes) groups, when compared to
the Control group (50 nodes). Crucially, at T3 among both cadmium
treatments CV & CC, putative pathogenic genera were abundant at the
expense of other non-pathogens, as demonstrated by multiple negative
correlations between these genera (Figure 3.4). Bacteroidetes at T1 in CC
and CV treatments (Supp. Table 3.5-a) drove also many of these negative
correlations (11 nodes in CC, 8 nodes in CV, and 3 nodes in Control) (Figure
3.4). Interestingly, Bacteroidetes at T0 were significantly lower in skin
compared to gut and water for groups CC & CV & Control (Supp. Figure 3.5).
At T1, they significantly increased over time (Supp. Table 3.5-a) in skin for
CC& CV, not for control, and they show a significant higher abundance in skin
compared to water and gut communities for treatments groups (Supp. Figure
3.5).
3.5.5.4 Fragmentation of water microbial community networks
Water microbial community networks were fragmented, both over time and
treatment (Figure 3.5). There was a notable lack of Tenericutes (Mycoplasma)
in comparison to the skin and intestinal microbial communities’ networks. For
instance, only 2 to 3 nodes of Tenericutes (Mycoplasma) occurred in Control
and CC networks at T0 and T3, and with very low relative abundance. No
Mycoplasma nodes were observed in the CV network. Furthermore, the
number of network components was the highest in Control network at T0 and
T3, in CC at T1, and was always intermediate in the CV network (Supp. Table
109
3.4). Overall, topologies of the water microbial communities network
encompassed disconnected small independent hubs.
3.5.6 Stochasticity in water community assembly and
determinism in that of host microbiota
We applied Sloan’s model (Sloan et al. 2006) using a package developed by
Burns et al. (2016) to determine the relative importance of neutral and non-
neutral dispersal processes in community assembly. The percent of OTUs that
fit the neutrality model within a confidence interval of 95% tended to be
variable across time, community, and cadmium exposure regime (Figure
3.6). In gut and skin communities, less than 50% of the total number of OTUs
between time points (T0 – T1, or T1 - T3) were best explained by the neutral
model. This would suggest a deterministic role of the host in the recruitment
of the microbiota in every treatment. However, the percentage of neutral
OTUs was higher in the control treatment compared to the variable and
constant cadmium regimes. In water communities, the percentage of neutral
OTUs was more than 50% at each time point for control and at T0 and T1 for
both cadmium exposure regimes (CV and CC). At T3, the majority of OTUs
(>70%) were non-neutral for both cadmium exposure regimes (Figure 3.6).
Studying the relationship of OTU abundance with neutrality, our analysis
revealed that neutral OTU's percentage highly increased (25% to 80%) in
host and water communities when the rare OTUs were discarded from the
NLS models (Supp. Figure 3.6, Supp. Table 3.6).
3.6 Discussion
Our study evidenced salient differential shifts in community assembly across
three community types (environmental or host-associated), time and
xenobiotic exposure regimes. This highlights for the first time the relative
contribution of neutral and non-neutral factors in shaping the microbiota
during the early life-stages of an ecotoxicology vertebrate model. First,
significant increase of alpha diversity in skin and gut microbiota was detected
in both constant and gradual selection regimes. Then, at the community-
level, significant phylogenetic divergence was observed between the control
and treatment groups in the three community types at two specific time
points, T1 and T3. These two key time points were investigated further with
110
co-abundance network analysis. Frequent significant negative correlations
between taxa in both selection regimes in the skin and the increasing richness
of environmental bacterial strains, suggest a dysbiosis in the mucosal host-
associated microbiota. Furthermore, network connectivity under stress was
maintained by rare OTUs, while abundant OTUs were mainly composed of
opportunistic invaders such as Mycoplasma and other genera related to fish
pathogens like Aeromonas, Pseudomonas and Flavobacterium. Finally, to
predict the nature of evolutionary processes driving the metacommunity
changes, application of a non-linear least squares model (NLS) to the data
suggested that stochastic processes drove taxonomic drift in cadmium-free
water communities, whereas host-microbiota assembly evolved in a
deterministic (non-neutral) manner in treatments and the control.
3.6.1 Phylogenetic divergence at the community-level revealed
the impact of Cadmium exposure.
At T1, significant phylogenetic divergence occurred between treatments (CC
& CV) and control (Control). Most importantly, a taxonomic convergence
between treatments (CC & CV) not only in skin but also within the water
community occurred. This convergence mainly resulted from an invasion of
environmental bacterial strains in skin. The significant increase of
Bacteroidetes only in treatments groups (CC &CV) in skin communities
compared to water communities could strongly support this hypothesis
(Supplementary Figure 3.5). Such gain or loss of tissue-specific community
type suggests a disruption of the host’s ability to control the assembly of skin
microbiota, which is correlated with Cd exposure. This phenomenon is termed
“direct colonisation resistance” (Buffie and Pamer, 2013); however, we do
not exclude that this colonisation failure also resulted, at least partly, from a
host immune system failure (termed as “immune colonisation resistance”).
This compositional disruption translated into many negative correlations
between taxa in both selection regimes in the skin-associated microbiota at
T1. Furthermore, the impact of Cd exposure on skin community structure was
also observed at T3, where the phylogenetic distance became significantly
divergent, even between both selection regimes, where many negative
correlations were detected between taxa. In addition to the increasing
invasion of environmental bacterial strains in skin (i.e. failure of colonization
111
resistance, see (Buffie and Pamer 2013a), the rise of negative correlations
suggests a dysbiosis state of skin-associated microbiota (Vázquez-Baeza et
al. 2016). This dysbiosis might be associated not only with an increase in
evenness and phylogenetic convergence with the water bacterial community,
but also with the rise of antagonism among OTU co-abundance networks in
both selection regimes (CC & CV). Most of antagonism was mediated through
rare and abundant Tenericutes (Mycoplasma) and Proteobacteria. Depending
upon the strain, Mycoplasma are thought to be either fish pathogens,
opportunists, or innocuous commensals in fish (R. M. Brown, Wiens, and
Salinas 2018; Holben et al. 2002). In comparison to skin, the significant
divergence between control (Control) and treatments (CC, CV), and the rise
of negative correlations, appeared later in the gut community: at T3. This
delayed pattern of dysbiosis strongly suggests that the physiological impact
of cadmium exposure was mitigated more effectively within the gut. In fish
(and other vertebrates), the liver is the main organ to accumulate xenobiotics
including cadmium. Therefore, the late compositional shift in the gut
microbiota potentially occurred when bioaccumulation of cadmium within the
liver reached its maximum carrying capacity.
3.6.2 Gradual disconnection of abundant taxa from the main gut
interacting network.
Another noticeable compositional shift was the disconnection of abundant
taxa from the main gut interacting network, which was proportional to the
stress intensity (Fig. 3). By T3, the overall taxonomic network connectivity
was formed exclusively from rare OTUs. Contrastingly, abundant OTUs were
peripherals and disconnected from central hubs, mainly composed of putative
opportunistic invaders such as the Mycoplasmas and other genera
encompassing strains associated to fish pathogens, like Bacillus (>6% in CC
and CV). As observed in other fish species such as Atlantic salmon (Salmo
salar) (Llewellyn et al. 2016) and the longjaw mudsucker (Gillichythys
mirabilis) (Bano et al. 2007; Givens et al. 2015), Yellow Perch have intestinal
microflora dominated by Tenericutes (Mycoplasma sp.). It is therefore difficult
to conclude whether the increase of several Mycoplasma strains is beneficial
or not to the host. Concerning Bacillus, a similar increase was associated with
irritable bowel disease (IBD) in dogs and negatively correlated with bacterial
112
strains associated to healthy individuals (Vázquez-Baeza et al. 2016).
Interestingly, rare OTUs and negative correlations did not play an essential
role in water co-abundance networks, which were highly fragmented both in
the control and treatment regimes. The taxonomic composition of water
microbial communities was characterised by the low occurrence of
Tenericutes (Mycoplasma) compared to host communities.
3.6.3 Rare OTUs play a pivotal role in community assembly.
Rare OTUs have been demonstrated to play a pivotal role in community
assembly (Banerjee, Schlaeppi, and Heijden 2018; Fierer 2017; Jousset et al.
2017; M. D. J. Lynch and Neufeld 2015; Pester et al. 2010) either in
promoting homeostasis (Jousset et al., 2017) or dysbiosis (Hajishengallis et
al. 2012). Therefore, we have applied the non-linear least squares model
(NLS) to disentangle neutral and non-neutral evolutionary processes that
were at play in community assembly in both control and treatment groups.
The distribution of neutral OTUs across different relative abundance averages
(Supp. Figure 3.6) suggested a non-neutral role of rare OTUs in water and
host communities. Our data demonstrated that stochastic processes mainly
drove taxonomic drift in cadmium-free water communities (from 53% to 69%
of OTUs), in contrast to what was measured in cadmium-treated water at T3
(6% of OTUs for CV, 26% of OTUs for CC). Contrastingly, host-microbiota
assembly evolved under a deterministic (non-neutral) manner in both
experimental and control groups, although this trend was more salient in both
experimental groups at T3. The percentage of OTUs in the gut and skin that
fit the neutral model was significantly lower than that of non-neutral OTUs.
The assembly of gut microbial communities may have evolved under non-
neutral processes due not only to the cadmium as a disrupting factor in both
experimental groups, but also due to the selection imposed by the host
development, as documented in different fish species raised under normal
environmental conditions (Stephens et al. 2016; Sylvain and Derome 2017).
3.7 Conclusions
In this experimental evolution trial, the extensive involvement of rare taxa
throughout community assembly was highlighted by the alpha and beta
diversities, and more specifically by the pattern of correlational networks.
Furthermore, the rise of negative correlations between taxa was proportional
113
to the cadmium exposure, therefore providing a reliable indicator of
disturbance to the homeostasis of the host bacterial metacommunity. The
gradual taxonomic convergence between water- and skin-associated bacterial
communities across both cadmium exposure groups highlights the loss of the
colonization resistance capacity of the host. This was potentially due to
physiological stress experienced by the host: cadmium bioaccumulation in
Perch's liver has already been documented to disrupt host physiology (Pyle,
Rajotte, and Couture 2005a). Finally, mathematical modeling demonstrated
that stochastic processes such as drift drove taxonomic assembly of
cadmium-free water communities, whereas host-microbiota assembly mainly
evolved under deterministic (non-neutral) processes.
3.8 Perspective
The patterns of neutral and non-neutral assembly in contrasting types of
bacterial communities (i.e. one environmental and two host-associated)
described here provide novel key insights regarding our understanding of
evolutionary forces that are at play in shaping the host microbiota when
facing a sublethal environmental stress. Living organisms are currently facing
unprecedented levels of environmental stressors that impact their capacity to
cope with natural pathogens, essentially by altering their overall immune
defense. Therefore, there is an urgent need to accurately decipher the early
warning signals occurring at the first stages of xenobiotic exposure. By
highlighting the link between a loss of colonization resistance and dysbiosis
within the host (which in turn is known to induce an inflammatory response),
our results will be useful not only for the field of microbial ecology but also
for biomedical research, as dysbiosis of gut microbiome composition has been
shown to result in the onset of various inflammatory diseases such as
diabetes, IBD, Crohn disease, cancer, and obesity (Mathis and Benoist 2012;
Snedeker and Hay 2012; Tilg and Kaser 2011).
114
3.9 Tables
Table 3. 1 Statistics’ summary of Cadmium concentration variation over
time and treatments in water tanks and fish livers.
The average of Cadmium (Cd) concentrations in treatments is higher 3 to 4
folds than in the Control groups (Table 1.1A). The overtime variation of Cd
concentration treatments groups (summarized in Table 1.1B) demonstrated
significant accumulation of Cd in fish livers. In water, at T1 and T3, the Cd
concentration was significantly different between Control and treatments, but
it is not between treatments (CC, CV) at T3. Treatments comparisons against
Table 3.1-a Liver Water
Cadmium
average
concentration
(ng/ml)
T0 T1 T3 T0 T1 T3
Ctrl 0.086 0.099 0.15775 0.065 NA 0.075
CV 0.086 0.09 0.4 0.067 1.54 6.43
CC 0.086 0.107 0.5235 0.098 3.08 5.8
Table 3.1b.
Over time
comparisons
T0-T1 T0-T3 T1-T3 Comments
Tukey's
p.value
Tukey's
p.value
Tukey's
p.value
Ctrl.T0=CV.T0=CC.T0
(no Cadmium
Treatment) Ctrl-Ctrl 0.99899388 0.999999983 0.944720869
CV-CV 0.99808831 0.000348064 4.4923E-08
CC-CC 0.999838113 4.57148E-07 0
Table 3.1c.
Statistics tests
(Tukey, Wilcox)
T1 T3
Liver Water Liver Water
Groups
comparisons
Tukey's
p.value
Tukey's
p.value
Tukey's
p.value
Tukey's p.value
Ctrl-CV 1 0.0059 0.00002 0
Ctrl-CC 1 0.05 0.0009 0
CC-CV 1 0 0.0281 0.1304
Ctrl-CC-CV
(Kruskal−Wallis
Rank Sum
Test)
0.99 0 0.0002 0
115
the Control group showed significant changes in Cd concentration at T1 and
T3 in water tanks and T3 in fish livers and (Table 1.2C). Accumulated Cd
concentrations in water tanks are approximately the same in CC and CV at
T3.
Table 3. 2 Statistical summary of alpha-diversity changes over time
and treatments.
Table 3-2
Community Time T0-T1 T0-T3 T1-T3
Alpha-
diversity
per regime
Richness Evennes
s
Richness Evenness Richness Evenness
adj. p-value (BH) adj. p-value (BH) adj. p-value (BH)
Gut Ctrl 0 0 0 0 0.279 0.5707
CV 0 0 0 0 0.0054 0.5498
CC 0 0 0 0 0.0006 0.3929
Skin Ctrl 0 0.869 0 0.0001 0 0
CV 0 0.0002 0 0 0 0
CC 0 0.002 0 0.0025 0 0
Water Ctrl 0.9502 0.0245 0.4673 0.058 0.4948 0.818
CV 0.5479 1 0.0023 1 0.0012 1
CC 0.4313 0.4008 0.479 0.04 0.0547 0.172
Table 3-b
Community Regime Ctrl-CV Ctrl-CC CC-CV
Alpha-
Diversity/Tim
e
Richness Evennes
s
Richness Evenness Richness Evenness
Corrected p-value
(BH)
Corrected p-value
(BH)
Corrected p-value
(BH)
Gut T3 0.0304 > 0.05 0 > 0.05 0.0354 > 0.05
Skin T1 > 0.05 0.017 > 0.05 0.017 > 0.05 0.7415
Over time, the richness or evenness have significantly changed between
treatments and Control, in all type of communities. At each time point, the
richness in gut and skin microbial communities has changed substantially in
all groups (Control, CC and CV) except in the Control group of the gut at T3
(Table 1.2A). On the other hand, the evenness showed significant changes in
116
all host communities between T0 and T1 except in the Control group of skin,
then between T1 and T3 except all groups of Gut communities (Table 1.2A).
In the water microbial communities, evenness showed significant changes
over time only in the Control group between T0 and T1. The evenness has
significantly changed over time in the water, just in CC and CV between T1
and T3. The significant changes of alpha-diversity indexes difference between
treatments and Control were statistically compared (Table 1.2B) using rank
statistics tests (Kruskal-Wallis/Wilcoxon). The same statistics were used to
compare alpha-diversity overtime. Overall, the evenness in microbial skin
communities has significantly diverged between treatments and Control at
T1. The richness in Gut microbial communities was significantly different
between treatments and Control only at T3.
117
Table 3. 3 Phylogenetic divergence at the community level.
Time Groups
Permanova Multiple test correction Betadisper Mrpp
p value
ADONIS p values B-H
p values
dispersion
p value
MRPP
Ctrl = control regime; CC = concentration is constant; CV = concentration is variable
GUT T0 All groups 0.052 0.079
CC-Ctrl 0.042 0.0765 0.048 0.033
CC-CV 0.051 0.0765 0.141 0.049
Ctrl-CV 0.28 0.28 0.302 0.234
T1 All groups 0.151 0.954
CC-Ctrl 0.124 0.25 0.941 0.114
CC-CV 0.25 0.25 0.843 0.234
Ctrl-CV 0.247 0.25 0.783 0.224
T3 All groups 0.001 0.367
CC-Ctrl 0.001 0.0015 0.916 0.001
CC-CV 0.006 0.006 0.145 0.008
Ctrl-CV 0.001 0.0015 0.295 0.001
SKIN T0 All groups 0.016 0.54
CC-Ctrl 0.166 0.166 0.649 0.154
CC-CV 0.081 0.1215 0.256 0.075
Ctrl-CV 0.026 0.078 0.599 0.02
T1 All groups 0.001 0.156
CC-Ctrl 0 0.003 0.249 0.006
CC-CV 0.085 0.085 0.395 0.079
Ctrl-CV 0.002 0.003 0.07 0.006
T3 All groups 0.008 0.82
CC-Ctrl 0.035 0.049 0.619 0.049
CC-CV 0.049 0.049 0.586 0.045
Ctrl-CV 0.021 0.049 0.872 0.009
Water
T0 All groups 0.001 0.656
CC-Ctrl 0.018 0.027 1 0.009
CC-CV 0.002 0.006 0.291 0.002
Ctrl-CV 0.036 0.036 0.519 0.038
T1 All groups 0.006 0.307
CC-Ctrl 0.009 0.0135 0.851 0.01
CC-CV 0.136 0.136 0.073 0.113
Ctrl-CV 0.002 0.006 0.238 0.001
T3 All groups 0.001 0.036
CC-Ctrl 0.002 0.002 0.068 0.002
CC-CV 0.001 0.0015 0.02 0.001
Ctrl-CV 0.001 0.0015 0.546 0.001
The phylogenetic distances between OTUs were computed using Gunifrac
(distance Chen et al. 2012). The divergence between treatments and Control
was assessed using PERMANOVA and the homogeneity for group dispersions
118
(distance from centroid), was evaluated using two multivariate tests,
BETADISPER and Multi-Response Permutation Procedure (MRRP) of within
versus among group dissimilarities. The significance of divergence between
groups was measured using multiple correction tests with Benjamini-
Hochberg BH (p-value<0.05)). For Gut microbial community, the significant
phylogenetic divergence between all groups was observed at T3. For Skin
microbial community, the significance in phylogenetic divergence was
observed between Control-CV and Control-CC at T1, and between all groups
at T3. For water microbial community, the phylogenetic divergence between
groups was significantly different at each time point except between CC and
CV at T1.
119
3.10 Figures
Figure 3. 1 Linear variations of alpha-diversity over time and
between treatments explained by the linear mixed model in water and host-microbial communities
The richness and evenness variations were considered as response variables
in the linear model with mixed effects. The fixed effects were defined by time
and cadmium concentration (in water and liver), and tanks have a random
120
effect. Using the lme4 package in R, the linear model (M1= lmer
(diversity.index (Evenness or Richness) ~ Cd.Water*Cd.Liver + Time +
(1|Tank) ) demonstrates sTable 1.profiles of alpha-diversity over time in the
Control group compared to treatments in both water and host (skin, gut)
microbial communities. Constant Cadmium regime (CC) is in orange, variable
Cadmium regime (CV) is in Yellow, and Control (Control) is in green.
Figure 3. 2 Phylogenetic divergence at the community level
NMDS (non-metric Multi-Dimensional Scaling) plot of generalized Unifrac
distances showed the distribution of host-microbial samples based on the
121
phylogenetic content of their microbiota. The plot shows a significant
separation of sample groups according to treatment (see in Supp Table 1.5-
b, p-values of the PERMANOVA test indicating the significance of group
separations) at T1 in the skin, and at T3 in skin and gut.
Figure 3. 3 Correlational co-abundance networks of gut microbial
community.
The topology of the gut microbial community network at T3 revealed a
gradual loss of connectivity from between treatments and Control. These
networks are composed of 30, 46, 65 of CC (connected components or hubs),
respectively in Control, CV, and CC groups. Therefore, the nodes number
(NN) was lower in the Control group (ControlNN: 410) compared to Cd groups
networks (CVNN: 488; CCNN: 526). On the contrary, an average of neighbors
(AN) was higher in the Control network (Control AN: 4) than in Cd groups
networks (CVAN: 2.98; CCAN: 2.82). Each node size in the network is
proportional to the average of the OTUs relative abundance in all samples.
The dominant OTUs, Tenericutes (Mycoplasma), Proteobacteria
(Pseudomonas) and, Firmicutes (Bacillus) are peripherals in the network.
Only in the Control group, dominant OTUs were connected to the central hub.
In contrast, in the Cadmium groups, dominant OTUs were disconnected from
the central hub into small independent hubs. In all conditions, the rare taxa
or tiny nodes hold most networks connections. This network was built using
R and Cytoscape software, to overview the complete topological features of
networks see the supplementary Table 1.6.
122
Figure 3. 4 Correlational co-abundance networks of skin microbial community.
In microbial skin networks, each node size in the network is proportional to
the average of the OTUs relative abundance in all samples. At time1, the
proportion of neighbors (AN) was lower in Control network (Control AN: 4)
compared to Cd groups networks (CVAN: 10.7; CCAN: 24.6). The main
difference in CV and CC relative to Control was manifest in the higher
percentage of significant negative correlations observed in those groups (CV
neg. corr.: 6.3%; CC neg. corr.: 6.9%) compared to Control group (Control
neg. corr.: 2.88%). The OTUs of Bacteroides significantly higher in CC and
CV groups were mostly involved in the negative correlations (red edges in
networks). Despite their high relative abundance in all groups, both of rare
and most abundant Tenericutes (Mycoplasma) nodes (OTUs) were involved
in negative interactions in CV and CC groups, but not in the Control group. At
T3, there was a loss of significant interactions compared to T1, which might
be partially related to the decrease of OTU numbers between T1 and T3 (-
123
53.2 % in Control, - 57.7% in CV, and - 54.7 % in CC). In the Control group,
the invasion of Tenericutes (Mycoplasma) was not involved in negative
interactions. Rare OTUs from both phylum Firmicutes and Proteobacteria
represent most nodes in the central hub of the network. At T3, in the CV
group there were any negative correlations, and Tenericutes (Mycoplasma)
OTUs accounted among the rare taxa. In the CC group, negative interactions
represented over 6 % of the total edges of the network. This network was
built using R and Cytoscape software. To overview the complete topological
features of networks, see the supplementary Table 1.6.
Figure 3. 5 Correlational co-abundance networks of water microbial community.
124
Water microbial community networks displayed fragmented interactions, over
time and between treatments — the topology of water microbial encompassed
sub-networks disconnected in small independent hubs. The number of
independent hubs was the highest in Control network at T0 and T3, at T3 in
CC, and it was always intermediate in CV network (see the supplementary
Table 1.3). This network was built using R and Cytoscape software.
Figure 3. 6 Bar plots of neutral OTUs change at community and
meta-community levels.
According to the non-linear least squares model (NLS), the proportion of OTUs
that fit neutrality model within a confidence interval of 95% tended to be
variable between communities across time, and cadmium treatments (Figure
2.6). In gut and skin communities, less than 50% of the total number of OTUs
over time was explained by the neutral model. However, the percentage of
neutral OTUs is higher in the Control treatment compared to the variable and
constant cadmium regimes. In the water communities, the percentage of
neutral OTUs is more than 50% at each time point. For cadmium CV and CC
at T3, the majority of OTUs (>70%) are non-neutral. At the meta-community
level including water and host-communities as a whole system, the
percentage of non-neutral OTUs was always higher than neutral OTUs in
125
treatments and Control. The goodness of fit (R2) was lower than 0.5 in all
comparisons except in Control water communities.
3.11 Supplementary Figures
Supplementary Figure 3.1. Box plots of alpha-diversity variations
over time and between treatments in host and water microbial
communities.
The box plot of richness and evenness variations showed different trends
between treatments and Control. In the gut (left), the alpha-diversity showed
the same tendency in all groups, except at time3. In skin (middle), the
evenness at T1 was higher in Cadmium treatments compared to Control while
the opposite produced for richness at T3. In water (right), the evenness and
richness were intermediate in the Control group compared to variable and
constant Cadmium selection treatments, except the evenness which was the
highest in Control group at T1. Constant Cadmium regime (CC) is in orange,
variable Cadmium regime (CV) is in Yellow, and Control (Control) is in green.
126
Supplementary Figure 3.2. Heatmaps of significant taxonomic
variation at the phylum level.
127
This Figure represented with three heatmaps the significant changes over
time of taxonomic composition at the phylum Level in the gut (2a.), in the
skin (2b.) and water (2c.). The hierarchical clustering of the relative
abundance of phyla which significantly changed over time was performed
using Ward's method and Bray–Curtis dissimilarity distance. Vegan package
and pheatmap () function in R were used to produce these heatmaps.
Supplementary Figure 3.3. Heatmaps of significant taxonomic
variation at the genus level.
128
This Figure represented with three heatmaps the significant overtime changes
of taxonomic composition at the genus Level in GMC (Gut Microbial
Community) (2a.), in SMC (Skin Microbial Community) (2b.) and WMC (Water
Microbial Community) (2c.). The hierarchical clustering of the relative
abundance of phyla which significantly changed over time was performed
using Ward's method and Bray–Curtis dissimilarity distance. Vegan package
and pheatmap () function in R were used to produce these heatmaps.
Supplementary Figure 3.4. Phylogenetic divergence at the
community-level in water microbiome.
NMDS (non-metric Multi-Dimensional
Scaling) plot of generalized Unifrac distances
showing the distribution of host-microbial
samples based on the phylogenetic content of
their microbiota. The plot shows a significant
separation of sample groups according to
treatment (see in Supp. Table 1.5-b, p-values
of the PERMANOVA test indicating the
significance of group separations) at each
time point, but treatments samples are closer
to each other than the Control group at T1
and T3.
129
Supplementary Figure 3.5 Boxplots of Bacteroidetes variation
overtime and between treatments.
At T0, the relative abundance of Bacteroidetes was significantly lower in Skin
compared to water and gut microbial communities. However, at T1,
Bacteroidetes abundance was significantly higher (Wilcox ‘s P-value <0.05)
in Skin communities only in treatments groups (CC and CV), while in Ctrl
group they showed any significant variation.
130
Supplementary Figure 3.6 Distribution of abundance versus
neutrality in the whole metacommunity , host and microbial communities
This Figure showed the distribution variation of neutral OTUs percentage (Y
axis) predicted by NLS models using 12 cutoffs of relative abundance
averages (X axis) in the entire metacommunity at T1 and T3.
131
3.12 Supplementary Material
Supplementary Table 1. ANOVA summary of metals concentrations in
water tanks and Perca flavescens fish livers. Tukey and Wilcoxon’s tests
showed that the Cadmium concentration in fish liver and has significantly
changed between treatments and Control; Constant Cadmium regime (CC),
Variable Cadmium regime (CV), and Control (Control) and over time. No
significant changes were observed between Zinc and Cooper between
treatments.
Supplementary Table 2. Summary of alpha-diversity over time and
treatments. This table summarises averages of richness and evenness in the
host (skin and gut) and water microbial communities at each time point
between treatments and Control groups.
Supplementary Table 3. Statistical summary of co-abundant OTUs in
correlational networks. The percentage of OTUs involved in significant
correlations was relatively higher in water compared to skin and gut microbial
communities’ networks. In gut microbial networks, this percentage decreased
between 78 % in the Control group to 69% in the CV group, and 63% in CC.
In microbial skin networks at T1, the significant part of OTUs varied between
89 %, 90 % and 97% in Control, CV and CC networks. At T3, the percentage
of OTUs in skin microbial communities varied between 56 %, 31 % and 54%
In water microbial communities, this percentage ranged between 46%, 9 %,
and 37% at T0; 31%, 33% and 34% at T1; then 51%, 7%, and 75% at T3,
respectively in Control, CV and CC networks. Per time and treatment, all the
networks nodes (OTUs) average relative abundance are reported in the
second sheet of this table.
Supplementary Table 4. Summary of correlation networks features in host
and water microbial communities. The networks of gut microbial communities
are composed of 30, 46, 65 of CC (connected components or hubs),
respectively in Control, CV, and CC groups (see sheet 1). Therefore, the
nodes number (NN) was lower in the Control group (ControlNN: 410)
compared to Cd regimes networks (CVNN:488; CCNN:526). On the contrary,
an average of neighbours (AN) was higher in the Control network (Control AN:
4) than in Cd groups networks (CVAN:2.98; CCAN:2.82). Skin microbial
132
networks were composed of 47, 10, 3 of CC (connected components),
respectively in Control, CV, and CC groups (see sheet 1). The number of CC
was inversely proportional to nodes number, low in Control group network
(ControlNN: 651) in comparison with Cd groups networks (CV NN: 661; CC
NN:759). Similarly, the average of neighbours (AN) was lower in Control
network (Control AN: 4) compared to Cd groups networks (CVAN: 10.7; CCAN:
24.6). This connectivity higher in CV and CC at T1 was proportional to the
higher percentage of strong negative correlations (R neg. Corr. < -0.6; B-H
p-value < 0.05), observed in those groups (CV neg. corr.: 6.3%; CC neg.
corr.: 6.9%) compared to Control group (Control neg. corr.: 2.88%). Water
microbial networks displayed variables features over time and between
regimes (see sheet 2).
Supplementary Table 5-a. Significant taxonomic changes in water and
host-microbial communities over time and between treatments. In the gut,
the significant changes in the relative abundance between T0 and T1 were
detected for Synergistetes in CC and Tenericutes in CC and CV. Later,
between T1 and T3, the relative abundance of four phyla, Euryarchaeota,
Firmicutes, Proteobacteria, and Tenericutes has significantly changed in all
groups. In skin, between T0 and T1, the significant changes were observed
for Firmicutes and Fibrobacteres in the CV; Bacteroidetes and Fusobacteria in
CC and CV. However, between T1 and T3, the significant changes of relative
abundance concerned Actinobacteria in CC and CV; Proteobacteria,
Tenericutes, Bacteroidetes, and Firmicutes in all groups. In water, between
T0 and T1, the significant changes were observed for Bacteroidetes in
Control; Proteobacteria in the CV; Firmicutes and Tenericutes in all groups.
However, between T1 and T3, the significant changes concerned
Proteobacteria and Tenericutes in CV, and Firmicutes in all groups. Two phyla
showed significant changes at long-term between T0 and T3, Fibrobacteres
in Control, and Actinobacteria in Control and CV.
Supplementary Table 5-b. Temporal taxonomic changes in host and water
communities. The Significant temporal variation of taxonomic composition
involved 24 genera in Gut microbial community, 33 in skin microbial
community and 19 in water microbial community. The corrected p-values of
133
paired Wilcoxon signed rank sum test are summarized in Supplementary
Table 5-a.
Supplementary Table 6. Features of water and host-microbial communities’
correlational networks per time and treatment. This table is composed of
multiple tables; each network features are represented in an independent
table. The most important features are the degree (connectivity) of nodes,
closeness centrality, betweenness centrality and their size proportional to
their average relative abundance in all samples.
Supplementary File 1. A brief description of the sample groups of host
microbiota and water.
Supplementary file 2. Materials and Methods (supplementary information)
Supplementary file 3. DNA extraction from TRIZOL from the organic phase
after RNA extraction
134
Chapitre 4: Community recovery dynamics in
yellow perch microbiome after gradual and
constant metallic perturbations
Cheaib, B., Seghouani, H., Ijaz, U.Z. et al. Community recovery dynamics in
yellow perch microbiome after gradual and constant metallic perturbations.
Microbiome 8, 14 (2020). https://doi.org/10.1186/s40168-020-0789-0
135
4.1 Resumé
Contexte : Les processus éco-évolutifs régissant l'assemblage microbien
post-perturbation restent mal étudiés, en particulier dans les systèmes hôte-
microbiome. La résilience de la structure d’une communauté microbienne
dépend non seulement du type, de la durée, de l'intensité et du gradient de
perturbation, mais également de la structure initiale de la communauté, et
du type de l'habitat. Dans cette étude, les microbiotes cutané et intestinal de
la Perchaude (Perca flavescens), ainsi que de l’eau ont été caractérisés avant,
pendant (90 jours) et après (60 jours) l’exposition au cadmium. Les
microbiotes de la peau, des intestins et des réservoirs d'eau dans les groupes
de contrôle et de traitement ont été caractérisés avant, pendant et après
l'exposition au cadmium en utilisant des librairies d'ADNr 16s et une
technologie de séquençage à haut débit (Illumina, Miseq).
Résultats : Les mesures spectrométriques ont révélé une augmentation de
bioaccumulation de cadmium dans le foie, même deux mois après l'arrêt de
l’exposition au cadmium. Le gradient de sélection montre des effets
différentiels sur la résilience, notamment par l’augmentation de diversité
alpha, et des transitions taxonomiques, ainsi qu'une divergence significative
de la diversité fonctionnelle et phylogénétique. La résilience des
communautés cutanées a montré une prolifération importante d'agents
pathogènes opportunistes du poisson (genre Flavobacterium). La neutralité
montre plus d’impact sur les communautés du groupe contrôle que ceux
exposés au cadmium. Le microbiome de la perche a atteint un état stable
alternatif dans la peau et une trajectoire de récupération presque complète
dans les communautés microbiennes intestinales.
Conclusions :
La perturbation métallique à court terme du développement des poissons
montre des répercussions à long terme pour la santé de l'hôte. La résilience
des communautés microbiennes après une exposition aux métaux dépend de
l'ampleur de l'exposition (constante, progressive) et de la nature de la niche
écologique (eau, peau et intestin). Dans l'ensemble, nos résultats montrent
que l'assemblage microbien pendant la résilience de du microbiote a été à la
fois orchestré par des processus neutres et déterministes.
136
4.2 Abstract
Background: The eco-evolutionary processes ruling post-disturbance
microbial assembly remain poorly studied, particularly in host-microbiome
systems. The community recovery depends not only on the type, duration,
intensity and gradient of disturbance, but also on the initial community
structure, phylogenetic composition, legacy, and habitat (soil, water, host).
In this study, yellow perch (Perca flavescens) juveniles were exposed over 90
days to constant and gradual sublethal doses of cadmium chloride.
Afterwards, the exposure of aquaria tank system to cadmium was ceased for
60 days. The skin, gut and water tank microbiomes in control and treatment
groups, were characterized before, during and after the cadmium exposure
using 16s rDNA libraries and high throughput sequencing technology
(Illumina, Miseq).
Results: Our data exhibited a long-term bioaccumulation of cadmium salts
in liver even after two months since ceasing the exposure. The gradient of
cadmium disturbance had differential effects on the perch microbiota
recovery, including increases in evenness, taxonomic composition shifts, as
well as functional and phylogenetic divergence. The perch microbiome
reached an alternative stable state in the skin and a nearly complete recovery
trajectories in the gut communities. The recovery of skin communities showed
a significant proliferation of opportunistic fish pathogens (i.e.
Flavobacterium). Our findings provide evidence that neutral processes were
a much more significant contributor to microbial community turnover in
control treatments than in those treated with cadmium, suggesting the role
of selective processes in driving community recovery.
Conclusions: The short-term metallic disturbance of fish development has
important long-term implications for host health. The recovery of microbial
communities after metallic exposure depends on the magnitude of exposure
(constant, gradual), and the nature of the ecological niche (water, skin, and
gut). The skin and gut microbiota of fish exposed to constant concentrations
of cadmium (CC) were closer to the control negative than those exposed to
the gradual concentrations (CV). Overall, our results show that the microbial
assembly during the community recovery were both orchestrated by neutral
and deterministic processes.
137
4.3 Introduction
Resilience refers to the capacity of a natural ecosystem to maintain a stable
state after facing different exogenous disturbances, both in terms of
amplitude and frequency (Holling 1973).Introduced first by Holling (1973),
the concept of resilience was redefined to incorporate the idea of recovery
following a temporary disruption (Pimm 1984; V. Grimm and Wissel 1997),
not simply the ability to resist this disturbance in the first place (Hodgson,
McDonald, and Hosken 2015). Both ecological concepts, ‘resistance’ and
‘recovery’, were simultaneously considered as measurable components that
together represent resilience (Hodgson, McDonald, and Hosken 2015). In
other microbial studies, the term ‘resistance’ is synonymous with
resilience(Ziegler et al. 2017) using Holling‘s definition. Notwithstanding,
‘sensitivity’ (inverse of resistance) is also sometimes used to represent the
degree to which a community changes in response to disturbance (Shade et
al. 2012). The recovery rate, time to reach an equilibrium state, and the
distance to an alternate stable state are quantitative measures that can be
used to compare the resilience (Cabrol et al. 2016; Hodgson, McDonald, and
Hosken 2015; Ingrisch and Bahn 2018; Shade et al. 2012) and improve our
understanding of ecosystem recovery (Scheffer et al. 2015; Ziegler et al.
2017). In this study we will employ the term ‘recovery’ to describe the pattern
of eco-evolutionary change that occurs when a community returns to an
alternative stable state.
The recovery of microbial communities depends on the type, duration,
intensity and variability of a disturbance. More importantly, microbial
recovery can be impacted by the initial community structure, phylogenetic
composition, legacy, and the type of habitat (soil, water, host). After
antibiotic treatment, the complete recovery of initial bacterial community
composition is rarely achieved, as reported in various host-microbiota
systems from honeybees (Raymann, Shaffer, and Moran 2017) to humans
(Dethlefsen and Relman 2011). The incomplete recovery of gut microbiota
ecosystems after antibiotic administration results in a shift of the microbial
composition to an alternative equilibrium called an “alternate stable state”
(Shade et al. 2012; C. A. Lozupone et al. 2012; F. Sommer et al. 2017). This
compositional shift occurs when resistance or recovery is weak, and/or when
138
the intensity of disturbance is high. Although the understanding of factors
that drive such regime shifts to an alternative equilibrium in microbial
ecosystems will have tremendous impacts in various fields of application (e.g.
personalized medicine, agriculture, bioremediation), this phenomenon is still
poorly studied.
The relative roles of ecological and evolutionary processes in the recovery of
the structure of microbial communities are still to decipher. Theoretically, the
nature of these processes can be neutral (stochastic) (Hubbell Stephen P.
2005; F. Sommer et al. 2017), or selective (deterministic) (Chase 2003;
Sloan et al. 2006), the latter being driven either by environmental filtering or
competitive exclusion (Cadotte et al. 2010; Webb et al. 2002), the former by
demographic sampling effects alone. In the context of community recovery,
a small number of studies revealed that deterministic processes drive
bacterial succession dynamics in a soil bacterial community disrupted either
by a depletion gradient of nutrients (Song et al. 2015), a thermal shock
(Jurburg et al. 2017),or a rainfall rehydration of dry soil (Placella, Brodie, and
Firestone 2012).
In the present study, we assessed the relative contribution of neutral and
deterministic processes in the recovery of the yellow perch (Perca flavescens)
microbiome assembly following an experimental metallic exposure gradient.
Polymetallic contamination in aquatic ecosystems mostly results from
exposure to acid mine drainages (AMD) occurring around the world (Wendt-
Potthoff and Koschorreck 2002; F. Wang et al. 2003; Gough et al. 2008;
Hudson-Edwards, Jamieson, and Lottermoser 2011; Moser and Weisse 2011;
Urbieta et al. 2012; Douglas et al. 2012; Masmoudi et al. 2013; Stankovic et
al. 2014; Valente et al. 2015; Cheaib et al. 2018). For instance, in the natural
Canadian lakes, the cadmium (Cd) concentration reaches 9 ppb (parts per
billion) in perch liver/water (Pyle, Rajotte, and Couture 2005b; Couture,
Rajotte, and Pyle 2008), and it has a clear quantitative impact on the perch
physiology, gene expression, and genotype diversity (Bougas et al. 2013b).
In the same polluted lake system studied by Couture et al. (2008), the
microbial communities’ assembly in water has evolved under chronic
exposure to a gradient of trace metals due to the AMD expelled in the
environment, leaving substantial genotypic signatures of adaption in the
139
taxonomic and functional repertories of AMD communities (Cheaib et al.
2018). Given that yellow perch juveniles can tolerate sublethal doses of
cadmium without encountering significant physiological damage or death
(Giguère et al. 2004; Campbell et al. 2005b), this host-microbiota model
system is well suited to study microbiota recovery following metal exposure
stress. In the laboratory, yellow perch juveniles underwent exposure to
sublethal doses of cadmium chloride (CdCl2), the accumulation of which was
tested in the water and within perch liver. The recovery of community
structure and function in water and host microbiome were then studied and
compared between constant and variable regimes of metallic stress, which
was defined by the levels of Cd detected in liver and water samples. To
distangle the effect of the xenobiotic from host development (Sylvain and
Derome 2017; Burns et al. 2016b) on bacterial strain recruitment ontogeny,
microbiota assembly was also assessed in stable conditions as a control
regime. Our expectation was that constant exposure to cadmium chloride,
due to its severe implications for host and microbial community physiology,
would impede community recovery most severely than in the gradual
exposure experimental group.
4.4 Methods
4.4.1 Fish rearing
The experiment is described in Figure 4.1. Briefly, there were two acclimation
periods: one in a standard container (1500L) and the second in 24 tanks (36
L) with an independent filtering system circuit for each aquarium. The fish
juveniles were reared within the same physicochemical conditions
(photoperiod, pH, ammonia, nitrogen dioxide). Throughout the experimental
period, to maintain viable conditions for perch in each water tank, fecal and
uneaten food particles were removed daily using specific pressing tubes for
each set of experimental conditions. A volume of 15 L of water was renewed
two times a week for each tank (Figure 4.1).
140
4.4.2 Exposure regimes to cadmium
Cd-treated and Control (Ctrl) tanks were designed into two cadmium chloride
exposure regimes (8 tanks per regime), and one negative control regime (8
tanks) (Schema 1). The yellow perch in treated tanks were exposed to
cadmium chloride (CdCl2) dissolved in water. Under the regime of constant
CdCl2 concentration exposure (CC), the cadmium chloride was initially added
at 0.8 ppb, then increased to reach a target theoretical concentration of 9
ppb (parts per billion) by the end of the first month (T1). The CdCl2
concentration was adjusted to 9 ppb every five days during two additional
months until the end of treatment (third month, T3) where the measured
concentration reached an average of 5.8 ppb. Under the regime of variable
CdCl2 concentration (CV), the CdCl2 was initially added at 0.6 ppb, then the
concentration was gradually increased every five days to meet the target
theoretical concentration of 9 ppb by the end of the third month. The
measured concentration reached an average of 6.8 ppb at the end of
treatment (third month, T3). The maximal CdCl2 concentration was settled at
9 g/mL, which was within range of concentrations detected in yellow perch
liver in contaminated Canadian lakes (Pyle, Rajotte, and Couture 2005b;
Couture, Rajotte, and Pyle 2008).
4.4.3 Recovery after the exposure to Cadmium
The cadmium administration was stopped after the third month (T3). The
experiment was extended two months (T5) after T3 to test the recovery of
microbiome assembly in water and host.
4.4.4 Host-microbiota and water sampling
Briefly, we selected 144 mucosa samples of skin (2 times x 3 regimes x 8
tanks x 3 replicates) and 144 gut (2 times x 3 regimes x 8 tanks x 3
replicates) samples corresponding to T0 (no cadmium) and T3 (ultima
cadmium treatment). Also, 48 water samples (2 times x 3 regimes x 8 tanks
x 1 technical replicate) from T0 and T3 were included. At the end of recovery
time (T5), 72 skin mucosa samples (1-time x 3 regimes x 8 tanks x 3
replicates) and 72 gut samples (1-time x 3 regimes x 8 tanks x 3 replicates)
were collected from the host. In addition, 240 samples (5 times x 3 regimes
x 8 tanks x 2 technical replicates) of water (2 L) microbial filter (0.22 μm)
141
were sampled between T3 and T5 at an interval time of 15 days corresponding
to five recovery time points (TR1, TR2, TR3, TR4, and T5).
4.4.5 Metal concentration in water and fish liver
Every week until the end of the CdCl2 exposure regimes, we measured the
concentration of trace metals of cadmium (Cd), zinc (Zn) and copper (Cu) in
the yellow perch liver and water tanks using the ICPMS (Ionization Coupled
Mass spectrometry) technology available at INRS (Institut National de la
Recherche Scientifique). For further details of the measurement of Cd in liver
preceded by acid digestion and lyophilisation see our under review study
Cheaib et al. (2019). Two-way analysis of variance (ANOVA), Tukey’s test,
and Wilcoxon rank test were applied to test the significance of cadmium
accumulation in liver and water over time, and between treatment groups.
4.4.6 DNA extraction, libraries preparation and 16S amplicons
sequencing
DNA was extracted using the Qiagen DNeasy blood and tissue kit for skin
mucosa, and TRIzol organic phase followed by BEB (back extraction buffer)
and PCI (phenol/chloroform/isoamyl alcohol 25:24:1) solutions for all gut
samples. The V3-V4 hypervariable region of the universal rDNA 16S gene
(Werner et al. 2012)was amplified using universal specific primers. The
libraries of amplicons were prepared using a set of 384 combinations of
adaptors, processed in one sequencing run, on an Illumina Miseq sequencing
machine. Reactions of PCR were verified by electrophoresis on 2% agarose
gel, purified and quantified by fluorescence for the double strand DNA
concentration using Quant-iT™ PicoGreen™ dsDNA Assay Kit (Thermo Fischer
Scientific).
4.4.7 Bioinformatics and biostatistics analyses
4.4.7.1 Reads preprocessing and OTUs clustering
Sequence analysis was performed with our bioinformatic pipeline as described
previously (Sylvain et al. 2016; Llewellyn et al. 2016). In the first instance,
we used SICKLE Version 1.2 to trim the reads (> Q30 Phred quality score)
followed by utilizing PANDASEQ Version 2.11 (Masella et al. 2012) assembler
for merging paired-end read into a single merged reads ( ~ 350 bp)
142
corresponding to the amplified 16S rRNA V3-V4 hypervariable region (347 F-
805 R). Based on an approach of De novo sequence clustering before the
taxonomic assignment, reads were clustered into OTUs at 97% identity with
USEARCH Version 9 (Edgar RC. 2010) and filtered out using UNOISE2
algorithm (Edgar 2016) to discard chimeric sequences, putatively produced
during PCR amplification cycles using OTUs were annotated using RDP
database as previously described in our pipeline (Sylvain et al. 2016;
Llewellyn et al. 2016). to. Community structure and composition of
metacommunities were analyzed across time and treatments by richness
(OTUs count), evenness (Shannon index) and the Gunifrac phylogenetic
distance (J. Chen et al. 2012) using the vegan (Oksanen et al. 2016) and
Rhea (Lagkouvardos et al. 2017) packages in R.
We then calculated the alpha-diversity indices (richness and evenness) and
beta-diversity (phylogenetic distance) differences between experimental
groups and used rank statistics tests (Kruskal-Wallis/Wilcoxon) to assess
their significance. The resulting p-values for pairwise comparisons in alpha
and beta-diversity were corrected for multiple testing using the Benjamini-
Hochberg method (Benjamini and Hoecheberg, 1995). Note that the Beta-
diversity was calculated using the generalized UniFrac metric (W. Chen et al.
2013), which considers both the dominant and the rare OTUs. The
permutational multivariate analysis of variance (henceforth referred to as
PERMANOVA) was applied to the Gunifrac distance matrices to explain the
sources of variations including the experimental groups. To test homogeneity
of variances, we performed the multivariate homogeneity test which a Multi-
Response Permutation Procedure (MRRP) of within versus among group
dissimilarities dispersions of Gunifrac distances. The Non-Metric Multi-
Dimensional Scaling (NMDS) was performed to visualize Gunifrac distances
in a reduced space with k=2 dimensions. For statistics comparison of one-
dimensional statistics of multiple groups, we used the non-parametric
Kruskal-Wallis Rank Sum test because of the strong assumption of the normal
distribution of OTUs abundance being rarely assumed.
The alpha-diversity variation across time and per treatment was predicted
and ploted with linear mixed effect models using the ratio of
richness/evenness as a response variable, time and cadmium concentrations
143
in water and liver as fixed effects, with the categorical variable tank as a
random effect.
Using the lmer R package, for water the model was used as following in R;
Model=lmer (Richness/Shannon.effective~Time+Cd.Water+(1|Tank),
data=mixdata, REML = TRUE), whereas for each host habitat (Skin, Gut),
we employed the following model;
Model_host=lmer(Richness/Shannon.effective~Time+Cd.Liver+Cd.Water+(
1|Tank), data=mixdata, REML = TRUE).
The confidence interval was then predicted using the predictInterval()
function in R.
4.4.7.2 Post-OTUs analysis, networks and function prediction.
Structure and diversity measures of groups (control & treatments) were
compared with rank statistics tests (Kruskal-Wallis/Wilcoxon) adjusted with
BH (Benjamin-Hochberg) test for multiple corrections, and the p-value <0.05
as a threshold of statistical significance. To understand the role of relative
abundance of OTUs on the similarities of community structure, correlation
networks of communities (samples) were constructed using the Spearman
coefficient as a robust approach of correlation detection (Weiss et al. 2016).
Significant positive and negative correlations were filtered and false discovery
rate (FDR) was assessed with B-H test for multiple corrections. Next, network
visualisation and analysis were performed with Cytoscape software (Shannon
et al. 2003).The network centrality was analysed using the “Network
Analyzer” plugin in Cytoscape. The betweenness centrality of a node was
calculated as the total number of the shortest paths from all nodes to all other
nodes that pass this node (Röttjers and Faust 2018). The centrality of the
nodes reflected their importance in transmitting information between hubs;
it does not depend on the feature of node degree, which describes the total
node connectivity. The size of nodes was proportional to the number of OTUs
in each sample, and the coefficient of the significant correlation between two
nodes was inversely proportional to the size of edge. Finally, functional
profiles of each community type at every time point were predicted using the
software TaxforFun (Aßhauer et al. 2015).
144
4.5 Neutral and deterministic models to asses the recovery of
community assembly.
In the null hypothesis, the neutral model (Sloan et al. 2006) (Sloan et al.
2006) assumes a beta distribution of OTU abundance. Using the non-linear
partial least square method (Burns et al. 2016b), which estimates the
migration rate (m) of OTUs from their source to a destination community, the
model predicts the frequencies of OTUs. The estimated migration rate (m) is
the probability that a random loss (death or emigration) of an OTU in a
destination community is replaced by dispersal from the source community.
Comparing the predicted versus observed frequencies, we can determine
which OTU fits the model in each host and water community, at every time
point, across both control and treatments groups. The goodness of fit to the
model was measured using the coefficient of determination R-square (R2
>0.5) within a confidence interval of 95%, where increased strength of
goodness of fit to the model suggests an essential role of stochastic processes
in the microbiome assembly.
4.6 Results
4.6.1 Cadmium concentration bioaccumulation in the fish liver
during recovery time
Interestingly, the concentration of cadmium ions measured with ICPMS
increased significantly in the fish liver even after two months from stopping
exposure. The Cd concentration increased from 0.4 ppb to 1 ppb in the
variable CdCl2 regime (CV) and from 0.5 ppb to 1.17 ppb in the constant
CdCl2 regime (CC). However, in the water, as expected, the Cd concentration
significantly decreased from 6.4 ppb to 1.06 ppb in CV, and from 5.8 ppb to
1.34 ppb in CC (Table 4.1a-b). Consequently, the accumulation of Cd in liver
and water was always significantly higher in treatments CC and CV compared
to the control group (Table 4.1-c). Similar Cd concentrations observed among
treatment groups CC and CV in the water at times T3 and T5 (expected at
maximum Cd concentration added in tanks) was observed in fish liver only at
time T5 (Table 4.1-c).
145
4.6.2 Genotypic signatures of community recovery
At the alpha-diversity level, to investigate how far diversity metrics could be
used as indicators for metacommunity structure recovery, both richness and
the evenness were calculated in water and host-microbial communities. In
the host microbiome, time had a significant effect on diversity measures
within all groups between times T3 and T5. The richness and evenness have
significantly increased overtime in skin microbiota, and significantly
decreased in the gut microbiota. Over the five recovery time points (TR1,
TR2, TR3, TR4, T5), temporal comparisons in the water microbial
communities associated with each experimental group did not show a
significant change of evenness for CC and CV, but did during TR2 – TR4 within
the control group (Ctrl). Within these communities, a significant change in
richness during the whole recovery period except TR2-TR3 was also found for
CC and CV (Additional Table 4.S1a-b).
In contrast, both richness and evenness in the control group of skin
microbiota significantly fluctuated over the recovery period (T3-T5, after
CdCl2 addition had stopped). At time T3, the pairwise comparison of CC and
CV against the control group (CC-Ctrl and CV-Ctrl) revealed significant
differences in microbial richness in the gut and evenness in the skin. At time
T5, statistical tests did not detect any significant change in diversity measures
between all groups for gut and skin microbiome (Additional Table 4.S1-c);
however, as at T3, the evenness of skin microbiome at T5 was significantly
divergent between cadmium treatments, (p-value = 0.0063) (Additional
Table 4.S1-c).
The comparative analysis of richness and evenness among water and host
microbiota showed convergent patterns of diversity between the water and
the skin communities before the disturbance and after the recovery
(Additional file 4.1: Figures 4.S1).
The predicted alpha-diversity values along with the fitted linear mixed effect
model for water communities showed a significant drop in treatment values
compared to the control group. On the other hand, for the host communities
(skin and gut), they increased under the selection regimes and decreased
during the recovery time span (Figure 4.2).
146
In summary, except for the linear mixed effect model results, the observed
patterns of alpha diversity metrics changes across the experiment did not
show a clear trend over the course of the experiment. Nonetheless, increasing
evenness and richness was a general trend for the skin, while decreasing
evenness and increasing richness was representative of the gut microbiome
community recovery.
Beta diversity (Gunifrac) between samples was compared using a
PERMANOVA and a multivariate test for variance homoscedasticity. By T3 –
at peak cadmium exposure - significant differences (p < 0.05) among
treatments were observed in all microbial communities of water and host
(Table 4.2 ; Additional file 4.2 : Figures 4.S2); and by T5, both variable (CV)
and constant (CC) cadmium exposure treatments retained differences in skin
communities compared to the controls despite the recovery period.
Surprisingly, given our expectation that cadmium exposure would have a
major impact on community recovery, a high similarity in the community
phylogenetic structure between the control and CC groups was detected
among gut microbial communities at T5. Beta-diversity between treatments
(CC, CV, Ctrl) was always significantly divergent at each time point in water
samples except for the observed convergence between CC and CV at recovery
time TR2 (Table 4.2; Additional file 4.2: Figures 4.S2). The comparison of
beta-diversity showed a community structure divergence (p-value<0.001)
between water, skin and the gut microbiota before the disturbance and after
the recovery (Additional file 4.3: Figures 4.S3). The results show that water
microbiome at time T3 is not representative enough of fish microbiome (see
the blue cluster in the phylogram of CC, page 2 of figures 4.S3). However, at
the recovery time T5, the water was not representative of the fish microbiome
in gradual selection regime (see the blue Cluster in the phylogram of CV, page
3 of figures 4.S3).
4.6.3 Microbial taxonomic composition change during recovery
At T5, no significant changes were observed between groups at the phyla
level in the water, but Actinobacteria in the gut, and both Euryarchaeota and
Tenericutes in the skin, were significantly different between control and
treatments (CC & CV). Additional Table 4.S2 details several taxa that showed
significant differential abundance between treatments (Ctrl, CV, CC). Of
147
particular importance putative pathogenic genus Flavobacterium was
significantly enriched in the skin for both groups of CdCl2 exposed fish at T5,
despite the recovery period. In the gut microbiome, Syntrophococcus was the
only genus to be significantly different between treatments (Figure 4.3;
Additional Table 4.S2). In the water, significant differences in taxonomic
abundance between CV and Ctrl were restricted to one genus (Kiloniella) at
recovery time TR1 and two genera (Marinobacter and Perlucidibaca) at T5.
No significant differences in taxonomic composition were detected between
CC and Ctrl in the water at T5. Overall, statistical analysis of taxonomic
composition dynamics over time within each treatment during the recovery
period revealed several minor differences (see Additional Tables 4.S3 for
more details).
On the other hand, the pairwise comparison of taxonomic composition
between different type of communities at each time point and for each
experimental group showed significant divergence between microbial
communities of gut, skin and water. At the recovery time (T5), Tenericutes,
Euryarchaeota, and Firmicutes were inherently associated (significantly up
regulated) with the gut microbiome; Actinobacteria and Bacteroidetes, on the
other hand were specific to skin microbiome; with Fibrobacteres and
Actinobacteria implicated with the water microbiome. Despite all this, the
Proteobacteria were found to be prevalent and common in water and skin
microbiome. At the selection time (T3), Fibrobacteria and Actinobacteria were
scarcely abundant and were picked up as differentially abundant (Additional
file 4.6: Figures 4.S6). To delineate the most relevant taxa (at the genus
level) significantly changing between the communities, we have performed a
pairwise test on an overall comparison of skin, gut and water at each time
point for different treatment groups. The results plotted in heatmaps in the
figure 4 clearly reveal that each community type has an inherent signature
and the corresponding proportions of genera differed between control and
treatments over time with high similarity between the CC and Ctrl at time T5
(Figure 4.4; Supplementary Tables 4.4).
Correlational analysis (Additional file 4.4: Figures 4.S4) revealed a positive
relationship between specific genera and concentrations of cadmium in perch
liver and water. In aquaria treated with CdCl2 (CV &CC), cadmium
148
concentrations in water and liver showed strong significant positive
correlations with seven genera from gut microbiome, each of which
represented a different phylum and had a strong negative correlation with
relative abundance of Mycoplasma. In the skin microbiome of both CC and
CV, 15 genera (Sphingomonas; Haloarcula; Legionella; Flavobacterium;
Ameyamaea; Dokdonella; Shigella; Massilia; Mycoplasma; Polaromonas;
Pseudomonas; Rhodobacter; Rhodococcus ; Shewanella ; Syntrophococcus)
showed significant profiles of positive or negative correlations with the Cd
concentrations in liver (Additional file 4.4: Figures 4.S4). Divergent profiles
between CC and CV were only observed for the correlations of Shewanella
and Syntrophococcus with Cadmium concentrations. Similar correlation
profiles between these groups were observed in the water (Additional file 4.5:
Figures 4.S5).
4.6.4 Correlational networks of host and water microbiome
In the host and water communities, the network analysis of samples
correlations showed a partitioned community distribution between treatments
groups, at time T3, and overlapping patterns during recovery period (Figure
4.6; Figure 4.7). At time T0, the correlational networks of host microbiome
showed unstructured topology with on average a fewer number of edges. An
edge can represents significant low correlations (Spearman’s Rho correlation
> 0.5) between samples from different groups with the node size proportional
to the richness of each sample. The topological distribution of nodes in the
network were further analyzed by comparing the betweenness centrality to
the eigenvalue centrality (Figure 4.8). The results indicate a shift in mean of
the centrality metrics between the control (which is higher) and the cadmium
selection regimes. The plots of eigenvalue centrality versus betweenness
centrality clearly reveal that these communities shift at times T3 for skin
microbiome, as well as at time T3 for Gut microbiome. The high betweenness
centrality observed in control reflects the efficiency of network centrality
measure to predict the effect of perturbation on the community structure
during the selection phase, but not during the recovery time as centrality
median shifts at T5 was not observed (Figure 4.8). The same centrality
analysis was obtained for water microbiome networks resulting in similar
patterns at time T3 (results not shown).
149
4.6.5 Recovery of microbial functional diversity a time T5.
At time T0, an ANOVA of functional richness within the metacommunity
showed a significantly higher average of functional diversity in gut and skin
microbiomes compared to water microbial communities. Surprisingly stable
in water communities, functional diversity did not show any significant
divergence among treatment and control group at T3 regardless of the
community type (skin, water, gut). The lack of treatment effect observed may
well have been masked by the strong influence of time over microbial
diversity (Cheaib et al. 2019 under review). However, at T5, the functional
diversity of skin microbiome was significantly higher in the control group than
in treatment groups according to the ANOVA (CC-CV(p-value) = 0.04; CV-
Ctrl(p-value) = 0.0055; CC-Ctrl(p-value) = 0.45) (Figure 4.5). In the gut
microbiota, no significant changes in functional diversity were detected
between treatments (CC-CV(p-value) = 0.3; CV-Ctrl (p-value) = 0.54; CC-
Ctrl (p-value) = 0.58).
4.6.6 The role of neutral and deterministic processes in the
recovery of host microbiota
The goodness of fit of host and water microbial communities to the non-linear
partial least square model (NLS) was high (R-square > 0.5), supporting the
theory of predominant neutrality (Additional Table 4.S5). To disentangle gut
and skin microbiota ontogeny from the cadmium effect, the NLS model was
deployed using the control as a reference. A comparison of observed versus
predicted OTU frequencies revealed that the percentage of neutral OTUs in
skin and gut microbiota (Figure 4.9) at the recovery time T5 is higher in
control group compared to those in treatments at T3 and T5. The same
analysis was undertaken in the water and the percentage of neutral versus
non-neutral OTUs showed the same trends across Ctrl, CC and CV at T5.
Overall, we noted a preponderance of OTUs that fitted the neutral model in
all comparisons. Majority of the OTUs that did not fit the neutral model were
assigned to Mycoplasma species (indeed no Mycoplasma sp. OTUs fitted the
neutral model), which can be seen in Figure 4.10 as well as Table 4.S6 and
Table 4.S7. The neutral process was much more prevalent in control group at
time T3 and T5 as compared to the treatment groups.
150
4.7 Discussion
Our data clearly show that long term bioaccumulation of cadmium occurs in
the in the Perca flavescens liver on exposure to aqueous cadmium salts. Our
data also showed that the cadmium persists at high concentrations even once
the treatment has been stopped for two months. We have already shown that
cadmium treatment clearly impacts both the skin and gut microbial
communities, as compared to controls (Cheaib et al 2019, under review).
Recovery was the focus of the current study, and microbial communities post-
exposure showed different routes to (and extents of) recovery in those
associated with the skin and gut once cadmium treatment was ceased. In the
skin, evenness – the extent to which different microbes in a community share
similar abundances – and richness increased during the recovery phase in
cadmium-treated fish. Beta-diversity comparisons, meanwhile, revealed
significant differences between all experimental cohorts (Ctrl, CC, CV) in
water and skin niches. Among gut microbial communities, decreasing richness
and increasing evenness was observed over the recovery period. Beta-
diversity metrics indicated few significant differences between cadmium and
control treatments. Crucially, by the end of the recovery period in the gut,
functional richness was comparable between tests and control, a potential
signal of full community recovery. We used models to assess the relative roles
of microbial assembly in the different groups. We found evidence that neutral
processes were a more prevalent contributor to microbial community
turnover in control treatments than in those treated with CdCl2 – likely
indicating the role of selective processes in driving community recovery.
Overall our data do not strongly support our prediction that the most extreme
cadmium exposure (CC) would lead to the least successful recovery. Instead,
CC and CV treatments, especially in the gut, demonstrated a good degree of
recovery, both in terms of both alpha and functional diversity.
At the end of the third month of exposure (T3), the cadmium concentration
in the liver was significantly higher in CC and CV than in the control group.
These concentration differences were still observed two months (T5) after the
gradual clearing of Cd started. The liver plays a major role in the
accumulation, excretion and biotransformation of contaminants like
metalloids (Silver and Phung 1996; M. A. Defo, Spear, and Couture 2014),
151
and bioaccumulated metals remain at high concentrations in the liver due to
its depuration function of other organs (such as gills and muscles) (Jezierska
and Witeska 2006). Long-term bioaccumulation of cadmium has been
documented in perch and other biological systems (Campbell et al. 2005b;
Giguère et al. 2006; Klinck et al. 2007; Xie et al. 2008; Nirola et al. 2016),
as has its effect on ecosystem services in soil and water (Cheaib et al. 2018;
Y. P. Chen et al. 2014) and metazoan gut ecosystems (Liu et al. 2014; S.
Zhang et al. 2015; Ba Qian et al. 2017; Šrut et al. 2019).This study not only
confirms the chronic bioaccumulative effect of Cd but also suggests that the
sequestered Cd in perch liver presumably cannot predict the regime of
exposure (CC, CV), as the concentration did not significantly vary in livers
between both regimes, CC and CV, at T5. Over the recovery period, the
concentration of Cd in water significantly decreased, but Cd was not
completely removed from the tank system since it has a strong affinity to the
tank silicon gaskets, and has a high competitiveness with Zn for the debris
of organic compounds always available in the water aquarium ecosystem
(Pinter and Stillman 2015).
The water microbial communities showed few differential abundances of taxa
differences during the recovery period (Figure 4.3). Furthermore, the
microbial functional diversity in water remained stable throughout the
experiment, and no significant differences between treatments were found
during the exposure or the recovery periods. However, the community beta-
diversity at the phylogenetical level between treatments (CC, CV, Ctrl)
showed significant difference at each time point, suggesting a pattern of
taxon-function decoupling as an adaptive strategy reported previously in
lacustrine water contaminated with cadmium (Cheaib et al. 2018).
To assess the yellow perch microbiome recovery, we examined alpha-
diversity (richness and evenness), beta-diversity (phylogenetic distance),
taxonomic composition and functional diversity (metabolic functions). Most of
these measurements are commonly used as community-wide metrics to
assess the recovery of microbial communities, for example in humans (Singh
et al. 2015; Dethlefsen and Relman 2011), soil(Jurburg et al. 2017; Griffiths
and Philippot 2013) and wastewater(De Vrieze et al. 2017) (Vrieze et al
2017).
152
In the skin microbiome, the disturbance intensity (cadmium gradient) had a
differential impact on the community recovery trajectories, resulting in a
significant difference of evenness (Additional Table S1c ) and functional
diversity (Figure 4.5) between CC & CV at time T5 (Table2-C). During the
gradual exposure regime (CV), the cadmium may provoke an endurance
effect on the skin microbiota which was progressively adapted to the cadmium
accumulating in the tank system, while within the constant exposure regime
(CC), abrupt diversity and taxonomic changes might have been triggered.
Gradual changes are evident under stress gradients, for example, within
bioreactors, the anaerobic microbiome has been shown to gradually adapt
following ammonium disturbance(Regueiro, Carballa, and Lema 2016).
Consequently, the significant divergence in the functional diversity between
CV-Ctrl and CV-CC, not between CC-Ctrl, perhaps indicate a unique adaptive
evolution signature of skin microbiome under CV regime. Therefore, the skin
communities from CV and CC may have followed a different recovery
trajectory after adaptation. Strikingly, the recovery of skin microbiota of the
most extreme exposure (CC) appeared to be the most successful, when
considering the convergence of richness, evenness and functional diversity
between CC and Ctrl. However, significant differences among CC, CV and Ctrl
in terms of phylogenetic divergence (Table 4.2) and taxonomic composition
shifts (Figure 4.3, Additional Table 4.S2; Additional Tables 4.S3; Additional
Tables 4.S4) suggest this recovery was incomplete. For instance, a significant
increase in fish pathogens like Flavobacterium, Legionella and opportunists
like Mycoplasma was detected in both cadmium groups (CC & CV) compared
to control. The relative abundance of Flavobacterium was significantly lower
in the control group with a low percentage (< 0.5 %). Perturbation with
cadmium can facilitate the proliferation of opportunistic pathogens, this
concern has been found in other studies of fish microbiota recovery after
exposure to antibiotic (Navarrete et al. 2008) and triclosan biocide (Narrowe
et al. 2015). Similar taxonomic changes in both exposure regimes (CC & CV)
were expected (Regueiro, Carballa, and Lema 2014). Overall, the cadmium
disturbance may cause a shift to an alternative stable state, demonstrating
differential and incomplete recovery of skin microbiota in CC and CV.
153
In the gut microbiome, the recovery routes were different; at time T5, there
was only a significant evenness convergence between CC and CV. Overall,
the few significant differences in taxonomy, as well as the phylogenetic
divergence (Additional Table S2) between CC-CV and Ctrl-CV, but not
between CC-Ctrl, suggests a full recovery of the gut microbiota in CC and
gradual recovery in CV. At the level of taxonomic composition, overall the
dominance of opportunists Tenericutes was also a feature of farmed Eurasian
perch (Perca fluviatilis) gut microbiota studied in a context of stress predation
(Zha et al. 2018), although they were not found in the wild Eurasian perch
(Bolnick et al. 2014).
In the skin and gut microbiota, the significant increase in diversity (evenness
and richness) over the recovery period (T3-T5) was consistent with the
diversity increase in other host-associated studies such as the recovery of the
fathead minnow gut microbiome from a low-level triclosan exposure (Narrowe
et al. 2015), the human intestinal microbiota post-infection (Singh et al.
2015) , the murine gut microbiome exposed to antibiotics in early life (Cho
et al. 2012), and the molasses wastewater (De Vrieze et al. 2017).Further,
the functional redundancy observed in all water and gut microbial
communities is a major adaptive strategy behind resistance and recovery
(Cheaib et al. 2018; De Vrieze et al. 2017). Lastly, the significant divergence
of skin and gut microbiota diversity over time within the control group
suggests persistent divergence from the initial community structure due to
microbiota ontogeny through the developmental stage of fish juveniles (Burns
et al. 2016b).
Our findings demonstrate a relative role of neutral processes shaping the
bacterial communities recovery following exposition to metallic stressors.
According to the neutral model fit, the percentage of neutral OTUs in skin and
gut microbiota was significantly higher in the control group compared to CdCl2
treated groups, which provides evidence that neutral processes are the major
contributor in the microbiota assembly in non-stressed yellow perch,
therefore suggesting that selective processes are at play in driving the
community recovery in stress-exposed groups. Furthermore, Mycoplasma sp.
are a dominant species in perch microbiome, implicated in literature for other
fish species (Llewellyn et al. 2016; Holben et al. 2002). The inability of neutral
154
models to explain the abundance of any OTUs for Mycoplasma sp. in the
current study suggests that these bacterial strains can quickly adapt to the
host environment. Our study is the first to investigate the relative importance
of neutrality and determinism in driving post-disturbance assembly of the
host-associated microbiome.
4.8 Conclusions
This study not only elucidates the long-term bioaccumulation effect of toxic
metals on biological systems but also suggests that the sequestered cadmium
in fish liver will not likely predict the magnitude of exposure regime (constant
or variable). The effect of cadmium exposure on microbial communities is
also varying and dependent on the nature of the host it is originating from.
Surprisingly, after recovery, skin and gut microbiota of fish exposed to
constant concentrations of cadmium (CC) were closer to the control group
than those exposed to the gradual concentrations (CV). In the skin, the
metallic perturbation caused a shift to an alternative stable state, leading to
an incomplete recovery and therefore, facilitating the proliferation of
opportunistic pathogens (like Flavobacterium). In the gut, the functional and
phylogenetic diversity measurements suggest a complete community
recovery in the CC group and gradual recovery in the CV group. The selective
pressure exerted by cadmium on host and water microbiota may have left
adaptive evolution patterns conserving functional diversity at the expense of
taxonomic diversity. In both skin and gut microbiota, the recovery was
associated with a significant increase of evenness and richness in skin and
vice versa in the gut. In the control group, as expected, the significant
divergence from the initial community structure confirms the dynamic of
bacterial strains through the developmental stage of fish juveniles.
Consequently, the community recovery was affected by both cadmium
pressure and host development. In addition, our results have shown that the
microbial assembly rules during the community recovery were both
orchestrated by neutral and deterministic processes. In the water, community
recovery was driven by a substantial role of phylogenetic structuring resulting
from a combined pattern of stochasticity and cadmium-induced selective
pressure, in which the causality remains unknown. Further studies are needed
155
to quantify the interactions of neutrality and determinism in driving post-
disturbance assembly of host-associated microbiome during recovery.
4.9 Tables
Table 4. 1 Statistics of Cd concentrations in water and fish liver over time and treatments.
p value < = 0.05 : “*”
p value < = 0.001 : “***”
4.1-c T3 T5
Statistics tests (Tukey,
Wilcox) Liver Water Liver Water
Groups Tukey’s p
value
Wilcox’s p
value
Tukey’s p
value
Wilcox’s p
value
Ctrl-CV 0.00002 *** 0 *** 0.0000 *** 0.0002 ***
Ctrl-CC 0.0009 *** 0 *** 0.0000 *** 0.0002 ***
CC-CV 0.0281 * 0.1304 0.5362 0.1304
All (Kruskal−Wallis rank-
sum test) 0.0002 *** 0 *** 0.0000 *** 0.0003 ***
p value < = 0.05 : “*” ; p value < = 0.001 : “***”
These Tables summarise statistics of cadmium concentration variation over
time (Table 4.1.1B) and among treatments (Table 4.1.1C) in water tanks and
4.1-a Liver Water
Cadmium average concentration
(ng/ml) T0 T3 T5 T0 T3 T5
Ctrl 0.0860 0.1578 0.2330 0.0650 0.0750 0.0388 *
CV 0.0860 0.4000 1.0015 0.0670 6.4300 1.0600
CC 0.0860 0.5235 1.1700 0.0980 5.8000 1.3400
4.1-b Liver Water
Overtime
comparisons T0–T3 T0–t5 T3–t5 T0–t3 T0–t5 T3–t5
Groups Tukey’s p value Wilcox’s p value
Ctrl-Ctrl 1.0000 0.8253 0.8112 1.0000 1.0000 1.0000
Cv-CV 0.0003 *** 0.0000 *** 0.0000 *** 0.0000 *** 0.0000 *** 0.0000 ***
CC-CC 0.0000 *** 0.0000 *** 0.0000 *** 0.0000 *** 0.0000 *** 0.0000 ***
156
fish livers. Table 4.1.1-a illustrated the average cadmium (Cd) concentrations
in each group.
Table 4. 2 Phylogenetic divergence in host and water microbiomes.
Time Groups
Permanova Multiple test correction Betadisper Mrpp
p value
ADONIS p values B-H
p values
dispersion
p value
MRPP
Ctrl = control regime; CC = concentration is constant; CV = concentration is variable
Gut
T0
All groups 0.052 0.079 0.279
CC-Ctrl 0.042 * 0.077 0.048* 0.033 *
CC-CV 0.051 0.077 0.141 0.049 *
Ctrl-CV 0.280* 0.280 0.302 0.234
T3
All groups 0.001** 0.367 0.647
CC-Ctrl 0.001** 0.002 ** 0.916 0.001***
CC-CV 0.006 ** 0.006 ** 0.145 0.008 *
Ctrl-CV 0.001*** 0.002 ** 0.295 0.001 ***
T5
All groups 0.006 * 0.217 0.199
CC-Ctrl 0.135 0.135 0.923 0.104
CC-CV 0.005 ** 0.012 * 0.084 0.009**
Ctrl-CV 0.008 ** 0.012 * 0.130 0.008 **
Skin
T0
All groups 0.016 * 0.540 0.500
CC-Ctrl 0.166 0.166 0.649 0.154
CC-CV 0.081 0.122 0.256 0.075
Ctrl-CV 0.026 * 0.078 0.599 0.020 *
T3
All groups 0.008 ** 0.820 0.580
CC-Ctrl 0.035* 0.049 * 0.619 0.049 *
CC-CV 0.049 * 0.049 * 0.586 0.045 *
Ctrl-CV 0.021* 0.049 * 0.872 0.009 **
T5
All groups 0.001*** 0.380 0.001 ***
CC-Ctrl 0.011 * 0.011 * 0.749 0.008 **
CC-CV 0.001*** 0.002 ** 0.307 0.002 **
Ctrl-CV 0.001 *** 0.002 ** 0.236 0.001 **
Water
T0
All groups 0.001 *** 0.656 0.886
CC-Ctrl 0.018 * 0.027 * 1.000 0.009 **
CC-CV 0.002 ** 0.006 ** 0.291 0.002 **
Ctrl-CV 0.036 * 0.036 * 0.519 0.038 *
T3
All groups 0.001 *** 0.036 * 0.596
CC-Ctrl 0.002 ** 0.002** 0.068 0.002 **
CC-CV 0.001 ** 0.002 ** 0.020 * 0.001 ***
Ctrl-CV 0.001 *** 0.002 ** 0.546 0.001 ***
157
Time Groups
Permanova Multiple test correction Betadisper Mrpp
p value
ADONIS p values B-H
p values
dispersion
p value
MRPP
TR1
All groups 0.001 *** 0.214 0.108
CC-Ctrl 0.001 *** 0.002 ** 0.305 0.002 **
CC-CV 0.001 *** 0.002 ** 0.064 0.001 ***
Ctrl-CV 0.007 ** 0.007 ** 0.653 0.007 **
TR2
All groups 0.001 *** 0.639 0.381
CC-Ctrl 0.003 ** 0.005 ** 0.395 0.003 **
CC-CV 0.096 0.096 0.892 0.077
Ctrl-CV 0.001 *** 0.003 ** 0.391 0.001 ***
TR3
All groups 0.001 *** 0.561 0.629
CC-Ctrl 0.001 *** 0.003 ** 0.343 0.001 ***
CC-CV 0.003 ** 0.005 ** 0.536 0.002 **
Ctrl-CV 0.007 ** 0.007 ** 0.620 0.006 **
TR4
All groups 0.001 *** 0.498 0.031 *
CC-Ctrl 0.009 ** 0.009 ** 0.536 0.007 **
CC-CV 0.006 ** 0.009 ** 0.179 0.009 **
Ctrl-CV 0.004 ** 0.009 ** 0.578 0.002 **
T5
All groups 0.001 *** 0.408 0.027 *
CC-Ctrl 0.002 ** 0.002 ** 0.331 0.001 **
CC-CV 0.001 *** 0.002 ** 0.578 0.003 **
Ctrl-CV 0.002 ** 0.002 ** 0.438 0.001 **
The phylogenetic distances between OTUs were computed using Gunifrac
distance (see “Methods” section). The divergence between treatments and
control was assessed using PERMANOVA and the homogeneity for group
dispersions (distance from centroid) was evaluated using two multivariate
tests, BETADISPER and multi-response permutation procedure (MRRP) of
within versus among group dissimilarities. The significance of divergence
between groups was measured by applying multiple correction tests with
Benjamini-Hochberg BH (p value < 0.05).
p value < = 0.001 : “***”, p value < = 0.01 : “**”, p value < = 0.05 : “*”.
158
4.10 Figures
Figure 4. 1 Schematic illustration of the perch microbiome recovery
experiment.
Figure 4. 2 Alpha-diversity dynamics in the water and perch microbiome.
Predicted alpha-diversity plots by linear mixed model. Alpha-diversity in
water and host-microbial communities over time and among treatments is
predicted using the linear mixed model. The richness/evenness ratio were
considered as response variables, the fixed effects were defined by time and
cadmium concentration (in water and liver), and tanks were taken as random
effects. Over time, the predicted alpha diversity in host microbial
communities (skin, gut) highlights stable trends of the Control group
compared to the treatments. However, all groups of the water microbial
communities decrease overtime. Constant cadmium regime (CC) is in orange,
variable cadmium regime (CV) is in yellow, and control (control) is in green.
159
Figure 4. 3 Taxonomic composition dynamics of host communities
Stacked Barplots show the most abundant taxa (>0.5%) overtime in the gut,
skin and water microbiomes. The genera that significantly changed among
treatments and control at T5 are summarised in Additional Table S2.
Figure 4. 4 Heatmaps of differential abundance among host and water communities
This figure from left to right includes 9 heatmaps of the significant taxonomic
fingerprints at the genus level between gut, the skin and the water at times
T0 (first column), T3 (second column) and T5 (third column) in the control
(first row), the CV (second row), and the CC (third row) groups. The
hierarchical clustering of the relative abundance of phyla which significantly
changed over time was performed using Ward's method and Bray–Curtis
160
dissimilarity distance. Vegan package and pheatmap () function in R were
used.
Figure 4. 5 Function diversity dynamics in host and water microbiome.
Boxplots of functions profiles were predicted from the matrices of taxa count
using the software Tax4Fun. The statistical significance (p-value< 0.05)
found using ANOVA followed by FDR (False Discovery Rate) test are
represented with asterisks points (0.001: "***", 0.01: "**", 0.05: "*").
161
Figure 4. 6 Recovery dynamics of the networks of host communities.
The networks organization is based on nodes betweenness centrality among
treatments and Control. Unstructured patterns in the networks were observed
at T0. Node size represents sample richness. The strength of correlation
(Spearman Correlation from 0.3 to 1) between two nodes is inversely
proportional to the size of the edge. This network was built using R and
Cytoscape software. Constant Cadmium samples (CC) are in orange, Variable
Cadmium samples (CV) are in Yellow, and Control (Control) samples are in
green.
162
Figure 4. 7 Recovery dynamics of the network of water communities
The networks organization at every resilience time TR1, TR2, TR3, TR4 and
WT5 is based on nodes betweenness centrality among treatments and
Control. The networks modules easliy distinguishable between groups since
T0. The size of nodes (sample richness) at the beginning of TR1 and at the
end of the time TR4 showed shifts in the community richness. The strength
of correlation (Corr. Spearman from 0.5 to 1) between two nodes is inversely
proportional to the size of the edge. This network was built using R and
Cytoscape software. Constant Cadmium samples (CC) are in orange, Variable
Cadmium samples (CV) are in Yellow, and Control (Control) samples are in
green.
163
Figure 4. 8 Centrality plots of host microbiome networks
This figure summarizes relationships of betweenness centrality versus
eigenvalue centrality of host microbiome networks among treatments and at
each time point. The results show evidence of shift in centrality medians
between the control regime (which is higher) and the cadmium selection
regimes. The plots of eigenvalue centrality versus betweenness centrality
clearly reveal that centrality shift at time T3 for skin and gut microbiome.
164
Figure 4. 9 Percentage of neutral OTUs over time and treatment.
165
Using the non-linear least squares model (NLS), the percentage of OTUs that
fit the neutral model within a confidence interval of 95% showed variable
trends between communities across time and treatments. A goodness of fit
R2> 0.5 was considered as the significant threshold of neutrality fit. The
cadmium treatment invoked stochasticity in the water communities, while in
gut and skin communities, the percentage of neutral OTUs remained higher
in the control compared to treatments.
Figure 4. 10 Demographic variation of metacommunity neutrality
across water and host microbiome
This figure summarizes the scatterplots of neutral model fitting the whole
metacommunity ( gut skin and water) at times T0 (first column), T3 (second
column) and T5 (third column) in the control (first row), the CV (second row),
and the CC (third row) groups. Neutral OTUs are shown in black, non-neutral
are depicted in grey, whilst the red is Mycoplasma sp. OTUs. We see no
Mycoplasma sp. OTUs that fit the neutral model in the whole metacommunity.
166
4.11 Supplementary figures
Additional file 4.1: Figures 4.S1. Dynamic of alpha-diversity divergence
between host and water communities. The significant ANOVA results of alpha
diversity between water (W), Skin(S) and Gut (Gut) communities in Control,
CV and CC groups before and during disturbance, and after recovery period
are represented with asterisks on the boxplots (0.001 : "***", 0.01 : "**",
0.05 : "*").
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM1_ESM.pdf
Additional file 4.2: Figures 4.S2. Beta-diversity divergence at the
treatment level. This file combines all the NMDS (non-metric Multi-
Dimensional Scaling) plots showing first two dimensions in the ordination of
when using generalized Unifrac distance measure of water and host-microbial
communities. The NMDS plots and PERMANOVA revealed a significant
separation between different treatments and control (for the pairwise, see
Table 2 for adjusted p-values after Benjamini-Hochberg correction in
PERMANOVA and MRPP tests) at T0, T3, and T5 for skin and gut microbiota,
and at T0, T3, TR1-TR4, and T5 for water microbial communities.
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM2_ESM.pdf
Additional file 4.3: Figures 4.S3. Beta-diversity divergence at the
community level. This file combines all the NMDS (non-metric Multi-
Dimensional Scaling) plots and phylograms based on generalized Unifrac
distances between water and host-microbial communities. The NMDS plots
and PERMANOVA revealed a significant separation of among all type of
communities per time (T0, T3, and T5) and treatment (Control, CC, CV).
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM3_ESM.pdf
Additional file 4.4: Figures 4.S4. Heatmaps of cadmium with taxa
diversity and composition in host and water communities. The
correlations indicate a gradient from positive (blue) to negative (red) along a
colour gradient, with rows representing diversity measures (richness,
167
evenness) as well as cadmium concentrations, and columns indicating
taxonomic levels. The gut microbiome in the constant CdCl2 (CC) and variable
CdCl2 (CV) regimes showed a negative correlation between Mycoplasma and
diversity indices. A strong positive correlation between Actinomycetales is
noticeable in the CV. For the skin microbiome, not only Actinomycetales, but
also Burkholderiales, and Chromatiales showed strong positive correlations
with CV. This figure was produced using the Rhea package.
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM4_ESM.pdf
Additional file 4.5: Figures 4.S5. Heatmaps of cadmium with taxa
diversity and composition in water during the recovery period. The
correlations indicate a gradient from positive (blue) to negative (red) along a
colour gradient, with rows representing diversity measures (richness,
evenness) as well as cadmium concentrations, and columns indicating
taxonomic levels. In the constant CdCl2 (CC) and variable CdCl2 (CV) regimes,
correlations of cadmium with taxa abundance showed variable profiles over
time. This figure was produced using the Rhea package.
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM5_ESM.pdf
Additional file 4.6: Figures 4.S6. Heat trees and stacked bar plots of
water and host microbiome structure.
This figure summarizes pairwise comparison of the community composition
of water and each of the host communities for different treatments (Ctrl, CC
and CV). Additionally, stacked bar plots of relative abundance at phylum
level are provided for each community (water, skin, gut). The non-grey
coloring (which category the branches are upregulated in) indicates
significant differences in terms of log median ratios for samples from different
habitats (Gut, Skin and Water) as determined by a Wilcox rank-sum test
followed by a Benjamini-Hochberg (FDR) correction for multiple testing. The
heat trees were built using metacoder and stacked barplots were produced
using the Rhea package.
168
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM6_ESM.pdf
4.12 Supplementary Material
Additional Table 4.S1. Alpha-diversity dynamics over time and
treatments. This statistical summary reveals richness or evenness changes
over time (table 1 a, b) and between control and treatments (table 1c) in
water and host communities. The significant changes of alpha-diversity
indices between treatments and control were statistically tested using
Kruskal-Wallis and Wilcoxon tests by applying Benjamini-Hochberg
correction. The same statistics were used to compare alpha-diversity over
time.
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM7_ESM.xlsx
Additional Table 4.S2. Statistical summary of taxa divergence
between treatments and control after recovery. This table summarises
significant taxonomic changes in water and host-microbial communities
between control and treatments using the Fisher test, and by applying
Benjamini-Hochberg correction.
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM9_ESM.xlsx
Additional Tables 4.S3. Statistical summary of taxa divergence over
time in host and water microbial communities after recovery. These
table summarise significant taxonomic changes over time in water and host-
microbial communities using Kruskal-Wallis and Wilcoxon tests, and by
applying Benjamini-Hochberg correction.
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM10_ESM.xlsx
Additional Tables 4.S4. Statistical summary of differential abundance
between water and host microbial communities at all taxonomic
levels. These table summarise significant taxonomic changes between water
169
and each of the host communities (skin and gut) at times T0 (sheet1), T3
(sheet2), and T5 (Sheet3) using Kruskal-Wallis and Wilcoxon tests, and by
applying Benjamini-Hochberg correction.
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM10_ESM.xlsx
Additional Table 5.S5. Statistics of the neutral model in host and
water microbiomes. This table summarises the neutral model fit based on
the following parameters; the migration rate (m.ci) within 95% of the
confidence interval, the goodness of fit(R2), number of samples, richness,
abundance cutoff, percentage of % neutral OTUs and non-neutral OTUs.
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM11_ESM.xlsx
Additional Table 6.S6: List of OTUs that accounted for those that did not
fit the neutral model
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM12_ESM.xlsx
Additional Table 7.S7: List of OTUs that accounted for those that did not
fit the neutral model and were assigned to Mycoplasma species.
https://static-content.springer.com/esm/art%3A10.1186%2Fs40168-020-
0789-0/MediaObjects/40168_2020_789_MOESM13_ESM.xlsx
170
Discussion et conclusions générales
Dans le cadre de ce projet de recherche, nous avons détecté des signatures
génotypiques de stratégies adaptatives de communautés microbiennes dans
un contexte de perturbations métalliques et ce en étudiant deux modèles
environnementaux et un système d’hôte-associé. Dans le deuxième chapitre,
un système microbien sous perturbation environnementale a été étudié. Les
communautés microbiennes lacustres ont été échantillonnées trois ans après
l’arrêt des exploitations minières. Ces mines se trouvaient à proximité du
bassin versant des lacs étudiés et leur exploitation a duré environ 60 ans. Ce
qui fait de ce système lacustre un excellent modèle naturel de pression
anthropique pour étudier la résilience des communautés microbiennes
exposées à un gradient de perturbations tels que les métaux traces. De plus,
un lac témoin non pollué a été utilisé comme contrôle négatif.
En utilisant une approche métagénomique de séquençage dans les deux types
de modèles microbiens (environnemental et hôte-associé), des empreintes
méta-génotypiques ont été identifiées sous forme de transitions
compositionnelles dans le répertoire taxonomique et fonctionnel des
communautés microbiennes. Les résultats du modèle environnemental
anthropique montrent une forte sélection exercée sur les gènes de résistance
aux métaux lourds. À l'échelle compositionnelle taxonomique des
métacommunautés, des changements évidents sont marqués par un
différentiel d’abondance relative des taxons. Ces changements du répertoire
taxonomique ont été beaucoup plus marqués au niveau du genre, suggérant
une adaptation « de type communauté » au gradient métallique dans chaque
niche écologique lacustre. À l'échelle compositionnelle des fonctions, nous
avons observé une faible adaptation révélée par l'érosion de voies
métaboliques (Carbohydrates et biosynthèse des protéines). À travers le
gradient métallique, l’abondance relative de certaines catégories
fonctionnelles, telles que la réponse au stress, la régulation, le métabolisme
des protéines et la résistance métallique, était beaucoup plus élevée dans les
lacs pollués que dans les lacs non pollués.
Les relations entre la diversité des taxons et des fonctions indiquent des
empreintes adaptatives persistantes lors de la phase de résilience du système
171
microbien environnemental. En analysant la relation entre les signatures
taxonomiques et fonctionnelles des communautés lacustres, nous avons
détecté un découplage entre la diversité taxonomique et le répertoire
fonctionnel dans les lacs pollués, ce qui est un indice clair d’adaptation,
potentiellement via le transfert horizontal de gènes (Burke et al. 2011; Green,
Bohannan, and Whitaker 2008; Louca, Parfrey, and Doebeli 2016; Louca et
al. 2016; Ram et al. 2005; Hemme et al. 2016). Ce découplage révèle une
redondance fonctionnelle malgré une divergence taxonomique, ce qui met en
évidence que la diversité microbienne à l’échelle taxonomique ne reflète pas
nécessairement la diversité des services écosystémiques dans les
environnements fortement et graduellement perturbés. Ces résultats
suggèrent pour la première fois, à l’aide des méthodes basées sur des
corrélations canoniques et de régularisation (González and Déjean 2012), un
découplage taxon-fonction au sein de communautés microbiennes naturelles
adaptées à un gradient de contamination polymétallique.
Cette étude propose une vision préliminaire du phénomène de découplage.
Des études supplémentaires seront nécessaires pour comprendre de manière
plus approfondie la nature de la convergence des communautés microbiennes
dans le contexte d’un gradient sélectif. L’investissement de l’évolution
moléculaire des gènes de résistance aux métaux serait une bonne piste pour
tester l’hypothèse de HGT. Des données de séquençage avec une couverture
plus profonde seront nécessaires pour assembler des génomes complets à
partir de métagénomes. Les séquences des génomes complets permettent de
mesurer le taux d’évolution des gènes de résistance aux métaux par rapport
à d’autres gènes vitaux à l’échelle d’un génome entier. Finalement le balayage
sélectif de gènes de résistance mérite notre attention car un tel évènement
de purge de diversité n’a pas encore été caractérisé à l’heure actuelle chez
des communautés microbiennes dans le milieu naturel (Doolittle and
Zhaxybayeva 2009).
Le modèle de communautés hôte-associé sous perturbation a été caractérisé
expérimentalement. Nous avons modélisé l’environnement naturel au
laboratoire en conditions artificielles d’intoxication similaires à l’environnent
anthropique. Nous avons donc étudié l’impact de la toxicité du Cadmium (Cd)
sur les microbiotes intestinal et cutané de la Perchaude (Perca flavescens)
172
reconnu comme un poisson modèle en écotoxicologie. La Perchaude s’adapte
au faible gradient de traces polymétalliques (Bourret et al. 2008; Bélanger-
Deschênes et al. 2013). En dépit de sa tolérance, plusieurs études
d’écotoxicologie ont mis en évidence l’impact des traces du Cd sur cet
organisme, notamment au niveau de la diversité génétique de ses
populations, de l’expression de gènes du système immunitaire et des voies
de biosynthèse de l’acide rétinoïque, primordial pour sa croissance et son
développement (Bourret et al. 2008; Pierron et al. 2009; Bélanger-Deschênes
et al. 2013; Bougas et al. 2013a; Michel A. Defo et al. 2014).
Le stress métallique aurait potentiellement des répercussions sur les
fonctions cellulaires, physiologiques et hormonales qui sont sous contrôle du
microbiote. Dans notre expérience, nous avons utilisé des doses non létales
de cadmium comme agent de sélection sans provoquer de dommages
physiologiques importants ni entraîner la mortalité de l’hôte. Nous avons
étudié la composition, l’assemblage, la structure et les réseaux de corrélation
microbienne de trois niches écologiques : l’eau, le mucus cutané et le mucus
intestinal de la Perchaude et ce, en réponse à une exposition au Cd selon
deux régimes ; constant et graduel. Cette expérience d’évolution
expérimentale nous a permis d’étudier la résistance et la résilience des
communautés microbiennes, qui ont respectivement constitués, les troisième
et quatrième chapitres de la thèse.
Dans le troisième chapitre, l’étude de la résistance a été conduite avec une
approche métagénomique d’amplicons suivie par des mesures de la diversité
alpha et beta, puis avec des réseaux de co-abondance et des modèles
d’écologie microbienne. Les résultats montrent selon le régime de
perturbations, des signatures adaptatives différentes entre le mucus cutané
et le mucus intestinal à l’échelle de la composition, de la structure et de la
dynamique d’assemblage du microbiote de la Perchaude. Premièrement, les
propriétés topologiques des réseaux de co-abondance montrent non
seulement l'étendue de l’implication de taxons rares dans l'assemblage de
microbiotes, mais aussi des corrélations négatives entre les taxons qui
augmentent proportionnellement aux concentrations du Cd introduites dans
le système, c’est-à-dire plus élevées dans le régime constant que dans le
régime graduel. L’augmentation de la diversité alpha, la divergence
173
phylogénétique révélée par l’analyse de la diversité bêta, et la convergence
taxonomique entre les communautés bactériennes de l'eau et la peau sous
les deux régimes d’exposition au Cd (constant et graduel) mettent en
évidence que l'hôte avait perdu la capacité de résister à la colonisation
bactérienne environnementale. Ce phénomène de «résistance à la
colonisation» (Buffie and Pamer 2013b) résultait théoriquement en partie
d'un dysfonctionnement du système immunitaire de l'hôte appelé «résistance
de la colonisation immunitaire».
Sous les deux régimes de sélection, la composition des communautés
microbiennes a été marquée par la prévalence des souches opportunistes,
telles que Mycoplasma, Pseudomonas, ainsi que d’autres souches connues en
aquaculture telle que le Falvobacterium, dont certaines sont pathogènes (R.
M. Brown, Wiens, and Salinas 2018; Holben et al. 2002).
Les perturbations induites par le Cd infligent à l’hôte des lésions hépatiques
identifiées par microscopie à haute résolution (résultats non présentés), avec
une bioaccumulation significative du cadmium dans le foie révélé par
spectrométrie de masse ionique à haute précision. Nos résultats confirment
bien les conclusions des études rapportées sur les perturbations induites par
le cadmium sur la physiologie non seulement de la Perchaude (Ponton et al.
2016; L. D. Kraemer, Campbell, and Hare 2006; Giguère et al. 2006; Klinck
et al. 2007), mais aussi chez d’autres espèces de poissons comme le Tilapia,
la Blennie paon et la Truite (Zhai et al. 2017; Naïja et al. 2017; Niyogi et al.
2004). Également, la dysbiose du microbiote de l’hôte causée par des métaux
ont été davantage reportées chez des modèles animaux comme la Souris (S.
Zhang et al. 2015; Ba Qian et al. 2017) que dans l’aquaculture. Pour tester
si la perte de résistance de l’hôte à la colonisation bactérienne s’est
accompagnée par des processus stochastiques d’assemblage microbien dans
l’eau et chez l’hôte, un modèle mathématique basé sur la régression des
moindres carrées non linéaires, a été appliqué pour estimer la contribution
relative de ces processus. Les résultats montrent qu’en absence de cadmium,
la dérive avait essentiellement conduit l'assemblage taxonomique des
communautés de l'eau, alors qu’en présence de cadmium, cet assemblage
était sous l’influence principale de processus déterministes (non neutres). En
revanche, peu importe le niveau d’exposition au cadmium, l'assemblage des
174
microbiotes se développait principalement selon des processus non-neutres.
L’assemblage du microbiote intestinal peut avoir évolué selon des processus
non neutres dus non seulement au cadmium en tant que facteur de
perturbation dans les deux groupes expérimentaux, mais également à la
sélection imposée par le développement de l’hôte, comme le démontrent les
résultats du groupe contrôle. Ce phénomène a été observé chez différentes
espèces de poissons élevées en conditions normales (Inoue and Ushida 2003;
Burns et al. 2016b; Sylvain and Derome 2017).
Le dernier chapitre de cette thèse s’est penché sur l’étude de la dynamique
de la résilience en étudiant les signatures adaptatives et les processus
écologiques mis en œuvre sous les régimes d’exposition métallique constant
et graduel et dans les différentes niches écologiques. Les résultats montrent
non seulement un effet de bioaccumulation à long terme des métaux traces
sur le système biologique étudié, mais suggèrent également que le cadmium
séquestré dans le foie du poisson ne permettrait probablement pas de prédire
l'intensité du régime d'exposition (constant ou variable).
L’effet du cadmium sur les communautés microbiennes varie en fonction de
la biologie et la physiologie de l’hôte. Étonnamment, après deux mois de
résilience, les microbiotes cutanés et intestinaux des poissons exposés au
régime constant se rapprochaient davantage du groupe témoin que ceux
exposés au régime graduel. Dans le mucus cutané, les résultats montrent un
processus de résilience incomplet ce qui conduit à un état intermédiaire stable
et facilite ainsi la prolifération de souches opportunistes (comme
Flavobacterium). Par conséquent, la prévalence de souches opportunistes
observées durant la phase de résistance caractérise également la phase de
résilience du microbiote cutané. En revanche, la phase de résilience témoigne
d’une augmentation significative de la richesse du microbiote dans la peau et
inversement dans les intestins.
Au sein du microbiote intestinal, les mesures de la diversité fonctionnelle et
phylogénétique suggèrent une résilience presque complète de la
communauté sous le régime constant et une résilience intermédiaire sous le
régime graduel. Comme attendu à la fin de l’expérience de résilience, dans le
groupe témoin, la divergence significative par rapport à la structure initiale
de la métacommunauté du départ confirme la théorie de l’ontogenèse (Inoue
175
and Ushida 2003; Burns et al. 2016b; Sylvain and Derome 2017) du
microbiote et montre une dynamique de colonisation microbienne orchestrée
par l’hôte tout au long de son développement dès le stade juvénile. Cela
pourrait suggérer que la résilience du microbiote de la Perchaude dans les
deux régimes de sélection a été influencée à la fois par les forces sélectives
associées au cadmium et celles liées au développement de l'hôte. Nos
résultats ont montré que les règles de l'assemblage microbien lors de la phase
de résilience de la communauté semblent être influencées par un gradient de
processus neutres et déterministes. Des études de la colonisation
symbiotique seront nécessaires avec plusieurs séries temporelles pour
quantifier les contributions des processus neutres et déterministes dans la
structuration et de l'assemblage de microbiote durant la résilience.
Curieusement, la pression de sélection exercée par le cadmium sur les
communautés microbiennes de l'eau montre des empreintes adaptatives
témoignant une redondance dans la diversité fonctionnelle aux dépens de la
diversité taxonomique. Cette signature est semblable au découplage taxon-
fonction observé dans la première étude au sein des communautés
microbiennes lacustres sous haute pression de sélection polymétallique. Au
niveau du système hôte-microbiote, la dynamique semble être différente, la
résilience de la méta-communauté ayant été marquée par une structuration
phylogénétique implique une potentielle interaction des processus
stochastiques et déterministes.
En général, nos travaux offrent une perspective solide sur le phénomène de
découplage observé entre les répertoires taxonomique et fonctionnel au sein
des communautés microbiennes sous sélection graduelle. Des études
supplémentaires seraient nécessaires et complémentaires pour discerner
d’une manière plus approfondie les stratégies adaptatives des communautés
microbiennes dans l’environnement naturel et celles associées à l’hôte.
Le phénomène de découplage taxon-fonction observé durant la période de
résilience du microbiote cutané de Perchaude est issu d’une pression de
sélection graduelle. Notre nouvelle hypothèse serait que l’adaptation de
l’holobionte sous sélection graduelle impliquerait un rôle dominant des
transferts horizontaux (Panda et al. 2018; López-Madrigal and Gil 2017).
Plusieurs arguments de l’évolution microbienne peuvent appuyer cette
176
hypothèse. Par exemple, les symbiontes bactériens connus chez plusieurs
modèles de Métazoaires évoluent par réduction de taille de génome (Wolf and
Koonin 2013) et n’échappent pas aux transfert horizontaux de gènes
(Dufresne, Garczarek, and Partensky 2005; Mao, Yang, and Bennett 2018;
Tian et al. 2017; López-Madrigal and Gil 2017; Hendry et al. 2018 ; González-
Torres et al. 2019). Lorsque l’évolution implique une optimisation de taille,
les évènements de transfert horizontaux deviennent logiquement des sources
d’innovations adaptatives pour maintenir la coévolution avec leur hôte (B. P.
Brown and Wernegreen 2019; González-Torres et al. 2019). Nos résultats
montrent une forte dominance de souches opportunistes dans les microbiotes
tels que les Mycoplasmas. Ceux-ci se caractérisent par des génomes de très
petites tailles (Trussart et al. 2017).
De la résistance à la résilience, nos résultats montrent que le microbiote est
un outil de diagnostic environnemental puissant permettant d'évaluer la
capacité des écosystèmes à résister aux contaminants comme les métaux
traces. Les processus écologiques et évolutifs sous-jacents décrivent une
colonisation bactérienne stochastique dans le système Perchaude-microbiote
durant l’exposition initiale au Cadmium. L’identification de processus neutres
et non neutres régissant l'assemblage des différents types de communautés
bactériennes (environnementale et associées à l'hôte) offre de nouvelles
informations clés pour la compréhension des forces évolutives qui façonnent
la composition du microbiote de l'hôte face au stress environnemental (le
Cadmium). Les organismes vivants sont actuellement confrontés à des
niveaux sans précédent de facteurs de stress environnementaux qui ont une
incidence sur leur capacité à faire face aux agents pathogènes naturels,
essentiellement en modifiant leur défense immunitaire et physiologique. Par
conséquent, il est utile et nécessaire de déchiffrer avec précision les signaux
précoces d’alerte apparaissant aux premiers stades de l’exposition aux
xénobiotiques. En soulignant le lien entre perte de résistance à la colonisation
et dysbiose chez l'hôte (qui est connu pour induire une réponse
inflammatoire), nos résultats seront utiles non seulement pour le domaine de
l'écologie microbienne, mais également pour la recherche biomédicale,
notamment pour mieux comprendre le phénomène de la dysbiose. Il a été
démontré que l’altération de la composition du microbiote intestinal entrainait
177
l'apparition de diverses maladies inflammatoires telles que le diabète
(Upadhyaya and Banerjee 2015; Tilg and Moschen 2014; Tilg and Kaser
2011), la maladie de Crohn (Svolos et al. 2019) , le cancer (Helmink et al.
2019), l'obésité (Thaiss 2018), ainsi que plusieurs maladies mentales telle
que l’autisme.
D’une même importance, notre étude est l’une des rares à s’être focalisée sur
la phase de résilience du microbiote associé à l'hôte, analysée à l'aide de
modèles d’assemblage de communautés écologiques. Nous avons pu évaluer
la contribution de facteurs neutres et déterministes à la base de la résilience
de systèmes biologiques confrontés à un gradient de pressions de sélection
d’origine anthropique. De faibles concentrations de métaux traces peuvent
causer des dommages physiologiques et génétiques des systèmes
biologiques, et modifier considérablement leur microbiote, sans retour à l’état
initial ce qui rappelle les conséquences des traitements par des antibiotiques
chez l’Homme. En raison de leur complexité, de leur taille et de leur
composition, la modélisation des mutations dans le répertoire génétique
microbien reste une tâche ardue. Il sera intéressant de tester si l'état post-
perturbation est plus résilient face à une perturbation identique ou plus
sévère. Des nouvelles perturbations répétitives post-perturbation peuvent
aider à identifier les changements de la composition bactérienne à l’échelle
de ce qu’on appelle espèce. La caractérisation des biomarqueurs candidats
peut sans aucun doute contribuer à la biorestauration et la réhabilitation à
grande échelle (par exemple au Canada) des masses d’eau polluées et
contaminées par les rejets des industries minières. Au-delà de la limite des
outils d'écologie microbienne basés sur les informations d'abondance, des
études supplémentaires sont nécessaires pour produire des
données « Omics », à l’échelle fonctionnelle et métabolique afin de trouver
des cœurs de gènes interchangeables entre les communautés au cours de la
réponse post-stress. Les études de génomique et de génétique des
populations peuvent caractériser l’évolution des gènes de résistance aux
métaux traces et leur taux de mutation. De plus, pour améliorer la résilience
et la santé de l'aquaculture, des études supplémentaires seront nécessaires
pour comprendre l’évolution contemporaine autour de l’axe contaminant-
hôte-microbiote.
178
Bibliographie
Acinas, Silvia G., Luisa A. Marcelino, Vanja Klepac-Ceraj, and Martin F. Polz. 2004.
“Divergence and Redundancy of 16S RRNA Sequences in Genomes with
Multiple Rrn Operons.” Journal of Bacteriology 186 (9): 2629–35.
https://doi.org/10.1128/JB.186.9.2629-2635.2004.
Albertsen, Mads, Philip Hugenholtz, Adam Skarshewski, Kåre L. Nielsen, Gene W.
Tyson, and Per H. Nielsen. 2013. “Genome Sequences of Rare, Uncultured
Bacteria Obtained by Differential Coverage Binning of Multiple Metagenomes.”
Nature Biotechnology 31 (6): 533–38. https://doi.org/10.1038/nbt.2579.
Alburaki, M., B. Cheaib, L. Quesnel, P.-L. Mercier, M. Chagnon, and N. Derome. 2016.
“Performance of Honeybee Colonies Located in Neonicotinoid-Treated and
Untreated Cornfields in Quebec.” Journal of Applied Entomology, June.
https://doi.org/10.1111/jen.12336.
Alleman, Austin, Kate L. Hertweck, and Srini Kambhampati. 2018. “Random Genetic
Drift and Selective Pressures Shaping the Blattabacterium Genome.” Scientific
Reports 8 (1): 1–12. https://doi.org/10.1038/s41598-018-31796-6.
Allen, Eric E., and Jillian F. Banfield. 2005. “Community Genomics in Microbial
Ecology and Evolution.” Nature Reviews Microbiology 3 (6): 489–98.
https://doi.org/10.1038/nrmicro1157.
Allesina, Stefano, and Jonathan M. Levine. 2011. “A Competitive Network Theory of
Species Diversity.” Proceedings of the National Academy of Sciences 108 (14):
5638–42. https://doi.org/10.1073/pnas.1014428108.
Allison, Steven D., and Jennifer B. H. Martiny. 2008. “Resistance, Resilience, and
Redundancy in Microbial Communities.” Proceedings of the National Academy
of Sciences 105 (Supplement 1): 11512–19.
https://doi.org/10.1073/pnas.0801925105.
Almaas, E., B. Kovács, T. Vicsek, Z. N. Oltvai, and A.-L. Barabási. 2004. “Global
Organization of Metabolic Fluxes in the Bacterium Escherichia Coli.” Nature
427 (6977): 839–43. https://doi.org/10.1038/nature02289.
Almeida, Welington I., Ricardo P. Vieira, Alexander Machado Cardoso, Cynthia B.
Silveira, Rebeca G. Costa, Alessandra M. Gonzalez, Rodolfo Paranhos, et al.
2008. “Archaeal and Bacterial Communities of Heavy Metal Contaminated
Acidic Waters from Zinc Mine Residues in Sepetiba Bay.” Extremophiles 13
(2): 263–71. https://doi.org/10.1007/s00792-008-0214-2.
Alonso, David, Rampal S. Etienne, and Alan J. McKane. 2006. “The Merits of Neutral
Theory.” Trends in Ecology & Evolution 21 (8): 451–57.
https://doi.org/10.1016/j.tree.2006.03.019.
Alpay, S. 2016. “Multidisciplinary Environmental Science Investigations Surrounding
the Former Aldermac Mine, Abitibi, Quebec: The Lac Dasserat Study Workshop
Summarized.” 7993. https://doi.org/10.4095/297747.
Álvarez, Analía, Santiago A. Catalano, and María Julia Amoroso. 2013. “Heavy Metal
Resistant Strains Are Widespread along Streptomyces Phylogeny.” Molecular
Phylogenetics and Evolution 66 (3): 1083–88.
https://doi.org/10.1016/j.ympev.2012.11.025.
Amann, R I, W Ludwig, and K H Schleifer. 1995. “Phylogenetic Identification and in
Situ Detection of Individual Microbial Cells without Cultivation.”
179
Microbiological Reviews 59 (1): 143–69.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC239358/.
Aminov, Rustam I. 2011. “Horizontal Gene Exchange in Environmental Microbiota.”
Frontiers in Microbiology 2 (July). https://doi.org/10.3389/fmicb.2011.00158.
Amir, Amnon, Daniel McDonald, Jose A. Navas-Molina, Evguenia Kopylova, James T.
Morton, Zhenjiang Zech Xu, Eric P. Kightley, et al. 2017. “Deblur Rapidly
Resolves Single-Nucleotide Community Sequence Patterns.” MSystems 2 (2):
e00191-16. https://doi.org/10.1128/mSystems.00191-16.
Ana Segura, Lázaro Molina. 2014. “PLASMID-MEDIATED TOLERANCE
TOWARDS ENVIRONMENTAL POLLUTANTS.” Microbiology Spectrum 2
(4): PLAS-0013-2013. https://doi.org/10.1128/microbiolspec.PLAS-0013-2013.
Andam, Cheryl P., David Williams, and J. Peter Gogarten. 2010. “Natural Taxonomy in
Light of Horizontal Gene Transfer.” Biology & Philosophy 25 (4): 589–602.
https://doi.org/10.1007/s10539-010-9212-8.
Anderson, James B., Jason Funt, Dawn Anne Thompson, Snehit Prabhu, Amanda
Socha, Caroline Sirjusingh, Jeremy R. Dettman, et al. 2010. “Determinants of
Divergent Adaptation and Dobzhansky-Muller Interaction in Experimental
Yeast Populations.” Current Biology 20 (15): 1383–88.
https://doi.org/10.1016/j.cub.2010.06.022.
Andersson, J. O., and S. G. Andersson. 2001. “Pseudogenes, Junk DNA, and the
Dynamics of Rickettsia Genomes.” Molecular Biology and Evolution 18 (5):
829–39. https://doi.org/10.1093/oxfordjournals.molbev.a003864.
Aßhauer, Kathrin P., Bernd Wemheuer, Rolf Daniel, and Peter Meinicke. 2015.
“Tax4Fun: Predicting Functional Profiles from Metagenomic 16S RRNA Data.”
Bioinformatics 31 (17): 2882–84. https://doi.org/10.1093/bioinformatics/btv287.
Azizishirazi, Ali, William A. Dew, Berenice Bougas, Mehdi Dashtban, Louis
Bernatchez, and Greg G. Pyle. 2014. “Chemosensory Mediated Behaviors and
Gene Transcription Profiles in Wild Yellow Perch (Perca Flavescens) from
Metal Contaminated Lakes.” Ecotoxicology and Environmental Safety 106
(August): 239–45. https://doi.org/10.1016/j.ecoenv.2014.04.045.
Ba Qian, Li Mian, Chen Peizhan, Huang Chao, Duan Xiaohua, Lu Lijun, Li Jingquan, et
al. 2017. “Sex-Dependent Effects of Cadmium Exposure in Early Life on Gut
Microbiota and Fat Accumulation in Mice.” Environmental Health Perspectives
125 (3): 437–46. https://doi.org/10.1289/EHP360.
Baker, G.C., J.J. Smith, and D.A. Cowan. 2003. “Review and Re-Analysis of Domain-
Specific 16S Primers.” Journal of Microbiological Methods 55 (3): 541–55.
https://doi.org/10.1016/j.mimet.2003.08.009.
Baldwin, R. L., and K. C. Donovan. 1998. “Modeling Ruminant Digestion and
Metabolism.” In Mathematical Modeling in Experimental Nutrition, edited by
Andrew J. Clifford and Hans-Georg Müller, 325–43. Advances in Experimental
Medicine and Biology. Boston, MA: Springer US. https://doi.org/10.1007/978-
1-4899-1959-5_21.
Balter, Michael. 2013. “Archaeologists Say the ‘Anthropocene’ Is Here—But It Began
Long Ago.” Science 340 (6130): 261–62.
https://doi.org/10.1126/science.340.6130.261.
Banci, Lucia, Ivano Bertini, Simone Ciofi-Baffoni, Leonardo Gonnelli, and Xun-Cheng
Su. 2003. “Structural Basis for the Function of the N-Terminal Domain of the
ATPase CopA from Bacillus Subtilis.” The Journal of Biological Chemistry 278
(50): 50506–13. https://doi.org/10.1074/jbc.M307389200.
180
Banerjee, Samiran, Klaus Schlaeppi, and Marcel G. A. van der Heijden. 2018.
“Keystone Taxa as Drivers of Microbiome Structure and Functioning.” Nature
Reviews Microbiology 16 (9): 567–76. https://doi.org/10.1038/s41579-018-
0024-1.
Bano, Nasreen, Angelique DeRae Smith, William Bennett, Lydia Vasquez, and James
T. Hollibaugh. 2007. “Dominance of Mycoplasma in the Guts of the Long-
Jawed Mudsucker, Gillichthys Mirabilis, from Five California Salt Marshes.”
Environmental Microbiology 9 (10): 2636–41. https://doi.org/10.1111/j.1462-
2920.2007.01381.x.
Barberán, Albert, Antoni Fernández-Guerra, Brendan J. M. Bohannan, and Emilio O.
Casamayor. 2012. “Exploration of Community Traits as Ecological Markers in
Microbial Metagenomes.” Molecular Ecology 21 (8): 1909–17.
https://doi.org/10.1111/j.1365-294X.2011.05383.x.
Behlau, Franklin, Blanca I. Canteros, Gerald V. Minsavage, Jeffrey B. Jones, and James
H. Graham. 2011. “Molecular Characterization of Copper Resistance Genes
from Xanthomonas Citri Subsp. Citri and Xanthomonas Alfalfae Subsp.
Citrumelonis▿.” Applied and Environmental Microbiology 77 (12): 4089–96.
https://doi.org/10.1128/AEM.03043-10.
Bejan, Dorin, and Nigel J. Bunce. 2015. “Acid Mine Drainage: Electrochemical
Approaches to Prevention and Remediation of Acidity and Toxic Metals.”
Journal of Applied Electrochemistry 45 (12): 1239–54.
https://doi.org/10.1007/s10800-015-0884-2.
Bélanger-Deschênes, Sébastien, Patrice Couture, Peter G. C. Campbell, and Louis
Bernatchez. 2013. “Evolutionary Change Driven by Metal Exposure as Revealed
by Coding SNP Genome Scan in Wild Yellow Perch (Perca Flavescens).”
Ecotoxicology 22 (5): 938–57. https://doi.org/10.1007/s10646-013-1083-8.
Belkaid, Yasmine, and Timothy Hand. 2014. “Role of the Microbiota in Immunity and
Inflammation.” Cell 157 (1): 121–41. https://doi.org/10.1016/j.cell.2014.03.011.
Belkaid, Yasmine, and Oliver J. Harrison. 2017. “Homeostatic Immunity and the
Microbiota.” Immunity 46 (4): 562–76.
https://doi.org/10.1016/j.immuni.2017.04.008.
Belkaid, Yasmine, and Julia A. Segre. 2014. “Dialogue between Skin Microbiota and
Immunity.” Science 346 (6212): 954–59.
https://doi.org/10.1126/science.1260144.
Bell, Graham. 2001. “Neutral Macroecology.” Science 293 (5539): 2413–18.
https://doi.org/10.1126/science.293.5539.2413.
Benjamini, Yoav, and Yosef Hochberg. 1995. “Controlling the False Discovery Rate: A
Practical and Powerful Approach to Multiple Testing.” Journal of the Royal
Statistical Society. Series B (Methodological) 57 (1): 289–300.
https://www.jstor.org/stable/2346101.
Benson, Dennis A., Mark Cavanaugh, Karen Clark, Ilene Karsch-Mizrachi, David J.
Lipman, James Ostell, and Eric W. Sayers. 2013. “GenBank.” Nucleic Acids
Research 41 (Database issue): D36-42. https://doi.org/10.1093/nar/gks1195.
Birky, C W, and J B Walsh. 1988. “Effects of Linkage on Rates of Molecular
Evolution.” Proceedings of the National Academy of Sciences of the United
States of America 85 (17): 6414–18.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC281982/.
Bissett, Andrew, Mark V. Brown, Steven D. Siciliano, and Peter H. Thrall. 2013.
“Microbial Community Responses to Anthropogenically Induced Environmental
181
Change: Towards a Systems Approach.” Edited by Marcel Holyoak. Ecology
Letters 16 (May): 128–39. https://doi.org/10.1111/ele.12109.
Bittar, Thais B., Aron Stubbins, Armando A.H. Vieira, and Kenneth Mopper. 2015.
“Characterization and Photodegradation of Dissolved Organic Matter (DOM)
from a Tropical Lake and Its Dominant Primary Producer, the Cyanobacteria
Microcystis Aeruginosa.” Marine Chemistry 177 (December): 205–17.
https://doi.org/10.1016/j.marchem.2015.06.016.
Bobay, Louis-Marie, and Howard Ochman. 2017. “The Evolution of Bacterial Genome
Architecture.” Frontiers in Genetics 8.
https://doi.org/10.3389/fgene.2017.00072.
Boisvert, Sébastien, Frédéric Raymond, Elénie Godzaridis, François Laviolette, and
Jacques Corbeil. 2012. “Ray Meta: Scalable de Novo Metagenome Assembly
and Profiling.” Genome Biology 13 (12): R122. https://doi.org/10.1186/gb-2012-
13-12-r122.
Bolnick, Daniel I., Lisa K. Snowberg, Philipp E. Hirsch, Christian L. Lauber, Elin Org,
Brian Parks, Aldons J. Lusis, Rob Knight, J. Gregory Caporaso, and Richard
Svanbäck. 2014. “Individual Diet Has Sex-Dependent Effects on Vertebrate Gut
Microbiota.” Nature Communications 5 (July): 4500.
https://doi.org/10.1038/ncomms5500.
Bougas, Bérénice, Eric Normandeau, Fabien Pierron, Peter G. C. Campbell, Louis
Bernatchez, and Patrice Couture. 2013a. “How Does Exposure to Nickel and
Cadmium Affect the Transcriptome of Yellow Perch (Perca Flavescens)--
Results from a 1000 Candidate-Gene Microarray.” Aquatic Toxicology
(Amsterdam, Netherlands) 142–143 (October): 355–64.
https://doi.org/10.1016/j.aquatox.2013.09.009.
Bougas, Bérénice, Eric Normandeau, Fabien Pierron, Peter G.C. Campbell, Louis
Bernatchez, and Patrice Couture. 2013b. “How Does Exposure to Nickel and
Cadmium Affect the Transcriptome of Yellow Perch (Perca Flavescens) –
Results from a 1000 Candidate-Gene Microarray.” Aquatic Toxicology 142–143
(October): 355–64. https://doi.org/10.1016/j.aquatox.2013.09.009.
Bourret, Vincent, Patrice Couture, Peter G.C. Campbell, and Louis Bernatchez. 2008.
“Evolutionary Ecotoxicology of Wild Yellow Perch (Perca Flavescens)
Populations Chronically Exposed to a Polymetallic Gradient.” Aquatic
Toxicology 86 (1): 76–90. https://doi.org/10.1016/j.aquatox.2007.10.003.
Boutin, Sébastien, Mohamed Alburaki, Pierre-Luc Mercier, Pierre Giovenazzo, and
Nicolas Derome. 2015. “Differential Gene Expression between Hygienic and
Non-Hygienic Honeybee (Apis Mellifera L.) Hives.” BMC Genomics 16 (1):
500. https://doi.org/10.1186/s12864-015-1714-y.
Bouvier, Thierry C., and Paul A. del Giorgio. 2002. “Compositional Changes in Free-
Living Bacterial Communities along a Salinity Gradient in Two Temperate
Estuaries.” Limnology and Oceanography 47 (2): 453–70.
https://doi.org/10.4319/lo.2002.47.2.0453.
Bowen, Jennifer L., Bess B. Ward, Hilary G. Morrison, John E. Hobbie, Ivan Valiela,
Linda A. Deegan, and Mitchell L. Sogin. 2011. “Microbial Community
Composition in Sediments Resists Perturbation by Nutrient Enrichment.” The
ISME Journal 5 (9): 1540–48. https://doi.org/10.1038/ismej.2011.22.
Bray, J. Roger, and J. T. Curtis. 1957. “An Ordination of the Upland Forest
Communities of Southern Wisconsin.” Ecological Monographs 27 (4): 325–49.
https://doi.org/10.2307/1942268.
182
Bridges, Kristin N., Yan Zhang, Thomas E. Curran, Jason T. Magnuson, Barney J.
Venables, Katherine E. Durrer, Michael S. Allen, and Aaron P. Roberts. 2018.
“Alterations to the Intestinal Microbiome and Metabolome of Pimephales
Promelas and Mus Musculus Following Exposure to Dietary Methylmercury.”
Environmental Science & Technology 52 (15): 8774–84.
https://doi.org/10.1021/acs.est.8b01150.
Brill, W. J. 1975. “Regulation and Genetics of Bacterial Nitrogen Fixation.” Annual
Review of Microbiology 29 (1): 109–29.
https://doi.org/10.1146/annurev.mi.29.100175.000545.
Brockhurst, Michael A., Nick Colegrave, and Daniel E. Rozen. 2011. “Next-Generation
Sequencing as a Tool to Study Microbial Evolution.” Molecular Ecology 20 (5):
972–80. https://doi.org/10.1111/j.1365-294X.2010.04835.x.
Brown, Bryan P., and Jennifer J. Wernegreen. 2019. “Genomic Erosion and Extensive
Horizontal Gene Transfer in Gut-Associated Acetobacteraceae.” BMC Genomics
20 (1): 472. https://doi.org/10.1186/s12864-019-5844-5.
Brown, Ryan M., Gregory D. Wiens, and Irene Salinas. 2018. “Analysis of the Gut and
Gill Microbiome of Resistant and Susceptible Lines of Rainbow Trout
(Oncorhynchus Mykiss).” Fish & Shellfish Immunology 86 (December): 497–
506. https://doi.org/10.1016/j.fsi.2018.11.079.
Bryant, Jessica A., Frank J. Stewart, John M. Eppley, and Edward F. DeLong. 2012.
“Microbial Community Phylogenetic and Trait Diversity Declines with Depth in
a Marine Oxygen Minimum Zone.” Ecology 93 (7): 1659–73.
https://doi.org/10.1890/11-1204.1.
Buchfink, Benjamin, Chao Xie, and Daniel H. Huson. 2015. “Fast and Sensitive Protein
Alignment Using DIAMOND.” Nature Methods 12 (1): 59–60.
https://doi.org/10.1038/nmeth.3176.
Buffie, Charlie G., and Eric G. Pamer. 2013a. “Microbiota-Mediated Colonization
Resistance against Intestinal Pathogens.” Nature Reviews Immunology 13 (11):
790–801. https://doi.org/10.1038/nri3535.
———. 2013b. “Microbiota-Mediated Colonization Resistance against Intestinal
Pathogens.” Nature Reviews. Immunology 13 (11): 790–801.
https://doi.org/10.1038/nri3535.
Burke, C., P. Steinberg, D. Rusch, S. Kjelleberg, and T. Thomas. 2011. “Bacterial
Community Assembly Based on Functional Genes Rather than Species.”
Proceedings of the National Academy of Sciences 108 (34): 14288–93.
https://doi.org/10.1073/pnas.1101591108.
Burns, Adam R., W. Zac Stephens, Keaton Stagaman, Sandi Wong, John F. Rawls,
Karen Guillemin, and Brendan JM Bohannan. 2016a. “Contribution of Neutral
Processes to the Assembly of Gut Microbial Communities in the Zebrafish over
Host Development.” The ISME Journal 10 (3): 655.
https://doi.org/10.1038/ismej.2015.142.
Burns, Adam R, W Zac Stephens, Keaton Stagaman, Sandi Wong, John F Rawls, Karen
Guillemin, and Brendan JM Bohannan. 2016b. “Contribution of Neutral
Processes to the Assembly of Gut Microbial Communities in the Zebrafish over
Host Development.” The ISME Journal 10 (3): 655–64.
https://doi.org/10.1038/ismej.2015.142.
Bush, William S., and Jason H. Moore. 2012. “Chapter 11: Genome-Wide Association
Studies.” PLOS Computational Biology 8 (12): e1002822.
https://doi.org/10.1371/journal.pcbi.1002822.
183
Cabrol, Léa, Franck Poly, Luc Malhautier, Thomas Pommier, Catherine Lerondelle,
Willy Verstraete, Anne-Sophie Lepeuple, Jean-Louis Fanlo, and Xavier Le
Roux. 2016. “Management of Microbial Communities through Transient
Disturbances Enhances the Functional Resilience of Nitrifying Gas-Biofilters to
Future Disturbances.” Environmental Science & Technology 50 (1): 338–48.
https://doi.org/10.1021/acs.est.5b02740.
Cadotte, Marc W., T. Jonathan Davies, James Regetz, Steven W. Kembel, Elsa Cleland,
and Todd H. Oakley. 2010. “Phylogenetic Diversity Metrics for Ecological
Communities: Integrating Species Richness, Abundance and Evolutionary
History.” Ecology Letters 13 (1): 96–105. https://doi.org/10.1111/j.1461-
0248.2009.01405.x.
Callahan, Benjamin J., Paul J. McMurdie, and Susan P. Holmes. 2017. “Exact Sequence
Variants Should Replace Operational Taxonomic Units in Marker-Gene Data
Analysis.” The ISME Journal 11 (12): 2639–43.
https://doi.org/10.1038/ismej.2017.119.
Callahan, Benjamin J., Paul J. McMurdie, Michael J. Rosen, Andrew W. Han, Amy Jo
A. Johnson, and Susan P. Holmes. 2016. “DADA2: High-Resolution Sample
Inference from Illumina Amplicon Data.” Nature Methods 13 (7): 581–83.
https://doi.org/10.1038/nmeth.3869.
Camacho, Christiam, George Coulouris, Vahram Avagyan, Ning Ma, Jason
Papadopoulos, Kevin Bealer, and Thomas L. Madden. 2009. “BLAST+:
Architecture and Applications.” BMC Bioinformatics 10 (December): 421.
https://doi.org/10.1186/1471-2105-10-421.
Campbell, Peter G. C., Anik Giguère, Emmanuelle Bonneris, and Landis Hare. 2005a.
“Cadmium-Handling Strategies in Two Chronically Exposed Indigenous
Freshwater Organisms—the Yellow Perch (Perca Flavescens) and the Floater
Mollusc (Pyganodon Grandis).” Aquatic Toxicology, Mechanisms in Metal
Toxicology, 72 (1–2): 83–97. https://doi.org/10.1016/j.aquatox.2004.11.023.
———. 2005b. “Cadmium-Handling Strategies in Two Chronically Exposed
Indigenous Freshwater Organisms—the Yellow Perch (Perca Flavescens) and
the Floater Mollusc (Pyganodon Grandis).” Aquatic Toxicology, Mechanisms in
Metal Toxicology, 72 (1): 83–97. https://doi.org/10.1016/j.aquatox.2004.11.023.
Cao, Yang, Xiaofei Zheng, Fei Li, and Xiaochen Bo. 2015. “Mmnet: An R Package for
Metagenomics Systems Biology Analysis.” BioMed Research International
2015: 167249. https://doi.org/10.1155/2015/167249.
Caporaso, J. G., C. L. Lauber, W. A. Walters, D. Berg-Lyons, C. A. Lozupone, P. J.
Turnbaugh, N. Fierer, and R. Knight. 2011. “Global Patterns of 16S RRNA
Diversity at a Depth of Millions of Sequences per Sample.” Proceedings of the
National Academy of Sciences 108 (Supplement_1): 4516–22.
https://doi.org/10.1073/pnas.1000080107.
Castillo-Ramírez, Santiago, Simon R. Harris, Matthew T. G. Holden, Miao He, Julian
Parkhill, Stephen D. Bentley, and Edward J. Feil. 2011. “The Impact of
Recombination on DN/DS within Recently Emerged Bacterial Clones.” PLOS
Pathog 7 (7): e1002129. https://doi.org/10.1371/journal.ppat.1002129.
“CCA: An R Package to Extend Canonical Correlation Analysis | González | Journal of
Statistical Software.” n.d. Accessed October 21, 2016.
https://www.jstatsoft.org/article/view/v023i12.
Cesare, Andrea Di, Ester Eckert, and Gianluca Corno. 2016. “Co-Selection of Antibiotic
and Heavy Metal Resistance in Freshwater Bacteria.” Journal of Limnology 75
(s2). https://www.jlimnol.it/index.php/jlimnol/article/view/jlimnol.2016.1198.
184
Cha, J. S., and D. A. Cooksey. 1991. “Copper Resistance in Pseudomonas Syringae
Mediated by Periplasmic and Outer Membrane Proteins.” Proceedings of the
National Academy of Sciences of the United States of America 88 (20): 8915–19.
Chao, Anne. 1984. “Nonparametric Estimation of the Number of Classes in a
Population.” Scandinavian Journal of Statistics 11 (4): 265–70.
http://www.jstor.org/stable/4615964.
Chao, Anne, and Shen-Ming Lee. 1992. “Estimating the Number of Classes via Sample
Coverage.” Journal of the American Statistical Association 87 (417): 210–17.
https://doi.org/10.2307/2290471.
Chao, Anne, and Mark C. K. Yang. 1993. “Stopping Rules and Estimation for
Recapture Debugging with Unequal Failure Rates.” Biometrika 80 (1): 193–201.
https://doi.org/10.2307/2336768.
Charlesworth, Brian, and Nick Barton. 2004. “Genome Size: Does Bigger Mean
Worse?” Current Biology: CB 14 (6): R233-235.
https://doi.org/10.1016/j.cub.2004.02.054.
Chase, Jonathan M. 2003. “Community Assembly: When Should History Matter?”
Oecologia 136 (4): 489–98. https://doi.org/10.1007/s00442-003-1311-7.
Chase, Jonathan M., and Mathew A. Leibold. 2003. Ecological Niches: Linking
Classical and Contemporary Approaches. Interspecific Interactions. Chicago:
University of Chicago Press.
Cheaib, Bachar, Malo Le Boulch, Pierre-Luc Mercier, and Nicolas Derome. 2018.
“Taxon-Function Decoupling as an Adaptive Signature of Lake Microbial
Metacommunities Under a Chronic Polymetallic Pollution Gradient.” Frontiers
in Microbiology 9. https://doi.org/10.3389/fmicb.2018.00869.
Chen, I.-Min A., Ken Chu, Krishna Palaniappan, Manoj Pillay, Anna Ratner, Jinghua
Huang, Marcel Huntemann, et al. 2019. “IMG/M v.5.0: An Integrated Data
Management and Comparative Analysis System for Microbial Genomes and
Microbiomes.” Nucleic Acids Research 47 (D1): D666–77.
https://doi.org/10.1093/nar/gky901.
Chen, Jun, Kyle Bittinger, Emily S. Charlson, Christian Hoffmann, James Lewis, Gary
D. Wu, Ronald G. Collman, Frederic D. Bushman, and Hongzhe Li. 2012.
“Associating Microbiome Composition with Environmental Covariates Using
Generalized UniFrac Distances.” Bioinformatics (Oxford, England) 28 (16):
2106–13. https://doi.org/10.1093/bioinformatics/bts342.
Chen, Lin-xing, Li-nan Huang, Celia Méndez-García, Jia-liang Kuang, Zheng-shuang
Hua, Jun Liu, and Wen-sheng Shu. 2016. “Microbial Communities, Processes
and Functions in Acid Mine Drainage Ecosystems.” Current Opinion in
Biotechnology 38 (April): 150–58. https://doi.org/10.1016/j.copbio.2016.01.013.
Chen, Wei, Clarence K. Zhang, Yongmei Cheng, Shaowu Zhang, and Hongyu Zhao.
2013. “A Comparison of Methods for Clustering 16S RRNA Sequences into
OTUs.” Edited by Maurizio Casiraghi. PLoS ONE 8 (8): e70837.
https://doi.org/10.1371/journal.pone.0070837.
Chen, Yi Ping, Qiang Liu, Yong Jun Liu, Feng An Jia, and Xin Hua He. 2014.
“Responses of Soil Microbial Activity to Cadmium Pollution and Elevated
CO2.” Scientific Reports 4 (March): 4287. https://doi.org/10.1038/srep04287.
Chesson, Peter. 2000. “Mechanisms of Maintenance of Species Diversity.” Annual
Review of Ecology and Systematics 31 (1): 343–66.
https://doi.org/10.1146/annurev.ecolsys.31.1.343.
185
Chesson, Peter L., and Robert R. Warner. 1981. “Environmental Variability Promotes
Coexistence in Lottery Competitive Systems.” The American Naturalist 117 (6):
923–43. https://doi.org/10.1086/283778.
Cho, Ilseung, Shingo Yamanishi, Laura Cox, Barbara A. Methé, Jiri Zavadil, Kelvin Li,
Zhan Gao, et al. 2012. “Antibiotics in Early Life Alter the Murine Colonic
Microbiome and Adiposity.” Nature 488 (7413): 621–26.
https://doi.org/10.1038/nature11400.
Chopp, D L, M J Kirisits, B. Moran, and M R Parsek. 2002. “A Mathematical Model of
Quorum Sensing in a Growing Bacterial Biofilm.” Journal of Industrial
Microbiology and Biotechnology 29 (6): 339–46.
https://doi.org/10.1038/sj.jim.7000316.
Connon, Stephanie A., and Stephen J. Giovannoni. 2002. “High-Throughput Methods
for Culturing Microorganisms in Very-Low-Nutrient Media Yield Diverse New
Marine Isolates.” Applied and Environmental Microbiology 68 (8): 3878–85.
https://doi.org/10.1128/aem.68.8.3878-3885.2002.
Connor, Nora, Johannes Sikorski, Alejandro P. Rooney, Sarah Kopac, Alexander F.
Koeppel, Andrew Burger, Scott G. Cole, et al. 2010. “Ecology of Speciation in
the Genus Bacillus.” Applied and Environmental Microbiology 76 (5): 1349–58.
https://doi.org/10.1128/AEM.01988-09.
Costello, Elizabeth K., Keaton Stagaman, Les Dethlefsen, Brendan J. M. Bohannan, and
David A. Relman. 2012. “The Application of Ecological Theory Toward an
Understanding of the Human Microbiome.” Science 336 (6086): 1255–62.
https://doi.org/10.1126/science.1224203.
Couture, Patrice, and Puja Rajender Kumar. 2003. “Impairment of Metabolic Capacities
in Copper and Cadmium Contaminated Wild Yellow Perch (Perca Flavescens).”
Aquatic Toxicology 64 (1): 107–20. https://doi.org/10.1016/S0166-
445X(03)00028-6.
Couture, Patrice, James W. Rajotte, and Gregory Pyle. 2008. “Seasonal and Regional
Variations in Metal Contamination and Condition Indicators in Yellow Perch
(Perca Flavescens) along Two Polymetallic Gradients. III. Energetic and
Physiological Indicators.” http://www.uleth.ca/dspace/handle/10133/3445.
Coyte, K. Z., J. Schluter, and K. R. Foster. 2015. “The Ecology of the Microbiome:
Networks, Competition, and Stability.” Science 350 (6261): 663–66.
https://doi.org/10.1126/science.aad2602.
Crall, James D., Callin M. Switzer, Robert L. Oppenheimer, Ashlee N. Ford Versypt,
Biswadip Dey, Andrea Brown, Mackay Eyster, et al. 2018. “Neonicotinoid
Exposure Disrupts Bumblebee Nest Behavior, Social Networks, and
Thermoregulation.” Science 362 (6415): 683–86.
https://doi.org/10.1126/science.aat1598.
Croom, Edward. 2012. “Chapter Three - Metabolism of Xenobiotics of Human
Environments.” In Progress in Molecular Biology and Translational Science,
edited by Ernest Hodgson, 112:31–88. Toxicology and Human Environments.
Academic Press. https://doi.org/10.1016/B978-0-12-415813-9.00003-9.
Crow, James F. 2010. “Wright and Fisher on Inbreeding and Random Drift.” Genetics
184 (3): 609–11. https://doi.org/10.1534/genetics.109.110023.
Crutzen, P.J., and E.F. Stoermer. 2000. “The Anthropocene.” Global Change
Newsletter, 2000, sec. no41.
Curnow, R. N. 1985. “Mathematical Models in Agriculture. A Quantitative Approach to
Problems in Agriculture and Related Sciences: France, J. and Thornley, J. H. M.
186
Butterworths, London, 1984. Price: £35·00.” Agricultural Systems 18 (1): 65–
66. https://doi.org/10.1016/0308-521X(85)90058-7.
Dabbs, Eric R., and Gillian J. Sole. 1988. “Plasmid-Borne Resistance to Arsenate,
Arsenite, Cadmium, and Chloramphenicol in a Rhodococcus Species.”
Molecular and General Genetics MGG 211 (1): 148–54.
https://doi.org/10.1007/BF00338406.
Dagan, T., Y. Artzy-Randrup, and W. Martin. 2008. “Modular Networks and
Cumulative Impact of Lateral Transfer in Prokaryote Genome Evolution.”
Proceedings of the National Academy of Sciences 105 (29): 10039–44.
https://doi.org/10.1073/pnas.0800679105.
Dalia, Ankur B., Kimberley D. Seed, Stephen B. Calderwood, and Andrew Camilli.
2015. “A Globally Distributed Mobile Genetic Element Inhibits Natural
Transformation of Vibrio Cholerae.” Proceedings of the National Academy of
Sciences 112 (33): 10485–90. https://doi.org/10.1073/pnas.1509097112.
Daubin, Vincent, and Nancy A. Moran. 2004. “Comment on ‘The Origins of Genome
Complexity.’” Science (New York, N.Y.) 306 (5698): 978; author reply 978.
https://doi.org/10.1126/science.1098469.
Dautremepuits, Claire, David J. Marcogliese, Andrée D. Gendron, and Michel Fournier.
2009. “Gill and Head Kidney Antioxidant Processes and Innate Immune System
Responses of Yellow Perch (Perca Flavescens) Exposed to Different
Contaminants in the St. Lawrence River, Canada.” Science of The Total
Environment 407 (3): 1055–64. https://doi.org/10.1016/j.scitotenv.2008.10.004.
Dawson, K. J. 1998. “Evolutionarily Stable Mutation Rates.” Journal of Theoretical
Biology 194 (1): 143–57. https://doi.org/10.1006/jtbi.1998.0752.
De Vrieze, Jo, Marlies E. R. Christiaens, Diego Walraedt, Arno Devooght, Umer
Zeeshan Ijaz, and Nico Boon. 2017. “Microbial Community Redundancy in
Anaerobic Digestion Drives Process Recovery after Salinity Exposure.” Water
Research 111 (March): 109–17. https://doi.org/10.1016/j.watres.2016.12.042.
DeAngelis, Donald L., and Wolf M. Mooij. 2005. “Individual-Based Modeling of
Ecological and Evolutionary Processes.” Annual Review of Ecology, Evolution,
and Systematics 36 (1): 147–68.
https://doi.org/10.1146/annurev.ecolsys.36.102003.152644.
Debroas, Didier, Jean-François Humbert, François Enault, Gisèle Bronner, Michael
Faubladier, and Emmanuel Cornillot. 2009. “Metagenomic Approach Studying
the Taxonomic and Functional Diversity of the Bacterial Community in a
Mesotrophic Lake (Lac Du Bourget - France).” Environmental Microbiology 11
(9): 2412–24. https://doi.org/10.1111/j.1462-2920.2009.01969.x.
Defo, M. A., P. A. Spear, and P. Couture. 2014. “Consequences of Metal Exposure on
Retinoid Metabolism in Vertebrates: A Review.” Toxicology Letters 225 (1): 1–
11. https://doi.org/10.1016/j.toxlet.2013.11.024.
Defo, Michel A., Louis Bernatchez, Peter G.C. Campbell, and Patrice Couture. 2014.
“Waterborne Cadmium and Nickel Impact Oxidative Stress Responses and
Retinoid Metabolism in Yellow Perch.” Aquatic Toxicology 154 (September):
207–20. https://doi.org/10.1016/j.aquatox.2014.05.024.
Delgado-Baquerizo, Manuel, Angela M. Oliverio, Tess E. Brewer, Alberto Benavent-
González, David J. Eldridge, Richard D. Bardgett, Fernando T. Maestre, Brajesh
K. Singh, and Noah Fierer. 2018. “A Global Atlas of the Dominant Bacteria
Found in Soil.” Science 359 (6373): 320–25.
https://doi.org/10.1126/science.aap9516.
187
Desoeuvre, Angélique, Corinne Casiot, and Marina Héry. 2016. “Diversity and
Distribution of Arsenic-Related Genes Along a Pollution Gradient in a River
Affected by Acid Mine Drainage.” Microbial Ecology 71 (3): 672–85.
https://doi.org/10.1007/s00248-015-0710-8.
Dethlefsen, Les, and David A. Relman. 2011. “Incomplete Recovery and Individualized
Responses of the Human Distal Gut Microbiota to Repeated Antibiotic
Perturbation.” Proceedings of the National Academy of Sciences 108
(Supplement 1): 4554–61. https://doi.org/10.1073/pnas.1000087107.
Devarajan, Naresh, Amandine Laffite, Neil D. Graham, Maria Meijer, Kandasamy
Prabakar, Josué I. Mubedi, Vicky Elongo, et al. 2015. “Accumulation of
Clinically Relevant Antibiotic-Resistance Genes, Bacterial Load, and Metals in
Freshwater Lake Sediments in Central Europe.” Environmental Science &
Technology 49 (11): 6528–37. https://doi.org/10.1021/acs.est.5b01031.
Di Prisco, G., V. Cavaliere, D. Annoscia, P. Varricchio, E. Caprio, F. Nazzi, G.
Gargiulo, and F. Pennacchio. 2013. “Neonicotinoid Clothianidin Adversely
Affects Insect Immunity and Promotes Replication of a Viral Pathogen in Honey
Bees.” Proceedings of the National Academy of Sciences 110 (46): 18466–71.
https://doi.org/10.1073/pnas.1314923110.
Dimkpa, C.o., D. Merten, A. Svatoš, G. Büchel, and E. Kothe. 2009. “Siderophores
Mediate Reduced and Increased Uptake of Cadmium by Streptomyces Tendae
F4 and Sunflower (Helianthus Annuus), Respectively.” Journal of Applied
Microbiology 107 (5): 1687–96. https://doi.org/10.1111/j.1365-
2672.2009.04355.x.
Doebeli, Michael. 2011. Adaptive Diversification (MPB-48). Princeton University
Press. https://doi.org/10.2307/j.ctt7rgw4.
Doebeli, Michael, and Iaroslav Ispolatov. 2010. “Complexity and Diversity.” Science
328 (5977): 494–97. https://doi.org/10.1126/science.1187468.
Donachie, W. D., and K. J. Begg. 1970. “Growth of the Bacterial Cell.” Nature 227
(5264): 1220–24. https://doi.org/10.1038/2271220a0.
Doolittle, W. Ford, and Olga Zhaxybayeva. 2009. “On the Origin of Prokaryotic
Species.” Genome Research 19 (5): 744–56.
https://doi.org/10.1101/gr.086645.108.
Dos Santos, Patricia C., Zhong Fang, Steven W. Mason, João C. Setubal, and Ray
Dixon. 2012. “Distribution of Nitrogen Fixation and Nitrogenase-like Sequences
amongst Microbial Genomes.” BMC Genomics 13 (1): 162.
https://doi.org/10.1186/1471-2164-13-162.
Doublet, Vincent, Maureen Labarussias, Joachim R. de Miranda, Robin F. A. Moritz,
and Robert J. Paxton. 2015. “Bees under Stress: Sublethal Doses of a
Neonicotinoid Pesticide and Pathogens Interact to Elevate Honey Bee Mortality
across the Life Cycle: Pesticide-Pathogen Interactions in Honey Bees.”
Environmental Microbiology 17 (4): 969–83. https://doi.org/10.1111/1462-
2920.12426.
Douglas, Thomas A., Lisa L. Loseto, Robie W. Macdonald, Peter Outridge, Aurélien
Dommergue, Alexandre Poulain, Marc Amyot, et al. 2012. “The Fate of
Mercury in Arctic Terrestrial and Aquatic Ecosystems, a Review.”
Environmental Chemistry 9 (4): 321. https://doi.org/10.1071/EN11140.
Drake, James A. 1991. “Community-Assembly Mechanics and the Structure of an
Experimental Species Ensemble.” The American Naturalist 137 (1): 1–26.
http://www.jstor.org/stable/2462154.
188
Drossel, BARBARA, PAUL G. Higgs, and ALAN J. Mckane. 2001. “The Influence of
Predator–Prey Population Dynamics on the Long-Term Evolution of Food Web
Structure.” Journal of Theoretical Biology 208 (1): 91–107.
https://doi.org/10.1006/jtbi.2000.2203.
Dufresne, Alexis, Laurence Garczarek, and Frédéric Partensky. 2005. “Accelerated
Evolution Associated with Genome Reduction in a Free-Living Prokaryote.”
Genome Biology 6 (2): R14. https://doi.org/10.1186/gb-2005-6-2-r14.
Dutilh, Bas E, Cristiane C Thompson, Ana CP Vicente, Michel A Marin, Clarence Lee,
Genivaldo GZ Silva, Robert Schmieder, et al. 2014. “Comparative Genomics of
274 Vibrio Cholerae Genomes Reveals Mobile Functions Structuring Three
Niche Dimensions.” BMC Genomics 15 (1): 654. https://doi.org/10.1186/1471-
2164-15-654.
Dykhuizen, Daniel E. 1990. “Experimental Studies of Natural Selection in Bacteria.”
Annual Review of Ecology and Systematics 21 (1): 373–98.
https://doi.org/10.1146/annurev.es.21.110190.002105.
Ebrahimi, Ali, Julia Schwartzman, and Otto X. Cordero. 2019. “Cooperation and Spatial
Self-Organization Determine Ecosystem Function for Polysaccharide-Degrading
Bacteria.” BioRxiv, May, 640961. https://doi.org/10.1101/640961.
Edelstein-Keshet, Leah. 2005. Mathematical Models in Biology. Classics in Applied
Mathematics. Society for Industrial and Applied Mathematics.
https://doi.org/10.1137/1.9780898719147.
Edgar, Robert C. 2016. “UNOISE2: Improved Error-Correction for Illumina 16S and
ITS Amplicon Sequencing.” BioRxiv, October, 081257.
https://doi.org/10.1101/081257.
———. 2018. “Updating the 97% Identity Threshold for 16S Ribosomal RNA OTUs.”
Bioinformatics 34 (14): 2371–75. https://doi.org/10.1093/bioinformatics/bty113.
Edwards, K. J, and D. A Bazylinski. 2008. “Intracellular Minerals and Metal Deposits
in Prokaryotes.” Geobiology 6 (3): 309–17. https://doi.org/10.1111/j.1472-
4669.2008.00156.x.
Elena, Santiago F., and Richard E. Lenski. 2003. “Evolution Experiments with
Microorganisms: The Dynamics and Genetic Bases of Adaptation.” Nature
Reviews Genetics 4 (6): 457–69. https://doi.org/10.1038/nrg1088.
Ellison, Christopher E., Charles Hall, David Kowbel, Juliet Welch, Rachel B. Brem, N.
L. Glass, and John W. Taylor. 2011. “Population Genomics and Local
Adaptation in Wild Isolates of a Model Microbial Eukaryote.” Proceedings of
the National Academy of Sciences of the United States of America 108 (7):
2831–36. https://doi.org/10.1073/pnas.1014971108.
Etienne, Rampal S. 2005. “A New Sampling Formula for Neutral Biodiversity.”
Ecology Letters 8 (3): 253–60. https://doi.org/10.1111/j.1461-
0248.2004.00717.x.
Ewens, Warren J. 2004. Mathematical Population Genetics 1: Theoretical Introduction.
2nd ed. Interdisciplinary Applied Mathematics, Mathematical Population
Genetics. New York: Springer-Verlag.
https://www.springer.com/gp/book/9780387201917.
Falkowski, Paul G., Tom Fenchel, and Edward F. Delong. 2008. “The Microbial
Engines That Drive Earth’s Biogeochemical Cycles.” Science 320 (5879): 1034–
39. https://doi.org/10.1126/science.1153213.
Farahpour, Farnoush, Mohammadkarim Saeedghalati, Verena Brauer, and Daniel
Hoffmann. 2018. “Trade-off Shapes Diversity in Eco-Evolutionary Dynamics.”
ELife. August 17, 2018. https://doi.org/10.7554/eLife.36273.
189
Faust, Karoline, Franziska Bauchinger, Béatrice Laroche, Sophie de Buyl, Leo Lahti,
Alex D. Washburne, Didier Gonze, and Stefanie Widder. 2018. “Signatures of
Ecological Processes in Microbial Community Time Series.” Microbiome 6 (1):
120. https://doi.org/10.1186/s40168-018-0496-2.
Faust, Karoline, and Jeroen Raes. 2012. “Microbial Interactions: From Networks to
Models.” Nature Reviews Microbiology 10 (8): 538–50.
https://doi.org/10.1038/nrmicro2832.
Feist, Adam M., Markus J. Herrgård, Ines Thiele, Jennie L. Reed, and Bernhard Ø
Palsson. 2009. “Reconstruction of Biochemical Networks in Microorganisms.”
Nature Reviews. Microbiology 7 (2): 129–43.
https://doi.org/10.1038/nrmicro1949.
Ferrer, Jordi, Clara Prats, and Daniel López. 2008. “Individual-Based Modelling: An
Essential Tool for Microbiology.” Journal of Biological Physics 34 (1–2): 19–
37. https://doi.org/10.1007/s10867-008-9082-3.
Fierer, Noah. 2017. “Embracing the Unknown: Disentangling the Complexities of the
Soil Microbiome.” Nature Reviews Microbiology 15 (10): 579–90.
https://doi.org/10.1038/nrmicro.2017.87.
Fierer, Noah, Christian L. Lauber, Kelly S. Ramirez, Jesse Zaneveld, Mark A. Bradford,
and Rob Knight. 2012. “Comparative Metagenomic, Phylogenetic and
Physiological Analyses of Soil Microbial Communities across Nitrogen
Gradients.” The ISME Journal 6 (5): 1007–17.
https://doi.org/10.1038/ismej.2011.159.
Fierer, Noah, Jonathan W. Leff, Byron J. Adams, Uffe N. Nielsen, Scott Thomas Bates,
Christian L. Lauber, Sarah Owens, Jack A. Gilbert, Diana H. Wall, and J.
Gregory Caporaso. 2012. “Cross-Biome Metagenomic Analyses of Soil
Microbial Communities and Their Functional Attributes.” Proceedings of the
National Academy of Sciences 109 (52): 21390–95.
https://doi.org/10.1073/pnas.1215210110.
Fitzpatrick, Mark J., Elah Feder, Locke Rowe, and Marla B. Sokolowski. 2007.
“Maintaining a Behaviour Polymorphism by Frequency-Dependent Selection on
a Single Gene.” Nature 447 (7141): 210–12.
https://doi.org/10.1038/nature05764.
Fleischmann, R. D., M. D. Adams, O. White, R. A. Clayton, E. F. Kirkness, A. R.
Kerlavage, C. J. Bult, J. F. Tomb, B. A. Dougherty, and J. M. Merrick. 1995.
“Whole-Genome Random Sequencing and Assembly of Haemophilus
Influenzae Rd.” Science (New York, N.Y.) 269 (5223): 496–512.
https://doi.org/10.1126/science.7542800.
Flemming, Hans-Curt, Jost Wingender, Ulrich Szewzyk, Peter Steinberg, Scott A. Rice,
and Staffan Kjelleberg. 2016. “Biofilms: An Emergent Form of Bacterial Life.”
Nature Reviews Microbiology 14 (9): 563–75.
https://doi.org/10.1038/nrmicro.2016.94.
Forsberg, Kevin J., Sanket Patel, Molly K. Gibson, Christian L. Lauber, Rob Knight,
Noah Fierer, and Gautam Dantas. 2014. “Bacterial Phylogeny Structures Soil
Resistomes across Habitats.” Nature 509 (7502): 612–16.
https://doi.org/10.1038/nature13377.
Foster, James A., Stephen M. Krone, and Larry J. Forney. 2008. “Application of
Ecological Network Theory to the Human Microbiome.” Research article.
Interdisciplinary Perspectives on Infectious Diseases. 2008.
https://doi.org/10.1155/2008/839501.
190
Fraser, D J, L K Weir, L Bernatchez, M M Hansen, and E B Taylor. 2011. “Extent and
Scale of Local Adaptation in Salmonid Fishes: Review and Meta-Analysis.”
Heredity 106 (3): 404–20. https://doi.org/10.1038/hdy.2010.167.
Frost, Laura S., Raphael Leplae, Anne O. Summers, and Ariane Toussaint. 2005.
“Mobile Genetic Elements: The Agents of Open Source Evolution.” Nature
Reviews Microbiology 3 (9): 722–32. https://doi.org/10.1038/nrmicro1235.
Fuhrer, Tobias, Eliane Fischer, and Uwe Sauer. 2005. “Experimental Identification and
Quantification of Glucose Metabolism in Seven Bacterial Species.” Journal of
Bacteriology 187 (5): 1581–90. https://doi.org/10.1128/JB.187.5.1581-
1590.2005.
Gac, Mickaël Le, Jessica Plucain, Thomas Hindré, Richard E. Lenski, and Dominique
Schneider. 2012. “Ecological and Evolutionary Dynamics of Coexisting
Lineages during a Long-Term Experiment with Escherichia Coli.” Proceedings
of the National Academy of Sciences 109 (24): 9487–92.
https://doi.org/10.1073/pnas.1207091109.
Garcia-Etxebarria, Koldo, Marc Garcia-Garcerà, and Francesc Calafell. 2014.
“Consistency of Metagenomic Assignment Programs in Simulated and Real
Data.” BMC Bioinformatics 15 (1): 90. http://www.biomedcentral.com/1471-
2105/15/90.
Giguère, Anik, Peter G. C. Campbell, Landis Hare, and Patrice Couture. 2006. “Sub-
Cellular Partitioning of Cadmium, Copper, Nickel and Zinc in Indigenous
Yellow Perch (Perca Flavescens) Sampled along a Polymetallic Gradient.”
Aquatic Toxicology 77 (2): 178–89.
https://doi.org/10.1016/j.aquatox.2005.12.001.
Giguère, Anik, Peter GC Campbell, Landis Hare, D Gordon McDonald, and Joseph B
Rasmussen. 2004. “Influence of Lake Chemistry and Fish Age on Cadmium,
Copper, and Zinc Concentrations in Various Organs of Indigenous Yellow Perch
(Perca Flavescens).” Canadian Journal of Fisheries and Aquatic Sciences 61
(9): 1702–16. https://doi.org/10.1139/f04-100.
Gillespie, J. J., J. S. Johnston, J. J. Cannone, and R. R. Gutell. 2006. “Characteristics of
the Nuclear (18S, 5.8 S, 28S and 5S) and Mitochondrial (12S and 16S) RRNA
Genes of Apis Mellifera (Insecta: Hymenoptera): Structure, Organization, and
Retrotransposable Elements.” Insect Molecular Biology 15 (5): 657–686.
http://onlinelibrary.wiley.com/doi/10.1111/j.1365-2583.2006.00689.x/full.
Gillham, Nicholas W. 2015. “The Battle Between the Biometricians and the
Mendelians: How Sir Francis Galton’s Work Caused His Disciples to Reach
Conflicting Conclusions About the Hereditary Mechanism.” Science &
Education 24 (1): 61–75. https://doi.org/10.1007/s11191-013-9642-1.
Gillings, Michael R. 2013. “Evolutionary Consequences of Antibiotic Use for the
Resistome, Mobilome and Microbial Pangenome.” Frontiers in Microbiology 4.
https://doi.org/10.3389/fmicb.2013.00004.
Ginzburg, Lev R., H. Reşit Akçakaya, and Junhyong Kim. 1988. “Evolution of
Community Structure: Competition.” Journal of Theoretical Biology 133 (4):
513–23. https://doi.org/10.1016/S0022-5193(88)80338-2.
Giovannoni, Stephen J., H. James Tripp, Scott Givan, Mircea Podar, Kevin L. Vergin,
Damon Baptista, Lisa Bibbs, et al. 2005. “Genome Streamlining in a
Cosmopolitan Oceanic Bacterium.” Science 309 (5738): 1242–45.
https://doi.org/10.1126/science.1114057.
191
Givens, Ce, B Ransom, N Bano, and Jt Hollibaugh. 2015. “Comparison of the Gut
Microbiomes of 12 Bony Fish and 3 Shark Species.” Marine Ecology Progress
Series 518 (January): 209–23. https://doi.org/10.3354/meps11034.
Glass, Elizabeth M., Jared Wilkening, Andreas Wilke, Dionysios Antonopoulos, and
Folker Meyer. 2010. “Using the Metagenomics RAST Server (MG-RAST) for
Analyzing Shotgun Metagenomes.” Cold Spring Harbor Protocols 2010 (1):
pdb.prot5368. https://doi.org/10.1101/pdb.prot5368.
Glassing, Angela, Scot E. Dowd, Susan Galandiuk, Brian Davis, Jeffrey R. Jorden, and
Rodrick J. Chiodini. 2015. “Changes in 16s RNA Gene Microbial Community
Profiling by Concentration of Prokaryotic DNA.” Journal of Microbiological
Methods 119 (December): 239–42. https://doi.org/10.1016/j.mimet.2015.11.001.
Glassman, Sydney I., and Jennifer B. H. Martiny. 2018. “Broadscale Ecological Patterns
Are Robust to Use of Exact Sequence Variants versus Operational Taxonomic
Units.” MSphere 3 (4): e00148-18. https://doi.org/10.1128/mSphere.00148-18.
Goffeau, A., B. G. Barrell, H. Bussey, R. W. Davis, B. Dujon, H. Feldmann, F. Galibert,
et al. 1996. “Life with 6000 Genes.” Science 274 (5287): 546–67.
https://doi.org/10.1126/science.274.5287.546.
Goldfarb, Katherine C., Ulas Karaoz, China A. Hanson, Clark A. Santee, Mark A.
Bradford, Kathleen K. Treseder, Matthew D. Wallenstein, and Eoin L. Brodie.
2011. “Differential Growth Responses of Soil Bacterial Taxa to Carbon
Substrates of Varying Chemical Recalcitrance.” Terrestrial Microbiology 2: 94.
https://doi.org/10.3389/fmicb.2011.00094.
González, Ignacio, and Sébastien Déjean. 2012. CCA: Canonical Correlation Analysis
(version 1.2). https://cran.r-project.org/web/packages/CCA/index.html.
González, Ignacio, Sébastien Déjean, Pascal Martin, and Alain Baccini. 2008. “CCA:
An R Package to Extend Canonical Correlation Analysis.” Journal of Statistical
Software 23 (12). https://doi.org/10.18637/jss.v023.i12.
González-Torres, Pedro, Francisco Rodríguez-Mateos, Josefa Antón, and Toni
Gabaldón. 2019. “Impact of Homologous Recombination on the Evolution of
Prokaryotic Core Genomes.” MBio 10 (1): e02494-18.
https://doi.org/10.1128/mBio.02494-18.
Good, Benjamin H., Michael J. McDonald, Jeffrey E. Barrick, Richard E. Lenski, and
Michael M. Desai. 2017. “The Dynamics of Molecular Evolution Over 60,000
Generations.” Nature 551 (7678): 45–50. https://doi.org/10.1038/nature24287.
Goodwin, Sara, John D. McPherson, and W. Richard McCombie. 2016. “Coming of
Age: Ten Years of next-Generation Sequencing Technologies.” Nature Reviews
Genetics 17 (6): 333–51. https://doi.org/10.1038/nrg.2016.49.
Gough, Heidi L., Amy L. Dahl, Melissa A. Nolan, Jean-François Gaillard, and David A.
Stahl. 2008. “Metal Impacts on Microbial Biomass in the Anoxic Sediments of a
Contaminated Lake.” Journal of Geophysical Research: Biogeosciences 113
(G2): G02017. https://doi.org/10.1029/2007JG000566.
Gould, Alison L., Vivian Zhang, Lisa Lamberti, Eric W. Jones, Benjamin Obadia,
Nikolaos Korasidis, Alex Gavryushkin, Jean M. Carlson, Niko Beerenwinkel,
and William B. Ludington. 2018. “Microbiome Interactions Shape Host
Fitness.” Proceedings of the National Academy of Sciences 115 (51): E11951–
60. https://doi.org/10.1073/pnas.1809349115.
Gouy, M., and C. Gautier. 1982. “Codon Usage in Bacteria: Correlation with Gene
Expressivity.” Nucleic Acids Research 10 (22): 7055–74.
https://doi.org/10.1093/nar/10.22.7055.
192
Gravel, Dominique, Thomas Bell, Claire Barbera, Thierry Bouvier, Thomas Pommier,
Patrick Venail, and Nicolas Mouquet. 2011. “Experimental Niche Evolution
Alters the Strength of the Diversity-Productivity Relationship.” Nature 469
(7328): 89–92. https://doi.org/10.1038/nature09592.
Green, Jessica L., Brendan J. M. Bohannan, and Rachel J. Whitaker. 2008. “Microbial
Biogeography: From Taxonomy to Traits.” Science 320 (5879): 1039–43.
https://doi.org/10.1126/science.1153475.
Griffiths, Bryan S., and Laurent Philippot. 2013. “Insights into the Resistance and
Resilience of the Soil Microbial Community.” FEMS Microbiology Reviews 37
(2): 112–29. https://doi.org/10.1111/j.1574-6976.2012.00343.x.
Grimm, V., and Christian Wissel. 1997. “Babel, or the Ecological Stability Discussions:
An Inventory and Analysis of Terminology and a Guide for Avoiding
Confusion.” Oecologia 109 (3): 323–34.
https://doi.org/10.1007/s004420050090.
Grimm, Volker, and Steven F. Railsback. 2005. Individual-Based Modeling and
Ecology: STU-Student edition. Princeton University Press.
http://www.jstor.org/stable/j.ctt5hhnk8.
Guerraggio, Angelo, and Giovanni Paoloni. 2013. Vito Volterra. Berlin Heidelberg:
Springer-Verlag. https://www.springer.com/gp/book/9783642272622.
Guo, Feng, Feng Ju, Lin Cai, and Tong Zhang. 2013. “Taxonomic Precision of
Different Hypervariable Regions of 16S RRNA Gene and Annotation Methods
for Functional Bacterial Groups in Biological Wastewater Treatment.” Edited by
André O. Hudson. PLoS ONE 8 (10): e76185.
https://doi.org/10.1371/journal.pone.0076185.
Gupta, Radhey S., and Emily Lorenzini. 2007. “Phylogeny and Molecular Signatures
(Conserved Proteins and Indels) That Are Specific for the Bacteroidetes and
Chlorobi Species.” BMC Evolutionary Biology 7 (May): 71.
https://doi.org/10.1186/1471-2148-7-71.
Gurney, W. S. C, and R. M Nisbet. 1998. Ecological Dynamics. New York : Oxford
University Press. https://trove.nla.gov.au/version/29163357.
Haller, Benjamin C. 2014. “Theoretical and Empirical Perspectives in Ecology and
Evolution: A Survey.” BioScience 64 (10): 907–16.
https://doi.org/10.1093/biosci/biu131.
Hammers, Matthew D., Michael J. Taormina, Matthew M. Cerda, Leticia A. Montoya,
Daniel T. Seidenkranz, Raghuveer Parthasarathy, and Michael D. Pluth. 2015.
“A Bright Fluorescent Probe for H2S Enables Analyte-Responsive, 3D Imaging
in Live Zebrafish Using Light Sheet Fluorescence Microscopy.” Journal of the
American Chemical Society 137 (32): 10216–23.
https://doi.org/10.1021/jacs.5b04196.
Hanemaaijer, Mark, Wilfred F. M. Röling, Brett G. Olivier, Ruchir A. Khandelwal, Bas
Teusink, and Frank J. Bruggeman. 2015. “Systems Modeling Approaches for
Microbial Community Studies: From Metagenomics to Inference of the
Community Structure.” Frontiers in Microbiology 6: 213.
https://doi.org/10.3389/fmicb.2015.00213.
Harcombe, William R., William J. Riehl, Ilija Dukovski, Brian R. Granger, Alex Betts,
Alex H. Lang, Gracia Bonilla, et al. 2014. “Metabolic Resource Allocation in
Individual Microbes Determines Ecosystem Interactions and Spatial Dynamics.”
Cell Reports 7 (4): 1104–15. https://doi.org/10.1016/j.celrep.2014.03.070.
Harris, K., T. L. Parsons, U. Z. Ijaz, L. Lahti, I. Holmes, and C. Quince. 2017. “Linking
Statistical and Ecological Theory: Hubbell #x0027;s Unified Neutral Theory of
193
Biodiversity as a Hierarchical Dirichlet Process.” Proceedings of the IEEE 105
(3): 516–29. https://doi.org/10.1109/JPROC.2015.2428213.
Hellweger, Ferdi L., Robert J. Clegg, James R. Clark, Caroline M. Plugge, and Jan-
Ulrich Kreft. 2016. “Advancing Microbial Sciences by Individual-Based
Modelling.” Nature Reviews. Microbiology 14 (7): 461–71.
https://doi.org/10.1038/nrmicro.2016.62.
Helmink, Beth A., M. A. Wadud Khan, Amanda Hermann, Vancheswaran
Gopalakrishnan, and Jennifer A. Wargo. 2019. “The Microbiome, Cancer, and
Cancer Therapy.” Nature Medicine 25 (3): 377–88.
https://doi.org/10.1038/s41591-019-0377-7.
Hemme, Christopher L, Ye Deng, Terry J Gentry, Matthew W Fields, Liyou Wu,
Soumitra Barua, Kerrie Barry, et al. 2010. “Metagenomic Insights into Evolution
of a Heavy Metal-Contaminated Groundwater Microbial Community.” The
ISME Journal 4 (5): 660–672.
Hemme, Christopher L., Stefan J. Green, Lavanya Rishishwar, Om Prakash, Angelica
Pettenato, Romy Chakraborty, Adam M. Deutschbauer, et al. 2016. “Lateral
Gene Transfer in a Heavy Metal-Contaminated-Groundwater Microbial
Community.” MBio 7 (2): e02234-15. https://doi.org/10.1128/mBio.02234-15.
Hendry, Tory A., Lindsay L. Freed, Dana Fader, Danté Fenolio, Tracey T. Sutton, and
Jose V. Lopez. 2018. “Ongoing Transposon-Mediated Genome Reduction in the
Luminous Bacterial Symbionts of Deep-Sea Ceratioid Anglerfishes.” MBio 9
(3): e01033-18. https://doi.org/10.1128/mBio.01033-18.
Henry, Christopher S, Matthew DeJongh, Aaron A Best, Paul M Frybarger, Ben Linsay,
and Rick L Stevens. 2010. “High-Throughput Generation, Optimization and
Analysis of Genome-Scale Metabolic Models.” Nature Biotechnology 28 (9):
977–82. https://doi.org/10.1038/nbt.1672.
Henson, Michael A. 2015. “Genome-Scale Modeling of Microbial Metabolism with
Temporal and Spatial Resolution.” Biochemical Society Transactions 43 (6):
1164–71. https://doi.org/10.1042/BST20150146.
Hershberg, Ruth. 2015. “Mutation—The Engine of Evolution: Studying Mutation and
Its Role in the Evolution of Bacteria.” Cold Spring Harbor Perspectives in
Biology 7 (9). https://doi.org/10.1101/cshperspect.a018077.
Hershberg, Ruth, Hua Tang, and Dmitri A. Petrov. 2007. “Reduced Selection Leads to
Accelerated Gene Loss in Shigella.” Genome Biology 8 (8): R164.
https://doi.org/10.1186/gb-2007-8-8-r164.
Heys, Chloe, Anne Lizé, Frances Blow, Lewis White, Alistair Darby, and Zenobia J.
Lewis. 2018. “The Effect of Gut Microbiota Elimination in Drosophila
Melanogaster: A How‐to Guide for Host–Microbiota Studies.” Ecology and
Evolution 8 (8): 4150–61. https://doi.org/10.1002/ece3.3991.
Hodgson, Dave, Jenni L. McDonald, and David J. Hosken. 2015. “What Do You Mean,
‘Resilient’?” Trends in Ecology & Evolution 30 (9): 503–6.
https://doi.org/10.1016/j.tree.2015.06.010.
Hoh, C. Y., and R. Cord-Ruwisch. 1996. “A Practical Kinetic Model That Considers
Endproduct Inhibition in Anaerobic Digestion Processes by Including the
Equilibrium Constant.” Biotechnology and Bioengineering 51 (5): 597–604.
https://doi.org/10.1002/(SICI)1097-0290(19960905)51:5<597::AID-
BIT12>3.0.CO;2-F.
Holben, W.E., P. Williams, M. Saarinen, L.K. Särkilahti, and J.H.A. Apajalahti. 2002.
“Phylogenetic Analysis of Intestinal Microflora Indicates a Novel Mycoplasma
194
Phylotype in Farmed and Wild Salmon.” Microbial Ecology 44 (2): 175–85.
https://doi.org/10.1007/s00248-002-1011-6.
Holling, C S. 1973. “Resilience and Stability of Ecological Systems.” Annual Review of
Ecology and Systematics 4 (1): 1–23.
https://doi.org/10.1146/annurev.es.04.110173.000245.
Holmes, Alison H., Luke S. P. Moore, Arnfinn Sundsfjord, Martin Steinbakk, Sadie
Regmi, Abhilasha Karkey, Philippe J. Guerin, and Laura J. V. Piddock. 2016.
“Understanding the Mechanisms and Drivers of Antimicrobial Resistance.” The
Lancet 387 (10014): 176–87. https://doi.org/10.1016/S0140-6736(15)00473-0.
Hooper, Sean D., Konstantinos Mavromatis, and Nikos C. Kyrpides. 2009. “Microbial
Co-Habitation and Lateral Gene Transfer: What Transposases Can Tell Us.”
Genome Biology 10 (4): R45. https://doi.org/10.1186/gb-2009-10-4-r45.
Hooper, Sean D., Jeroen Raes, Konrad U. Foerstner, Eoghan D. Harrington, Daniel
Dalevi, and Peer Bork. 2008. “A Molecular Study of Microbe Transfer between
Distant Environments.” PLoS ONE 3 (7): e2607.
https://doi.org/10.1371/journal.pone.0002607.
Hoostal, Matthew J., M. Gabriela Bidart-Bouzat, and Juan L. Bouzat. 2008. “Local
Adaptation of Microbial Communities to Heavy Metal Stress in Polluted
Sediments of Lake Erie.” FEMS Microbiology Ecology 65 (1): 156–68.
https://doi.org/10.1111/j.1574-6941.2008.00522.x.
Huang, Li-Nan, Jia-Liang Kuang, and Wen-Sheng Shu. 2016. “Microbial Ecology and
Evolution in the Acid Mine Drainage Model System.” Trends in Microbiology
24 (7): 581–93. https://doi.org/10.1016/j.tim.2016.03.004.
Hubbell, S. P. 1997. “A Unified Theory of Biogeography and Relative Species
Abundance and Its Application to Tropical Rain Forests and Coral Reefs.” Coral
Reefs 16 (1): S9–21. https://doi.org/10.1007/s003380050237.
Hubbell, Stephen P. 2001. The Unified Neutral Theory of Biodiversity and
Biogeography (MPB-32). Princeton University Press.
Hubbell Stephen P. 2005. “Neutral Theory in Community Ecology and the Hypothesis
of Functional Equivalence.” Functional Ecology 19 (1): 166–72.
https://doi.org/10.1111/j.0269-8463.2005.00965.x.
Hubbell, Stephen P. 2006. “Neutral Theory and the Evolution of Ecological
Equivalence.” Ecology 87 (6): 1387–1398.
http://www.esajournals.org/doi/abs/10.1890/0012-
9658(2006)87%5B1387:NTATEO%5D2.0.CO%3B2.
Hudson-Edwards, Karen A., and Bernhard Dold. 2015. “Mine Waste Characterization,
Management and Remediation.” Minerals 5 (1): 82–85.
https://doi.org/10.3390/min5010082.
Hudson-Edwards, Karen A., Heather E. Jamieson, and Bernd G. Lottermoser. 2011.
“Mine Wastes: Past, Present, Future.” Elements 7 (6): 375–80.
https://doi.org/10.2113/gselements.7.6.375.
Huson, D. H., A. F. Auch, J. Qi, and S. C. Schuster. 2007. “MEGAN Analysis of
Metagenomic Data.” Genome Research 17 (3): 377–86.
https://doi.org/10.1101/gr.5969107.
Hutchinson, G. E. 1961. “The Paradox of the Plankton.” The American Naturalist 95
(882): 137–45. https://doi.org/10.1086/282171.
Ingrisch, Johannes, and Michael Bahn. 2018. “Towards a Comparable Quantification of
Resilience.” Trends in Ecology & Evolution 33 (4): 251–59.
https://doi.org/10.1016/j.tree.2018.01.013.
195
Inoue, Ryo, and Kazunari Ushida. 2003. “Development of the Intestinal Microbiota in
Rats and Its Possible Interactions with the Evolution of the Luminal IgA in the
Intestine.” FEMS Microbiology Ecology 45 (2): 147–53.
https://doi.org/10.1016/S0168-6496(03)00134-X.
Iverson, Vaughn, Robert M. Morris, Christian D. Frazar, Chris T. Berthiaume, Rhonda
L. Morales, and E. Virginia Armbrust. 2012. “Untangling Genomes from
Metagenomes: Revealing an Uncultured Class of Marine Euryarchaeota.”
Science 335 (6068): 587–590.
http://science.sciencemag.org/content/335/6068/587.short.
Jaccard, Paul. 1912. “The Distribution of the Flora in the Alpine Zone.1.” New
Phytologist 11 (2): 37–50. https://doi.org/10.1111/j.1469-8137.1912.tb05611.x.
Jackson, Togwell A., Sophia Vlaar, Nien Nguyen, Gary G. Leppard, and Turlough M.
Finan. 2015. “Effects of Bioavailable Heavy Metal Species, Arsenic, and Acid
Drainage from Mine Tailings on a Microbial Community Sampled Along a
Pollution Gradient in a Freshwater Ecosystem.” Geomicrobiology Journal 32
(8): 724–50. https://doi.org/10.1080/01490451.2014.969412.
Janda, J. Michael, and Sharon L. Abbott. 2007. “16S RRNA Gene Sequencing for
Bacterial Identification in the Diagnostic Laboratory: Pluses, Perils, and
Pitfalls.” Journal of Clinical Microbiology 45 (9): 2761–64.
https://doi.org/10.1128/JCM.01228-07.
Jaspers, Elke, and Jörg Overmann. 2004. “Ecological Significance of Microdiversity:
Identical 16S RRNA Gene Sequences Can Be Found in Bacteria with Highly
Divergent Genomes and Ecophysiologies.” Applied and Environmental
Microbiology 70 (8): 4831–39. https://doi.org/10.1128/AEM.70.8.4831-
4839.2004.
Jayathilake, Pahala Gedara, Prashant Gupta, Bowen Li, Curtis Madsen, Oluwole
Oyebamiji, Rebeca González-Cabaleiro, Steve Rushton, et al. 2017. “A
Mechanistic Individual-Based Model of Microbial Communities.” PLOS ONE
12 (8): e0181965. https://doi.org/10.1371/journal.pone.0181965.
Jemielita, Matthew, Michael J. Taormina, Adam R. Burns, Jennifer S. Hampton, Annah
S. Rolig, Karen Guillemin, and Raghuveer Parthasarathy. 2014. “Spatial and
Temporal Features of the Growth of a Bacterial Species Colonizing the
Zebrafish Gut.” MBio 5 (6): e01751-14. https://doi.org/10.1128/mBio.01751-14.
Jensen, Jeffrey D., Bret A. Payseur, Wolfgang Stephan, Charles F. Aquadro, Michael
Lynch, Deborah Charlesworth, and Brian Charlesworth. 2019. “The Importance
of the Neutral Theory in 1968 and 50 Years on: A Response to Kern and Hahn
2018.” Evolution 73 (1): 111–14. https://doi.org/10.1111/evo.13650.
Jeraldo, Patricio, Maksim Sipos, Nicholas Chia, Jennifer M. Brulc, A. Singh Dhillon,
Michael E. Konkel, Charles L. Larson, et al. 2012. “Quantification of the
Relative Roles of Niche and Neutral Processes in Structuring Gastrointestinal
Microbiomes.” Proceedings of the National Academy of Sciences 109 (25):
9692–98. https://doi.org/10.1073/pnas.1206721109.
Jezierska, Barbara, and Małgorzata Witeska. 2006. “THE METAL UPTAKE AND
ACCUMULATION IN FISH LIVING IN POLLUTED WATERS.” In Soil and
Water Pollution Monitoring, Protection and Remediation, edited by Irena
Twardowska, Herbert E. Allen, Max M. Häggblom, and Sebastian Stefaniak,
107–14. NATO Science Series. Springer Netherlands.
Jie, Shiqi, Mingming Li, Min Gan, Jianyu Zhu, Huaqun Yin, and Xueduan Liu. 2016.
“Microbial Functional Genes Enriched in the Xiangjiang River Sediments with
196
Heavy Metal Contamination.” BMC Microbiology 16 (1): 179.
https://doi.org/10.1186/s12866-016-0800-x.
Johansen, Jeffrey R., Jan Mareš, Nicole Pietrasiak, Markéta Bohunická, Jan Zima Jr,
Lenka Štenclová, and Tomáš Hauer. 2017. “Highly Divergent 16S RRNA
Sequences in Ribosomal Operons of Scytonema Hyalinum (Cyanobacteria).”
PLOS ONE 12 (10): e0186393. https://doi.org/10.1371/journal.pone.0186393.
Johnson, D.Barrie, and Kevin B. Hallberg. 2003. “The Microbiology of Acidic Mine
Waters.” Research in Microbiology 154 (7): 466–73.
https://doi.org/10.1016/S0923-2508(03)00114-1.
Johnson, Katerina V.-A., and Kevin R. Foster. 2018. “Why Does the Microbiome
Affect Behaviour?” Nature Reviews Microbiology 16 (10): 647–55.
https://doi.org/10.1038/s41579-018-0014-3.
Johnson, Zackary I., Erik R. Zinser, Allison Coe, Nathan P. McNulty, E. Malcolm S.
Woodward, and Sallie W. Chisholm. 2006. “Niche Partitioning Among
Prochlorococcus Ecotypes Along Ocean-Scale Environmental Gradients.”
Science 311 (5768): 1737–40. https://doi.org/10.1126/science.1118052.
Jost, Lou. 2006. “Entropy and Diversity.” Oikos 113 (2): 363–75.
https://doi.org/10.1111/j.2006.0030-1299.14714.x.
———. 2007. “Partitioning Diversity into Independent Alpha and Beta Components.”
Ecology 88 (10): 2427–39. https://doi.org/10.1890/06-1736.1.
Jousset, Alexandre, Christina Bienhold, Antonis Chatzinotas, Laure Gallien, Angélique
Gobet, Viola Kurm, Kirsten Küsel, et al. 2017. “Where Less May Be More:
How the Rare Biosphere Pulls Ecosystems Strings.” The ISME Journal 11 (4):
853–62. https://doi.org/10.1038/ismej.2016.174.
Jurburg, Stephanie D., Inês Nunes, James C. Stegen, Xavier Le Roux, Anders Priemé,
Søren J. Sørensen, and Joana Falcão Salles. 2017. “Autogenic Succession and
Deterministic Recovery Following Disturbance in Soil Bacterial Communities.”
Scientific Reports 7 (1). https://doi.org/10.1038/srep45691.
Kanehisa, Minoru, Susumu Goto, Yoko Sato, Masayuki Kawashima, Miho Furumichi,
and Mao Tanabe. 2014. “Data, Information, Knowledge and Principle: Back to
Metabolism in KEGG.” Nucleic Acids Research 42 (Database issue): D199-205.
https://doi.org/10.1093/nar/gkt1076.
Kawecki, Tadeusz J., and Dieter Ebert. 2004. “Conceptual Issues in Local Adaptation.”
Ecology Letters 7 (12): 1225–41. https://doi.org/10.1111/j.1461-
0248.2004.00684.x.
Kejnovsky, Eduard, and Edward N. Trifonov. 2016. “Horizontal Transfer - Imperative
Mission of Acellular Life Forms, Acytota.” Mobile Genetic Elements 6 (2):
e1154636. https://doi.org/10.1080/2159256X.2016.1154636.
Kembel, Steven W., Martin Wu, Jonathan A. Eisen, and Jessica L. Green. 2012.
“Incorporating 16S Gene Copy Number Information Improves Estimates of
Microbial Diversity and Abundance.” Edited by Christian von Mering. PLoS
Computational Biology 8 (10): e1002743.
https://doi.org/10.1371/journal.pcbi.1002743.
Kermack, William Ogilvy, A. G. McKendrick, and Gilbert Thomas Walker. 1927. “A
Contribution to the Mathematical Theory of Epidemics.” Proceedings of the
Royal Society of London. Series A, Containing Papers of a Mathematical and
Physical Character 115 (772): 700–721. https://doi.org/10.1098/rspa.1927.0118.
Kessick, M. A. 1974. “Letter: The Kinetics of Bacterial Growth.” Biotechnology and
Bioengineering 16 (11): 1545–47. https://doi.org/10.1002/bit.260161111.
197
Khachane, Amit N., Kenneth N. Timmis, and Vítor A. P. Martins dos Santos. 2007.
“Dynamics of Reductive Genome Evolution in Mitochondria and Obligate
Intracellular Microbes.” Molecular Biology and Evolution 24 (2): 449–56.
https://doi.org/10.1093/molbev/msl174.
Khandelwal, Ruchir A., Brett G. Olivier, Wilfred F. M. Röling, Bas Teusink, and Frank
J. Bruggeman. 2013. “Community Flux Balance Analysis for Microbial
Consortia at Balanced Growth.” PLOS ONE 8 (5): e64567.
https://doi.org/10.1371/journal.pone.0064567.
Kimura, M. 1991. “The Neutral Theory of Molecular Evolution: A Review of Recent
Evidence.” Idengaku Zasshi 66 (4): 367–86.
Kimura, Motoo. 1967. “On the Evolutionary Adjustment of Spontaneous Mutation
Rates.” Genetical Research 9 (1): 23–34.
https://doi.org/10.1017/S0016672300010284.
Kingsland, Sharon. 2015. “Alfred J. Lotka and the Origins of Theoretical Population
Ecology.” Proceedings of the National Academy of Sciences 112 (31): 9493–95.
https://doi.org/10.1073/pnas.1512317112.
Klapper, Isaac., and Jack. Dockery. 2010. “Mathematical Description of Microbial
Biofilms.” SIAM Review 52 (2): 221–65. https://doi.org/10.1137/080739720.
Klinck, J. S., W. W. Green, R. S. Mirza, S. R. Nadella, M. J. Chowdhury, C. M. Wood,
and G. G. Pyle. 2007. “Branchial Cadmium and Copper Binding and Intestinal
Cadmium Uptake in Wild Yellow Perch (Perca Flavescens) from Clean and
Metal-Contaminated Lakes.” Aquatic Toxicology, A Tribute to Rick Playle: The
Interface of Toxicology, Physiology, and Modeling in Improving Water Quality
Regulations for Metals, 84 (2): 198–207.
https://doi.org/10.1016/j.aquatox.2007.03.024.
Konopka, A. 1984. “Microbial Growth: Growth of the Bacterial Cell.” Science (New
York, N.Y.) 224 (4650): 738–39. https://doi.org/10.1126/science.224.4650.738.
Körstgens, V, H. -C Flemming, J Wingender, and W Borchard. 2001. “Uniaxial
Compression Measurement Device for Investigation of the Mechanical Stability
of Biofilms.” Journal of Microbiological Methods 46 (1): 9–17.
https://doi.org/10.1016/S0167-7012(01)00248-2.
Koschorreck, Matthias. 2008. “Microbial Sulphate Reduction at a Low PH.” FEMS
Microbiology Ecology 64 (3): 329–42. https://doi.org/10.1111/j.1574-
6941.2008.00482.x.
Kraemer, Lisa D., Peter G. C. Campbell, and Landis Hare. 2006. “Seasonal Variations
in Hepatic Cd and Cu Concentrations and in the Sub-Cellular Distribution of
These Metals in Juvenile Yellow Perch (Perca Flavescens).” Environmental
Pollution 142 (2): 313–25. https://doi.org/10.1016/j.envpol.2005.10.004.
Kraemer, Susanne A., and Primrose J. Boynton. 2017. “Evidence for Microbial Local
Adaptation in Nature.” Molecular Ecology 26 (7): 1860–76.
https://doi.org/10.1111/mec.13958.
Kreft, J. U., C. Picioreanu, J. W. Wimpenny, and M. C. van Loosdrecht. 2001.
“Individual-Based Modelling of Biofilms.” Microbiology (Reading, England)
147 (Pt 11): 2897–2912. https://doi.org/10.1099/00221287-147-11-2897.
Kuang, Jia-Liang, Li-Nan Huang, Lin-Xing Chen, Zheng-Shuang Hua, Sheng-Jin Li,
Min Hu, Jin-Tian Li, and Wen-Sheng Shu. 2013. “Contemporary Environmental
Variation Determines Microbial Diversity Patterns in Acid Mine Drainage.” The
ISME Journal 7 (5): 1038–50. https://doi.org/10.1038/ismej.2012.139.
Kuang, Jialiang, Linan Huang, Zhili He, Linxing Chen, Zhengshuang Hua, Pu Jia,
Shengjin Li, et al. 2016. “Predicting Taxonomic and Functional Structure of
198
Microbial Communities in Acid Mine Drainage.” The ISME Journal, March.
https://doi.org/10.1038/ismej.2015.201.
Kumar, Kishore R., Mark J. Cowley, and Ryan L. Davis. 2019. “Next-Generation
Sequencing and Emerging Technologies.” Seminars in Thrombosis and
Hemostasis, May. https://doi.org/10.1055/s-0039-1688446.
Kuo, Chih-Horng, Nancy A. Moran, and Howard Ochman. 2009. “The Consequences of
Genetic Drift for Bacterial Genome Complexity.” Genome Research 19 (8):
1450–54. https://doi.org/10.1101/gr.091785.109.
Lacroix, Alexandra, and Alice Hontela. 2004. “A Comparative Assessment of the
Adrenotoxic Effects of Cadmium in Two Teleost Species, Rainbow Trout,
Oncorhynchus Mykiss, and Yellow Perch, Perca Flavescens.” Aquatic
Toxicology 67 (1): 13–21. https://doi.org/10.1016/j.aquatox.2003.11.010.
Lagkouvardos, Ilias, Sandra Fischer, Neeraj Kumar, and Thomas Clavel. 2017. “Rhea:
A Transparent and Modular R Pipeline for Microbial Profiling Based on 16S
RRNA Gene Amplicons.” PeerJ 5 (January): e2836.
https://doi.org/10.7717/peerj.2836.
Lander, E. S., L. M. Linton, B. Birren, C. Nusbaum, M. C. Zody, J. Baldwin, K. Devon,
et al. 2001. “Initial Sequencing and Analysis of the Human Genome.” Nature
409 (6822): 860–921. https://doi.org/10.1038/35057062.
Langenheder, Silke, Eva S. Lindström, and Lars J. Tranvik. 2005. “Weak Coupling
between Community Composition and Functioning of Aquatic Bacteria.”
Limnology and Oceanography 50 (3): 957–67.
https://doi.org/10.4319/lo.2005.50.3.0957.
Langenheder, Silke, and Anna J. Székely. 2011. “Species Sorting and Neutral Processes
Are Both Important during the Initial Assembly of Bacterial Communities.” The
ISME Journal 5 (7): 1086–94. https://doi.org/10.1038/ismej.2010.207.
Langille, Morgan G I, Jesse Zaneveld, J Gregory Caporaso, Daniel McDonald, Dan
Knights, Joshua A Reyes, Jose C Clemente, et al. 2013. “Predictive Functional
Profiling of Microbial Communities Using 16S RRNA Marker Gene
Sequences.” Nature Biotechnology 31 (9): 814–21.
https://doi.org/10.1038/nbt.2676.
Laplante, Karine, Sébastien Boutin, and Nicolas Derome. 2013. “Parallel Changes of
Taxonomic Interaction Networks in Lacustrine Bacterial Communities Induced
by a Polymetallic Perturbation.” Evolutionary Applications 6 (4): 643–59.
https://doi.org/10.1111/eva.12050.
Laplante, Karine, and Nicolas Derome. 2011. “Parallel Changes in the Taxonomical
Structure of Bacterial Communities Exposed to a Similar Environmental
Disturbance: Parallel Changes in the Structure of Bacterial Communities.”
Ecology and Evolution 1 (4): 489–501. https://doi.org/10.1002/ece3.37.
Larkin, Alyse A., and Adam C. Martiny. 2017. “Microdiversity Shapes the Traits, Niche
Space, and Biogeography of Microbial Taxa.” Environmental Microbiology
Reports 9 (2): 55–70. https://doi.org/10.1111/1758-2229.12523.
Larson, Greger, Dolores R. Piperno, Robin G. Allaby, Michael D. Purugganan, Leif
Andersson, Manuel Arroyo-Kalin, Loukas Barton, et al. 2014. “Current
Perspectives and the Future of Domestication Studies.” Proceedings of the
National Academy of Sciences 111 (17): 6139–46.
https://doi.org/10.1073/pnas.1323964111.
Lavoie, Michel, Claude Fortin, and Peter G.C. Campbell. 2012. “Influence of Essential
Elements on Cadmium Uptake and Toxicity in a Unicellular Green Alga: The
199
Protective Effect of Trace Zinc and Cobalt Concentrations.” Environmental
Toxicology and Chemistry 31 (7): 1445–52. https://doi.org/10.1002/etc.1855.
Lemos, Leandro N., Roberta R. Fulthorpe, Eric W. Triplett, and Luiz F. W. Roesch.
2011. “Rethinking Microbial Diversity Analysis in the High Throughput
Sequencing Era.” Journal of Microbiological Methods 86 (1): 42–51.
https://doi.org/10.1016/j.mimet.2011.03.014.
Levy, Roie, and Elhanan Borenstein. 2013. “Metabolic Modeling of Species Interaction
in the Human Microbiome Elucidates Community-Level Assembly Rules.”
Proceedings of the National Academy of Sciences 110 (31): 12804–9.
https://doi.org/10.1073/pnas.1300926110.
———. 2014. “Metagenomic Systems Biology and Metabolic Modeling of the Human
Microbiome: From Species Composition to Community Assembly Rules.” Gut
Microbes 5 (2): 265–70. https://doi.org/10.4161/gmic.28261.
Li, An-Dong, Li-Guan Li, and Tong Zhang. 2015. “Exploring Antibiotic Resistance
Genes and Metal Resistance Genes in Plasmid Metagenomes from Wastewater
Treatment Plants.” Frontiers in Microbiology 6 (September).
https://doi.org/10.3389/fmicb.2015.01025.
Li, Wendy, Yali Yuan, Yao Xia, Yang Sun, Yinglei Miao, and Sam Ma. 2018. “A
Cross-Scale Neutral Theory Approach to the Influence of Obesity on
Community Assembly of Human Gut Microbiome.” Frontiers in Microbiology
9. https://doi.org/10.3389/fmicb.2018.02320.
Lidicker, William Z. 1979. “A Clarification of Interactions in Ecological Systems.”
BioScience 29 (8): 475–77. https://doi.org/10.2307/1307540.
Lima-Mendez, Gipsi, Karoline Faust, Nicolas Henry, Johan Decelle, Sébastien Colin,
Fabrizio Carcillo, Samuel Chaffron, et al. 2015. “Ocean Plankton. Determinants
of Community Structure in the Global Plankton Interactome.” Science (New
York, N.Y.) 348 (6237): 1262073. https://doi.org/10.1126/science.1262073.
Lind, Peter A., and Dan I. Andersson. 2008. “Whole-Genome Mutational Biases in
Bacteria.” Proceedings of the National Academy of Sciences 105 (46): 17878–
83. https://doi.org/10.1073/pnas.0804445105.
Lind, Peter A., Christina Tobin, Otto G. Berg, Charles G. Kurland, and Dan I.
Andersson. 2010. “Compensatory Gene Amplification Restores Fitness after
Inter-Species Gene Replacements.” Molecular Microbiology 75 (5): 1078–89.
https://doi.org/10.1111/j.1365-2958.2009.07030.x.
Liu, Yehao, Yuhui Li, Kaiyong Liu, and Jie Shen. 2014. “Exposing to Cadmium Stress
Cause Profound Toxic Effect on Microbiota of the Mice Intestinal Tract.” PLOS
ONE 9 (2): e85323. https://doi.org/10.1371/journal.pone.0085323.
Llewellyn, Martin S, Philip McGinnity, Melanie Dionne, Justine Letourneau, Florian
Thonier, Gary R Carvalho, Simon Creer, and Nicolas Derome. 2016. “The
Biogeography of the Atlantic Salmon (Salmo Salar) Gut Microbiome.” The
ISME Journal 10 (5): 1280–84. https://doi.org/10.1038/ismej.2015.189.
Löhr, Ansje J., Anniet M. Laverman, Martin Braster, Nico M. van Straalen, and Wilfred
F. M. Röling. 2006. “Microbial Communities in the World’s Largest Acidic
Volcanic Lake, Kawah Ijen in Indonesia, and in the Banyupahit River
Originating from It.” Microbial Ecology 52 (4): 609–18.
https://doi.org/10.1007/s00248-006-9068-2.
López-Madrigal, Sergio, and Rosario Gil. 2017. “Et Tu, Brute? Not Even Intracellular
Mutualistic Symbionts Escape Horizontal Gene Transfer.” Genes 8 (10).
https://doi.org/10.3390/genes8100247.
200
Louca, Stilianos, Saulo M. S. Jacques, Aliny P. F. Pires, Juliana S. Leal, Diane S.
Srivastava, Laura Wegener Parfrey, Vinicius F. Farjalla, and Michael Doebeli.
2016. “High Taxonomic Variability despite Stable Functional Structure across
Microbial Communities.” Nature Ecology & Evolution 1 (1): s41559-016-0015–
0016. https://doi.org/10.1038/s41559-016-0015.
Louca, Stilianos, Laura Wegener Parfrey, and Michael Doebeli. 2016. “Decoupling
Function and Taxonomy in the Global Ocean Microbiome.” Science 353 (6305):
1272–77. https://doi.org/10.1126/science.aaf4507.
Lowe, Winsor H., Ryan P. Kovach, and Fred W. Allendorf. 2017. “Population Genetics
and Demography Unite Ecology and Evolution.” Trends in Ecology & Evolution
32 (2): 141–52. https://doi.org/10.1016/j.tree.2016.12.002.
Lozupone, C., and R. Knight. 2005. “UniFrac: A New Phylogenetic Method for
Comparing Microbial Communities.” Applied and Environmental Microbiology
71 (12): 8228–35. https://doi.org/10.1128/AEM.71.12.8228-8235.2005.
Lozupone, Catherine A., and Rob Knight. 2015. “The Unifrac Significance Test Is
Sensitive to Tree Topology.” BMC Bioinformatics 16 (1).
https://doi.org/10.1186/s12859-015-0640-y.
Lozupone, Catherine A., Jesse I. Stombaugh, Jeffrey I. Gordon, Janet K. Jansson, and
Rob Knight. 2012. “Diversity, Stability and Resilience of the Human Gut
Microbiota.” Nature 489 (7415): 220–30. https://doi.org/10.1038/nature11550.
Ludwig, Marcus, Tiing Tiing Chua, Chyue Yie Chew, and Donald A. Bryant. 2015.
“Fur-Type Transcriptional Repressors and Metal Homeostasis in the
Cyanobacterium Synechococcus Sp. PCC 7002.” Microbial Physiology and
Metabolism, 1217. https://doi.org/10.3389/fmicb.2015.01217.
Luria, S. E., and M. Delbrück. 1943. “Mutations of Bacteria from Virus Sensitivity to
Virus Resistance.” Genetics 28 (6): 491–511.
Lynch, M. 2016. “Mutation and Human Exceptionalism: Our Future Genetic Load.”
Genetics 202 (3): 869–75. https://doi.org/10.1534/genetics.115.180471.
Lynch, Michael. 2006. “Streamlining and Simplification of Microbial Genome
Architecture.” Annual Review of Microbiology 60 (1): 327–49.
https://doi.org/10.1146/annurev.micro.60.080805.142300.
———. 2010. “Evolution of the Mutation Rate.” Trends in Genetics: TIG 26 (8): 345–
52. https://doi.org/10.1016/j.tig.2010.05.003.
Lynch, Michael, and John S. Conery. 2003. “The Origins of Genome Complexity.”
Science (New York, N.Y.) 302 (5649): 1401–4.
https://doi.org/10.1126/science.1089370.
Lynch, Michael D. J., and Josh D. Neufeld. 2015. “Ecology and Exploration of the Rare
Biosphere.” Nature Reviews Microbiology 13 (4): 217–29.
https://doi.org/10.1038/nrmicro3400.
Maddamsetti, Rohan, Richard E. Lenski, and Jeffrey E. Barrick. 2015. “Adaptation,
Clonal Interference, and Frequency-Dependent Interactions in a Long-Term
Evolution Experiment with Escherichia Coli.” Genetics 200 (2): 619–31.
https://doi.org/10.1534/genetics.115.176677.
Magoč, Tanja, and Steven L. Salzberg. 2011. “FLASH: Fast Length Adjustment of
Short Reads to Improve Genome Assemblies.” Bioinformatics (Oxford,
England) 27 (21): 2957–63. https://doi.org/10.1093/bioinformatics/btr507.
Magurran, Anne E., and Brian J. McGill, eds. 2011. Biological Diversity: Frontiers in
Measurement and Assessment. Oxford ; New York: Oxford University Press.
201
Maharjan, Ram, Shona Seeto, Lucinda Notley-McRobb, and Thomas Ferenci. 2006.
“Clonal Adaptive Radiation in a Constant Environment.” Science 313 (5786):
514–17. https://doi.org/10.1126/science.1129865.
Maignien, L., E. A. DeForce, M. E. Chafee, A. M. Eren, and S. L. Simmons. 2014.
“Ecological Succession and Stochastic Variation in the Assembly of Arabidopsis
Thaliana Phyllosphere Communities.” MBio 5 (1).
https://doi.org/10.1128/mBio.00682-13.
Mao, Meng, Xiushuai Yang, and Gordon M. Bennett. 2018. “Evolution of Host Support
for Two Ancient Bacterial Symbionts with Differentially Degraded Genomes in
a Leafhopper Host.” Proceedings of the National Academy of Sciences of the
United States of America 115 (50): E11691–700.
https://doi.org/10.1073/pnas.1811932115.
Marcogliese, David J., Lila Gagnon Brambilla, François Gagné, and Andrée D.
Gendron. 2005. “Joint Effects of Parasitism and Pollution on Oxidative Stress
Biomarkers in Yellow Perch Perca Flavescens.” Diseases of Aquatic Organisms
63 (1): 77–84. https://doi.org/10.3354/dao063077.
Marcogliese, David J., Claire Dautremepuits, Andrée D. Gendron, and Michel Fournier.
2010. “Interactions between Parasites and Pollutants in Yellow Perch
(Perca Flavescens) in the St. Lawrence River, Canada: Implications for
Resistance and Tolerance to Parasites.” Canadian Journal of Zoology 88 (3):
247–58. https://doi.org/10.1139/Z09-140.
Martiny, Adam C., Kathleen Treseder, and Gordon Pusch. 2013. “Phylogenetic
Conservatism of Functional Traits in Microorganisms.” The ISME Journal 7 (4):
830–38. https://doi.org/10.1038/ismej.2012.160.
Martiny, Jennifer B. Hughes, Brendan J. M. Bohannan, James H. Brown, Robert K.
Colwell, Jed A. Fuhrman, Jessica L. Green, M. Claire Horner-Devine, et al.
2006. “Microbial Biogeography: Putting Microorganisms on the Map.” Nature
Reviews Microbiology 4 (2): 102–12. https://doi.org/10.1038/nrmicro1341.
Masella, Andre P., Andrea K. Bartram, Jakub M. Truszkowski, Daniel G. Brown, and
Josh D. Neufeld. 2012. “PANDAseq: Paired-End Assembler for Illumina
Sequences.” BMC Bioinformatics 13 (1): 31.
http://www.biomedcentral.com/1471-2105/13/31/.
Masmoudi, Salma, Nhung Nguyen-Deroche, Aurore Caruso, Habib Ayadi, Annick
Morant-Manceau, Gérard Tremblin, MartineBertrand, and Benoît Schoefs. 2013.
“Cadmium, Copper, Sodium and Zinc Effects on Diatoms: From Heaven to Hell
— a Review.” Cryptogamie, Algologie 34 (2): 185–225.
https://doi.org/10.7872/crya.v34.iss2.2013.185.
Mathiesen, Joachim, Namiko Mitarai, Kim Sneppen, and Ala Trusina. 2011.
“Ecosystems with Mutually Exclusive Interactions Self-Organize to a State of
High Diversity.” Physical Review Letters 107 (18): 188101.
https://doi.org/10.1103/PhysRevLett.107.188101.
Mathis, Diane, and Christophe Benoist. 2012. “The Influence of the Microbiota on
Type-1 Diabetes: On the Threshold of a Leap Forward in Our Understanding.”
Immunological Reviews 245 (1): 239–49. https://doi.org/10.1111/j.1600-
065X.2011.01084.x.
Matthews, Thomas J., and Robert J. Whittaker. 2014. “Neutral Theory and the Species
Abundance Distribution: Recent Developments and Prospects for Unifying
Niche and Neutral Perspectives.” Ecology and Evolution, May, n/a-n/a.
https://doi.org/10.1002/ece3.1092.
202
Maughan, Heather, and Geraldine Van der Auwera. 2011. “Bacillus Taxonomy in the
Genomic Era Finds Phenotypes to Be Essential Though Often Misleading.”
Infection, Genetics and Evolution 11 (5): 789–97.
https://doi.org/10.1016/j.meegid.2011.02.001.
Mavromatis, Konstantinos, Natalia Ivanova, Kerrie Barry, Harris Shapiro, Eugene
Goltsman, Alice C McHardy, Isidore Rigoutsos, et al. 2007. “Use of Simulated
Data Sets to Evaluate the Fidelity of Metagenomic Processing Methods.” Nature
Methods 4 (6): 495–500. https://doi.org/10.1038/nmeth1043.
May, R. M. 1976. “Simple Mathematical Models with Very Complicated Dynamics.”
Nature 261 (5560): 459–67. https://doi.org/10.1038/261459a0.
Mayali, Xavier, Peter K. Weber, Shalini Mabery, and Jennifer Pett-Ridge. 2014.
“Phylogenetic Patterns in the Microbial Response to Resource Availability:
Amino Acid Incorporation in San Francisco Bay.” PLoS ONE 9 (4).
https://doi.org/10.1371/journal.pone.0095842.
McCafferty, Jonathan, Marcus Mühlbauer, Raad Z. Gharaibeh, Janelle C. Arthur,
Ernesto Perez-Chanona, Wei Sha, Christian Jobin, and Anthony A. Fodor. 2013.
“Stochastic Changes over Time and Not Founder Effects Drive Cage Effects in
Microbial Community Assembly in a Mouse Model.” The ISME Journal 7 (11):
2116–25. https://doi.org/10.1038/ismej.2013.106.
McGill, Brian J., Brian A. Maurer, and Michael D. Weiser. 2006. “Empirical Evaluation
of Neutral Theory.” Ecology 87 (6): 1411–23. https://doi.org/10.1890/0012-
9658(2006)87[1411:EEONT]2.0.CO;2.
Mehrshad, Maliheh, Mohammad Ali Amoozegar, Rohit Ghai, Seyed Abolhassan
Shahzadeh Fazeli, and Francisco Rodriguez-Valera. 2016. “Genome
Reconstruction from Metagenomic Datasets Reveals Novel Microbes in the
Brackish Waters of the Caspian Sea.” Applied and Environmental Microbiology,
January. https://doi.org/10.1128/AEM.03381-15.
Miki, Takeshi, Taichi Yokokawa, and Kazuaki Matsui. 2014. “Biodiversity and
Multifunctionality in a Microbial Community: A Novel Theoretical Approach to
Quantify Functional Redundancy.” Proc. R. Soc. B 281 (1776): 20132498.
https://doi.org/10.1098/rspb.2013.2498.
Mira, A., H. Ochman, and N. A. Moran. 2001. “Deletional Bias and the Evolution of
Bacterial Genomes.” Trends in Genetics: TIG 17 (10): 589–96.
https://doi.org/10.1016/s0168-9525(01)02447-7.
Morrison-Whittle, Peter, and Matthew R. Goddard. 2015. “Quantifying the Relative
Roles of Selective and Neutral Processes in Defining Eukaryotic Microbial
Communities.” The ISME Journal 9 (9): 2003–11.
https://doi.org/10.1038/ismej.2015.18.
Morrissey, Ember M., Rebecca L. Mau, Egbert Schwartz, J. Gregory Caporaso, Paul
Dijkstra, Natasja van Gestel, Benjamin J. Koch, et al. 2016. “Phylogenetic
Organization of Bacterial Activity.” The ISME Journal 10 (9): 2336–40.
https://doi.org/10.1038/ismej.2016.28.
Moser, Michael, and Thomas Weisse. 2011. “The Most Acidified Austrian Lake in
Comparison to a Neutralized Mining Lake.” Limnologica 41 (4): 303–15.
https://doi.org/10.1016/j.limno.2011.01.002.
Mota, Rita, Sara B. Pereira, Marianna Meazzini, Rui Fernandes, Arlete Santos, Caroline
A. Evans, Roberto De Philippis, Phillip C. Wright, and Paula Tamagnini. 2015.
“Effects of Heavy Metals on Cyanothece Sp. CCY 0110 Growth, Extracellular
Polymeric Substances (EPS) Production, Ultrastructure and Protein Profiles.”
203
Journal of Proteomics 120 (April): 75–94.
https://doi.org/10.1016/j.jprot.2015.03.004.
Motta, Erick V. S., Kasie Raymann, and Nancy A. Moran. 2018. “Glyphosate Perturbs
the Gut Microbiota of Honey Bees.” Proceedings of the National Academy of
Sciences 115 (41): 10305–10. https://doi.org/10.1073/pnas.1803880115.
Moxon, E. Richard. 2011. “Darwin, Microbes and Evolution by Natural Selection.” In
Hot Topics in Infection and Immunity in Children VII, edited by Nigel Curtis,
Adam Finn, and Andrew J. Pollard, 77–86. Advances in Experimental Medicine
and Biology. New York, NY: Springer. https://doi.org/10.1007/978-1-4419-
7185-2_6.
Muegge, Brian D., Justin Kuczynski, Dan Knights, Jose C. Clemente, Antonio
González, Luigi Fontana, Bernard Henrissat, Rob Knight, and Jeffrey I. Gordon.
2011. “Diet Drives Convergence in Gut Microbiome Functions across
Mammalian Phylogeny and within Humans.” Science (New York, N.Y.) 332
(6032): 970–74. https://doi.org/10.1126/science.1198719.
Naïja, Azza, Patrick Kestemont, Benoit Chénais, Zohra Haouas, Ronny Blust, Ahmed
Noureddine Helal, and Justine Marchand. 2017. “Cadmium Exposure Exerts
Neurotoxic Effects in Peacock Blennies Salaria Pavo.” Ecotoxicology and
Environmental Safety 143 (September): 217–27.
https://doi.org/10.1016/j.ecoenv.2017.05.041.
Narrowe, Adrienne B., Munira Albuthi-Lantz, Erin P. Smith, Kimberly J. Bower,
Timberley M. Roane, Alan M. Vajda, and Christopher S. Miller. 2015.
“Perturbation and Restoration of the Fathead Minnow Gut Microbiome after
Low-Level Triclosan Exposure.” Microbiome 3 (1): 6.
https://doi.org/10.1186/s40168-015-0069-6.
Navarrete, Paola, Pamela Mardones, Rafael Opazo, Romilio Espejo, and Jaime Romero.
2008. “Oxytetracycline Treatment Reduces Bacterial Diversity of Intestinal
Microbiota of Atlantic Salmon.” Journal of Aquatic Animal Health 20 (3): 177–
83. https://doi.org/10.1577/H07-043.1.
Navarro, Claudio A., Diego von Bernath, and Carlos A. Jerez. 2013. “Heavy Metal
Resistance Strategies of Acidophilic Bacteria and Their Acquisition: Importance
for Biomining and Bioremediation.” Biol Res 46: 363–371.
http://biologicalresearch.cl/BR_papers/v46n4_art07.pdf.
Nei, M., and T. Gojobori. 1986. “Simple Methods for Estimating the Numbers of
Synonymous and Nonsynonymous Nucleotide Substitutions.” Molecular
Biology and Evolution 3 (5): 418–26.
https://doi.org/10.1093/oxfordjournals.molbev.a040410.
Nemergut, Diana R., Steven K. Schmidt, Tadashi Fukami, Sean P. O’Neill, Teresa M.
Bilinski, Lee F. Stanish, Joseph E. Knelman, et al. 2013. “Patterns and Processes
of Microbial Community Assembly.” Microbiology and Molecular Biology
Reviews 77 (3): 342–56. https://doi.org/10.1128/MMBR.00051-12.
Ni, Caiying, Dean J. Horton, Junpeng Rui, Michael W. Henson, Yumei Jiang, Xiaolan
Huang, and Deric R. Learman. 2016. “High Concentrations of Bioavailable
Heavy Metals Impact Freshwater Sediment Microbial Communities.” Annals of
Microbiology 66 (3): 1003–12. https://doi.org/10.1007/s13213-015-1189-8.
Nichols, D., K. Lewis, J. Orjala, S. Mo, R. Ortenberg, P. O’Connor, C. Zhao, P. Vouros,
T. Kaeberlein, and S. S. Epstein. 2008. “Short Peptide Induces an ‘Uncultivable’
Microorganism To Grow In Vitro.” Applied and Environmental Microbiology 74
(15): 4889–97. https://doi.org/10.1128/AEM.00393-08.
204
Niehus, Rene, Sara Mitri, Alexander G. Fletcher, and Kevin R. Foster. 2015. “Migration
and Horizontal Gene Transfer Divide Microbial Genomes into Multiple Niches.”
Nature Communications 6 (1): 1–9. https://doi.org/10.1038/ncomms9924.
Nilsson, A. I., S. Koskiniemi, S. Eriksson, E. Kugelberg, J. C. D. Hinton, and D. I.
Andersson. 2005. “Bacterial Genome Size Reduction by Experimental
Evolution.” Proceedings of the National Academy of Sciences of the United
States of America 102 (34): 12112–16.
https://doi.org/10.1073/pnas.0503654102.
Nirola, Ramkrishna, Mallavarapu Megharaj, Christopher Saint, Rupak Aryal,
Palanisami Thavamani, Kadiyala Venkateswarlu, Ravi Naidu, and Simon
Beecham. 2016. “Metal Bioavailability to Eisenia Fetida through Copper Mine
Dwelling Animal and Plant Litter, a New Challenge on Contaminated
Environment Remediation.” International Biodeterioration & Biodegradation,
Challenges in Environmental Science and Engieering – 2015, 113 (September):
208–16. https://doi.org/10.1016/j.ibiod.2016.03.007.
Niyogi, S, P Couture, G Pyle, D G McDonald, and C M Wood. 2004. “Acute Cadmium
Biotic Ligand Model Characteristics of Laboratory-Reared and Wild Yellow
Perch (Perca Flavescens) Relative to Rainbow Trout (Oncorhynchus Mykiss).”
Canadian Journal of Fisheries and Aquatic Sciences 61 (6): 942–53.
https://doi.org/10.1139/f04-044.
Nordstrom, D. Kirk. 2011. “Mine Waters: Acidic to Circmneutral.” Elements 7 (6):
393–98. https://doi.org/10.2113/gselements.7.6.393.
Nowak, Martin. 2006. “Evolutionary Dynamics.” 2006.
https://www.hup.harvard.edu/catalog.php?isbn=9780674023383.
Ochman, Howard, and Liliana M. Davalos. 2006. “The Nature and Dynamics of
Bacterial Genomes.” Science 311 (5768): 1730–33.
https://doi.org/10.1126/science.1119966.
O’Dwyer, James P., Steven W. Kembel, and Thomas J. Sharpton. 2015. “Backbones of
Evolutionary History Test Biodiversity Theory for Microbes.” Proceedings of
the National Academy of Sciences 112 (27): 8356–61.
https://doi.org/10.1073/pnas.1419341112.
Ofiţeru, Irina Dana, Mary Lunn, Thomas P. Curtis, George F. Wells, Craig S. Criddle,
Christopher A. Francis, and William T. Sloan. 2010. “Combined Niche and
Neutral Effects in a Microbial Wastewater Treatment Community.” Proceedings
of the National Academy of Sciences 107 (35): 15345–15350.
https://www.pnas.org/content/107/35/15345.full.
Oh, Sung T., and Alastair D. Martin. 2007. “Thermodynamic Equilibrium Model in
Anaerobic Digestion Process.” Biochemical Engineering Journal 34 (3): 256–
66. https://doi.org/10.1016/j.bej.2006.12.011.
Oksanen, Jari, F. Guillaume Blanchet, Michael Friendly, Roeland Kindt, Pierre
Legendre, Dan McGlinn, Peter R. Minchin, et al. 2016. “Community Ecology
Package. R Packag. 2.3-3,” September.
https://doi.org/10.4135/9781412971874.n145.
Olendorf, Robert, F. Helen Rodd, David Punzalan, Anne E. Houde, Carla Hurt, David
N. Reznick, and Kimberly A. Hughes. 2006. “Frequency-Dependent Survival in
Natural Guppy Populations.” Nature 441 (7093): 633–36.
https://doi.org/10.1038/nature04646.
Oliveira-Pinto, Clarisse, Patrícia S. Costa, Mariana P. Reis, Edmar Chartone-Souza, and
Andréa M. A. Nascimento. 2016. “Diversity of Gene Cassettes and the
Abundance of the Class 1 Integron-Integrase Gene in Sediment Polluted by
205
Metals.” Extremophiles 20 (3): 283–89. https://doi.org/10.1007/s00792-016-
0820-3.
Outten, F. W., D. L. Huffman, J. A. Hale, and T. V. O’Halloran. 2001. “The
Independent Cue and Cus Systems Confer Copper Tolerance during Aerobic and
Anaerobic Growth in Escherichia Coli.” The Journal of Biological Chemistry
276 (33): 30670–77. https://doi.org/10.1074/jbc.M104122200.
Overbeek, Ross, Robert Olson, Gordon D. Pusch, Gary J. Olsen, James J. Davis, Terry
Disz, Robert A. Edwards, et al. 2014. “The SEED and the Rapid Annotation of
Microbial Genomes Using Subsystems Technology (RAST).” Nucleic Acids
Research 42 (Database issue): D206–14. https://doi.org/10.1093/nar/gkt1226.
Pal, C., J. Bengtsson-Palme, C. Rensing, E. Kristiansson, and D. G. J. Larsson. 2014.
“BacMet: Antibacterial Biocide and Metal Resistance Genes Database.” Nucleic
Acids Research 42 (D1): D737–43. https://doi.org/10.1093/nar/gkt1252.
Panda, Arup, Michel Drancourt, Tamir Tuller, and Pierre Pontarotti. 2018. “Genome-
Wide Analysis of Horizontally Acquired Genes in the Genus Mycobacterium.”
Scientific Reports 8 (1): 1–13. https://doi.org/10.1038/s41598-018-33261-w.
Parks, Donovan H., and Robert G. Beiko. 2010. “Identifying Biologically Relevant
Differences between Metagenomic Communities.” Bioinformatics 26 (6): 715–
21. https://doi.org/10.1093/bioinformatics/btq041.
Parks, Donovan H., Michael Imelfort, Connor T. Skennerton, Philip Hugenholtz, and
Gene W. Tyson. 2015. “CheckM: Assessing the Quality of Microbial Genomes
Recovered from Isolates, Single Cells, and Metagenomes.” Genome Research 25
(7): 1043–55. https://doi.org/10.1101/gr.186072.114.
Parks, Donovan H., Gene W. Tyson, Philip Hugenholtz, and Robert G. Beiko. 2014.
“STAMP: Statistical Analysis of Taxonomic and Functional Profiles.”
Bioinformatics (Oxford, England) 30 (21): 3123–24.
https://doi.org/10.1093/bioinformatics/btu494.
Parnell, J. Jacob, Todd A. Crowl, Bart C. Weimer, and Michael E. Pfrender. 2009.
“Biodiversity in Microbial Communities: System Scale Patterns and
Mechanisms.” Molecular Ecology 18 (7): 1455–62.
https://doi.org/10.1111/j.1365-294X.2009.04128.x.
Pearson, Karl, and Olaus Magnus Friedrich Erdmann Henrici. 1896. “VII. Mathematical
Contributions to the Theory of Evolution.—III. Regression, Heredity, and
Panmixia.” Philosophical Transactions of the Royal Society of London. Series A,
Containing Papers of a Mathematical or Physical Character 187 (January):
253–318. https://doi.org/10.1098/rsta.1896.0007.
Pei, A. Y., W. E. Oberdorf, C. W. Nossa, A. Agarwal, P. Chokshi, E. A. Gerz, Z. Jin, et
al. 2010. “Diversity of 16S RRNA Genes within Individual Prokaryotic
Genomes.” Applied and Environmental Microbiology 76 (12): 3886–97.
https://doi.org/10.1128/AEM.02953-09.
Pelletier, Fanie, and David W. Coltman. 2018. “Will Human Influences on Evolutionary
Dynamics in the Wild Pervade the Anthropocene?” BMC Biology 16 (1).
https://doi.org/10.1186/s12915-017-0476-1.
Pérez Carrascal, Olga M., David VanInsberghe, Soledad Juárez, Martin F. Polz, Pablo
Vinuesa, and Víctor González. 2016. “Population Genomics of the Symbiotic
Plasmids of Sympatric Nitrogen-Fixing Rhizobium Species Associated with
Phaseolus Vulgaris.” Environmental Microbiology 18 (8): 2660–76.
https://doi.org/10.1111/1462-2920.13415.
Pérez, María Teresa, Carina Rofner, and Ruben Sommaruga. 2015. “Dissolved Organic
Monomer Partitioning among Bacterial Groups in Two Oligotrophic Lakes.”
206
Environmental Microbiology Reports 7 (2): 265–72.
https://doi.org/10.1111/1758-2229.12240.
Perez-Rodriguez, Fernando, and Antonio Valero. 2013. Predictive Microbiology in
Foods. SpringerBriefs in Food, Health, and Nutrition. New York: Springer-
Verlag. https://www.springer.com/gp/book/9781461455196.
Pernthaler, Jakob. 2013. “Freshwater Microbial Communities.” In The Prokaryotes,
edited by Eugene Rosenberg, Edward F. DeLong, Stephen Lory, Erko
Stackebrandt, and Fabiano Thompson, 97–112. Berlin, Heidelberg: Springer
Berlin Heidelberg. http://link.springer.com/10.1007/978-3-642-30123-0_40.
Pester, Michael, Norbert Bittner, Pinsurang Deevong, Michael Wagner, and Alexander
Loy. 2010. “A ‘Rare Biosphere’ Microorganism Contributes to Sulfate
Reduction in a Peatland.” The ISME Journal 4 (12): 1591–1602.
https://doi.org/10.1038/ismej.2010.75.
Philippot, Laurent, Siv G. E. Andersson, Tom J. Battin, James I. Prosser, Joshua P.
Schimel, William B. Whitman, and Sara Hallin. 2010. “The Ecological
Coherence of High Bacterial Taxonomic Ranks.” Nature Reviews Microbiology
8 (7): 523–29. https://doi.org/10.1038/nrmicro2367.
Pierron, Fabien, Vincent Bourret, Jérôme St-Cyr, Peter G. C. Campbell, Louis
Bernatchez, and Patrice Couture. 2009. “Transcriptional Responses to
Environmental Metal Exposure in Wild Yellow Perch (Perca Flavescens)
Collected in Lakes with Differing Environmental Metal Concentrations (Cd, Cu,
Ni).” Ecotoxicology 18 (5): 620–31. https://doi.org/10.1007/s10646-009-0320-7.
Pimm, Stuart L. 1984. “The Complexity and Stability of Ecosystems.” Nature 307
(5949): 321. https://doi.org/10.1038/307321a0.
Pinter, Tyler B. J., and Martin J. Stillman. 2015. “Kinetics of Zinc and Cadmium
Exchanges between Metallothionein and Carbonic Anhydrase.” Biochemistry 54
(40): 6284–93. https://doi.org/10.1021/acs.biochem.5b00912.
Placella, Sarah A., Eoin L. Brodie, and Mary K. Firestone. 2012. “Rainfall-Induced
Carbon Dioxide Pulses Result from Sequential Resuscitation of Phylogenetically
Clustered Microbial Groups.” Proceedings of the National Academy of Sciences
109 (27): 10931–36. https://doi.org/10.1073/pnas.1204306109.
Pokrovsky, O.S., R.E. Martinez, S.V. Golubev, E.I. Kompantseva, and L.S. Shirokova.
2008. “Adsorption of Metals and Protons on Gloeocapsa Sp. Cyanobacteria: A
Surface Speciation Approach.” Applied Geochemistry 23 (9): 2574–88.
https://doi.org/10.1016/j.apgeochem.2008.05.007.
Ponton, Dominic E., Antoine Caron, Landis Hare, and Peter G.C. Campbell. 2016.
“Hepatic Oxidative Stress and Metal Subcellular Partitioning Are Affected by
Selenium Exposure in Wild Yellow Perch (Perca Flavescens).” Environmental
Pollution 214 (July): 608–17. https://doi.org/10.1016/j.envpol.2016.04.051.
Preheim, Sarah P., Allison R. Perrotta, Jonathan Friedman, Chris Smilie, Ilana Brito,
Mark B. Smith, and Eric Alm. 2013. “Computational Methods for High-
Throughput Comparative Analyses of Natural Microbial Communities.” In
Methods in Enzymology, 531:353–70. Elsevier.
http://linkinghub.elsevier.com/retrieve/pii/B9780124078635000186.
Price, Morgan N., Adam P. Arkin, and Eric J. Alm. 2006. “The Life-Cycle of Operons.”
PLOS Genetics 2 (6): e96. https://doi.org/10.1371/journal.pgen.0020096.
Pyle, Gregory G., James W. Rajotte, and Patrice Couture. 2005a. “Effects of Industrial
Metals on Wild Fish Populations along a Metal Contamination Gradient.”
Ecotoxicology and Environmental Safety 61 (3): 287–312.
https://doi.org/10.1016/j.ecoenv.2004.09.003.
207
———. 2005b. “Effects of Industrial Metals on Wild Fish Populations along a Metal
Contamination Gradient.” Ecotoxicology and Environmental Safety 61 (3): 287–
312. https://doi.org/10.1016/j.ecoenv.2004.09.003.
R Foundation for Statistical Computing. 2008. R: A Language and Environment for
Statistical Computing. Vienna, Austria. https://www.r-project.org.
Raes, Jeroen, Ivica Letunic, Takuji Yamada, Lars Juhl Jensen, and Peer Bork. 2011.
“Toward Molecular Trait-Based Ecology through Integration of
Biogeochemical, Geographical and Metagenomic Data.” Molecular Systems
Biology 7 (March): 473. https://doi.org/10.1038/msb.2011.6.
Railsback, Steven F., and Volker Grimm. 2011. Agent-Based and Individual-Based
Modeling: A Practical Introduction. Princeton University Press.
http://www.jstor.org/stable/j.ctt7sns7.
Raine, Nigel E. 2018. “Pesticide Affects Social Behavior of Bees.” Science 362 (6415):
643–44. https://doi.org/10.1126/science.aav5273.
Ram, Rachna J., Nathan C. VerBerkmoes, Michael P. Thelen, Gene W. Tyson, Brett J.
Baker, Robert C. Blake, Manesh Shah, Robert L. Hettich, and Jillian F. Banfield.
2005. “Community Proteomics of a Natural Microbial Biofilm.” Science 308
(5730): 1915–20. https://doi.org/10.1126/science. 1109070.
Raman, Karthik, and Nagasuma Chandra. 2009. “Flux Balance Analysis of Biological
Systems: Applications and Challenges.” Briefings in Bioinformatics 10 (4): 435–
49. https://doi.org/10.1093/bib/bbp011.
Rawls, John F., Buck S. Samuel, and Jeffrey I. Gordon. 2004. “Gnotobiotic Zebrafish
Reveal Evolutionarily Conserved Responses to the Gut Microbiota.”
Proceedings of the National Academy of Sciences of the United States of
America 101 (13): 4596–4601. https://doi.org/10.1073/pnas.0400706101.
Raymann, Kasie, Zack Shaffer, and Nancy A. Moran. 2017. “Antibiotic Exposure
Perturbs the Gut Microbiota and Elevates Mortality in Honeybees.” PLOS
Biology 15 (3): e2001861. https://doi.org/10.1371/journal.pbio.2001861.
Reeve, Wayne G., Ravi P. Tiwari, Neetin B. Kale, Michael J. Dilworth, and Andrew R.
Glenn. 2002. “ActP Controls Copper Homeostasis in Rhizobium
Leguminosarum Bv. Viciae and Sinorhizobium Meliloti Preventing Low PH-
Induced Copper Toxicity.” Molecular Microbiology 43 (4): 981–91.
Regueiro, Leticia, Marta Carballa, and Juan M. Lema. 2014. “Outlining Microbial
Community Dynamics during Temperature Drop and Subsequent Recovery
Period in Anaerobic Co-Digestion Systems.” Journal of Biotechnology 192
(December): 179–86. https://doi.org/10.1016/j.jbiotec.2014.10.007.
———. 2016. “Microbiome Response to Controlled Shifts in Ammonium and LCFA
Levels in Co-Digestion Systems.” Journal of Biotechnology 220 (February): 35–
44. https://doi.org/10.1016/j.jbiotec.2016.01.006.
Rho, M., H. Tang, and Y. Ye. 2010. “FragGeneScan: Predicting Genes in Short and
Error-Prone Reads.” Nucleic Acids Research 38 (20): e191–e191.
https://doi.org/10.1093/nar/gkq747.
Robinson, Courtney J., Brendan J. M. Bohannan, and Vincent B. Young. 2010. “From
Structure to Function: The Ecology of Host-Associated Microbial
Communities.” Microbiology and Molecular Biology Reviews : MMBR 74 (3):
453–76. https://doi.org/10.1128/MMBR.00014-10.
Rocha, Eduardo P. C. 2018. “Neutral Theory, Microbial Practice: Challenges in
Bacterial Population Genetics.” Molecular Biology and Evolution 35 (6): 1338–
47. https://doi.org/10.1093/molbev/msy078.
208
Rohart, Florian, Benoît Gautier, Amrit Singh, and Kim-Anh Lê Cao. 2017. “MixOmics:
An R Package for ‘omics Feature Selection and Multiple Data Integration.”
PLOS Computational Biology 13 (11): e1005752.
https://doi.org/10.1371/journal.pcbi.1005752.
Rosindell, James, Stephen P. Hubbell, and Rampal S. Etienne. 2011. “The Unified
Neutral Theory of Biodiversity and Biogeography at Age Ten.” Trends in
Ecology & Evolution 26 (7): 340–48. https://doi.org/10.1016/j.tree.2011.03.024.
Rosindell, James, Stephen P. Hubbell, Fangliang He, Luke J. Harmon, and Rampal S.
Etienne. 2012. “The Case for Ecological Neutral Theory.” Trends in Ecology &
Evolution 27 (4): 203–8. https://doi.org/10.1016/j.tree.2012.01.004.
Röttjers, Lisa, and Karoline Faust. 2018. “From Hairballs to Hypotheses-Biological
Insights from Microbial Networks.” FEMS Microbiology Reviews 42 (6): 761–
80. https://doi.org/10.1093/femsre/fuy030.
Ryan Gregory, T. 2002. “Genome Size and Developmental Complexity.” Genetica 115
(1): 131–46. https://doi.org/10.1023/A:1016032400147.
Ryman, Jennie E., Jillian L. A. Van Walleghem, and Paul J. Blanchfield. 2008.
“Methylmercury Levels in a Parasite (<Emphasis Type="Italic">Apophallus
Brevis</Emphasis> Metacercariae) and Its Host, Yellow Perch (<Emphasis
Type="Italic">Perca Flavescens</Emphasis>).” Aquatic Ecology 42 (3): 495–
501. https://doi.org/10.1007/s10452-007-9104-4.
Salcher, Michaela M. 2013. “Same Same but Different: Ecological Niche Partitioning of
Planktonic Freshwater Prokaryotes.” Journal of Limnology 73 (s1).
https://doi.org/10.4081/jlimnol.2014.813.
Sánchez-Andrea, Irene, Nuria Rodríguez, Ricardo Amils, and José Luis Sanz. 2011.
“Microbial Diversity in Anaerobic Sediments at Río Tinto, a Naturally Acidic
Environment with a High Heavy Metal Content.” Applied and Environmental
Microbiology 77 (17): 6085–93. https://doi.org/10.1128/AEM.00654-11.
Sandifer, Paul A., and Ariana E. Sutton-Grier. 2014. “Connecting Stressors, Ocean
Ecosystem Services, and Human Health.” Natural Resources Forum 38 (3):
157–67. https://doi.org/10.1111/1477-8947.12047.
Scheffer, Marten, Stephen R. Carpenter, Vasilis Dakos, and Egbert H. van Nes. 2015.
“Generic Indicators of Ecological Resilience: Inferring the Chance of a Critical
Transition.” Annual Review of Ecology, Evolution, and Systematics 46 (1): 145–
67. https://doi.org/10.1146/annurev-ecolsys-112414-054242.
Schloss, Patrick D., Sarah L. Westcott, Thomas Ryabin, Justine R. Hall, Martin
Hartmann, Emily B. Hollister, Ryan A. Lesniewski, et al. 2009. “Introducing
Mothur: Open-Source, Platform-Independent, Community-Supported Software
for Describing and Comparing Microbial Communities.” Applied and
Environmental Microbiology 75 (23): 7537–41.
https://doi.org/10.1128/AEM.01541-09.
Schmidt, Thomas S. B., João F. Matias Rodrigues, and Christian von Mering. 2014.
“Ecological Consistency of SSU RRNA-Based Operational Taxonomic Units at
a Global Scale.” PLOS Comput Biol 10 (4): e1003594.
https://doi.org/10.1371/journal.pcbi.1003594.
Schmidt, Victor T., Katherine F. Smith, Donald W. Melvin, and Linda A. Amaral‐
Zettler. 2015. “Community Assembly of a Euryhaline Fish Microbiome during
Salinity Acclimation.” Molecular Ecology 24 (10): 2537–50.
https://doi.org/10.1111/mec.13177.
Schomburg, Ida, Antje Chang, Christian Ebeling, Marion Gremse, Christian Heldt,
Gregor Huhn, and Dietmar Schomburg. 2004. “BRENDA, the Enzyme
209
Database: Updates and Major New Developments.” Nucleic Acids Research 32
(Database issue): D431-433. https://doi.org/10.1093/nar/gkh081.
Shade, Ashley, Hannes Peter, Steven D. Allison, Didier Baho, Mercé Berga, Helmut
Buergmann, David H. Huber, et al. 2012. “Fundamentals of Microbial
Community Resistance and Resilience.” Frontiers in Microbiology 3.
https://doi.org/10.3389/fmicb.2012.00417.
Shannon, Paul, Andrew Markiel, Owen Ozier, Nitin S. Baliga, Jonathan T. Wang,
Daniel Ramage, Nada Amin, Benno Schwikowski, and Trey Ideker. 2003.
“Cytoscape: A Software Environment for Integrated Models of Biomolecular
Interaction Networks.” Genome Research 13 (11): 2498–2504.
https://doi.org/10.1101/gr.1239303.
Shapiro, B. Jesse. 2014. “Signatures of Natural Selection and Ecological Differentiation
in Microbial Genomes.” In Ecological Genomics, 339–359. Springer.
http://link.springer.com/chapter/10.1007/978-94-007-7347-9_17.
Shapiro, B. Jesse, Jonathan Friedman, Otto X. Cordero, Sarah P. Preheim, Sonia C.
Timberlake, Gitta Szabó, Martin F. Polz, and Eric J. Alm. 2012. “Population
Genomics of Early Events in the Ecological Differentiation of Bacteria.” Science
(New York, N.Y.) 336 (6077): 48–51. https://doi.org/10.1126/science.1218198.
Shapiro, B. Jesse, and Martin F. Polz. 2014. “Ordering Microbial Diversity into
Ecologically and Genetically Cohesive Units.” Trends in Microbiology 22 (5):
235–47. https://doi.org/10.1016/j.tim.2014.02.006.
Sharp, Paul M., Elizabeth Bailes, Russell J. Grocock, John F. Peden, and R. Elizabeth
Sockett. 2005. “Variation in the Strength of Selected Codon Usage Bias among
Bacteria.” Nucleic Acids Research 33 (4): 1141–53.
https://doi.org/10.1093/nar/gki242.
Sherrill-Mix, Scott, Kevin McCormick, Abigail Lauder, Aubrey Bailey, Laurie
Zimmerman, Yingying Li, Jean-Bosco N. Django, et al. 2018. “Allometry and
Ecology of the Bilaterian Gut Microbiome.” MBio 9 (2): e00319-18.
https://doi.org/10.1128/mBio.00319-18.
Shou, Wenying, Carl T. Bergstrom, Arup K. Chakraborty, and Frances K. Skinner.
2015. “Theory, Models and Biology.” ELife 4 (July): e07158.
https://doi.org/10.7554/eLife.07158.
Silva, Genivaldo Gueiros Z., Kevin T. Green, Bas E. Dutilh, and Robert A. Edwards.
2015. “SUPER-FOCUS: A Tool for Agile Functional Analysis of Shotgun
Metagenomic Data.” Bioinformatics, btv584.
http://bioinformatics.oxfordjournals.org/content/early/2015/10/08/bioinformatics
.btv584.short.
Silver, Simon, and Le T. Phung. 1996. “BACTERIAL HEAVY METAL
RESISTANCE: New Surprises.” Annual Review of Microbiology 50 (1): 753–
89. https://doi.org/10.1146/annurev.micro.50.1.753.
Simpson, E. H. 1949. “Measurement of Diversity.” Nature 163 (4148): 688–688.
https://doi.org/10.1038/163688a0.
Singh, Pallavi, Tracy K. Teal, Terence L. Marsh, James M. Tiedje, Rebekah Mosci,
Katherine Jernigan, Angela Zell, et al. 2015. “Intestinal Microbial Communities
Associated with Acute Enteric Infections and Disease Recovery.” Microbiome 3
(1). https://doi.org/10.1186/s40168-015-0109-2.
Siviter, Harry, Julia Koricheva, Mark J. F. Brown, and Ellouise Leadbeater. 2018.
“Quantifying the Impact of Pesticides on Learning and Memory in Bees.”
Journal of Applied Ecology 55 (6): 2812–21. https://doi.org/10.1111/1365-
2664.13193.
210
Sloan, William T., Mary Lunn, Stephen Woodcock, Ian M. Head, Sean Nee, and
Thomas P. Curtis. 2006. “Quantifying the Roles of Immigration and Chance in
Shaping Prokaryote Community Structure.” Environmental Microbiology 8 (4):
732–40. https://doi.org/10.1111/j.1462-2920.2005.00956.x.
Sloan, William T., Stephen Woodcock, Mary Lunn, Ian M. Head, and Thomas P. Curtis.
2007. “Modeling Taxa-Abundance Distributions in Microbial Communities
Using Environmental Sequence Data.” Microbial Ecology 53 (3): 443–55.
https://doi.org/10.1007/s00248-006-9141-x.
Sluijs, Jeroen P van der, Noa Simon-Delso, Dave Goulson, Laura Maxim, Jean-Marc
Bonmatin, and Luc P Belzunces. 2013. “Neonicotinoids, Bee Disorders and the
Sustainability of Pollinator Services.” Current Opinion in Environmental
Sustainability 5 (3–4): 293–305. https://doi.org/10.1016/j.cosust.2013.05.007.
Smillie, Chris S., Mark B. Smith, Jonathan Friedman, Otto X. Cordero, Lawrence A.
David, and Eric J. Alm. 2011. “Ecology Drives a Global Network of Gene
Exchange Connecting the Human Microbiome.” Nature 480 (7376): 241–44.
https://doi.org/10.1038/nature10571.
Smith, John Maynard, and John Haigh. 1974. “The Hitch-Hiking Effect of a Favourable
Gene.” Genetics Research 23 (1): 23–35.
https://doi.org/10.1017/S0016672300014634.
Snedeker, Suzanne M., and Anthony G. Hay. 2012. “Do Interactions between Gut
Ecology and Environmental Chemicals Contribute to Obesity and Diabetes?”
Environmental Health Perspectives 120 (3): 332–39.
https://doi.org/10.1289/ehp.1104204.
Sokurenko, Evgeni V., Michael Feldgarden, Elena Trintchina, Scott J. Weissman,
Serine Avagyan, Sujay Chattopadhyay, James R. Johnson, and Daniel E.
Dykhuizen. 2004. “Selection Footprint in the FimH Adhesin Shows
Pathoadaptive Niche Differentiation in Escherichia Coli.” Molecular Biology
and Evolution 21 (7): 1373–83. https://doi.org/10.1093/molbev/msh136.
Sommer, Felix, Jacqueline Moltzau Anderson, Richa Bharti, Jeroen Raes, and Philip
Rosenstiel. 2017. “The Resilience of the Intestinal Microbiota Influences Health
and Disease.” Nature Reviews Microbiology 15 (10): 630–38.
https://doi.org/10.1038/nrmicro.2017.58.
Sommer, U., and B. Worm. 2002. “Synthesis: Back to Santa Rosalia, or No Wonder
There Are So Many Species.” In Competition and Coexistence, edited by Ulrich
Sommer and Boris Worm, 207–18. Ecological Studies. Berlin, Heidelberg:
Springer Berlin Heidelberg. https://doi.org/10.1007/978-3-642-56166-5_8.
Song, Hyun-Seob, William R. Cannon, Alexander S. Beliaev, and Allan Konopka.
2014. “Mathematical Modeling of Microbial Community Dynamics: A
Methodological Review.” Processes 2 (4): 711–52.
https://doi.org/10.3390/pr2040711.
Song, Hyun-Seob, Ryan S. Renslow, Jim K. Fredrickson, and Stephen R. Lindemann.
2015. “Integrating Ecological and Engineering Concepts of Resilience in
Microbial Communities.” Frontiers in Microbiology 6 (December).
https://doi.org/10.3389/fmicb.2015.01298.
Soucek, Pavel. 2011. “Xenobiotics.” In Encyclopedia of Cancer, edited by Manfred
Schwab, 3964–67. Berlin, Heidelberg: Springer Berlin Heidelberg.
https://doi.org/10.1007/978-3-642-16483-5_6276.
Soyer, Orkun S., ed. 2012. Evolutionary Systems Biology. Advances in Experimental
Medicine and Biology. New York: Springer-Verlag.
https://www.springer.com/gp/book/9781461435662.
211
Šrut, Maja, Sebastian Menke, Martina Höckner, and Simone Sommer. 2019.
“Earthworms and Cadmium – Heavy Metal Resistant Gut Bacteria as Indicators
for Heavy Metal Pollution in Soils?” Ecotoxicology and Environmental Safety
171 (April): 843–53. https://doi.org/10.1016/j.ecoenv.2018.12.102.
Staley, James T., and Allan Konopka. 1985. “Measurement of in Situ Activities of
Nonphotosynthetic Microorganisms in Aquatic and Terrestrial Habitats.” Annual
Review of Microbiology 39 (1): 321–46.
https://doi.org/10.1146/annurev.mi.39.100185.001541.
Stankovic, Srdjan, Ivana Moric, Aleksandar Pavic, Branka Vasiljevic, Barrie Johnson,
and Vladica Cvetkovic. 2014. “Investigation of the Microbial Diversity of an
Extremely Acidic, Metal-Rich Water Body (Lake Robule, Bor, Serbia).” Journal
of the Serbian Chemical Society 79 (6): 729–41.
https://doi.org/10.2298/JSC130605071S.
Starikova, Irina, Klaus Harms, Pål Haugen, Tracy T. M. Lunde, Raul Primicerio, Ørjan
Samuelsen, Kaare M. Nielsen, and Pål J. Johnsen. 2012. “A Trade-off between
the Fitness Cost of Functional Integrases and Long-Term Stability of Integrons.”
Edited by Xavier Didelot. PLoS Pathogens 8 (11): e1003043.
https://doi.org/10.1371/journal.ppat.1003043.
Stegen, James C, Xueju Lin, Allan E Konopka, and James K Fredrickson. 2012.
“Stochastic and Deterministic Assembly Processes in Subsurface Microbial
Communities.” The ISME Journal 6 (9): 1653–64.
https://doi.org/10.1038/ismej.2012.22.
Stephens, W Zac, Adam R Burns, Keaton Stagaman, Sandi Wong, John F Rawls, Karen
Guillemin, and Brendan J M Bohannan. 2016. “The Composition of the
Zebrafish Intestinal Microbial Community Varies across Development.” The
ISME Journal 10 (3): 644–54. https://doi.org/10.1038/ismej.2015.140.
Stokes, Hatch W., and Michael R. Gillings. 2011. “Gene Flow, Mobile Genetic
Elements and the Recruitment of Antibiotic Resistance Genes into Gram-
Negative Pathogens: Lateral Gene Transfer and Antibiotic Resistance
Evolution.” FEMS Microbiology Reviews 35 (5): 790–819.
https://doi.org/10.1111/j.1574-6976.2011.00273.x.
Streten-Joyce, Claire, Judy Manning, Karen S. Gibb, Brett A. Neilan, and David L.
Parry. 2013. “The Chemical Composition and Bacteria Communities in Acid
and Metalliferous Drainage from the Wet–Dry Tropics Are Dependent on
Season.” Science of The Total Environment 443 (January): 65–79.
https://doi.org/10.1016/j.scitotenv.2012.10.024.
Sugihara, George, and Robert M. May. 1990. “Nonlinear Forecasting as a Way of
Distinguishing Chaos from Measurement Error in Time Series.” Nature 344
(6268): 734–41. https://doi.org/10.1038/344734a0.
Sung, W., M. S. Ackerman, S. F. Miller, T. G. Doak, and M. Lynch. 2012. “Drift-
Barrier Hypothesis and Mutation-Rate Evolution.” Proceedings of the National
Academy of Sciences 109 (45): 18488–92.
https://doi.org/10.1073/pnas.1216223109.
Svolos, Vaios, Richard Hansen, Ben Nichols, Christopher Quince, Umer Z. Ijaz,
Rodanthi T. Papadopoulou, Christine A. Edwards, et al. 2019. “Treatment of
Active Crohn’s Disease With an Ordinary Food-Based Diet That Replicates
Exclusive Enteral Nutrition.” Gastroenterology 156 (5): 1354-1367.e6.
https://doi.org/10.1053/j.gastro.2018.12.002.
Sylvain, François-Étienne, Bachar Cheaib, Martin Llewellyn, Tiago Gabriel Correia,
Daniel Barros Fagundes, Adalberto Luis Val, and Nicolas Derome. 2016. “PH
212
Drop Impacts Differentially Skin and Gut Microbiota of the Amazonian Fish
Tambaqui (Colossoma Macropomum).” Scientific Reports 6 (August): 32032.
https://doi.org/10.1038/srep32032.
Sylvain, François-Étienne, and Nicolas Derome. 2017. “Vertically and Horizontally
Transmitted Microbial Symbionts Shape the Gut Microbiota Ontogenesis of a
Skin-Mucus Feeding Discus Fish Progeny.” Scientific Reports 7 (1): 5263.
https://doi.org/10.1038/s41598-017-05662-w.
Syvanen, Michael. 2012. “Evolutionary Implications of Horizontal Gene Transfer.”
Annual Review of Genetics 46 (1): 341–58. https://doi.org/10.1146/annurev-
genet-110711-155529.
Taxis, Tasia M., Sara Wolff, Sarah J. Gregg, Nicholas O. Minton, Chiqian Zhang,
Jingjing Dai, Robert D. Schnabel, et al. 2015. “The Players May Change but the
Game Remains: Network Analyses of Ruminal Microbiomes Suggest
Taxonomic Differences Mask Functional Similarity.” Nucleic Acids Research 43
(20): 9600–9612. https://doi.org/10.1093/nar/gkv973.
Thaiss, Christoph A. 2018. “Microbiome Dynamics in Obesity.” Science 362 (6417):
903–4. https://doi.org/10.1126/science.aav6870.
Theis, Kevin R., Nolwenn M. Dheilly, Jonathan L. Klassen, Robert M. Brucker, John F.
Baines, Thomas C. G. Bosch, John F. Cryan, et al. 2016. “Getting the
Hologenome Concept Right: An Eco-Evolutionary Framework for Hosts and
Their Microbiomes.” MSystems 1 (2). https://doi.org/10.1128/mSystems.00028-
16.
Thorpe, Harry A., Sion C. Bayliss, Laurence D. Hurst, and Edward J. Feil. 2017.
“Comparative Analyses of Selection Operating on Nontranslated Intergenic
Regions of Diverse Bacterial Species.” Genetics 206 (1): 363–76.
https://doi.org/10.1534/genetics.116.195784.
Tian, Ren-Mao, Weipeng Zhang, Lin Cai, Yue-Him Wong, Wei Ding, and Pei-Yuan
Qian. 2017. “Genome Reduction and Microbe-Host Interactions Drive
Adaptation of a Sulfur-Oxidizing Bacterium Associated with a Cold Seep
Sponge.” MSystems 2 (2): e00184-16. https://doi.org/10.1128/mSystems.00184-
16.
Tilg, Herbert, and Arthur Kaser. 2011. “Gut Microbiome, Obesity, and Metabolic
Dysfunction.” The Journal of Clinical Investigation 121 (6): 2126–32.
https://doi.org/10.1172/JCI58109.
Tilg, Herbert, and Alexander R. Moschen. 2014. “Microbiota and Diabetes: An
Evolving Relationship.” Gut 63 (9): 1513–21. https://doi.org/10.1136/gutjnl-
2014-306928.
Tilman, D. 1982. “Resource Competition and Community Structure.” Monographs in
Population Biology 17: 1–296.
Tilman, David. 2004. “Niche Tradeoffs, Neutrality, and Community Structure: A
Stochastic Theory of Resource Competition, Invasion, and Community
Assembly.” Proceedings of the National Academy of Sciences 101 (30): 10854–
61. https://doi.org/10.1073/pnas.0403458101.
Trussart, Marie, Eva Yus, Sira Martinez, Davide Baù, Yuhei O. Tahara, Thomas Pengo,
Michael Widjaja, et al. 2017. “Defined Chromosome Structure in the Genome-
Reduced Bacterium Mycoplasma Pneumoniae.” Nature Communications 8
(March). https://doi.org/10.1038/ncomms14665.
Tucker, Marlee A., Katrin Böhning-Gaese, William F. Fagan, John M. Fryxell, Bram
Van Moorter, Susan C. Alberts, Abdullahi H. Ali, et al. 2018. “Moving in the
213
Anthropocene: Global Reductions in Terrestrial Mammalian Movements.”
Science 359 (6374): 466–69. https://doi.org/10.1126/science.aam9712.
“UniProt: The Universal Protein Knowledgebase.” 2017. Nucleic Acids Research 45
(Database issue): D158–69. https://doi.org/10.1093/nar/gkw1099.
Upadhyaya, Smitha, and Gautam Banerjee. 2015. “Type 2 Diabetes and Gut
Microbiome: At the Intersection of Known and Unknown.” Gut Microbes 6 (2):
85–92. https://doi.org/10.1080/19490976.2015.1024918.
Urbieta, M. Sofía, E. González Toril, A. Aguilera, M. Alejandra Giaveno, and E.
Donati. 2012. “First Prokaryotic Biodiversity Assessment Using Molecular
Techniques of an Acidic River in Neuquén, Argentina.” Microbial Ecology 64
(1): 91–104. https://doi.org/10.1007/s00248-011-9997-2.
Valente, T., M. J. Rivera, S. F. P. Almeida, C. Delgado, P. Gomes, J. A. Grande, M. L.
de la Torre, and M. Santisteban. 2015. “Characterization of Water Reservoirs
Affected by Acid Mine Drainage: Geochemical, Mineralogical, and Biological
(Diatoms) Properties of the Water.” Environmental Science and Pollution
Research 23 (7): 6002–11. https://doi.org/10.1007/s11356-015-4776-0.
Vanwonterghem, Inka, Paul D Jensen, Paul G Dennis, Philip Hugenholtz, Korneel
Rabaey, and Gene W Tyson. 2014. “Deterministic Processes Guide Long-Term
Synchronised Population Dynamics in Replicate Anaerobic Digesters.” The
ISME Journal 8 (10): 2015–28. https://doi.org/10.1038/ismej.2014.50.
Vázquez-Baeza, Yoshiki, Embriette R. Hyde, Jan S. Suchodolski, and Rob Knight.
2016. “Dog and Human Inflammatory Bowel Disease Rely on Overlapping yet
Distinct Dysbiosis Networks.” Nature Microbiology 1 (October): 16177.
https://doi.org/10.1038/nmicrobiol.2016.177.
Vellend, Mark. 2010. “Conceptual Synthesis in Community Ecology.” The Quarterly
Review of Biology 85 (2): 183–206.
Venkataraman, Arvind, Christine M. Bassis, James M. Beck, Vincent B. Young, Jeffrey
L. Curtis, Gary B. Huffnagle, and Thomas M. Schmidt. 2015. “Application of a
Neutral Community Model To Assess Structuring of the Human Lung
Microbiome.” MBio 6 (1): e02284-14. https://doi.org/10.1128/mBio.02284-14.
Vieira-Silva, Sara, and Eduardo P. C. Rocha. 2010. “The Systemic Imprint of Growth
and Its Uses in Ecological (Meta)Genomics.” PLOS Genetics 6 (1): e1000808.
https://doi.org/10.1371/journal.pgen.1000808.
Vincent, Antony T., Nicolas Derome, Brian Boyle, Alexander I. Culley, and Steve J.
Charette. 2017. “Next-Generation Sequencing (NGS) in the Microbiological
World: How to Make the Most of Your Money.” Journal of Microbiological
Methods, What’s next in microbiology methods? Emerging methods, 138 (July):
60–71. https://doi.org/10.1016/j.mimet.2016.02.016.
Vincent, Antony T., Jeff Gauthier, Nicolas Derome, and Steve J. Charette. 2019. “The
Rise and Fall of Antibiotics in Aquaculture.” In Microbial Communities in
Aquaculture Ecosystems: Improving Productivity and Sustainability, edited by
Nicolas Derome, 1–19. Cham: Springer International Publishing.
https://doi.org/10.1007/978-3-030-16190-3_1.
Volant, Aurélie, Odile Bruneel, Angélique Desoeuvre, Marina Héry, Corinne Casiot,
Noëlle Bru, Sophie Delpoux, et al. 2014. “Diversity and Spatiotemporal
Dynamics of Bacterial Communities: Physicochemical and Other Drivers along
an Acid Mine Drainage.” FEMS Microbiology Ecology 90 (1): 247–63.
https://doi.org/10.1111/1574-6941.12394.
214
Volterra, Vito. 1928. “Variations and Fluctuations of the Number of Individuals in
Animal Species Living Together.” ICES Journal of Marine Science 3 (1): 3–51.
https://doi.org/10.1093/icesjms/3.1.3.
Vymazal, Jan. 1987. “Toxicity and Accumulation of Cadmium with Respect to Algae
and Cyanobacteria: A Review.” Toxicity Assessment 2 (4): 387–415.
https://doi.org/10.1002/tox.2540020403.
Wang, Fushun, Congqiang Liu, Xiaobing Liang, and Zhongqing Wei. 2003.
“Remobilization of Trace Metals Induced by Microbiological Activities near
Sediment-Water Interface, Aha Lake, Guiyang.” Chinese Science Bulletin 48
(21): 2352–56. https://doi.org/10.1360/03wd0013.
Wang, Kai, Demin Zhang, Jinbo Xiong, Xinxin Chen, Jialai Zheng, Changju Hu, Yina
Yang, and Jianlin Zhu. 2015. “Response of Bacterioplankton Communities to
Cadmium Exposure in Coastal Water Microcosms with High Temporal
Variability.” Applied and Environmental Microbiology 81 (1): 231–40.
https://doi.org/10.1128/AEM.02562-14.
Wanner, O., and W. Gujer. 1986. “A Multispecies Biofilm Model.” Biotechnology and
Bioengineering 28 (3): 314–28. https://doi.org/10.1002/bit.260280304.
Ward, David M., Mary M. Bateson, Michael J. Ferris, Michael Kühl, Andrea Wieland,
Alex Koeppel, and Frederick M. Cohan. 2006. “Cyanobacterial Ecotypes in the
Microbial Mat Community of Mushroom Spring (Yellowstone National Park,
Wyoming) as Species-like Units Linking Microbial Community Composition,
Structure and Function.” Philosophical Transactions of the Royal Society of
London. Series B, Biological Sciences 361 (1475): 1997–2008.
https://doi.org/10.1098/rstb.2006.1919.
Waters, Colin N., Jan Zalasiewicz, Colin Summerhayes, Anthony D. Barnosky,
Clément Poirier, Agnieszka Gałuszka, Alejandro Cearreta, et al. 2016. “The
Anthropocene Is Functionally and Stratigraphically Distinct from the Holocene.”
Science 351 (6269): aad2622. https://doi.org/10.1126/science.aad2622.
Webb, Campbell O., David D. Ackerly, Mark A. McPeek, and Michael J. Donoghue.
2002. “Phylogenies and Community Ecology.” Annual Review of Ecology and
Systematics 33 (1): 475–505.
https://doi.org/10.1146/annurev.ecolsys.33.010802.150448.
Weiss, Sophie, Will Van Treuren, Catherine Lozupone, Karoline Faust, Jonathan
Friedman, Ye Deng, Li Charlie Xia, et al. 2016. “Correlation Detection
Strategies in Microbial Data Sets Vary Widely in Sensitivity and Precision.” The
ISME Journal 10 (7): 1669–81. https://doi.org/10.1038/ismej.2015.235.
Wendt-Potthoff, K., and M. Koschorreck. 2002. “Functional Groups and Activities of
Bacteria in a Highly Acidic Volcanic Mountain Stream and Lake in Patagonia,
Argentina.” Microbial Ecology 43 (1): 92–106. https://doi.org/10.1007/s00248-
001-1030-8.
Werner, Jeffrey J., Omry Koren, Philip Hugenholtz, Todd Z. DeSantis, William A.
Walters, J. Gregory Caporaso, Largus T. Angenent, Rob Knight, and Ruth E.
Ley. 2012. “Impact of Training Sets on Classification of High-Throughput
Bacterial 16s RRNA Gene Surveys.” The ISME Journal 6 (1): 94–103.
https://doi.org/10.1038/ismej.2011.82.
Widder, Stefanie, Rosalind J. Allen, Thomas Pfeiffer, Thomas P. Curtis, Carsten Wiuf,
William T. Sloan, Otto X. Cordero, et al. 2016. “Challenges in Microbial
Ecology: Building Predictive Understanding of Community Function and
Dynamics.” The ISME Journal 10 (11): 2557–68.
https://doi.org/10.1038/ismej.2016.45.
215
Wielgoss, Sébastien, Jeffrey E. Barrick, Olivier Tenaillon, Michael J. Wiser, W. James
Dittmar, Stéphane Cruveiller, Béatrice Chane-Woon-Ming, Claudine Médigue,
Richard E. Lenski, and Dominique Schneider. 2013. “Mutation Rate Dynamics
in a Bacterial Population Reflect Tension between Adaptation and Genetic
Load.” Proceedings of the National Academy of Sciences of the United States of
America 110 (1): 222–27. https://doi.org/10.1073/pnas.1219574110.
Wiles, Travis J., Matthew Jemielita, Ryan P. Baker, Brandon H. Schlomann, Savannah
L. Logan, Julia Ganz, Ellie Melancon, Judith S. Eisen, Karen Guillemin, and
Raghuveer Parthasarathy. 2016. “Host Gut Motility Promotes Competitive
Exclusion within a Model Intestinal Microbiota.” PLOS Biology 14 (7):
e1002517. https://doi.org/10.1371/journal.pbio.1002517.
Wilke, Andreas, Jared Bischof, Travis Harrison, Tom Brettin, Mark D’Souza, Wolfgang
Gerlach, Hunter Matthews, et al. 2015. “A RESTful API for Accessing
Microbial Community Data for MG-RAST.” Edited by Paul P. Gardner. PLoS
Computational Biology 11 (1): e1004008.
https://doi.org/10.1371/journal.pcbi.1004008.
Wills, Christopher, Richard Condit, Robin B. Foster, and Stephen P. Hubbell. 1997.
“Strong Density- and Diversity-Related Effects Help to Maintain Tree Species
Diversity in a Neotropical Forest.” Proceedings of the National Academy of
Sciences 94 (4): 1252–57. https://doi.org/10.1073/pnas.94.4.1252.
Wintersdorff, Christian J. H. von, John Penders, Julius M. van Niekerk, Nathan D.
Mills, Snehali Majumder, Lieke B. van Alphen, Paul H. M. Savelkoul, and Petra
F. G. Wolffs. 2016. “Dissemination of Antimicrobial Resistance in Microbial
Ecosystems through Horizontal Gene Transfer.” Frontiers in Microbiology 7
(February). https://doi.org/10.3389/fmicb.2016.00173.
Woese, C R. 1987. “Bacterial Evolution.” Microbiological Reviews 51 (2): 221–71.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC373105/.
Wolf, Yuri I, and Eugene V Koonin. 2013. “Genome Reduction as the Dominant Mode
of Evolution.” Bioessays 35 (9): 829–37.
https://doi.org/10.1002/bies.201300037.
Wong, Sandi, and John F. Rawls. 2012. “Intestinal Microbiota Composition in Fishes Is
Influenced by Host Ecology and Environment.” Molecular Ecology 21 (13):
3100–3102. http://onlinelibrary.wiley.com/doi/10.1111/j.1365-
294X.2012.05646.x/full.
Wong, Sandi, W. Zac Stephens, Adam R. Burns, Keaton Stagaman, Lawrence A.
David, Brendan J. M. Bohannan, Karen Guillemin, and John F. Rawls. 2015.
“Ontogenetic Differences in Dietary Fat Influence Microbiota Assembly in the
Zebrafish Gut.” MBio 6 (5): e00687-15. https://doi.org/10.1128/mBio.00687-15.
Woodcock, Stephen, Christopher J. Van Der Gast, Thomas Bell, Mary Lunn, Thomas P.
Curtis, Ian M. Head, and William T. Sloan. 2007. “Neutral Assembly of
Bacterial Communities: Neutral Assembly of Bacterial Communities.” FEMS
Microbiology Ecology 62 (2): 171–80. https://doi.org/10.1111/j.1574-
6941.2007.00379.x.
Wu, Xiangyang, Samuel J. Cobbina, Guanghua Mao, Hai Xu, Zhen Zhang, and Liuqing
Yang. 2016. “A Review of Toxicity and Mechanisms of Individual and Mixtures
of Heavy Metals in the Environment.” Environmental Science and Pollution
Research, March, 1–16. https://doi.org/10.1007/s11356-016-6333-x.
Xie, Lingtian, Daniel Lambert, Caitrin Martin, Daniel J. Cain, Samuel N. Luoma, and
David Buchwalter. 2008. “Cadmium Biodynamics in the Oligochaete
216
Lumbriculus Variegatus and Its Implications for Trophic Transfer.” Aquatic
Toxicology 86 (2): 265–71. https://doi.org/10.1016/j.aquatox.2007.11.008.
Yamada, Takuji, Ivica Letunic, Shujiro Okuda, Minoru Kanehisa, and Peer Bork. 2011.
“IPath2.0: Interactive Pathway Explorer.” Nucleic Acids Research 39 (Web
Server issue): W412-415. https://doi.org/10.1093/nar/gkr313.
Yan, Qingyun, Jinjin Li, Yuhe Yu, Jianjun Wang, Zhili He, Joy D. Van Nostrand,
Megan L. Kempher, et al. 2016. “Environmental Filtering Decreases with Fish
Development for the Assembly of Gut Microbiota: Gut Microbiota Assembly
across Fish Development.” Environmental Microbiology 18 (12): 4739–54.
https://doi.org/10.1111/1462-2920.13365.
Yang, Jianguo, Xiaqing Xie, Mingxuan Yang, Ray Dixon, and Yi-Ping Wang. 2017.
“Modular Electron-Transport Chains from Eukaryotic Organelles Function to
Support Nitrogenase Activity.” Proceedings of the National Academy of
Sciences 114 (12): E2460–65. https://doi.org/10.1073/pnas.1620058114.
Yeh, Yi-Chun, Pedro R. Peres‐Neto, Shiao-Wei Huang, Yung-Chih Lai, Chen-Yi Tu,
Fuh-Kwo Shiah, Gwo-Ching Gong, and Chih-hao Hsieh. 2015. “Determinism of
Bacterial Metacommunity Dynamics in the Southern East China Sea Varies
Depending on Hydrography.” Ecography 38 (2): 198–212.
https://doi.org/10.1111/ecog.00986.
Yoshida, Takehito, Stephen P. Ellner, Laura E. Jones, Brendan J. M. Bohannan, Richard
E. Lenski, and Nelson G. Hairston Jr. 2007. “Cryptic Population Dynamics:
Rapid Evolution Masks Trophic Interactions.” PLOS Biology 5 (9): e235.
https://doi.org/10.1371/journal.pbio.0050235.
Yoshida, Takehito, Laura E. Jones, Stephen P. Ellner, Gregor F. Fussmann, and Nelson
G. Hairston. 2003. “Rapid Evolution Drives Ecological Dynamics in a Predator–
Prey System.” Nature 424 (6946): 303–6. https://doi.org/10.1038/nature01767.
Zaccaria, Marco, Sandra Dedrick, and Babak Momeni. 2017. “Modeling Microbial
Communities: A Call for Collaboration between Experimentalists and
Theorists.” Processes 5 (4): 53. https://doi.org/10.3390/pr5040053.
Zeng, Qinglong, Jeet Sukumaran, Steven Wu, and Allen Rodrigo. 2015. “Neutral
Models of Microbiome Evolution.” PLOS Computational Biology 11 (7):
e1004365. https://doi.org/10.1371/journal.pcbi.1004365.
Zeng, Qinglong, Steven Wu, Jeet Sukumaran, and Allen Rodrigo. 2017. “Models of
Microbiome Evolution Incorporating Host and Microbial Selection.”
Microbiome 5 (September): 127. https://doi.org/10.1186/s40168-017-0343-x.
Zeng, Xiangfeng, Xijuan Chen, and Jie Zhuang. 2015. “The Positive Relationship
between Ocean Acidification and Pollution.” Marine Pollution Bulletin 91 (1):
14–21. https://doi.org/10.1016/j.marpolbul.2014.12.001.
Zha, Yinghua, Alexander Eiler, Frank Johansson, and Richard Svanbäck. 2018. “Effects
of Predation Stress and Food Ration on Perch Gut Microbiota.” Microbiome 6
(1): 28. https://doi.org/10.1186/s40168-018-0400-0.
Zhai, Qixiao, Leilei Yu, Tianqi Li, Jiamin Zhu, Chengcheng Zhang, Jianxin Zhao, Hao
Zhang, and Wei Chen. 2017. “Effect of Dietary Probiotic Supplementation on
Intestinal Microbiota and Physiological Conditions of Nile Tilapia (Oreochromis
Niloticus) under Waterborne Cadmium Exposure.” Antonie van Leeuwenhoek
110 (4): 501–13. https://doi.org/10.1007/s10482-016-0819-x.
Zhang, Erica, and James C. Nieh. 2015. “The Neonicotinoid Imidacloprid Impairs
Honey Bee Aversive Learning of Simulated Predation.” Journal of Experimental
Biology 218 (20): 3199–3205. https://doi.org/10.1242/jeb.127472.
217
Zhang, Songbin, Yuanxiang Jin, Zhaoyang Zeng, Zhenzhen Liu, and Zhengwei Fu.
2015. “Subchronic Exposure of Mice to Cadmium Perturbs Their Hepatic
Energy Metabolism and Gut Microbiome.” Chemical Research in Toxicology 28
(10): 2000–2009. https://doi.org/10.1021/acs.chemrestox.5b00237.
Zhang, Tianyu. 2017. “Modeling Biofilms: From Genes to Communities.” Processes 5
(1): 5. https://doi.org/10.3390/pr5010005.
Zhaxybayeva, Olga, Pascal Lapierre, and J.Peter Gogarten. 2004. “Genome Mosaicism
and Organismal Lineages.” Trends in Genetics 20 (5): 254–60.
https://doi.org/10.1016/j.tig.2004.03.009.
Ziegler, Maren, Francois O. Seneca, Lauren K. Yum, Stephen R. Palumbi, and Christian
R. Voolstra. 2017. “Bacterial Community Dynamics Are Linked to Patterns of
Coral Heat Tolerance.” Nature Communications 8 (February): 14213.
https://doi.org/10.1038/ncomms14213.
Top Related