JÉRÔME LEMAÎTRE

DÉTERMINANTS DE LA RÉPARTITION DES OISEAUX ET DES MICROMAMMIFÈRES EN FORÊT BORÉALE

NATURELLE ET AMÉNAGÉE

Thèse présentée à la Faculté des études supérieures de l’Université Laval

dans le cadre du programme de doctorat en biologie pour l’obtention du grade de Philosophiae Doctor (Ph.D.)

DÉPARTEMENT DE BIOLOGIE FACULTÉ DES SCIENCES ET GÉNIE

UNIVERSITÉ LAVAL QUÉBEC

2009 © Jérôme Lemaître, 2009

i

Résumé

La compréhension des déterminants de la répartition des organismes est un thème central en

écologie. Mon objectif était de comprendre quatre mécanismes de répartition des oiseaux et des

micromammifères des forêts boréales naturelles et aménagées (i.e., sélection d’habitat,

parasitisme compétition et prédation). Alors que la relation entre la diversité et l’hétérogénéité de

l’habitat était établie, la contribution relative de la structure et de la composition de l’habitat était

incertaine. Mon étude a montré que la structure et la composition de l’habitat expliquaient la

même variance dans 96 assemblages d’oiseaux occupant les vieilles forêts boréales, bien que

certaines espèces répondaient uniquement à la structure ou à la composition de l’habitat. Ces

résultats étaient similaires à cinq échelles spatiales, soit pour des rayons de paysage allant de 100

à 1000 m autour des stations. Les stratégies de conservation devraient mettre autant d’emphase à

préserver la structure et la composition de l’habitat. Ensuite, j’ai démontré que le parasitisme par

l’œstre (Cuterebra spp) pourrait contribuer à la dynamique de 36 populations du campagnol à dos

roux (Myodes gapperi), le micromammifère le plus abondant en forêt boréale naturelle. La

probabilité de survie de 341 campagnols diminuait avec l’infection dans des conditions de stress

élevé. Le principal facteur responsable du risque individuel d’infection était l’abondance de

campagnols plutôt que les traits d’histoire naturelle ou les variables d’habitat. La croissance

estivale des populations était négativement reliée à la prévalence du parasite. Enfin, j’ai

déterminé quels mécanismes, parmi la prédation et la compétition, influençaient la répartition du

campagnol à dos roux en fonction d’un gradient d’intensité de l’exploitation forestière, à l’aide

d’une expérience sur l’effort d’approvisionnement dans 464 mangeoires réparties dans 29 sites.

J’ai trouvé que la compétition interspécifique avec les souris sylvestre (Peromyscus maniculatus)

semblait être le principal mécanisme responsable du déclin des populations de campagnols à dos

roux induit par l’exploitation forestière. En conclusion, cette thèse améliore la connaissance de

quatre mécanismes de répartition de la faune boréale, grâce à la combinaison originale de l’étude

des assemblages d’espèces, de la dynamique des populations et du comportement animal. Mes

résultats devraient fournir des outils utiles pour l’aménagement durable de nos forêts.

ii

Abstract

Understanding the determinants that modulate organism distribution is a central theme in

ecology. The objective of my thesis was to understand four mechanisms that may influence bird

and small mammal distributions in natural and managed boreal forests (i.e., habitat selection,

parasitism, competition, and predation). Whereas the positive relationship between habitat

heterogeneity and species diversity was well established, the relative contributions of habitat

structure and composition remain uncertain. My study showed that habitat structure and

composition explained a similar amount of variance in 96 bird assemblages of old-growth boreal

forests, despite the fact that some species were exclusively associated with structure or

composition. These results did not vary among the five spatial scales considered, i.e. from 100 to

1000 m radii at the vicinity of sampling stations. Therefore, conservation strategies should put

equal weight in preserving structural and compositional habitat attributes. Next, I demonstrated

that bot fly (Cuterebra spp) parasitism can play a major role in the dynamic of 36 populations of

the red-backed vole (Myodes gapperi), the most abundant small mammal species of the boreal

forest. The probability of survival of red-backed voles decreased with bot fly infection under

stressful conditions (n = 341 individuals). The main factor influencing the individual risk of

infection was red-backed vole abundance rather than life-history traits or habitat attributes.

Ultimately, population growth of red-backed voles over the summer was negatively related to bot

fly prevalence. Lastly, I determined what mechanisms, among predation, intra- and interspecific

competition, influenced red-backed vole distribution along a gradient of forest harvesting

intensity, using an experiment on the foraging behavior in 464 seed trays distributed in 29 sites. I

found that interspecific competition with deer mice (Peromyscus maniculatus) seemed to be the

main mechanism responsible for the decline in red-backed vole populations that was induced by

forest harvesting. To conclude, this thesis improves our knowledge of four mechanisms

influencing wildlife distribution in the boreal forest, owing to the original combination of the

study of species assemblages, population dynamics, and animal behaviour. My results should

provide useful tools for sustainable management of the boreal forest.

iii

Avant-propos

Ce doctorat est présenté sous la forme d’une thèse avec insertion de trois articles scientifiques. La

thèse inclut une introduction générale et une conclusion générale qui lient l’ensemble des articles,

bien que chacun puisse être lu indépendamment. En tant qu’auteur principal des trois articles, j’ai

élaboré les objectifs de recherche, j’ai planifié et réalisé la collecte de données, j’ai effectué les

analyses statistiques et j’ai rédigé les manuscrits. Mon directeur de thèse, Daniel Fortin, a

largement contribué aux étapes de planification de la recherche et de rédaction des manuscrits.

Marcel Darveau, mon codirecteur, a également beaucoup contribué à ces étapes. Ensemble, nous

avons rédigé le chapitre 1 qui a été soumis à Ecography.

Daniel et moi avons codirigé Pierre-Olivier Montiglio dans le cadre de son initiation à la

recherche. Pierre-Olivier s’est chargé des dissections de micromammifères et ses travaux ont

éventuellement été intégrés à l’ensemble des résultats du chapitre 2. Ce chapitre est publié dans

Oecologia et mes coauteurs sont Daniel, Pierre-Olivier et Marcel.

Douglas W. Morris (Lakehead University, ON) a contribué à la planification du design

expérimental et à la rédaction du chapitre 3. Daniel, Douglas et Marcel sont les coauteurs de ce

chapitre qui a été soumis à Oikos.

Finalement, je voudrais souligner ma contribution à un article hors thèse, celui d’Ermias Azeria

et al. (2009) publié dans Global Ecology and Biogeography et pour lequel je suis troisième

auteur. Pour cet article, j’ai planifié et réalisé l’ensemble de la collecte de données d’oiseaux ainsi

qu’une partie des données de végétation, j’ai préparé les bases de données pour les analyses

statistiques, j’ai rédigé les sections aire d’étude et collecte de données pour l’article, j’ai aidé

Ermias concernant les analyses de végétation et d’oiseaux et j’ai révisé le manuscrit.

iv

Remerciements

Cette thèse est l’aboutissement de plusieurs années de travail que je n’aurais jamais pu accomplir

seul. Je voudrais donc remercier tous ceux qui ont contribué de près ou de loin au succès de ce

projet. D’abord, je tiens à remercier mon directeur de thèse, Daniel Fortin, qui m’a offert un

projet à construire et qui était toujours disponible pour m’aider quand j’en avais besoin. À travers

ce projet et la supervision de Daniel, j’ai appris énormément sur la recherche et la rédaction

d’articles. Merci aussi à Marcel Darveau, mon codirecteur, qui a initié ce projet et qui m’a

supporté jusqu’à la fin. L’expérience complémentaire de mes codirecteurs a été un atout de taille

dans ce projet.

Douglas W. Morris m’a beaucoup aidé à comprendre les concepts et les expériences d’écologie

de l’approvisionnement. Ses commentaires avisés et ses encouragements continuels furent des

plus appréciés et je garde un excellent souvenir de notre collaboration. Merci à Pierre-Olivier

Montiglio qui a disséqué des centaines de micromammifères. J’ai beaucoup appris sur la

supervision d’étudiants en sa présence. André Desrochers a été l’œil externe indispensable sur

mon comité de doctorat et ses conseils et commentaires se sont avérés des plus avisés, merci

André. David Pothier, avec sa connaissance de la forêt boréale, m’a permis de réaliser tôt dans le

projet que certaines de mes hypothèses seraient difficilement vérifiables dans l’aire d’étude, et je

lui suis reconnaissant de m’avoir épargné des efforts inutiles. David, ainsi que Jean-Claude Ruel

ont contribué à faciliter une logistique de terrain relativement complexe. Je voudrais remercier

également Louis Imbeau qui a accepté d’être l’examinateur externe de cette thèse, ainsi que

David et André qui seront les examinateurs internes.

Tous les étudiants du laboratoire de Daniel Fortin m’ont aidé, à un moment donné ou à autre de

mon projet. Notamment, James Hodson s’est avéré une ressource essentielle pour sa vision de

l’écologie forestière, sa fluidité en anglais et pour ses qualités de guitariste. Nico Courbin est le

maître incontesté du GIS et, comme un secouriste, il répondait en moins de 5 minutes à mes

requêtes. Mélanie-Louise Leblanc m’a aidé avec les analyses multivariées mais surtout, par nos

longues discussions sur nos projets respectifs. Merci aussi à Kim Poitras qui a grandement

contribué au succès du chapitre 3 grâce à nos travaux conjoints sur les GUD. Enfin, merci aux

autres étudiants du labo pour leur révision d’article, leurs conseils et leur bonne humeur : Ermias

v

Azeria, Jean-Sébastien Babin, Sabrina Courant, Karine Dancose, Pierre Etcheverry, Marie-eve

Fortin, Mélina Houle, Philippe Janssen, Cheryl A. Johnson, Isabelle Plante et David Pinaud.

En plus des étudiants du labo, les révisions de William F. J. Parsons et de Ghislain Rompré ont

permis d’améliorer le chapitre 1. Libby Marschall a beaucoup contribuée à l’amélioration du

chapitre 2 ainsi que Moshew Wolf et deux réviseurs anonymes.

J’ai une énorme reconnaissance pour le dévouement et la volonté de mes assistants de terrain,

sans qui ce projet n’aurait jamais pu atteindre une telle ampleur en termes de collecte de données.

En particulier, je voudrais remercier Emilie Bilodeau et Maud Bouthillette qui ont été des

assistantes hors pair pour les collectes de données d’oiseaux, de micromammifères et de

végétation. Merci aussi à Antoine Bourke, Sophie Brugerolle et Frédéric Lavoie. Également,

Louis Demers-Rousseau, Valérie Lecomte, Alexandra Dufresne, Rémi Lesmerises et Dominique

Trudeau m’ont impressionné par leur acharnement à transporter des dizaines de kilos de sable

dans les sites d’échantillonnage. Ils ont été des compagnons de terrain inestimables. Merci à

Guillaume Côté pour son aide énorme concernant la collecte de données de végétation. Vincent

Laflèche et Christine Casabon ont été d’excellents coordinateurs de terrain et m’ont beaucoup

facilité la vie. Je tiens à remercier Marc Saint-Onge (Kruger), Charles Gauthier et Charles

Warren (Abitibi Consol.), Daniel Gagnon (Bowater) et David Trudel (Arbec) qui nous ont permis

de faire nos recherches malgré les conditions économiques difficiles de l’industrie forestière et

nous ont évité de finir rôtis par les grands feux de 2005.

Je remercie bien sûr ma famille et mes amis qui m’ont offert un support inébranlable. En

particulier (car je ne peux pas tous vous nommer), merci à Emilie qui me supporte dans mon

doctorat et dans mes autres entreprises. Merci à Éliam, qui a vu le jour durant mon doctorat, et

avec qui j’apprends à m’arrêter pour profiter des joies de l’enfance. Merci à mes parents et à ma

grand-mère qui m’ont toujours supporté malgré la distance. Enfin, merci à ma plus fidèle

assistante de terrain, ma chienne Nayah. Elle supporte mes longues journées de rédaction par des

soupirs qui en disent long. Merci aussi à Pico, joyeuse gerbille, qui a goulûment contribué à mes

premières expériences de GUD.

vi

Finalement, je voudrais remercier les partenaires financiers de cette étude : la Chaire de recherche

CRSNG – Université Laval en sylviculture et faune, la Fondation canadienne pour l’innovation

(FCI) et le Fond québécois de la recherche sur la nature et les technologies (FQRNT).

Enfin, merci à tous ceux que j’ai oubliés par pure inadvertance…

vii

À Éliam, Emilie et Andrée

viii

Table des matières

Résumé ............................................................................................................................................. i Abstract............................................................................................................................................ ii Avant-propos .................................................................................................................................. iii Remerciements ............................................................................................................................... iv Table des matières ........................................................................................................................ viii Liste des tableaux ............................................................................................................................ x Liste des figures............................................................................................................................. xii Introduction générale ....................................................................................................................... 1

La biodiversité et l’humain.......................................................................................................... 1 Cadre de travail............................................................................................................................ 2 Sélection d’habitat et hétérogénéité de l’habitat.......................................................................... 6

Contribution relative de la structure et de la composition de l’habitat.................................... 7 Répartition et assemblage des espèces .................................................................................... 8 Référence écologique .............................................................................................................. 9

Interactions biologiques : parasitisme ......................................................................................... 9 Potentiel régulateur du parasitisme........................................................................................ 10 Cycle de vie de l’œstre .......................................................................................................... 12

Interactions biologiques : prédation et compétition .................................................................. 13 Altération de l’habitat ............................................................................................................ 13 Comportement et conservation.............................................................................................. 14 Hypothèses et prédictions...................................................................................................... 16

Aire d’étude : forêt boréale de l’est du Québec......................................................................... 17 Objectifs et organisation de la thèse .......................................................................................... 19

Chapitre 1 .............................................................................................................................. 19 Chapitre 2 .............................................................................................................................. 19 Chapitre 3 .............................................................................................................................. 19

Modèles d’étude : oiseaux et micromammifères....................................................................... 20 Chapitre 1. Multiscale assessment of the influence of habitat structure and composition on the distribution of boreal birds ............................................................................................................ 22

Résumé ...................................................................................................................................... 23 Abstract...................................................................................................................................... 24 Introduction ............................................................................................................................... 25 Materials and methods............................................................................................................... 26

Study area and bird sampling ................................................................................................ 26 Habitat structure and composition......................................................................................... 27 Community analyses.............................................................................................................. 28 Individual species analyses.................................................................................................... 29

Results ....................................................................................................................................... 30 Bird assemblage..................................................................................................................... 30 Probability of occurrence of individual bird species ............................................................. 31

Discussion.................................................................................................................................. 33 Relative contribution of habitat structure and composition................................................... 33 Habitat heterogeneity and spatial scale ................................................................................. 34 Characterizing habitat heterogeneity ..................................................................................... 35

ix

Conservation implications ..................................................................................................... 36 Chapitre 2. Bot fly parasitism of the red-backed vole: host survival, infection risk, and population growth............................................................................................................................................ 51

Résumé ...................................................................................................................................... 52 Abstract...................................................................................................................................... 53 Introduction ............................................................................................................................... 54 Materials and methods............................................................................................................... 58

Study area and sampling design ............................................................................................ 58 Sampling of red-backed voles and bot flies........................................................................... 58 Attributes of sampling site..................................................................................................... 59 Statistical analysis.................................................................................................................. 60

Results ....................................................................................................................................... 62 Bot fly infection and red-backed vole’s survival and reproduction ...................................... 63 Distance traveled by recaptured red-backed voles according to infection status .................. 63 Link between infection risk, habitat attributes, life-history traits, and host abundance ........ 64 Bot fly prevalence and growth rate of red-backed vole populations over the summer ......... 64

Discussion.................................................................................................................................. 65 Consequences of bot fly infection on red-backed vole survival and reproduction ............... 65 Abiotic factors influencing vole survival .............................................................................. 67 Infection patterns of bot flies in red-backed voles ................................................................ 68 Consequences of bot fly infection on growth rate of red-backed vole populations over the summer .................................................................................................................................. 69

Chapitre 3. Deer mice mediate red-backed vole behavior and abundance along a gradient of habitat alteration ............................................................................................................................ 84

Résumé ...................................................................................................................................... 85 Abstract...................................................................................................................................... 86 Introduction ............................................................................................................................... 87 Material and methods ................................................................................................................ 89

Study area and experimental design ...................................................................................... 89 Giving-up densities................................................................................................................ 90 Small-mammal abundance .................................................................................................... 90 Habitat variables .................................................................................................................... 91 Statistical analyses ................................................................................................................. 92

Results ....................................................................................................................................... 93 Habitat principal components................................................................................................ 93 Mean GUD and ∆ GUD......................................................................................................... 94 Small-mammal abundance .................................................................................................... 95

Discussion.................................................................................................................................. 95 Competition ........................................................................................................................... 96 Predation costs....................................................................................................................... 96

Conclusion générale .................................................................................................................... 105 Contribution relative de la structure et de la composition de l’habitat.................................... 105 Potentiel régulateur des populations par un parasite ............................................................... 107 Réaction du campagnol à dos roux à l’aménagement forestier ............................................... 109 Implications pour la conservation............................................................................................ 110 Avenues de recherche.............................................................................................................. 112

Bibliographie ............................................................................................................................... 115

x

Liste des tableaux

Table 1. Proportion of point count stations occupied by bird species in a matrix of old-growth

boreal forest of the Côte-Nord region of Québec, Canada (n = 96 stations). Common names,

scientific names, and acronyms are provided. ....................................................................... 38

Table 2. Area under the ROC curve (AUC) reflecting the performance of multiple logistic

regressions evaluating the probability of occurrence of 25 boreal bird species as a function

of habitat composition variables, habitat structure variables, and spatial variables evaluated

at five spatial scales, i.e., r = 100 m, 250 m, 500 m, 750 m, and 1000 m. A higher AUC

indicates a stronger link between the probability of occurrence for that species and habitat

attributes. AUC values < 0.70 indicate poor model performance. ........................................ 40

Table 3. Models of the probability of occurrence of bird species as a function of habitat structure

and composition, as well as spatial attributes. Regression coefficient (β) is provided in the

first row for each species and percentage of variance explained by environmental variables

is given in parentheses in the second row. The sum (∑) of variance explained by habitat

structure, habitat composition, and spatial attributes is also presented in the second row for

each species. Here, only the best model of five spatial scales (r = 100 m, 250 m, 500 m, 750

m, and 1000 m), i.e., the model with the highest AUC value (see Table 2 for AUC values as

a function of spatial scale), is presented for each of the 20 species that responded to habitat

or spatial attributes. Five species had AUC < 0.7 at each of the five spatial scales, indicating

a poor model fit, and their models are therefore not presented. ............................................ 42

Table 4. Parameter estimates, standard errors, and p-values of factors contained in all models

with ∆BIC < 2 for (1) the probability of survival in live traps of 341 red-backed voles

captured at 36 sites in the boreal forest of the Côte-Nord region of Québec, Canada; (2) the

probability of bot fly infection for red-backed voles; (3) the abundance of bot fly larvae in

50 dead red-backed voles; and (4) the growth of 36 red-backed vole populations between

July and August. Model selections 1 to 3 were based on GLMM with sites as a random

effect and model selection 4 was based on GLM. Parameter estimates of the top-ranking

models are presented, along with model fit (AUC, R2 or pseudo-R2) and Akaike weight (wi).

Complete model selections are presented in Appendix 1...................................................... 70

xi

Table 5. Loadings and proportion of explained variance of the PCA conducted on 12 habitat

variables measured in 29 pairs of adjacent harvested and natural habitats (n = 58 habitats) in

the boreal forest of eastern Quebec. Stars indicate the level of significance of each loading

according to a bootstrap validation........................................................................................ 98

Table 6. Multivariable GLMM of mean GUD, ∆GUD (i.e. GUD in open minus GUD in covered

food patch), and red-backed vole abundance as a function of habitat alteration (PC2),

moisture conditions (PC1), and the presence of deer mice in pairs, a dummy variable having

no deer mouse as the reference value. Pseudo-R2 was 0.56 for vole abundance, 0.58 for

mean GUD and 0.08 for ∆GUD. ......................................................................................... 100

xii

Liste des figures

Figure 1. Organigramme de la méthode scientifique utilisée pour comprendre la répartition d’une

espèce donnée. Tiré de Krebs (1985, p. 39). ........................................................................... 4

Figure 2. Point count station locations and (inset) general location of the study area in Québec,

Canada. .................................................................................................................................. 46

Figure 3. Venn diagram summarizing the variance partitioning analysis of the CCA of the species

data matrix as a function of the three structural attributes, the three compositional attributes,

and the three spatial attributes. The proportion of the variance in species assemblage

explained by all the nine variables was 14.8%. The diagram shows how this 14.8% of

variance is partitioned among independent and joint effects of habitat composition, habitat

structure, and spatial attributes. The sum of the percentages presented in the diagram is

100%...................................................................................................................................... 47

Figure 4. Ordination plot of a canonical correspondence analysis (CCA) conducted on 25 bird

species (acronyms are defined in Table 1) as a function of habitat structure, composition,

and spatial attributes of 96 sites located in a matrix of old-growth boreal forest in the Côte-

Nord region of Québec, Canada. a) Position of the 96 sites (open circles) in the ordination

space; an arrow’s length and its angle represents the strength of the correlation between

environmental variables and CCA axes. DECID1 and DECID0 are the centroids of the

dummy variable “presence of deciduous stands in the landscape”. b) Position of the 25 bird

species in the ordination space. ............................................................................................. 49

Figure 5. Mean (± s.e.) proportion of variance explained by composition, structure, and spatial

attributes in models of the probability of occurrence for each of 20 bird species in a matrix

of old-growth boreal forest. Proportion of variance explained by each environmental

variable was calculated using hierarchical partitioning. a) Summed proportion of variance

explained by the three variables of habitat structure (proportion of dense, open, and sparse

habitats in the landscape), the three variables habitat composition (proportion of

mixedwood, coniferous, and deciduous in the landscape), and the three spatial attributes

(latitude, altitude, and longitude). b) Proportion of variance explained by each of the nine

environmental variables......................................................................................................... 50

xiii

Figure 6. Number of a. female and b. male red-backed voles (Myodes gapperi) that were captured

alive (white boxes) and dead (grey boxes) according to their infection status with bot fly

(Cuterebra spp.) parasites. The 341 individuals were sampled at 36 sites in the boreal forest

of the Côte-Nord region of Québec, Canada, from 12 to 22 July and from 12 to 22 August

2004. ...................................................................................................................................... 72

Figure 7. Probability of survival in live traps of red-backed voles as a function of bot fly

infection, host sex, and body condition. Open circles represent raw data and curves represent

the binomial fit according to model S1 (see Table 4 for estimates and standard errors and

Appendix 1 for an explanation of model S1). ....................................................................... 73

Figure 8. a) Probability of bot fly infections as a function of vole abundance (no per 100 trap

nights) in August based on low, mean or high latitude and b) abundance of bot flies per host.

Open circles represent raw data and the curve represent the fits according to model a. PI1

and b. A1 (see Table 4 for estimates and standard errors and Appendix 1 for an explanation

of model PI1 and A1). In panel b the inclusion or exclusion of the outlier, i.e. 13 larvae in a

host, yielded similar results. .................................................................................................. 74

Figure 9. Relationship between the abundance of red-backed voles (number per 100 trap nights)

in August and in July at 36 individuals sites sampled in the boreal forest of the Côte-Nord

region of Québec, Canada. The equation of the linear regression is: y = 17 .8 + 0.5 x, p =

0.002. ..................................................................................................................................... 75

Figure 10. Short-term growth rate of red-backed vole populations between July and August 2004.

Open circles represent raw data and lines represent the linear fit of the model PGR1 (see

Table 4 for estimates and standard errors). Tree and sapling basal areas were categorized as

low, i.e. first quartile where first quartile was 17.9 m2/ha for trees and 2.3 m2/ha for saplings

and high, i.e. third quartile where third quartile was 31.0 m2/ha for trees and 5.5 m2/ha for

saplings. ................................................................................................................................. 76

Figure 11. Location of the study area in the Côte-Nord region of Quebec, Canada. Black dots in

the top-right panel indicate habitat pair location (n = 29 pairs). ......................................... 101

Figure 12. PCA conducted on 12 habitat variables in 29 pairs of adjacent harvested and natural

forests (n = 58 habitats). Habitats are represented in the PCA space on the left panel with

black dots representing natural habitats and white dots representing harvested habitats. The

xiv

right panel presents loadings of the PCA with arrow’s direction and length representing the

strength of the correlation between each habitat variable and principal components. PC1

represents a moisture gradient, from xeric to mesic habitats, and PC2 represents a habitat

alteration gradient, from natural stands to stands harvested with high intensity................. 102

Figure 13. Interactions between red-backed voles and deer mice. (a) mean giving-up density

(GUD); (b) ∆GUD (i.e. GUD in open minus GUD in the paired covered food patch); and (c)

patterns of abundance in xeric and mesic habitats as a function of habitat alteration (PC2),

moisture conditions (PC1). Lines represent model fits (see Table 5 for estimates and p-

values). Low PC1, i.e. xeric habitats, refers to the 1st quartile of PC1 while high PC1, i.e.

mesic habitats, refers to the 3rd quartile of PC1................................................................... 104

1

Introduction générale

La biodiversité et l’humain

Le mot biodiversité est une contraction du terme diversité biologique (Wilson & Peter 1989). Elle

peut-être définie de plusieurs façons, dépendamment du point de vue environnemental, social, ou

économique que l’on adopte (Wilson & Peter 1989, Gaston & Spicer 2004). La définition de la

biodiversité qui est la plus fonctionnelle sur le plan scientifique est probablement la suivante : la

variété de structures et de fonctions dans les formes de vie, au niveau de la génétique, des

populations, des communautés et des écosystèmes (Sandlund et al. 1992).

Depuis des millénaires, à l’instar de la plupart des espèces occupant cette planète, nous tirons

profit de la biodiversité pour subvenir à nos besoins (chasse, pêche, cueillette). Cependant, la

similarité s’arrête ici, puisque les modifications que nous apportons à l’environnement ne se

comparent en rien à celles faites par les autres organismes vivants, et cette différence est encore

plus dramatique depuis l’avènement de l’ère industrielle. À la fin des années 90, nous avions déjà

séquestré 40 % de la productivité primaire des écosystèmes terrestres, et ce chiffre continue

d’augmenter (Vitousek et al. 1997, Tilman 2000).

À l’heure actuelle, encore trop peu de citoyens et de gouvernements réalisent pleinement

l’énormité de nos impacts sur l’environnement et notre dépendance à la biodiversité (Loreau et

al. 2001, Gaston 2003, Myers 2003, Sekercioglu et al. 2004). De ce fait, les budgets alloués aux

recherches sur la biodiversité sont encore bien faibles en comparaison de ceux alloués à d’autres

sciences, comme la médecine ou la biotechnologie, ralentissant d’autant plus notre acquisition de

connaissances sur la biodiversité et les mécanismes qui la régissent (Hubbell 2001, Gaston &

Spicer 2004). Pour contrer cette tendance, de nombreuses initiatives tentent de conscientiser les

citoyens à la conservation de la biodiversité en menant des campagnes de sensibilisation aux

valeurs environnementales, sociales et économiques de la biodiversité (Gaston & Spicer 2004,

Kremen 2005).

2

La valeur la plus essentielle de la biodiversité est sa valeur intrinsèque ; chaque élément de la

biodiversité est important, indépendamment de son utilité pour notre espèce (Soulé 1986, Wilson

& Peter 1989). Cette valeur intrinsèque trouve cependant moins de résonnance dans la population

que sa valeur utilitaire, qui assume que les millions d’espèces peuplant la planète sont là pour

servir principalement les objectifs économiques de notre seule espèce (Meadows 1990). On a

classifié la valeur utilitaire selon quatre types de ressources que la biodiversité fournit à

l’humain : agricoles, génétiques, médicales et industrielles (Myers 1983). Chacune de ces

ressources génère des profits annuels estimés à 40 milliards de dollars américains (Myers 1983).

À beaucoup d’égards, l’avancement des recherches sur la biodiversité se trouve à l’étape à

laquelle se trouvait la médecine au Moyen-âge : on dissèque les corps (écosystèmes) pour

identifier les organes (espèces) qui se trouvent à l’intérieur (Hubbell 2001). En effet, on a

seulement identifié une infime partie des organismes vivants occupants notre planète, avec un

fort débalancement taxonomique en faveur des vertébrés, et un fort débalancement géographique

en faveur de l’Europe et de l’Amérique du Nord (May 1988, Dobson et al. 2008). D’ailleurs,

l’identification des points chauds de biodiversité (hotspots), i.e. régions à la fois réservoirs de

biodiversité et également menacées de destruction, est un thème de recherche très actif en

biologie de la conservation (e.g., Bush et al. 2004, Hopper & Gioia 2004, Latimer et al. 2005).

À l’heure actuelle, on est encore bien loin de comprendre les mécanismes qui régissent les

interactions entre les êtres vivants et leur environnement, une compréhension pourtant essentielle

à la conservation de la biodiversité (Caro 2007, Morris et al. 2009). Ce manque de connaissance

est difficilement justifiable en regard de la menace grandissante à la biodiversité que présente le

développement de notre civilisation (Vitousek et al. 1997, Tilman 2000). Cette thèse s’inscrit

dans ce besoin urgent de mieux comprendre les mécanismes qui régissent la biodiversité dans les

milieux naturels et les milieux perturbés par l’humain.

Cadre de travail

Bien qu’essentielles à la conservation, les recherches descriptives visant à identifier les points

chauds de biodiversité ne permettent pas nécessairement de comprendre les mécanismes qui

3

régissent la biodiversité (Caro 2007, Morris et al. 2009). Ultimement, la compréhension de ces

mécanismes devrait permettre d’améliorer nos politiques de conservation et de réduire notre

empreinte écologique sur les écosystèmes de la planète (Caro 1999, Caro 2007). Selon Krebs

(1985), l’influence de ces mécanismes sur la répartition des espèces, et par extension sur les

patrons de biodiversité, peut être représentée d’une manière hiérarchique (Fig. 1). Dans la

présente section, je vais expliquer brièvement comment ces mécanismes peuvent influencer la

répartition des espèces et je vais présenter les grandes lignes de ma thèse. Dans les sections

subséquentes de l’introduction, je vais expliquer pourquoi je m’intéresse à chacun de ces

mécanismes. Je vais exposer l’état des connaissances, identifier les besoins de recherche,

présenter les objectifs de la thèse et caractériser l’aire d’étude. Finalement, je vais justifier le

choix des espèces étudiées.

La dispersion est le premier mécanisme qui peut limiter la répartition d’une espèce et les patrons

de biodiversité (Fig. 1). En effet, beaucoup d’espèces n’occupent pas la totalité de leur aire

potentielle de répartition alors qu’elles pourraient survivre et se reproduire ailleurs que dans les

territoires qu’elles occupent (Krebs 1985). Les mécanismes de dispersion ont été largement

étudiés par le passé (e.g., Desrochers & Hannon 1997, Gautestad & Mysterud 2005, Robinet et

al. 2008, Bost et al. 2009, Cronin 2009) et ont mené à des théories très influentes en écologie,

comme la théorie de la biogéographie des îles (MacArthur & Wilson 1967), la théorie des

métapopulations (Levins 1968) ou la théorie neutre de la biodiversité (Hubbell 2001). Dans mon

aire d’étude, une matrice de forêt relativement peu perturbée (Bouchard et al. 2008), la sélection

d’habitat (2e mécanisme, Fig. 1) et les interactions biologiques (3e mécanisme, Fig. 1) devraient

avoir une influence prépondérante sur la répartition des espèces animales puisque ce type de

matrice devrait favoriser davantage les processus de dispersion que les matrices d’habitat

perturbés (Bender et al. 2003, Westerberg & Wennergren 2003).

4

Figure 1. Organigramme de la méthode scientifique utilisée pour comprendre la répartition d’une

espèce donnée. Tiré de Krebs (1985, p. 39).

Dans le premier chapitre, j’ai étudié le comportement de sélection d’habitat des oiseaux nicheurs

dans une matrice naturelle de forêt boréale à structure irrégulière, située dans la région de la

Côte-Nord du Québec (2e mécanisme, Fig. 1). Le comportement de sélection d’habitat, comme

beaucoup de comportements, est le fruit de la sélection naturelle, qui pousse les individus à

adopter des comportements qui maximisent leur valeur adaptative (fitness) (Darwin 1859). Ainsi,

la répartition des organismes dans un territoire n’est pas uniforme ; la densité de population est

généralement plus élevée dans les habitats qui permettent de maximiser la valeur adaptative des

individus (e.g., Fretwell & Lucas 1970, Rosenzweig 1981, Abrams 2000, Morris 2003). Nos

connaissances des mécanismes qui influencent la répartition de la faune dans la forêt boréale à

structure irrégulière sont relativement limitées (e.g., Azeria et al. 2009b, Courbin et al. 2009,

Janssen et al. 2009). Or, il est nécessaire de comprendre rapidement ces mécanismes à l’état

naturel car, alors qu’une grande partie de cette forêt est vierge, les coupes forestières gagnent du

terrain (Bouchard et al. 2008). L’information recueillie devrait permettre d’établir une référence

écologique pour les patrons de diversité aviaire dans la forêt boréale de l’est du Québec.

5

Dans le second chapitre, j’ai profité d’un fort taux de parasitisme par l’œstre (Cuterebra spp,

Diptera : Cuterebridae) pour étudier l’influence de ce mécanisme sur la répartition du campagnol

à dos roux de Gapper (Myodes gapperi) en forêt boréale naturelle (3e mécanisme, Fig. 1). Le

campagnol à dos roux est l’espèce de micromammifère la plus abondante de la forêt boréale et sa

répartition a été largement étudiée (e.g., Pearce & Venier 2005b, Boonstra & Krebs 2006, Elias et

al. 2006, Morris & Mukherjee 2007a). Cependant, peu d’études ont tenté de comprendre l’impact

des parasites sur la répartition de cette espèce (Bowman 2000, Pearce & Venier 2005a). De façon

générale, le parasitisme pourrait jouer un rôle majeur dans la répartition des espèces (e.g.,

Anderson & May 1978, Hudson et al. 1992, Dobson et al. 2008) mais sa contribution a été

largement sous-évaluée par le passé (Møller 2005, Lively 2006), notamment car ses effets son

fréquemment masqués par la compétition ou la prédation (Hatcher et al. 2006).

Dans le troisième chapitre, j’ai étudié l’influence de la compétition et la prédation sur la

répartition du campagnol à dos roux dans les forêts aménagées (3e mécanisme, Fig. 1). Alors que

l’on a largement étudié les patrons de répartition des espèces en fonction de l’altération

anthropique de l’habitat (e.g., Sullivan & Boateng 1996, Drapeau et al. 2000, Guénette & Villard

2005, Preston & Harestad 2007) ainsi que les effets de la compétition et de la prédation sur la

répartition des espèces dans les milieux naturels (e.g., Stenseth et al. 1996, Hanski et al. 2001,

Gilg et al. 2003, Huitu et al. 2005, Boonstra & Krebs 2006), rares sont les études qui ont

explicitement tenté de comprendre le lien de causalité entre l’altération anthropique de l’habitat et

la répartition des espèces (Caro 1999, Caro 2007, Morris et al. 2009). Or, cette compréhension

permettrait de développer des outils de conservation efficaces.

Finalement, les facteurs physico-chimiques auraient également un rôle à jouer dans la répartition

des espèces (4e mécanisme, Fig. 1). À l’échelle de la planète, la température et l’humidité sont

deux des principaux facteurs qui limitent la répartition et l’assemblage des espèces (e.g., Hawkins

2001, Mittelbach et al. 2001, Worm & Duffy 2003, Bini et al. 2004, Currie et al. 2004). Dans

cette thèse, j’ai tenu compte de facteurs physiques comme la température (chapitre 2) ou encore

la latitude (chapitre 1, chapitre 2).

6

Sélection d’habitat et hétérogénéité de l’habitat

Comme on l’a vu, la sélection de l’habitat est un des principaux processus qui régit la répartition

des espèces et les patrons de biodiversité (Rosenzweig 1991, Pulliam 2000). C’est un processus

hiérarchique dans lequel un organisme sélectionne son domaine vital et ses microsites

d’alimentation, de nidification ou de protection contre les prédateurs (Johnson 1980, Boyce

2006). On s’accorde généralement sur le fait que l’hétérogénéité de l’habitat a une grande

influence sur la sélection de l’habitat (Hutchinson 1957, Hurlbert 2004, Tews et al. 2004) ainsi

que sur les interactions biologiques (e.g., Danielson 1991, Cronin 2009). Cependant, on s’entend

moins sur la définition même du terme (Tews et al. 2004). Dépendamment du groupe

taxonomique et de la résolution spatiale, l’hétérogénéité de l’habitat peut-être caractérisée par

l’architecture d’une espèce de plante (e.g., Lawton 1983) ou bien par les patrons de composition

et de configuration du paysage (e.g., Böhning-Gaese 1997). De plus, au sein d’un assemblage

d’espèces, par exemple les arthropodes, l’hétérogénéité de l’habitat peut référer à une échelle

relativement fine comme la « complexité des bordures » (Haslett 1997) ou à une échelle

continentale comme « l’hétérogénéité topographique » (Kerr & Packer 1997).

D’un point de vue faunique, l’hétérogénéité de l’habitat pourrait être définie comme la variation

spatio-temporelle de la structure et de la composition de l’habitat (Rosenzweig 1981, Abrams

2000). La structure réfère à l’arrangement spatial des différents attributs de l’écosystème (e.g.,

espacement des arbres, nombre de niveaux de canopée) et informerait les animaux quant à la

disponibilité de couvert contre les prédateurs et les intempéries (Robinson & Holmes 1984). Elle

influencerait également l’efficacité avec laquelle les animaux se déplacent à travers

l’environnement pour échapper aux prédateurs et pour s’alimenter (Robinson & Holmes 1984).

La composition de l’habitat, quant à elle, réfère à la diversité des espèces végétales et elle

influencerait directement et indirectement l’abondance et le type de ressources alimentaires

disponibles (Holmes & Robinson 1981).

Selon la théorie des niches écologiques, l’hétérogénéité de l’habitat devrait augmenter la diversité

en espèces (Simpson 1949, Hutchinson 1957, MacArthur & Wilson 1967, Lack 1969).

Cependant, on explique encore mal la contribution relative de la structure et de la composition de

l’habitat sur la diversité. Plusieurs études ont trouvé que la composition avait significativement

7

plus d’influence que la structure de l’habitat sur la diversité (e.g., Rotenberry 1985, Lopez &

Moro 1997, Fleishman et al. 2003), mais au moins autant ont trouvé la relation opposée (e.g.,

Wiens & Rotenberry 1981, Wigley & Roberts 1997, Law & Chidel 2002). Enfin, d’autres études

ont noté un effet similaire de la structure et de la composition de l’habitat sur la diversité (e.g.,

Mac Nally 1990, Bersier & Meyer 1994).

Contribution relative de la structure et de la composition de l’habitat

Cette diversité de résultats pourrait être la conséquence de la variation inter-études des indices

utilisés pour caractériser la structure et la composition de l’habitat (revues dans Tews et al. 2004,

McElhinny et al. 2005) ou de la variation de réponse entre les groupes fonctionnels ou

taxonomiques étudiés (e.g., Janssen et al. 2009). Aussi, ces indices sont souvent

mathématiquement complexes et McElhinny et al. (2005) proposent d’utiliser des indices de

structure ou de composition plus simples pour obtenir des résultats plus cohérents entre les

études. Finalement, plusieurs études utilisent des indices de structure qui sont non indépendants

des indices de composition, ce qui pourrait biaiser les résultats (revues dans Tews et al. 2004,

McElhinny et al. 2005). En effet, plusieurs éléments de composition structurent également

l’habitat. Par exemple, la présence conjointe d’espèces tolérantes et intolérantes à l’ombre devrait

résulter dans l’établissement d’un peuplement à plusieurs niveaux de canopée (e.g., Spies &

Franklin 1991).

La variété des résultats obtenus dans les études précédentes pourrait également être une

conséquence de l’échelle spatiale (Wiens 1989, Kotliar & Wiens 1990, Levin 1992, Cushman &

McGarigal 2004). Certains avancent que la structure serait le principal facteur influençant la

répartition des espèces à échelle étendue alors que la composition de l’habitat serait le facteur le

plus déterminant à échelle fine (Wiens & Rotenberry 1981, Bersier & Meyer 1994, Deppe &

Rotenberry 2008). Cependant, ce patron est incertain car les indices utilisés pour caractériser la

structure et la composition de l’habitat diffèrent en fonction de l’échelle spatiale. En effet,

l’hétérogénéité de l’habitat est souvent caractérisée par une collecte de données sur le terrain à

échelle fine alors qu’elle est caractérisée par des études cartographiques à partir de systèmes

d’information géographique (SIG) à échelle étendue (e.g., Grand & Cushman 2003, Deppe &

Rotenberry 2008). Ainsi, l’utilisation des mêmes indices de composition et de structure pour

8

chaque échelle spatiale, i.e. variation de l’étendue de l’échelle spatiale en conservant le grain

constant (sensu Wiens 1989), devrait améliorer notre compréhension du rôle de l’échelle spatiale

dans la contribution relative de la structure et de la composition sur la répartition et l’assemblage

des espèces.

Répartition et assemblage des espèces

Une des principales méthodes d’analyse des assemblages d’espèces est le calcul de la diversité

alpha de l’assemblage à l’aide d’indices comme la richesse spécifique, la diversité de Simpson

(1949) ou encore la diversité de Shannon (Shannon & Waever 1949). Ces indices sont des outils

intéressants qui permettent de quantifier la diversité mais ils ne permettent pas capturer la

variabilité des réponses entre les espèces à l’intérieur des assemblages (Hurlbert 1971, Gotelli &

Colwell 2001). Par exemple, deux habitats peuvent avoir une même richesse mais une

composition en espèces très différente (Jonsson & Jonsell 1999, Ebenman et al. 2004, Mouillot et

al. 2005).

Ainsi, il pourrait être avantageux d’étudier simultanément la réponse individuelle des espèces

ainsi que leur assemblage, plutôt que de se fier uniquement aux indices de diversité (Grand &

Cushman 2003, Preston & Harestad 2007). Ces deux approches sont complémentaires car l’étude

de la répartition d’une espèce donnée informe davantage sur le processus de sélection d’habitat,

mais elle ne tient pas directement compte du contexte d’interaction de l’assemblage (Wiens 1989,

Prendergast et al. 1993, Johnson 2007). En effet, les espèces interagissent dans l’assemblage et

certaines espèces peuvent présenter des patrons de cooccurrence ou d’évitement (Azeria et al.

2009b, Courbin et al. 2009). De son côté, l’étude unique des assemblages fournit une information

générale sur l’assemblage des espèces en fonction des variables de l’environnement mais manque

souvent de précision sur les patrons de répartition spécifiques de chaque espèce en fonction des

variables de l’environnement (Grand & Cushman 2003, Preston & Harestad 2007).

Dans le chapitre 1, j’ai déterminé la contribution relative de la structure et de la composition de

l’habitat sur la répartition des espèces et sur l’assemblage des oiseaux nicheurs dans une matrice

de vieille forêt boréale et ce, à plusieurs échelles spatiales. Cette étude devrait notamment

9

contribuer à l’établissement d’une référence écologique pour les forêts boréales de l’est du

Québec.

Référence écologique

La majorité des études qui ont tenté de comprendre l’influence relative de la structure et de la

composition de l’habitat ont été réalisées dans des matrices d’habitats fortement altérées par les

perturbations anthropiques (revue dans Tews et al. 2004). Ces résultats pourraient donc ne pas

représenter l’influence de l’hétérogénéité naturelle de l’habitat sur la répartition des espèces. Il

serait pertinent d’étudier ces patrons dans des matrices d’habitat peu perturbées, qui pourraient

par ailleurs servir d’écosystème de référence (Schmiegelow & Mönkkönen 2002, Parrish et al.

2003, Hawkins 2006, Gibbons et al. 2008). En théorie, un écosystème de référence est un

écosystème dans lequel les patrons de biodiversité ne sont pas affectés par les perturbations

anthropiques. En pratique, tous les écosystèmes terrestres sont influencés par l’humain, à des

degrés plus ou moins importants, et les références écologiques sont plutôt obtenues à travers

l’étude des écosystèmes les moins perturbés d’une région donnée (Vitousek et al. 1997). Notre

aire d’étude présente ce potentiel puisque plus de la moitié de la superficie est encore vierge

(Bouchard et al. 2008). L’étude combinée de l’influence de la sélection de l’habitat et des

interactions biologiques sur la répartition des espèces devrait permettre d’améliorer notre

compréhension des mécanismes qui régissent la biodiversité et d’établir ainsi une référence

écologique solide.

Interactions biologiques : parasitisme

Comme la sélection d’habitat, les interactions biologiques jouent un rôle majeur dans la

répartition des organismes et les patrons de biodiversité (voir Fig. 1). Le parasitisme est une

interaction biologique dans laquelle une espèce parasite tire profit d’une espèce hôte en vivant

dans ou sur son corps (Anderson & May 1978). Un parasite peut passer la majeure partie de son

cycle vital en association directe avec l’hôte ou seulement quelques phases, étant le reste du

temps indépendant de l’hôte pour survivre (Anderson & May 1978). Durant les phases actives du

parasitisme, le parasite dépend étroitement de l’hôte pour acquérir l’énergie et les nutriments

10

essentiels à son métabolisme (Anderson & May 1978). Pour le parasite, l’interaction est donc

obligatoire, alors qu’elle est néfaste pour l’hôte. Cependant, même si le parasite réduit la valeur

adaptative de son hôte, il n’a pas d’avantage évolutif à le tuer puisque sa survie dépend de celle

de l’hôte (Lively 2006). En revanche, un parasitoïde est une espèce qui nécessite la mort d’un

hôte pour maximiser sa valeur adaptative (May et al. 1981). C’est le cas de plusieurs insectes qui

pondent leurs œufs à l’intérieur d’autres insectes. La larve du parasitoïde consomme l’hôte petit à

petit durant sa croissance (May et al. 1981).

Des études récentes suggèrent qu’environ 75 % des liens dans les réseaux trophiques impliquent

des parasites (revue dans Dobson et al. 2008). Malgré son importance, la contribution du

parasitisme sur la dynamique des populations a été relativement peu étudiée (voir revue dans

Møller 2005, Lively 2006). Une des principales raisons est la difficulté logistique d’étudier les

effets des parasites puisque leur suivi requiert des méthodes invasives, dont le sacrifice de l’hôte

(Albon et al. 2002). De plus, l’étude du parasitisme est compliquée par le fait que ses effets

pourraient être masqués par les effets de la prédation ou de la compétition (Hatcher et al. 2006).

En effet, la prédation ou la compétition peuvent être identifiées comme causes ultimes de la

mortalité d’un individu alors que la cause initiale serait un affaiblissement de l’hôte par le

parasite ou encore une réduction de l’appréhension de l’hôte (Hatcher et al. 2006). Malgré ces

difficultés techniques, de plus en plus de recherches semblent indiquer que les parasites auraient

le potentiel de réguler les populations d’hôtes (e.g. Witting 2000, Hanski et al. 2001, Turchin &

Hanski 2001, Eccard & Ylonen 2002, 2003, Gilg et al. 2006).

Potentiel régulateur du parasitisme

En pratique, pour qu’un mécanisme soit considéré comme régulateur des populations, il doit

répondre à des critères précis (Anderson & May 1978, May & Anderson 1978). Notamment, ce

mécanisme doit influencer la survie ou la reproduction d’une manière qui dépend de la densité

des individus (density-dependent), c’est-à-dire que le taux de mortalité ou de natalité causé par le

mécanisme varie selon la densité de la population. Le mécanisme doit également affecter une

grande portion de la population, incluant les individus ayant un fort potentiel à contribuer à

l’effort reproducteur de l’ensemble de a population (Anderson & May 1978, May & Anderson

1978).

11

Dans le chapitre 2, j’ai profité d’un taux de parasitisme élevé chez le campagnol à dos roux en

forêt boréale naturelle pour déterminer si les œstres avaient le potentiel de réguler les populations

de leurs hôtes. L’impact des œstres sur la survie des hôtes demeure ambigu à cause

d’observations conflictuelles. Par exemple, les œstres semblent réduire la survie du Campagnol

de Townsend (Microtus townsendii) (Boonstra et al. 1980), alors qu’ils augmenteraient la survie

des souris (Peromyscus spp) (Hunter et al. 1972, Clark & Kaufman 1990, Munger & Karasov

1991).

Le bénéfice apparent du parasite sur la survie des hôtes pourrait cependant être un artefact

résultant de la différence dans les probabilités de capture entre les micromammifères résidents et

ceux en transit (Wecker 1962, Hunter et al. 1972, mais voir Burns et al. 2005). Les hôtes

résidents auraient une probabilité plus élevée d’être infectés que les hôtes en transit, car les

femelles d’œstres gravides pondent leurs œufs sur la végétation proche des terriers, qui sont

principalement utilisés par les résidents (Catts 1982, Slansky 2007). Ainsi, la probabilité

d’infection plus élevée pour les résidents, combinée à leur plus haute probabilité de recapture que

les hôtes en transit, pourrait conduire à la conclusion erronée que les hôtes infectés ont une

probabilité de survie plus élevée (Wecker 1962, Hunter et al. 1972).

Pour clarifier la relation directe entre l’infection des œstres et la survie de l’hôte, il serait donc

préférable d’utiliser une mesure de la survie de l’hôte qui n’est pas reliée à la probabilité de

recapture et de contrôler les effets de la prédation et de la compétition directe (Wecker 1962,

Hunter et al. 1972). Une méthode appropriée pourrait être d’étudier la probabilité de survie

durant la nuit à l’intérieur des pièges à capture vivante. Dans cette méthode, la probabilité de

survie serait effectivement indépendante de la probabilité de recapture. En plus, les individus

capturés seraient également à l’abri de la prédation et de la compétition par interférence. La

réponse des micromammifères au confinement dans les pièges devraient correspondre à leur

réponse à des événements naturels en situation de stress élevé. En effet, le confinement dans les

pièges produit une réponse physiologique similaire à la réponse physiologique induite par les

risques naturels comme la prédation (i.e., une augmentation du niveau de corticostérone) chez le

campagnol des champs (Microtus pennsylvanicus) (Harper & Austad 2001, Fletcher & Boonstra

2006).

12

Finalement, l’étude des effets des œstres sur la survie et la reproduction de leurs hôtes est

possible du fait que les œstres sont parmi les rares endoparasites (i.e., parasites internes) que l’on

peu détecter par un examen externe de l’hôte. En effet, lors de son troisième stade de croissance,

la larve perce un pore respiratoire à travers la peau de l’hôte qui peut être facilement détecté par

un simple examen externe du rongeur.

Cycle de vie de l’œstre

Le cycle de vie de l’œstre est composé d’un stade d’œuf, de trois stades larvaires, d’un stade de

pupation et d’un stade adulte. Durant la phase adulte, l’œstre est un insecte volant dont la survie

est indépendante de celle de l’hôte (Catts 1982). La femelle gravide de l’œstre dépose ses œufs

dans un microsite propice à leur éclosion, généralement sur la végétation proche de l’entrée des

terriers de micromammifères (Slansky 2007). Les œufs sont enduits d’une substance adhésive qui

leur permet de coller aux poils d’un micromammifère dès le premier contact. Le contact avec la

chaleur du corps de l’hôte permet l’éclosion de l’œuf en premier stade larvaire (Catts 1982). La

larve, longue d’un ou deux millimètres, pénètre l’hôte par voie nasale ou orale (Catts 1982).

Durant deux semaines, elle migre jusqu’à son site de développement, la région inguinale, en

suivant les voies naturelles ou en ou perforant les organes (Gingrich 1981). La larve se développe

alors en stade 2 et 3 et perce son pore respiratoire. Finalement, trois semaines après l’éclosion, la

larve, qui mesure environ 1.5 cm, sort du corps de l’hôte par le pore respiratoire, laissant un trou

béant qui se refermera après quelques semaines (Gingrich 1981). La larve s’enfouit dans le sol où

elle fera sa pupe et passera l’hiver. L’été suivant, l’adulte émergera (Catts 1982).

Ce volet de ma thèse devrait permettre de mieux comprendre l’impact du parasitisme, une

interaction relativement peu étudiée, sur la dynamique des populations d’un micromammifère en

milieu naturel. Les deux autres interactions à l’étude, soient la compétition et la prédation (Fig.

1), ont été beaucoup plus étudiées que le parasitisme dans les milieux naturels (Møller 2005).

Cependant, on comprend encore mal comment l’altération de l’habitat modifie la compétition et

la prédation, influençant par ricochet la répartition des individus et la dynamique des populations

(Caro 1999, Caro 2007).

13

Interactions biologiques : prédation et compétition

Comme le parasitisme, la compétition et la prédation ont le potentiel d’affecter la répartition des

individus et la dynamique des populations (e.g., Stenseth et al. 1996, Hanski et al. 2001, Gilg et

al. 2003, Huitu et al. 2005, Boonstra & Krebs 2006, Fig. 1). Alors que les interactions

compétitives se produisent à l’intérieur d’un même niveau trophique, les interactions de prédation

prennent place lorsqu’une espèce d’un niveau trophique supérieur se nourrit d’une espèce d’un

niveau trophique inférieur (Holling 1959, Lima & Dill 1990).

On distingue deux principaux types de compétition, la compétition par exploitation et la

compétition par interférence. Dans la compétition par exploitation, des individus d’une même

espèce, ou de plusieurs espèces, utilisent une ressource commune ayant une disponibilité limitée

(Tilman 1990, Holt et al. 1994). Dans la compétition par interférence, des individus en quête de

nourriture agressent d’autres individus pour obtenir ou protéger leurs ressources alimentaires. Les

individus les moins dominants sont souvent repoussés en dehors des habitats les plus propices

(Eccard & Ylonen 2002, Schmidt et al. 2005). La prédation, quant à elle, affecte la répartition des

proies qui se distribuent généralement de manière à limiter leur risque de prédation, par exemple

en s’approvisionnement davantage dans les habitats couverts (Brown & Kotler 2004, Ripple &

Beschta 2004, Morris 2005).

Comme on l’a vu, peu d’études ont cherché à comprendre les effets de l’altération de l’habitat sur

la compétition et la prédation (Kotler et al. 2007, Morris et al. 2009). Pourtant, cette information

est cruciale pour comprendre et réduire notre empreinte écologique puisque l’altération de

l’habitat est l’un des plus importants impacts de notre espèce sur l’environnement et la

biodiversité (Vitousek et al. 1997, Gaston et al. 2003).

Altération de l’habitat

L’exploitation forestière, l’agriculture et l’urbanisation sont les trois principales formes

d’altération de l’habitat (Vitousek et al. 1997). L’exploitation forestière se distingue des deux

autres en ce sens que les perturbations qu’elle entraîne sur les écosystèmes sont plus difficiles à

cerner (McGarigal & McComb, 1995). Pour les micromammifères forestiers par exemple, les

14

paysages agricoles et urbains peuvent généralement être caractérisés par des études de contraste

simple à partir de matrices de paysage binaires, c’est-à-dire l’étude de zones boisées dans une

matrice perturbée. En revanche, l’études de ces micromammifères dans les paysages forestiers

nécessitent généralement des études de contrastes multiples à partir de matrices forestières

complexes, présentant un gradient de perturbations allant d’assemblages de peuplements sous

l’effet des seules perturbations naturelles (cycles de feux, chablis ou épidémies d’insectes

défoliateurs) jusqu’aux zones de production intensive de matière ligneuse que sont les plantations

(Boulet et al. 2000, Harvey et al. 2006).

À l’heure actuelle, on est capable de prédire dans une certaine mesure la répartition et

l’abondance des espèces en fonction de l’altération de l’habitat (e.g., Sullivan & Boateng 1996,

Drapeau et al. 2000, Guénette & Villard 2005, Preston & Harestad 2007). Cependant, alors que

l’abondance permet généralement de nous renseigner sur la qualité intrinsèque de l’habitat

(Fretwell & Lucas 1970, Morris 1987) (mais voir Van Horne 1983), elle ne permet pas

nécessairement de comprendre les causes sous-jacentes qui expliquent la réponse animale à

l’altération de l’habitat (McCollin 1998, Andruskiw et al. 2008). Par exemple, l’intensité de la

compétition pourrait être affectée directement par les changements dans la disponibilité des

ressources alimentaires ou indirectement par la colonisation des espèces compétitrices, alors que

la prédation pourrait varier à travers l’altération du couvert anti-prédateur ou l’abondance de

prédateurs (Caro 2007).

Comportement et conservation

L’étude du comportement des animaux devrait permettre d’améliorer notre compréhension des

liens de causalité entre l’altération de l’habitat et la répartition des organismes (Caro 2007,

Morris et al. 2009). L’avantage de l’étude du comportement en comparaison de l’étude de la

dynamique des populations est que les organismes répondent rapidement aux modifications de

leur environnement en changeant leur comportement de façon à maximiser leur valeur adaptative

compte tenu des nouvelles conditions (Morris 1999). Au contraire, les changements

démographiques suivant l’altération de l’habitat se produisent généralement après un certain délai

(Brooks et al. 1999). Ainsi, l’étude du comportement d’approvisionnement peut par exemple

nous renseigner quant aux impacts de l’altération de l’habitat sur la compétition ou le risque

15

prédation (e.g., Morris & Davidson 2000, Reed et al. 2005, Morris & Mukherjee 2007b,

Andruskiw et al. 2008). Ultimement, ces effets devraient se faire sentir sur la dynamique des

populations (Morris et al. 2009). L’étude du comportement permettrait en quelques sortes de

prédire la dynamique future des populations en réponse à l’altération de l’habitat (Morris et al.

2009).

Plusieurs études traitant du comportement d’approvisionnement se basent sur le théorème de la

valeur marginale (Charnov 1976), qui stipule qu’un individu qui s’approvisionne dans une

parcelle de nourriture devrait exploiter cette parcelle jusqu’à ce que les bénéfices énergétiques

gagnés par l’acquisition de nourriture égalisent les coûts d’approvisionnement accumulés dans la

parcelle. Les coûts d’approvisionnement sont la somme des coûts métaboliques

d’approvisionnement (C), des coûts de prédation (P) et des coûts des opportunités manquées

(COM) de ne pas s’engager dans des activités alternatives qui maximiseraient la valeur

adaptative, comme la recherche de partenaire sexuel ou l’alimentation dans des parcelles plus

profitables (Brown 1988).

La valeur marginale peut être estimée en mesurant la densité de nourriture laissée à l’abandon

(giving-up density, GUD) d’une parcelle dans laquelle la quantité initiale de nourriture est connue

(Brown 1988, Brown et al. 1992). En utilisant un design approprié, les différences locales entre

C, P et COM peuvent être déterminées à travers les différences dans les GUDs. Par exemple, on

peut estimer les coûts de prédation associés à l’approvisionnement en mesurant la différence de

GUD dans une paire expérimentale de parcelles de nourriture dont l’une est localisée sous un

épais couvert anti-prédateur et l’autre est placée en milieu plus ouvert (Brown et al. 1992, voir

applications dans Morris & Davidson 2000, Schmidt et al. 2005, Andruskiw et al. 2008).

L’approche des GUD m’a permis de tester des hypothèses spécifiques quant aux effets de

l’altération de l’habitat sur les coûts de compétition et de prédation reliés à l’approvisionnement

et ainsi d’améliorer notre compréhension de liens de causalité entre l’altération de l’habitat et la

répartition des espèces.

16

Hypothèses et prédictions

Dans le chapitre 3, j’ai émis deux hypothèses concernant les effets de l’altération de l’habitat sur

les coûts de compétition et de prédation reliés à l’approvisionnement du campagnol à dos roux,

en étudiant la réponse de l’espèce à un vaste gradient d’intensité de récolte ligneuse des

peuplements forestiers. La réponse numérique du campagnol à dos roux à l’altération de l’habitat

a été largement étudiée, en partie car il est l’espèce la plus abondante de l’assemblage de

micromammifères dans les forêts boréales de l’Amérique du Nord (Pearce & Venier 2005b,

Macdonald et al. 2006, Lemaître et al. 2009). Les conclusions à propos des impacts de

l’exploitation forestière sur les populations de campagnols à dos roux sont par contre peu

cohérentes. Plusieurs études ont documenté un déclin dans l’abondance suivant la coupe

forestière (e.g., Mills 1995, Sullivan et al. 1999, Darveau et al. 2001, Moses & Boutin 2001),

alors que d’autres ont détecté une relation opposée (e.g., Hayward et al. 1999, Suzuki & Hayes

2003, Homyack et al. 2005). L’étude du comportement d’approvisionnement de l’espèce pourrait

possiblement expliquer le manque de cohérence entre les études précédentes.

Le campagnol à dos roux est une espèce omnivore qui s’alimente largement sur les champignons

hypogés (Orrock & Pagels 2002), que l’on retrouve plus abondamment en forêt mature et vieille

(Johnson 1996). Ainsi, l’exploitation forestière devrait réduire l’abondance de sa principale

ressource alimentaire. De plus, la principale espèce compétitrice du campagnol à dos roux, la

souris sylvestre (Peromyscus maniculatus), est également omnivore (Morris 1996) et

s’approvisionne de façon très efficace, particulièrement en forêt perturbée (Suzuki & Hayes

2003, Fuller et al. 2004). L’exploitation forestière devrait donc augmenter la compétition entre

les campagnols et les souris. J’ai émis l’hypothèse que la qualité générale de l’habitat devrait

diminuer avec l’exploitation forestière pour le campagnol à dos roux. La prédiction est que le

taux d’ingestion de nourriture auquel un campagnol cesse d’exploiter une parcelle de nourriture

(quitting-harvest rate), estimé par la GUD moyenne d’un habitat, devrait diminuer avec

l’exploitation forestière. En d’autres mots, le campagnol devrait utiliser de façon plus intensive

chacune des parcelles de nourriture à mesure que l’intensité de l’exploitation forestière augmente.

La théorie de l’approvisionnement prédit que les coûts de prédation devraient augmenter avec la

réduction du couvert anti-prédateur (Verdolin 2006, While & McArthur 2006, Eccard et al.

17

2008), une réponse qui devrait cependant dépendre de la qualité de l’habitat. En effet, le principe

de protection des acquis de Clark (1994, voir Ydenberg et al. 1995, et Reed et al. 2005 pour

support empirique) stipule que les individus qui s’approvisionnent dans les habitats de haute

qualité devraient exhiber un fort comportement anti-prédateur car il serait plus coûteux pour ces

individus de se faire blesser ou tuer, en terme de valeur adaptative, en comparaison aux individus

s’approvisionnant dans les habitats de plus faible qualité. En effet, il devrait être difficile pour un

individu s’approvisionnant dans un habitat de faible qualité de combler ses besoins énergétiques.

De ce fait, cet individu pourrait être poussé à prendre plus de risque (i.e., en s’approvisionnant

dans les microsites ayant un risque de prédation plus élevé) pour combler ses besoins

énergétiques. Il aurait peu à perdre s’il était blessé ou tué car, de toute façon, il ne rencontrait pas

les besoins énergétiques de base qui permettent d’assurer sa survie. J’ai émis l’hypothèse que les

coûts de prédation devraient décroître plus dramatiquement à la suite de perturbations dans les

habitats de haute qualité (fournissant une valeur adaptative élevée) qu’à la suite de perturbations

dans les habitats de faible qualité. Enfin, la plus grande différence dans les GUD entre les

habitats devrait être plus reliée à la disponibilité de nourriture qu’au risque de prédation (Olsson

& Holmgren 1999, Olsson & Molokwu 2007). J’ai testé ces hypothèses dans des paires d’habitat

comprenant des coupes forestières adjacentes à des peuplements naturels représentatifs de la forêt

boréale irrégulière de l’est du Québec.

Aire d’étude : forêt boréale de l’est du Québec

La forêt boréale constitue 25 % de la superficie forestière résiduelle mondiale (Ressources

naturelles Canada 2009) et assure de ce fait un rôle majeur dans les processus planétaires qui

impliquent les forêts, comme la photosynthèse ou les cycles des nutriments (Begon et al. 2006).

Le Canada est un acteur clé de la conservation de la forêt boréale puisque la forêt boréale est le

propre des régions nordiques (Canada, Russie, Scandinavie, ou encore Alaska). Au Canada, la

forêt boréale occupe 35 % de la superficie du pays et représente 77 % des forêts (Ressources

naturelles Canada 2009). Plusieurs activités anthropiques menacent la forêt boréale dont

l’exploitation intensive de la matière ligneuse pour nos besoins en bois d’œuvre, en pâte et

papiers et en produits dérivés, ainsi que les transformations permanentes et l’activité humaine

18

liées à l’hydroélectricité, l’exploitation minière et l’exploitation pétrolière (Ressources naturelles

Canada 2009).

L’aire d’étude est localisée dans la portion Est de la forêt boréale canadienne, plus précisément

dans la région de la Côte-Nord du Québec. L’aire d’étude présente deux particularités. D’abord,

son historique de coupe est plus récent que bien d’autres forêts boréales. Ainsi, environ la moitié

de l’aire d’étude n’a pas encore été influencée par l’exploitation forestière (Bouchard et al. 2008).

À titre de comparaison, l’exploitation forestière intensive a débuté autours des années 1940 dans

la région du Lac Duparquet, en Abitibi-Témiscaminque (Drapeau et al. 2000). Cette différence

dans l’historique des coupes s’explique par le fait que les forêts localisées les plus proches des

centres urbains sont généralement celles qui sont exploitées en premier lieu (Imbeau et al. 2001).

En effet, mon aire d’étude est située entre 150 à 350 km au Nord de la plus proche ville, Baie-

Comeau. Ainsi, l’aire d’étude présente un fort potentiel de référence écologique pour la forêt

boréale de l’est du Canada ou à tout le moins, pour la forêt boréale de l’est du Québec.

De plus, l’aire d’étude subit l’influence d’un climat maritime, à cause de sa proximité avec

l’Océan Atlantique. Les précipitations annuelles se situent entre 1000 et 1400 mm alors que les

moyennes des températures annuelles sont entre -2.5 à 0.0 °C (Boucher et al. 2003). Les

précipitations abondantes ont pour effet d’allonger le cycle des feux, laissant davantage de place

aux perturbations secondaires (chablis, épidémies d’insectes) ainsi qu’à la mortalité naturelle des

arbres (Pham et al. 2004). Ces mécanismes favorisent le développement de peuplements multi-

cohortes (peuplements à structure irrégulière) par une dynamique de trouées où la mortalité d’un

ou plusieurs individus crée une ouverture dans la canopée, qui sera subséquemment occupée par

la régénération (McCarthy 2001). En revanche, ailleurs en forêt boréale, le climat est

généralement plus sec et le feu est le principal agent de perturbation (Pham et al. 2004). Les

peuplements sont alors des peuplements à structure régulière, les arbres étant généralement issus

de la même cohorte (Pham et al. 2004). La moyenne d’âge des peuplements de notre aire d’étude

est de plus de 270 ans (Bouchard et al. 2008). Les peuplements sont dominés par l’épinette noire

(Picea mariana) et le sapin baumier (Abies balsamea). L’épinette blanche (Picea glauca), le

bouleau blanc (Betula papyrifera), le peuplier faux-tremble (Populus tremuloides) et le pin gris

(Pinus banksiana) sont aussi abondants localement.

19

Objectifs et organisation de la thèse

L’objectif général de ma thèse était de caractériser les interactions entre les espèces et

l’hétérogénéité de l’habitat ainsi que de mieux comprendre trois des principales interactions intra-

et interspécifiques (i.e., parasitisme, compétition et prédation) en forêt boréale naturelle et

perturbée par l’humain. Pour rencontrer cet objectif, j’ai considéré trois niveaux hiérarchiques :

assemblage des espèces, dynamique des populations et comportement des individus. Plus

spécifiquement, j’ai adressé les questions suivantes :

Chapitre 1

Quelle est la contribution relative de la structure et de la composition de l’habitat, à plusieurs

échelles spatiales, sur la répartition et l’assemblage des espèces d’oiseaux nicheurs dans une

matrice de forêt boréale faiblement influencée par l’humain ? L’analyse des patrons de répartition

et des assemblages d’oiseaux nicheurs dans 96 peuplements naturels répartis dans un vaste

territoire de l’aire d’étude a permis de répondre à cette question.

Chapitre 2

Est-ce que les populations d’un parasite, l’œstre, peuvent influencer la survie, la reproduction et

ultimement la croissance des populations de campagnol à dos roux au cours de l’été ? J’ai

répondu à cette question grâce à des mesures d’abondance des populations, des mesures de traits

d’histoire naturelle des campagnols à dos roux et des mesures d’abondance de parasites à

l’intérieur des hôtes.

Chapitre 3

Comment l’intensité de l’exploitation forestière influence la compétition intra- et interspécifique

et les coûts de prédation chez le campagnol à dos roux dans un écosystème forestier également

occupé par la souris sylvestre, une espèce sympatrique ? J’ai utilisé des observations sur

l’abondance et le comportement d’approvisionnement du campagnol à dos roux pour répondre à

cette question.

20

Modèles d’étude : oiseaux et micromammifères

Les oiseaux nicheurs et les micromammifères occupent un rôle central dans l’écosystème. Ce

sont des proies essentielles pour plusieurs prédateurs terrestres et aviaires (Donovan et al. 1995;

Gilg et al. 2003; Cheveau et al. 2004). De plus, les espèces insectivores contribuent à la

régulation des densités de populations d’insectes, limitant ainsi les risques d’épidémies (Kroll &

Fleet 1979, Fayt et al. 2005), alors que les granivores contribuent à la dispersion des graines et

ainsi à la colonisation des végétaux (Halvorson 1982, Côté et al. 2003, Elias et al. 2006).

L’objectif d’échantillonnage du chapitre 1 était extensif : échantillonner un grand nombre de sites

répartis dans différents habitats et différentes régions de l’aire d’étude, afin de s’assurer de

capturer des patrons d’occurrence représentatifs de la biodiversité de la forêt boréale naturelle de

la Côte-Nord. Pour rencontrer cet objectif extensif, j’ai échantillonné les oiseaux nicheurs. En

effet, les oiseaux nicheurs sont les organismes les plus faciles et les moins coûteux à inventorier

car les espèces peuvent être identifiées directement par leur chant (Gunn et al. 2000). De plus, les

oiseaux sont très importants en forêt boréale puisqu’ils représentent plus de 70 % des espèces de

vertébrés de cette forêt (Niemi et al. 1998).

Le même objectif aurait été difficilement réalisable avec les micromammifères. Leur

identification nécessite fréquemment la capture vivante ou mortelle des individus. Ces méthodes

d’inventaire sont bien plus coûteuses en temps et en argent que l’échantillonnage par points

d’écoute des oiseaux, tout en risquant de réduire l’abondance des populations animales. À titre de

comparaison, mon équipe de terrain a échantillonné 96 sites pour les oiseaux contre seulement 36

sites pour les micromammifères lors de l’été 2004. De plus, notre inventaire des populations de

micromammifères a révélé que cet assemblage était est très peu diversifié, un résultat également

observé ailleurs en forêt boréale (Bowman et al. 2000, Pearce & Venier 2005b). Le campagnol à

dos roux était de loin l’espèce la plus abondante dans les peuplements naturels (moyenne ± écart-

type = 17.2 ± 10.0 individus par 100 nuits-pièges) et il occupait 100 % des 36 sites que nous

avons échantillonné. En revanche, aucune autre espèce de micromammifère n’avait une

abondance supérieure à 2 individus par 100 nuits-pièges. À titre d’information, ces espèces

étaient la souris sylvestre (fréquence d’occurrence = 5 % des 36 sites échantillonnés), la

musaraigne pygmée (Sorex hoyi, 8 %), le campagnol lemming-boréal (Synaptomys borealis, 10

21

%), le campagnol des rochers (Microtus chrotorrhinus, 13 %), le campagnol lemming de Cooper

(Synaptomys cooperi, 21 %), l’écureuil roux (Tamiasciurus hudsonicus, 34 %), le campagnol des

bruyères (Phenacomys ungava, 45 %) et la musaraigne cendrée (Sorex cinereus, 63 %).

La première année d’échantillonnage a permis l’étude du parasitisme chez le campagnol à dos

roux, qui a mené à la publication du chapitre 2 dans Oecologia. Nous avions alors remarqué un

fort taux de parasitisme chez les campagnols par des larves d’œstre. À son troisième stade de

croissance, la larve atteint 5 à 10 % de la masse d’un campagnol (Catts 1982). En comparaison,

un fœtus de neuf mois chez l’humain pèse généralement moins de 5 % de la masse de sa mère.

On a donc émis l’hypothèse qu’un parasite de cette envergure pourrait être un facteur influant

pour la dynamique des populations de campagnol à dos roux de la forêt boréale de la Côte-Nord.

L’étude de cette relation hôte-parasite était d’autant plus intéressante qu’elle était peu

documentée (Bowman 2000, Pearce & Venier 2005a) puisque la plupart des études nord-

américaines rapportaient une association des œstres avec les souris (Peromyscus) plutôt qu’avec

les campagnols (Smith 1977, Catts 1982, Slansky 2007).

L’objectif d’échantillonnage du chapitre 3 était intensif : étudier en détails les effets de

l’altération de l’habitat sur la compétition intra- et interspécifique ainsi que sur les coûts de

prédation à l’aide du suivi de comportements d’approvisionnement. De la même façon que

l’étude des oiseaux nicheurs cadrait bien avec l’objectif extensif du chapitre 1, l’étude de l’espèce

la plus abondante de micromammifère, i.e. le campagnol à dos roux, était idéale pour rencontrer

l’objectif intensif du chapitre 3. Comme mentionné plus haut, l’échantillonnage des

micromammifères nécessite par défaut des méthodes plus intensives que celui des oiseaux.

22

Chapitre 1. Multiscale assessment of the influence of habitat structure and composition on the distribution of boreal birds

Jérôme Lemaître*, Daniel Fortin* & Marcel Darveau†

*NSERC-Université Laval Industrial Research Chair in Silviculture and Wildlife, Département de Biologie, Université Laval, Québec, QC G1V 0A6, Canada.

†Ducks Unlimited Canada, 710 rue Bouvier, bureau 260, Québec, QC G2J 1C2, Canada.

Article soumis à Ecography en Septembre 2009

23

Résumé

Nous avons déterminé la contribution relative de la structure et de la composition de l’habitat

pour 25 espèces d’oiseaux boréaux dans une matrice de vieille forêt, à des échelles spatiales

allant de 100 à 1000 m de rayon. Des analyses canoniques des correspondances ont révélé que la

structure de l’habitat expliquait la même proportion de variance dans l’assemblage des espèces

que la composition, indépendamment de l’échelle spatiale. Des régressions logistiques de la

probabilité d’occurrence des espèces, robustes dans 90% des cas, ont indiqué que certaines

espèces étaient exclusivement associées à la structure ou à la composition de l’habitat. En plus de

fournir une référence écologique de la diversité aviaire dans l’est de la forêt boréale canadienne,

notre étude indique que la structure et la composition sont d’importances égales dans le

façonnement des assemblages d’oiseaux. Les stratégies de conservation devraient donc préserver

autant la structure que la composition de l’habitat.

24

Abstract

While the positive relationship between species diversity and habitat heterogeneity is well

established, the relative contributions of habitat structure and composition to this association

remain unclear. The use of simple heterogeneity indices, together with the consideration of

individual species at multiple scales, should enlighten the relationship. We assessed the relative

contributions of structural and compositional habitat attributes for 25 bird species in a matrix of

old-growth forest, at five spatial scales characterized by radii ranging from 100 to 1000 m. We

recorded species occurrence at 96 stations located in >5 ha old-growth stands in the boreal forest

of eastern Canada. We characterized habitat structure from the proportion of the five radii

comprised of dense, open, and sparse stands, and habitat composition from the proportions of

coniferous, mixedwood, and deciduous stands, while accounting for latitude, longitude, and

altitude. Partial canonical correspondence analyses revealed that habitat structure explained an

amount of variance (21.7%) in the bird assemblage similar to that of composition (21.6%),

regardless of spatial scale. Logistic regression of the probability of occurrence for individual

species yielded fair predictions (i.e., Area Under the receiver-operating characteristic Curve >

0.70) for 90% of the species. Some species were exclusively associated with structure or

composition, despite the fact that habitat structure and composition explained similar percentages

of variance in the bird assemblage. In addition to providing an ecological benchmark of bird

diversity in the eastern boreal forest of North America, our study indicates that structural and

compositional habitat attributes are both important in shaping bird assemblages. Our study

stresses that conservation strategies should consider the complexity of responses among species

within assemblages. Our observations can have far-reaching consequences, as they imply that, for

example, measures mitigating forest harvest effects on animal species in boreal ecosystems must

put equal weight on the various components of habitat heterogeneity.

25

Introduction

Niche theory predicts that habitat heterogeneity should increase species diversity (Hutchinson

1957, MacArthur 1964). Habitat heterogeneity generally relates to variation in habitat structure

and composition over space and time (Rosenzweig 1995, Abrams 2000). Structural habitat

attributes are often associated with shelters and escape routes from predators (Robinson &

Holmes 1984), whereas compositional attributes are generally linked to resource availability

(Holmes & Robinson 1981). The relative contributions of habitat structure and composition to

species diversity remain unclear, despite several studies. Some authors have found equal

contributions of habitat structure and composition to diversity (Mac Nally 1990, Bersier & Meyer

1994), while others have reported stronger effects of either structure (Fleishman et al. 2003) or

composition (Rotenberry 1985, Lee & Rotenberry 2005). Given the increasing pressure of human

activities on Earth’s ecosystems (Vitousek et al. 1997), there is an urgent need to understand the

determinants of biodiversity to orient management and conservation decisions (Parrish et al.

2003, Fuhlendorf et al. 2006).

McElhinny et al. (2005) suggested that previous biodiversity studies may have reached different

conclusions because of the variability and complexity of indices used to characterize habitat

structure and composition. The authors therefore have advocated for the use of simple indices.

Also, the diversity of results might simply reflect that diversity indices, such as richness or

Simpson’s diversity, may not fully capture the variability of responses among species within a

community (Gotelli & Colwell 2001, Azeria et al. 2009a). For example, two habitats may have

similar species richness, but be occupied by two distinct assemblages (Jonsson & Jonsell 1999,

Azeria et al. 2009a). To understand the relative role of habitat structure and composition, it may

thus be advantageous to study the simultaneous response of individual species and species

assemblage rather than to rely only on diversity indices (Grand & Cushman 2003, Preston &

Harestad 2007). Additionally, habitat composition and structure often covary in space because

some compositional attributes may also structure the habitat (Deppe & Rotenberry 2008). For

example, the joint presence of shade-tolerant and shade-intolerant plant species may result in

multilayer canopies in forest stands (Spies & Franklin 1991). Such dependency may also hamper

26

our ability to assess the relative contributions of habitat structure and composition to biodiversity

patterns (Smith et al. 2009).

Habitat selection is recognized as a hierarchical process that strongly shapes species distributions

in heterogeneous environments (Boyce 2006). Nonetheless, only a few studies have evaluated the

relative contribution of structural and compositional habitat attributes to biodiversity patterns

while accounting for scaling effects (e.g., Deppe & Rotenberry 2008, Janssen et al. 2009). Some

authors have proposed that habitat structure would be the key factor at coarse spatial scales,

whereas habitat composition would become the most important factor at fine spatial scales

(Wiens & Rotenberry 1981, Deppe & Rotenberry 2008). In this study, we test this hypothesis by

varying the extent of spatial scale, i.e., the size of the sampling area, without varying its grain,

i.e., the resolution of data within sampling areas (Wiens 1989).

Our main objective was to determine the relative contributions of habitat structure and

composition to species distribution and assemblage of boreal bird species in a matrix of old-

growth forest, at multiple spatial scales. We identified the spatial scale that explained the most

variance in bird species assemblages, and we determined the independent and joint effects of

habitat structure and composition, together with those of spatial attributes (i.e., latitude,

longitude, and altitude), on bird assemblages. We modeled the probability of occurrence of each

species as a function of habitat structure and habitat composition at five spatial scales, while

taking into account bird response to spatial attributes. We then evaluated the relative

contributions of structure and composition to the distribution of individual species.

Materials and methods

Study area and bird sampling

The study took place in the eastern spruce-moss subdomain of the boreal forest (Boucher et al.

2003), in the Côte-Nord region of Québec, Canada (50°5’ N, 68°8’ W; Fig. 2). The study area is

characterized by a maritime climate, with annual mean temperature ranging from -2.5 to 0.0 °C

and annual precipitation ranging from 1,000 to 1,400 mm (Boucher et al. 2003). The abundant

27

precipitation results in a long fire cycle (Bouchard et al. 2008) and in landscapes dominated by

old-growth forests (average stand age >270 year old, Bouchard et al. 2008). The study area is

therefore characterized by forest stands that are irregular in structure and composition (Boucher

et al. 2003). Dominant tree species were black spruce (Picea mariana) and balsam fir (Abies

balsamea). White spruce (Picea glauca), paper birch (Betula papyrifera), trembling aspen

(Populus tremuloides), and jack pine (Pinus banksiana), were also locally abundant.

All bird surveys were carried out in stands that had never been logged. All stands were >5 ha and

>70-years-old (range: 70 to >500 years old, Bouchard et al. 2008), with nearly 65% of them

being ≥120-years-old. Stand selection reflected the range of structure and composition of the

boreal forests found in the Côte-Nord region. Point count stations were located >150 m from

stand edge, and >2 km from any other point count station. We determined the geographic

position of the point count station (i.e., latitude, longitude, and altitude), with a GPS unit (GPS

76CxTM, Garmin Ltd, Olathe, Kansas, USA).

We sampled birds at 96 stations using 15-min point count surveys during the breeding season

(Verner 1988). Each station was sampled twice (two weeks apart) between 0400 h and 1000 h.

During each visit, observers waited 2 min to allow birds to resume normal activity and then

recorded all birds detected within 100 m. We sampled 53 sites between 11 June and 8 July 2004

and 43 sites between 1 June and 25 June 2005. A species was coded as present if at least one

individual was detected, and was considered as absent otherwise.

Habitat structure and composition

We used few and simple habitat variables that were more strongly related either to habitat

structure or to habitat composition, as suggested in several studies (e.g., Tischendorf 2001,

McElhinny et al. 2005, Smith et al. 2009). We considered three habitat composition classes, i.e.,

coniferous, mixedwood, and deciduous, and three habitat structure classes, i.e., dense, open, and

sparse stands. We used the proportion of the landscape covered by each of these six classes as our

landscape metric. Tischendorf (2001) showed that this metric was the most reliable of the most

commonly used landscape metrics, and that it could also accurately predict ecological processes

such as dispersal or immigration.

28

We characterized habitat structure and composition within the vicinity of our point count stations

from a classified LandsatTM image (year: 2000) having a 25 m resolution. The LandsatTM

image was originally classified into 49 land cover classes by the Canadian Forest Service, but we

simplified this classification by collapsing land cover classes in a way that made it possible to

characterize forest structure apart from forest composition. Pixels were classified as coniferous,

mixedwood, and deciduous when conifers covered >75%, 25-75%, and <25% of the pixel area,

respectively. Pixels were also classified as dense, open, and sparse stands when within-pixel tree

density was >60%, 10-25%, and < 10%, respectively. Pixels with a tree density <10% were not

classified as forest stands. We quantified the proportion of land covered by each of the structure

and composition classes at five scales, each being matched to a radius centred on the point count

stations: r = 100 m, 250 m, 500 m, 750 m, and 1000 m.

Habitat structure and composition variables were arcsine-transformed prior to analyses to meet

normality assumptions (Zar 1999), and the proportion of deciduous forest was transformed in a

dummy variable because of overabundance of zeros. In other words, the presence of deciduous

stands in the landscape was coded as one and the absence was coded as zero. Landscape

characterization and data compilation were performed using ArcGIS 9.1 (ESRI 2006).

Community analyses

We used canonical correspondence analysis (CCA) to assess which combination of the three

habitat structure variables (i.e., proportions of dense, open, and sparse stands), the three habitat

composition variables (i.e., proportions of coniferous, mixedwood, and deciduous stands), and

the three spatial attributes (i.e., altitude, latitude, and longitude) was most strongly related to bird

assemblage (ter Braak 1986). CCA can be represented by joint biplots of the species and site

ordination scores in which quantitative environmental variables are depicted as arrows (length

and direction indicating the strength of the correlation with CCA axes) and binary environmental

variables are shown as centroids (ter Braak 1986). We calculated the proportion of variance in

bird assemblage that was explained by CCA at each spatial scale, as the sum of eigenvalues of

the constrained axes (i.e., axes correlated with explanatory variables), divided by the sum of

eigenvalues from a correspondence analysis (CA) of the bird community (Borcard et al. 1992,

Drapeau et al. 2000). In addition, we used partial constrained CCA to evaluate the independent

29

and joint contributions of habitat structure, habitat composition, and spatial attributes to variation

in assemblage of bird species (ter Braak 1986, Borcard et al. 1992).

Individual species analyses

We used multiple logistic regression to model the probability of occurrence of each bird species

as a function of the above structural, compositional and spatial attributes, and for each of the five

spatial scales (r = 100 m, 250 m, 500 m, 750 m, and 1000 m) (Anderson et al. 2005, Nams et al.

2006). Only bird species that occurred at more than 5% of the point count stations were included

in the analyses. We built models using a backward procedure, i.e. we built the global model and

eliminated, step by step, variables that did not improve the overall model fit according to log-

likelihood ratio tests (α = 0.05) (Hosmer & Lemeshow 2000). We used hierarchical partitioning

analysis to estimate the proportion of variance in the probability of occurrence of species that was

explained by the structure, composition, and spatial attributes (Chevan & Sutherland 1991).

Model performance was evaluated using receiver-operating characteristic (ROC) (Hosmer &

Lemeshow 2000). A value of the area under the ROC curve (AUC) of 1.0 represents a perfect

model, whereas a value of 0.5 indicates the poorest model fit (Hosmer & Lemeshow 2000).

Prediction accuracy can be classified as “poor” (AUC < 0.70), “fair” (≥ 0.70), “good” (≥ 0.80),

and “excellent” (≥ 0.90) (Hosmer & Lemeshow 2000). We investigated whether or not

multicollinearity influenced our results by using both the pairwise correlation matrix among all

predictors and each predictor’s variance inflation factor (VIF) (Smith et al. 2009). VIF is a

positive value representing the overall correlation of each predictor with all others in a model.

VIF were calculated for each predictor as the inverse of the coefficient of non-determination

(1/[1-R2]) for a regression of that predictor on all other predictors (Smith et al. 2009). Generally,

a VIF > 10 indicates ‘‘severe’’ collinearity (Smith et al. 2009).

We determined whether spatial scale influenced the proportion of variance in the probability of

occurrence of bird species that was explained by habitat structure, composition, and spatial

attributes. We used Friedman’s test (1937) that allowed considering bird species as a random

(block) effect. Friedman’s test is the nonparametric equivalent of a one-way, completely

randomized block ANOVA design (Zar 1999), where the observations in each of b blocks

30

(number of bird species) are ranked across the a levels of the treatment factor (number of spatial

scales). The usual χ2r statistic for the test was converted to the FF statistic of Iman and Davenport

(1980), which is considered to be much less conservative than χ2r, and where degrees of freedom

of the fixed-effect factor were a-1, and degrees of freedom of the random-effect (block) factor

were (a-1)(b-1), with a and b being the number of fixed and random effects, respectively. Also,

we used rank-based Kruskal-Wallis tests (nonparametric one-way ANOVA) to compare the

proportion of variance in the probability of occurrence of bird species explained by structure,

composition, and spatial attributes, based on the best model for each species, i.e., the model

having the highest AUC among the five spatial scales for each species. We used these

nonparametric analyses because the proportion of variance explained by most of the

environmental variables had an overabundance of zeros: between 40 to 70% of species did not

respond to a given environmental variable. Analyses were conducted with the R software, version

2.9.2 (R Development Core Team 2006).

Results

Bird assemblage

CCA (n = 25 species) that was constrained by habitat structure, composition, and spatial

attributes explained 12.8% of the variation in bird assemblage at a 100 m radius, 14.5% at 250 m,

14.6% at 500 m, 14.7% at 750 m, and 14.8% at 1000 m. We only present results from the CCA at

the 1000 m radius in the following analyses because all spatial scales yielded similar results, i.e.,

the relative contribution of habitat structure, composition, and spatial attributes was not

influenced by the spatial scale.

Partial CCA revealed that habitat structure and composition explained similar proportions of the

variance in bird assemblage, with habitat structure alone accounting for 21.7% and habitat

composition for 21.6% of the variance explained (Fig. 3). The sum of independent and joint

effects of habitat structure and composition accounted for 53.9% of the variance explained (Fig.

3). Spatial attributes accounted for 32.3% of the variance explained in bird assemblage.

31

Hermit thrush, white-winged crossbill, and black-backed woodpecker (see Table 1 for scientific

names) exhibited a strong negative relationship with the first canonical axis (CCA1), which

indicates that these species were more closely associated with landscapes that were located at

high latitude and high altitude, and which were comprised of a high proportion of mixedwood

(Fig. 4). In contrast, magnolia warbler, Nashville warbler, chipping sparrow, least flycatcher, and

golden-crowned kinglet were all positively associated with CCA1 (Fig. 4). American goldfinch,

black-throated green warbler, and Tennessee warbler were strongly and negatively linked to

CCA2, which reveals that these species were closely associated with landscapes containing a

high proportion of dense and coniferous stands. Conversely, pine siskin and chipping sparrow

were positively associated with CCA2. American robin, Swainson’s thrush, magnolia warbler,

and least flycatcher were the species most closely associated with landscapes that included

deciduous stands. In contrast, white-winged crossbill was the species most closely associated

with landscapes where deciduous stands were absent (Fig. 4).

Probability of occurrence of individual bird species

The probability of occurrence of 20 of the 25 bird species was significantly influenced by a

combination of habitat structure, composition, and/or spatial attributes at one or several spatial

scales (Table 2). At all spatial scales, pairwise correlations between predictors were < 0.60. At all

scales and for all bird species, VIF of all predictors were <4.5, except the proportion of open

stands (mean ± S.D. of VIF = 14.0 ± 3.7) and coniferous stands (14.9 ± 4.5). When one of these

two predictors was removed from the regression, the VIF of the other dropped to 1.4 ± 0.3.

Nevertheless, overall conclusions regarding the relative contributions of habitat structure and

composition were robust to collinearity. Indeed, the removal of the four species whose

probability of occurrence was influenced by the combination of open and coniferous stands (i.e.,

American robin, golden-crowned kinglet, pine siskin, and Tennessee warbler) did not influence

the results of Friedman’s Tests and Kruskal-Wallis Test (Fig. 5, see Table 3 for models of the

probability of occurrence).

We found that the spatial scale did not influence the variance in the probability of occurrence of

bird species that was explained by the sum of three structural (FF, 4, 76 = 0.34, p = 0.86), the sum

of three compositional (FF, 4, 76 = 0.03, p = 0.99), and the sum of three spatial attributes (FF, 4, 76 =

32

0.51, p = 0.73). Also, the variance in the probability of occurrence of bird species that was

explained by each of the nine variables was not influenced by the spatial scale (range among the

nine variables: FF, 4, 76 = 0.02 to 0.91, p = 0.45 to 0.99). The sum of three structural, three

compositional, and three spatial attributes explained similar amount of variation in the probability

of occurrence of bird species (Kruskal-Wallis H = 0.7, df = 2, p = 0.70; Fig. 5a). Also, each of the

nine variables explained similar amount of variation in the probability of occurrence of bird

species (Kruskal-Wallis H = 12.3, df = 8, p = 0.13; Fig. 5b), but the proportion of dense stands

tended to explain more variance (mean ± s.e. = 0.22 ± 0.06) than other structural habitat attributes

(≤0.08 ± 0.03) and longitude tended to explain less variance (0.06 ± 0.03) than the other spatial

attributes (≥0.14 ± 0.04).

As observed with the community analyses, analyses of the probability of occurrence of bird

species indicated that compositional, structural and spatial attributes explained a similar amount

of variation in the bird distributions. These latter models also identified the main determinants of

individual species occurrences (Table 3). Five species (i.e., alder flycatcher, Lincoln’s sparrow,

chipping sparrow, magnolia warbler, and Swainson’s thrush) did not respond to structural habitat

attributes (Table 3), but their occurrence was linked instead to habitat composition. For example,

the probability of occurrence for the alder flycatcher and Lincoln’s sparrow increased with the

proportion of deciduous stands in landscapes, while the probability of occurrence for the chipping

sparrow decreased. Also, the probability of occurrence for the magnolia warbler was negatively

related with the proportion of mixedwood stands, whereas the opposite relationship was observed

for Swainson’s thrush (Table 3). We also identified five species (i.e., brown creeper, black-

throated green warbler, winter wren, black-backed woodpecker, and white-winged crossbill) that

did not respond to composition variables, but their probability of occurrence varied with habitat

structure. The probability of occurrence for the brown creeper, black-throated green warbler, and

winter wren increased with the increase in the proportion of dense stands, and the probability of

occurrence for the black-backed woodpecker and white-winged crossbill decreased with the

proportion of open stands (Table 3).

33

Discussion

Despite a 10-fold change in spatial extent, we did not detect an influence of spatial scale on the

relative contributions that habitat structure and composition make to the probability of occurrence

and the assemblage of boreal bird species. Structural and compositional habitat attributes

contributed similar amounts of variation in bird species assemblages, although we observed

species-specific responses to habitat heterogeneity. Habitat structure and composition thus would

seem to be equally important for the creation of ecological niches.

Relative contribution of habitat structure and composition

Our results support previous studies reporting that habitat structure and composition have

similarly strong effects on species diversity (Mac Nally 1990, Bersier & Meyer 1994). These

results were robust to the collinear effect of the proportion of dense and coniferous stands. The

probability of occurrence for only 20% (4/20) of the species was influenced by the combination

of these two collinear predictors, i.e., American robin, golden-crowned kinglet, pine siskin, and

Tennessee warbler. The inclusion or removal of these species from the analyses did not influence

the results, i.e., habitat structure and composition had similarly strong effects on the probability

of occurrence of bird species. We conclude that structural and compositional habitat attributes are

both important in shaping bird assemblages, possibly because they would provide different food

and shelter resources (Holmes & Robinson 1981, Robinson & Holmes 1984, Deppe &

Rotenberry 2008).

Within bird assemblages, we found that the relative contributions of structural and compositional

habitat attributes varied among species, a result previously reported in other avian studies (e.g.,

Villard et al. 1999, Schmiegelow & Mönkkönen 2002). For example, we determined that some

species were not related to habitat structure (alder flycatcher, Lincoln’s sparrow, chipping

sparrow, magnolia warbler, and Swainson’s thrush), whereas others were not associated with

habitat composition (brown creeper, black-throated green warbler, winter wren, black-backed

woodpecker, and white-winged crossbill). The patterns of species occurrence that we observed in

an old-growth boreal forest well support the current expectations from autecology. For example,

brown creeper strongly depends on large trees for nesting and, to a lesser extent, for feeding

34

(Imbeau et al. 1999, Poulin et al. 2008), and we observed that the species was exclusively

associated with structural attributes. Also, alder flycatcher is closely associated with young

deciduous stands and deciduous wetlands (Crête et al. 1995, Hobson & Schieck 1999), and we

found that the species was exclusively associated with compositional attributes. Azeria et al.

(2009b) described assemblages as complex mixtures of species responding differently to similar

types of habitat heterogeneity, mainly because of interspecific interactions. Here, we showed how

habitat selection can also shape bird species assemblages.

Habitat heterogeneity and spatial scale

We found that the relative contribution of habitat structure and composition on the probability of

occurrence and the assemblage of boreal bird species was independent of the spatial scale. In

contrast, Wiens and Rotenberry (1981) found that shrubsteppe birds were more strongly

influenced by habitat structure at broad scales and by habitat composition at finer scales. Deppe

and Rotenberry (2008) found a similar pattern for fall migratory birds at stopover sites. In our

study, we varied the extent of the spatial scale without varying the grain (sensu Wiens 1989),

consistent with recent studies on spatial scaling (Anderson et al. 2005, Nams et al. 2006). This

approach contrasts with commonly used multiscale assessments for which the nature of the

habitat attributes that are considered have varied across scales. Fine-scale measurements often

come from field sampling, whereas coarse-scale characterization is performed with mapping

from geographic information systems (GIS, e.g., Grand & Cushman 2003, Deppe & Rotenberry

2008, Janssen et al. 2009). Although the latter approach is useful in identifying habitat attributes

that can have a major influence on species diversity at various spatial scales (Wiens 1989), it

might not be ideal in evaluating how the relative contributions of habitat structure and

composition vary with spatial extent (Tews et al. 2004, McElhinny et al. 2005). The spatial

extent that we used in this study encompassed what are generally considered as local and

landscape scales from a forest bird perspective (e.g., Lichstein et al. 2002, Boscolo & Metzger

2009), i.e., from 100 m to 1000 m radii centred on point count stations, which corresponded to

sampling areas ranging from approximately 3 to 314 ha. Nonetheless, we did not detect an

influence of spatial scale on the relative contributions of habitat structure and composition on

bird assemblages. We conclude that interspecific variation linking bird assemblages to habitat

35

structure and composition has stronger effects than spatial scale effects. In fact, there was little to

be gained by considering habitat heterogeneity at spatial scales larger than 250 m for bird

assemblages.

Characterizing habitat heterogeneity

We used parsimonious indices to characterize habitat structure and composition, as recently

recommended (Tischendorf 2001, Tews et al. 2004, McElhinny et al. 2005). Habitat structure

was characterized using three variables (i.e., proportions of dense, open and sparse stands);

habitat composition was also characterized with three variables (i.e., proportions of coniferous,

mixedwood, and deciduous stands). We also took into account three spatial attributes (i.e.,

latitude, longitude, and altitude). This combination of nine variables explained 14.8% of the

variance in the bird assemblage. In comparison, Drapeau et al. (2000) explained 16.4% of the

variance in their boreal bird assemblage using 41 variables, and Hobson et al. (2000) explained

22.8% of the variance in their boreal bird assemblage using 26 variables. Our more parsimonious

approach therefore explained similar amounts of variance, and its simplicity should encourage

others to verify our conclusions on the relative contributions of habitat structure and composition

in different ecosystems and with other taxa (Tews et al. 2004, McElhinny et al. 2005, Smith et al.

2009).

At the individual species level, our approach yielded fairly robust predictions (i.e., AUC ≥ 0.70)

of the probability of occurrence for 80% (20/25) of the bird species. Three of the five species that

did not respond to habitat or spatial attributes were also the most frequent in the study area

(occurrence >70%), i.e., ruby-crowned kinglet, dark-eyed junco, and yellow-rumped warbler, a

result previously reported in the boreal forest of Québec (Crête et al. 1995, Imbeau et al. 1999).

These generalist species exhibit high plasticity in habitat selection and foraging decisions (Simon

et al. 2003). Hence, it is not surprising that we could not derive robust predictions of the

occurrence of these ubiquitous species. Therefore, our parsimonious models allowed us to build

robust predictions for 90% (20/22) of the species that were expected to respond selectively to

habitat heterogeneity.

36

The absence of relationships between the probability of occurrence of the red-breasted nuthatch

and habitat attributes is not uncommon. This species is a cyclic species, and its occurrence

patterns seem to be more closely linked to cone production and insect outbreaks than to habitat

attributes (Norris & Martin 2008). Lastly, the yellow-bellied flycatcher did not respond to habitat

heterogeneity or spatial attributes. The ecology of this species remains poorly documented. Burris

and Haney (2005) found that yellow-bellied flycatchers preferred blowdowns over control mature

forest in Minnesota, whereas King et al. (2008) found that the species was associated with mature

forest conditions in New Hampshire. Therefore, doubts still remain regarding the association

between the yellow-bellied flycatcher and attributes of habitat heterogeneity.

Conservation implications

We showed that the response of birds to natural habitat heterogeneity was highly species-specific.

Richness is a commonly-used measure to assess biodiversity hotspots (e.g., Prendergast et al.

1993, Virolainen et al. 2000) and to manage for biodiversity conservation (Fleishman et al.

2006). Our study stresses that comprehensive strategies for biodiversity conservation should

account for information on individual species because simple diversity indices may overlook the

complexity of responses among species within community assemblages (Gotelli & Colwell 2001,

Azeria et al. 2009a). Individual species information allowed us to understand, for example, why

the boreal chickadee is considered as a species of concern in the boreal forest by the Committee

on the Status of Endangered Wildlife in Canada (COSEWIC 2009). We determined that the

boreal chickadee was more likely to occupy areas with high proportions of dense and coniferous

stands, which are also the target of forestry companies. The occurrence patterns of boreal

chickadees that we observed in old-growth forests indicate that forest harvesting should be

detrimental to the species.

Moreover, our study provides a valuable ecological benchmark of bird diversity in the eastern

part of the North American boreal forest. Half of the study area was pristine (Bouchard et al.

2008), and our point count stations were located in >5 ha old-growth stands. While a few studies

have also documented bird diversity in old-growth boreal forests (Edenius & Sjöberg 1997 in

Sweden, Virkkala & Rajasärkkä 2006 in Finland, Azeria et al. 2009b in our study area), none of

them have specifically addressed the relative contributions of structural and compositional

37

attributes at multiple spatial scales. This information is valuable, however, in orienting

biodiversity conservation strategies because it improves our assessment of the effects of human

disturbance on the diversity of boreal birds. For example, when compared to clearcut harvesting,

partial cutting and retention harvests should have less of an impact on wildlife (Vanderwel et al.

2007). We suspect that even moderate intensity harvesting might be detrimental to bird species

that are strongly associated with structural attributes, such as the black-throated green warbler or

the brown creeper. The information of bird distribution patterns in pristine forest that we have

provided is required as baseline information to fully assess harvesting impacts. Also, tree planting

is a common silvicultural practice that simplifies habitat composition in the long run (Flemming

& Freedman 1998). Our baseline information suggests that plantations might affect species

depending on habitat composition, such as hermit thrush, Lincoln’s sparrow or Swainson’s

thrush. These predictions may be validated in a near future in our study area because the area

being harvested increases on a regular basis. In this context, our point count stations conducted in

virgin stands may provide the basis for a long-term survey on the impact of forest harvesting on

avian diversity.

38

Table 1. Proportion of point count stations occupied by bird species in a matrix of old-growth

boreal forest of the Côte-Nord region of Québec, Canada (n = 96 stations). Common names,

scientific names, and acronyms are provided.

Common name Scientific name Acronym Proportion

Ruby-crowned kinglet Regulus calendula RCKI 0.88

Swainson's thrush Catharus ustulatus SWTH 0.72

Yellow-rumped warbler Dendroica coronata YRWA 0.71

Dark-eyed junco Junco hyemalis DEJU 0.71

White-throated sparrow Zonotrichia albicollis WTSP 0.55

Winter wren Troglodytes troglodytes WIWR 0.52

Boreal chickadee Poecile hudsonicus BOCH 0.39

Red-breasted nuthatch Sitta canadensis RBNU 0.35

Golden-crowned kinglet Regulus satrapa GCKI 0.31

Magnolia warbler Dendroica magnolia MAWA 0.27

Brown creeper Certhia americana BRCR 0.27

Nashville warbler Vermivora ruficapilla NAWA 0.25

Hermit thrush Catharus guttatus HETH 0.22

Black-backed woodpecker Picoides arcticus BBWO 0.21

Chipping sparrow Spizella passerina CHSP 0.19

Tennessee warbler Vermivora peregrina TEWA 0.18

39

Pine siskin Carduelis pinus PISI 0.16

American robin Turdus migratorius AMRO 0.16

Yellow-bellied flycatcher Empidonax flaviventris YBFL 0.15

Alder flycatcher Empidonax alnorum ALFL 0.09

White winged crossbill Loxia leucoptera WWCR 0.08

Least flycatcher Empidonax minimus LEFL 0.08

American goldfinch Carduelis tristis AMGO 0.08

Lincoln's sparrow Melospiza lincolnii LISP 0.07

Black-throated green warbler Dendroica virens BTGW 0.07

40

Table 2. Area under the ROC curve (AUC) reflecting the performance of multiple logistic

regressions evaluating the probability of occurrence of 25 boreal bird species as a function of

habitat composition variables, habitat structure variables, and spatial variables evaluated at five

spatial scales, i.e., r = 100 m, 250 m, 500 m, 750 m, and 1000 m. A higher AUC indicates a

stronger link between the probability of occurrence for that species and habitat attributes. AUC

values < 0.70 indicate poor model performance.

Species Landscape radius (m)

100 250 500 750 1000

Strongest response at r = 100 m

Alder flycatcher 0.777 0.695 0.752 0.748 0.768

American robin 0.752 0.623 0.649 0.632 0.649

Least flycatcher 0.817 0.800 0.748 0.653 0.670

Nashville warbler 0.803 0.769 0.760 0.769 0.768

Strongest response at r = 250 m

American goldfinch 0.734 0.858 0.827 0.798 0.830

Brown creeper 0.739 0.750 0.709 0.705 0.698

Lincoln's sparrow 0.791 0.864 0.785 0.785 0.785

Swainson's thrush 0.500 0.725 0.500 0.680 0.675

Tennessee warbler 0.823 0.878 0.831 0.824 0.827

White-throated sparrow 0.728 0.746 0.741 0.731 0.703

Strongest response at r = 500 m

41

Black-backed woodpecker 0.696 0.757 0.765 0.725 0.732

Black-throated green warbler 0.738 0.857 0.872 0.823 0.812

Boreal chickadee 0.661 0.670 0.783 0.696 0.678

Winter wren 0.677 0.689 0.738 0.713 0.686

Strongest response at r = 750 m

White-winged crossbill 0.718 0.718 0.751 0.776 0.767

Strongest response at r = 1000 m

Chipping sparrow 0.779 0.802 0.789 0.813 0.821

Golden-crowned kinglet 0.656 0.720 0.736 0.788 0.793

Hermit thrush 0.710 0.744 0.769 0.767 0.782

Magnolia warbler 0.757 0.782 0.766 0.779 0.784

Pine siskin 0.700 0.747 0.742 0.774 0.804

Weak response

Dark-eyed junco 0.658 0.658 0.675 0.674 0.683

Red-breasted nuthatch 0.628 0.676 0.638 0.646 0.672

Ruby-crowned kinglet 0.691 0.643 0.643 0.643 0.650

Yellow-bellied flycatcher 0.500 0.500 0.500 0.656 0.689

Yellow-rumped warbler 0.634 0.500 0.500 0.592 0.500

4

2

Tab

le 3

. M

odel

s of

the

pro

babi

lity

of

occu

rren

ce o

f bi

rd s

peci

es a

s a

func

tion

of

habi

tat

stru

ctur

e an

d co

mpo

siti

on,

as w

ell

as s

pati

al a

ttri

bute

s.

Reg

ress

ion

coef

fici

ent

(β)

is p

rovi

ded

in t

he f

irst

row

for

eac

h sp

ecie

s an

d pe

rcen

tage

of

vari

ance

exp

lain

ed b

y en

viro

nmen

tal

vari

able

s is

giv

en i

n

pare

nthe

ses

in th

e se

cond

row

. The

sum

(∑

) of

var

ianc

e ex

plai

ned

by h

abit

at s

truc

ture

, hab

itat

com

posi

tion

, and

spa

tial

att

ribu

tes

is a

lso

pres

ente

d in

the

seco

nd r

ow f

or e

ach

spec

ies.

Her

e, o

nly

the

best

mod

el o

f fi

ve s

pati

al s

cale

s (r

= 1

00 m

, 25

0 m

, 50

0 m

, 75

0 m

, an

d 10

00 m

), i

.e.,

the

mod

el w

ith

the

high

est A

UC

val

ue (

see

Tab

le 2

for

AU

C v

alue

s as

a f

unct

ion

of s

pati

al s

cale

), is

pre

sent

ed f

or e

ach

of th

e 20

spe

cies

that

res

pond

ed to

hab

itat

or

spat

ial

attr

ibut

es. F

ive

spec

ies

had

AU

C <

0.7

at e

ach

of th

e fi

ve s

pati

al s

cale

s, in

dica

ting

a p

oor

mod

el f

it, a

nd th

eir

mod

els

are

ther

efor

e no

t pre

sent

ed.

Spe

cies

S

truc

ture

Com

posi

tion

Spa

tial

att

ribu

tes

D

ense

O

pen

Spa

rse

∑

C

onif

erou

s M

ixed

woo

d D

ecid

uous

∑

Alt

itud

e L

atit

ude

Lon

gitu

de

∑

Ald

er f

lyca

tche

r

-4

.366

3.

618

-3.7

63

(0)

(38.

9)

(24.

16)

(63.

06)

(3

6.94

) (3

6.94

)

Am

eric

an g

oldf

inch

-2.5

87

-1

7.13

1

-0

.009

5.

047

(13.

01)

(1

3.01

)

(23.

03)

(23.

03)

(1

2.57

) (5

1.39

)

(63.

96)

Am

eric

an r

obin

5.

249

2.03

4

-5.4

27

-2.6

9 -1

6.67

1

(1

4.48

) (2

0.27

)

(34.

75)

(3

4.24

) (9

.29)

(2

1.72

) (6

5.25

)

(0

)

Bla

ck-b

acke

d w

oodp

ecke

r -5

.598

-0

.282

5.

653

(6

6.5)

(5

.93)

(72.

43)

(0)

(2

7.57

) (2

7.57

)

Bla

ck-t

hroa

ted

gree

n w

arbl

er

21.0

81

2.04

-5

.611

4

3

(6

4.55

) (2

8.34

)

(92.

89)

(0)

(7

.11)

(7

.11)

Bor

eal c

hick

adee

2.

415

-3

.768

0.

138

-4.2

26

-1.9

13

-1

.446

(1

5.57

)

(13.

53)

(29.

1)

(2

5.96

) (1

2.66

) (2

4.81

) (6

3.43

)

(7

.48)

(7.4

8)

Bro

wn

cree

per

3.12

8 3.

036

0.00

6

(2

0.75

) (3

7.96

)

(58.

71)

(0)

(4

1.28

)

(4

1.28

)

Chi

ppin

g sp

arro

w

-0

.114

-1.7

97

(0)

(5

5.7)

(5

5.7)

(4

4.3)

(44.

3)

Gol

den-

crow

ned

king

let

-0.8

21

-0.6

73

-9.2

41

0.38

8 -8

.004

-2

.155

-4

.077

(1

9.1)

(1

1.67

) (1

7.76

) (4

8.53

)

(15.

72)

(4.1

)

(19.

82)

(22.

94)

(8.7

) (3

1.64

)

Her

mit

thru

sh

-0.8

27

6.

876

1.53

1

1.66

8

(4

8.82

)

(4

8.82

)

(2

8.25

) (9

.29)

(3

7.54

)

(1

3.64

)

(13.

64)

Lea

st f

lyca

tche

r 4.

171

-2.4

74

-1

6.39

(6

0.93

)

(6

0.93

)

(25.

87)

(1

3.2)

(3

9.07

)

(0

)

Lin

coln

's s

parr

ow

2.

302

-0.0

13

10

.925

(0)

(2

9.97

) (2

9.97

)

(35.

9)

(3

4.13

) (7

0.03

)

4

4

Mag

noli

a w

arbl

er

-6.0

86

-0

.006

-1

.546

(0)

(33.

48)

(3

3.48

)

(30.

86)

(35.

66)

(6

6.52

)

Nas

hvil

le w

arbl

er

2.95

1

2.38

8

-1

.267

-0

.011

(2

9.75

)

(7.2

5)

(37)

(11.

43)

(11.

43)

(5

1.57

)

(5

1.57

)

Pin

e si

skin

0.61

4

-0.2

78

-2

.097

(30.

67)

(3

0.67

)

(42.

28)

(42.

28)

(27.

05)

(2

7.05

)

Sw

ains

on's

thru

sh

2.72

1 -1

.839

0.

004

-1.1

78

(0)

(24.

05)

(41.

94)

(65.

99)

(1

8.24

) (1

5.77

)

(34.

01)

Ten

ness

ee w

arbl

er

-7.8

34

-12.

656

-5.6

22

10.2

63

10.7

66

-0

.016

3.

139

(1

0.5)

(6

.84)

(4

.22)

(2

1.56

)

(5.6

2)

(23.

32)

(2

8.94

)

(33.

1)

(16.

38)

(4

9.48

)

Whi

te-t

hroa

ted

spar

row

-4

.877

-4.4

16

1.

624

-1.2

47

(7

1.01

)

(11.

73)

(82.

74)

(6.2

5)

(6

.25)

(1

1.01

)

(11.

01)

Whi

te-w

inge

d cr

ossb

ill

-0

.367

0.

01

1.94

7

(24.

75)

(2

4.75

)

(0

)

(46.

13)

(29.

11)

(7

5.24

)

Win

ter

wre

n 2.

872

4.

436

-0

.004

-1

.085

4

5

(3

3.33

)

(29.

19)

(62.

52)

(0)

(1

4.9)

(2

2.59

)

(37.

49)

46

Figure 2. Point count station locations and (inset) general location of the study area in Québec,

Canada.

47

Composition

StructureSpatial

attributes

21.6%

21.7%32.3%

5.3%

1.5%

10.6%

6.7%

Composition

StructureSpatial

attributes

21.6%

21.7%32.3%

5.3%

1.5%

10.6%

6.7%

Figure 3. Venn diagram summarizing the variance partitioning analysis of the CCA of the species

data matrix as a function of the three structural attributes, the three compositional attributes, and

the three spatial attributes. The proportion of the variance in species assemblage explained by all

the nine variables was 14.8%. The diagram shows how this 14.8% of variance is partitioned

among independent and joint effects of habitat composition, habitat structure, and spatial

attributes. The sum of the percentages presented in the diagram is 100%.

48

-1.0 -0.5 0.0 0.5 1.0

-1.0

-0.5

0.0

0.5

CCA1

CCA2

DECID0

DECID1-1

0

LAT

LONG

ALT

CONIF

MIXED

DENSE

OPEN

SPARSElisp

btgw

wwcr

amgo

leflalfl

ybfl

amro

pisi

tewa

chsp

bbwoheth

nawabrcr

mawa

gcki

rbnuboch

wiwr

wtsp

dejuyrwaswth

rcki

-4 -2 0 2 4

-4-2

02

CCA1

CCA2

DECID0DECID1 0

LAT

LONG

ALT

CONIF

MIXED

DENSE

OPEN

SPARSE

a

b

49

Figure 4. Ordination plot of a canonical correspondence analysis (CCA) conducted on 25 bird

species (acronyms are defined in Table 1) as a function of habitat structure, composition, and

spatial attributes of 96 sites located in a matrix of old-growth boreal forest in the Côte-Nord

region of Québec, Canada. a) Position of the 96 sites (open circles) in the ordination space; an

arrow’s length and its angle represents the strength of the correlation between environmental

variables and CCA axes. DECID1 and DECID0 are the centroids of the dummy variable

“presence of deciduous stands in the landscape”. b) Position of the 25 bird species in the

ordination space.

50

Struct.

Comp.

Abiotic

Proportion of variance explained

0.0

0.1

0.2

0.3

0.4

0.5

dense

open

sparse

conifer

decid.

mixed

latitud.

altitut.

longit.

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Structure

Composition

Abiotic

a b

Spatialattributes

Structure Composition

Structure

Composition

Spatial attributesStruct.

Comp.

Abiotic

Proportion of variance explained

0.0

0.1

0.2

0.3

0.4

0.5

dense

open

sparse

conifer

decid.

mixed

latitud.

altitut.

longit.

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Structure

Composition

Abiotic

a b

Spatialattributes

Structure Composition

Structure

Composition

Spatial attributesStruct.

Comp.

Abiotic

Proportion of variance explained

0.0

0.1

0.2

0.3

0.4

0.5

dense

open

sparse

conifer

decid.

mixed

latitud.

altitut.

longit.

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Structure

Composition

Abiotic

a b

Struct.

Comp.

Abiotic

Proportion of variance explained

0.0

0.1

0.2

0.3

0.4

0.5

dense

open

sparse

conifer

decid.

mixed

latitud.

altitut.

longit.

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Structure

Composition

Abiotic

a

Struct.

Comp.

Abiotic

Proportion of variance explained

0.0

0.1

0.2

0.3

0.4

0.5

dense

open

sparse

conifer

decid.

mixed

latitud.

altitut.

longit.

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Structure

Composition

AbioticStruct.

Comp.

Abiotic

Proportion of variance explained

0.0

0.1

0.2

0.3

0.4

0.5

dense

open

sparse

conifer

decid.

mixed

latitud.

altitut.

longit.

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Structure

Composition

Abiotic

a b

Spatialattributes

Structure Composition

Structure

Composition

Spatial attributes

Figure 5. Mean (± s.e.) proportion of variance explained by composition, structure, and spatial

attributes in models of the probability of occurrence for each of 20 bird species in a matrix of old-

growth boreal forest. Proportion of variance explained by each environmental variable was

calculated using hierarchical partitioning. a) Summed proportion of variance explained by the

three variables of habitat structure (proportion of dense, open, and sparse habitats in the

landscape), the three variables habitat composition (proportion of mixedwood, coniferous, and

deciduous in the landscape), and the three spatial attributes (latitude, altitude, and longitude). b)

Proportion of variance explained by each of the nine environmental variables.

51

Chapitre 2. Bot fly parasitism of the red-backed vole: host survival, infection risk, and population growth

Jérôme Lemaître*, Daniel Fortin*, Pierre-Olivier Montiglio* & Marcel Darveau†

*NSERC–Université Laval industrial research chair in silviculture and wildlife, Département de biologie, Université Laval, Sainte-Foy, Québec, G1V 0A6, Canada.

†Ducks Unlimited Canada, 710 rue Bouvier, bureau 260, Québec, QC G2J 1C2, Canada.

© Reproduction autorisée de l’article Lemaître et al. (2009) Oecologia 159: 283-294

52

Résumé

Les parasites pourraient jouer un rôle important dans la dynamique des populations d’hôtes mais

les évidences empiriques sont peu nombreuses. Nous avons déterminé les effets du parasitisme

par les œstres (Cuterebra spp) sur 36 populations de campagnols à dos roux (Myodes gapperi).

Nous avons trouvé que la probabilité de survie de 341 campagnols diminuait avec l’infection par

les œstres dans des conditions de stress élevé. Il n’y avait pas d’effet de l’infection sur l’activité

reproductrice des femelles de campagnols. Le principal facteur responsable du risque individuel

d’infection et de l’abondance de larves parasites dans les campagnols était l’abondance de

campagnols plutôt que les traits d’histoire naturelle ou les variables d’habitat. La croissance

estivale des populations de campagnols était négativement reliée à la prévalence des œstres. Nous

concluons que les œstres ont le potentiel de réduire la survie d’une grande partie des populations

de campagnols à dos roux.

53

Abstract

Parasites can play an important role in the dynamics of host populations, but empirical evidence

remains sparse. We investigated the role of bot fly (Cuterebra spp.) parasitism in red-backed

voles (Myodes gapperi) by first assessing the impacts of the parasite on the probability of vole

survival under stressful conditions as well as on the reproductive activity of females. We then

identified the main factors driving both the individual risk of infection and the abundance of bot

flies inside red-backed voles. Finally, we evaluated the impacts of bot fly prevalence on the

growth rate of vole populations between mid-July and mid-August. Thirty-six populations of red-

backed voles were sampled in the boreal forest of Québec, Canada. The presence and the

abundance of parasites in voles, two host life-history traits (sex and body condition), three indices

of habitat complexity (tree basal area, sapling basal area, coarse woody debris volume), and vole

abundance were considered in models evaluating the effects of bot flies on host populations. We

found that the probability of survival of red-backed voles in live traps decreased with bot fly

infection. Both the individual risk of infection and the abundance of bot flies in red-backed voles

were driven mainly by vole abundance rather than by the two host life-history traits or the three

variables of habitat complexity. Parasitism had population consequences: bot fly prevalence was

linked to a decrease in short-term growth rate of vole populations over the summer. We found

that bot flies have the potential to reduce survival of red-backed voles, an effect that may apply to

large portions of populations.

54

Introduction

A central objective of population ecology is to identify factors controlling population dynamics.

Biological factors such as predation, competition, and parasitism can play this role. Theoretical

and empirical investigations have largely focused on the role of predation and competition on

population dynamics, leaving the contribution of parasitism relatively underappreciated (see

review in Møller 2005, Lively 2006). There are field observations, however, indicating that

parasites can control the populations of their hosts (e.g. Witting 2000, Hanski et al. 2001, Turchin

& Hanski 2001, Eccard & Ylonen 2002, 2003, Gilg et al. 2006). For example, parasitic nematodes,

such as Trichostrongylus tenuis and Ostertagia gruehneri, were identified as population regulators

of red grouse (Lagopus lagopus scoticus) (Dobson & Hudson 1992) and Svalbard reindeer

(Rangifer tarandus platyrhynchus) (Albon et al. 2002), respectively. To regulate host populations,

parasites should affect individual survival or reproduction in a density-dependent manner and they

should infect large portions of populations, including individuals having high potential to

contribute to the reproductive output of populations (Anderson & May 1978, May & Anderson

1978).

We investigated whether bot flies (Cuterebra spp.) can impact populations of the most abundant

small mammal of North American boreal forest, the southern red-backed vole (Myodes gapperi)

(Pearce & Venier 2005b, Macdonald et al. 2006). Whereas bot flies have long been thought to

parasitize mainly Peromyscus spp. (Smith 1977, Catts 1982, Slansky 2007), recent observations

indicate that the red-backed vole would be the primary host in eastern Canada (Bowman 2000,

Pearce & Venier 2005a). First, we examined the relationships between bot fly infection and red-

backed vole survival and reproduction. Second, by looking at biotic attributes influencing the

probability of infection and the abundance of parasites in voles, we evaluated whether bot flies had

the potential to infect large portions of red-backed vole populations. Finally, we assessed impacts

of bot flies on the growth rate of red-backed vole populations over the summer. Analyses also

included a search for density dependence to evaluate whether effects of bot flies were such that the

parasite could ultimately regulate vole populations.

55

The impact of bot flies on the survival of their hosts is ambiguous because of conflicting

observations. For example, bot flies seem to reduce the survival of Townsend’s voles (Microtus

townsendii) (Boonstra et al. 1980), but to increase that of Peromyscus (Hunter et al. 1972, Clark &

Kaufman 1990, Munger & Karasov 1991). The apparent benefit of the parasite on host survival,

however, could simply be an artifact resulting from differential trapping probabilities among

resident and transient small mammals (Wecker 1962, Hunter et al. 1972, but see Burns et al.

2005). Resident hosts would have a higher probability of infection than transients, because gravid

bot flies lay their eggs on vegetation near burrows and movement pathways that are mainly used

by residents (Catts 1982, Slansky 2007). The higher probability of infection for residents

combined with their higher probability of being recaptured than transients might lead to the faulty

conclusion that infected hosts have a higher probability of survival (Wecker 1962, Hunter et al.

1972). Also, once hosts are weakened by parasites, they could become more prone to predation or

to negative effects of competition (Albon et al. 2002). Predation or competition could then be

identified as having negative effects on individuals, while the ultimate cause might have been

parasitism. Predation and competition thus may conceal the impact of parasites on their hosts.

To clarify the direct relationship between bot fly infection and host survival, one would have to

use a metric of host survival that is unrelated to the probability of recapture and to control for

predation and direct competition. This may be achieved by studying host survival during live

trapping. The dependent variable would be whether the host is dead or alive following overnight

trap confinement. With such an experiment, the overnight survival should be independent from the

probability of recapture and, while inside the trap, individuals would be isolated from predation

and interference competition. Although trap confinement will not be experienced by most small

mammals, their response to the event can inform on the probability of survival under elevated

stress experienced in natural conditions. Trap confinement produces a physiological response (i.e.

increase of corticosterone levels) in meadow voles (Microtus pennsylvanicus) that is similar to the

response induced by natural threats, such as predation (Harper & Austad 2001, Fletcher &

Boonstra 2006). Survival in live traps would provide more valuable information on the potential

effects of parasites on their host if infected hosts do not spend more time in traps than uninfected

hosts, and if both groups experience similar air temperature during confinement.

56

While many studies have focused on the relationship between bot fly infection and host survival,

fewer have looked at the effects of bot flies on host reproduction. The few laboratory studies that

have addressed this issue found that bot flies impaired gonadal development of males (Wecker

1962, Timm & Cook 1979) whereas the number of embryos, corpora lutea, or placental scars did

not differ between infected and uninfected females (Timm & Cook 1979). Early field studies

showed that the proportion of reproductive individuals was reduced in populations subject to bot

fly infection (Wecker 1962, Brown 1965, Dunaway et al. 1967), but a recent study failed to detect

the relationship (Burns et al. 2005). In the field, the link between bot fly infection and

reproductive activity cannot be assessed for males because larvae of bot flies embed in the

inguinal region, causing scrotal swelling in males (Burns et al. 2005), while reproductive activity

is typically assessed by the position of testes (scrotal or abdominal).

Anderson and May (1978) stated that parasites have to infect large portions of populations to play

a regulatory role. In this study, we verified whether bot flies had the potential to infect large

portions of red-backed vole populations by looking at how bot flies responded to two key life-

history traits of this host: sex and body condition. The absence of selection for the sex of their host

would indicate that bot flies can infect large portions of vole populations. Indeed, their effect on

hosts would then no longer be restricted to a particular gender. We also investigated the

relationship between botfly infection and the body condition of red-backed voles because this life-

history trait is linked to health status and dominance in small mammals (Schulte-Hostedde et al.

2005). Non-selectivity of bot flies for body condition would indicate that the parasite can infect

large portions of host populations. Their effect would not be restricted, for example, to the weakest

hosts. On the other hand, bot flies could also have significant impacts on host populations by

selecting hosts with better body condition, as these individuals contribute the most to the

reproductive output of populations.

The infection by bot flies is a two-step process. First, gravid bot flies lay their eggs on suitable

microsites, usually on vegetation near burrow entrances of small mammals (Catts 1982). Second,

eggs stick to the passing hosts and hatch on contact with the body heat of small mammals. Larvae

penetrate the host body usually through the mouth or nose (Catts 1982). They migrate to their

development sites for approximately two weeks (usually the inguinal regions), grow there for

approximately one week, and they finally exit their host to pupate into the soil (Gingrich 1981).

57

The larvae emerge as adults the following summer (Catts 1982). Gravid bot flies should lay their

eggs where they are most likely to get in contact with small mammals. Movement pathways of

small mammals are related to habitat structure (e.g. woody debris on the ground, basal area of

trees, snags and saplings), and parasites thus could adjust egg laying patterns to habitat

complexity, with consequence that infection risk for hosts might also be linked to habitat

complexity (Hensley 1976, Cockle & Richardson 2003). Gravid bot flies should also select

habitats having high density of host individuals to increase the pool of potential hosts.

Ultimately, to clarify the potential influence of bot flies on red-backed vole populations, one

should test whether bot fly impacts on host survival or reproduction are density dependent and

whether these impacts turn into effects on host populations. This could be achieved by studying

the population growth rate of red-backed voles as a function of their infection rates. Burns et al.

(2005) showed that the parasite can decrease the growth rate of white-footed mouse populations.

To our knowledge, such information is not available for red-backed voles.

In this study, we first tested whether bot flies decreased the probability of survival of red-backed

voles in live traps in a density-dependent manner, i.e., whether the negative impact of bot flies on

survival increased with population density. Second we verified whether bot fly infection decreased

reproductive activity of red-backed vole females. Third we verified whether bot flies had the

potential to infect large portions of red-backed vole’s population by examining the relationship

between both the probability of infection and the abundance of bot flies in hosts, and the

combination of three habitat complexity variables (i.e., volume of coarse woody debris, basal area

of trees and snags, and basal area of saplings), host abundance, as well as two key host life-history

traits (i.e., sex and body condition). Finally, we evaluated factors influencing the growth rate of

red-backed vole populations between mid-July and mid-August, and determined whether

population growth was related to the prevalence of bot fly infection within the host populations.

58

Materials and methods

Study area and sampling design

The study took place in a boreal forest of the Côte-Nord region of Québec, Canada (50°5’ N, 68°8’

W). The study area is characterized by a maritime climate, with annual precipitation ranging from

1000 to 1400 mm and annual mean temperature ranging from -2.5 to 0.0 °C (Boucher et al. 2003).

Forest stands are dominated by black spruce (Picea mariana) or by a combination of black spruce

and balsam fir (Abies balsamea). We sampled 36 populations of red-backed voles along a 150-km

latitudinal gradient. To insure that we were dealing with distinct populations, individual sites were

at least 2 km apart, which is at least twice the dispersal distance for small mammals, i.e. 10 m to 1

km (Sutherland et al. 2000).

At each of the 36 sites, 14 aluminum Sherman™ live traps (7.7 × 8.8 × 23.0 cm; H. B. Sherman

Traps, Inc., Tallahassee, Florida) were placed every 10 m along two 60-m parallel transects that

were 30 m apart, i.e. seven traps per transect. Two trapping sessions were conducted: 12 to 22 July

2004 and 12 to 22 August 2004. Trapping at individual sites lasted three days, and traps were set

and inspected at dawn. Peanut butter, an apple slice and a cotton ball were provided in equal

quantities inside each trap. Conifer branches were placed on the top of traps to provide some

protection against the rain as well as convective and radiative heat loss. In the third objective, we

investigated which factors were related to bot fly abundance in hosts. As this variable is

determined through the dissections of dead small mammals, we increased our sampling effort of

dead individuals by placing one pitfall trap (23 × 23 × 24 cm) 15 m away from one end of each

transect, i.e. two pitfall traps per site.

Sampling of red-backed voles and bot flies

Small mammals captured in both types of traps were identified to species, sexed, weighed (to the

nearest 0.5 g using a 300 g Pesola, Baar, Switzerland), and measured for body length (including

tail length, ± 5 mm using a 200 mm caliper). Small mammals were ear-tagged with a unique tag

number (style 1005-1 from National Band & Tag Company, Newport, KY, USA), before being

released. Gestating and lactating females, as well as females with open vagina, were considered as

59

reproductively active, whereas females with closed vagina were considered as reproductively

inactive.

All individuals were inspected for the presence of bot fly warbles. Warbles are the encapsulated

pocket produced by a vertebrate host in response to the presence of a bot in its subdermal tissue

(Slansky 2007). Warbles are obvious and easily detected during a visual inspection of small

mammals. All bot flies were most likely C. grisea, as it should be the only species present in the

study area (Sabrosky 1986). Individuals captured in pitfalls traps and those that died in live traps

were dissected to determine the abundance of bot fly larvae inside hosts. We searched for larvae

(development stages one to three) under a binocular microscope along the known migratory route

in their hosts, i.e. trachea, thorax, abdominal cavity and reproductive organs (Hunter & Webster

1973, Gingrich 1981). Subcutaneous tissues were also inspected to detect uncommon infection

sites. The dissections confirmed the validity of our assessment of parasitism based on warbles. Of

the 137 dissected red-backed voles, 51 had warble pores. Among those individuals, 96% also had

at least one internal larva. For voles that did not have warble pores, only 2% were infected.

Attributes of sampling site

To evaluate the link between infection risk and habitat complexity, we considered three habitat

attributes: tree basal area (m²/ha), sapling basal area (m²/ha), and volume of coarse woody debris

(i.e., CWD volume; m³/ha). At each of the 36 sites, tree basal area was estimated from the

diameter at breast height (dbh) of all trees and snags (i.e., > 9 cm dbh) found inside three circular

plots of 400 m² (total area: 1200 m²), one at the centre of the trapping site, and the other two

located 60 m from the central plot. Sapling basal area was estimated from the dbh of the saplings

(i.e., ≤ 9 cm dbh and ≥ 1.30 m height) in two linear plots (2 × 20 m) located within each of the

three 400 m² circular plots. CWD volume was the total volume of coarse woody debris (> 9 cm)

lying on the ground, inside the 400-m² central plot. The length and diameter of each portion of

coarse woody debris located inside the sampling plot were used to estimate the volume of debris

based on the equation of the frustum of a cone. The air temperature was recorded at the trapping

site each day of trap inspection.

60

Statistical analysis

We determined which combination of factors could best explain the following dependent

variables: (1) the probability of survival of red-backed voles following overnight trap confinement,

(2) the reproductive activity of female voles, (3) the infection probability of voles by bot flies, (4)

the abundance of bot flies inside individual hosts, and (5) the change in the size of vole

populations between mid-July and mid-August.

First, factors related to the probability of survival of red-backed voles in live traps were tested

using generalized linear mixed effects models (GLMM) with binomial distributions. The dataset

included all red-backed voles, dead or alive, captured in live traps. The dependent variable was a

binary variable: individual dead (coded as zero) or alive (coded as one) when the trap was

inspected. Survival of recaptured individuals was determined from the last event of capture.

Because patterns of survival in live traps of individuals captured at a given site may not be

independent, sites were included as a random effect in GLMM. The following independent

variables were considered in the analysis: bot fly, sex, body condition, vole abundance in August,

latitude, and air temperature.

The presence of bot flies was a binary variable coded as one when the red-backed vole had at least

one bot fly warble and as zero otherwise. Sex was a dummy variable with the reference category

being female. Following Schulte-Hostedde et al. (2005), body condition of red-backed voles was

evaluated using an index calculated from the residuals of an ordinary least square regression of

body mass as a function of body length (R² = 0.73, p < 0.001). Body mass of voles was corrected

to account for mass of bot fly larvae (Smith 1977, Cramer & Cameron 2006). Latitude was

included to take into account potential spatial trends along the North-South axis of the study area.

Air temperature at individual sites was included in the analysis to control for potential temperature

effects on vole survival.

Vole abundance was calculated separately for July and for August as the minimum number alive

(MNA), the number of different individuals captured in live traps at a site per 100 trap-nights.

Following Beauvais & Buskirk (1999), i.e. traps occupied by red-backed voles or traps where no

individual were captured were counted as one trap-night, traps occupied by recaptured individuals,

by species other than the red-backed vole, or traps that did not function were not counted, and

61

sprung traps were counted as one half a trap-night. We restricted the analyses of host-parasite

relationships to our August samples because we did not detect any bot fly infections in July.

Second, factors related to the reproductive activity of females were modeled using GLMM with

binomial distributions and site as a random effect. The dataset included all female red-backed

voles captured in live and pitfall traps. The dependent variable was binary and considered whether

a given female was reproductively active (coded as one) or inactive (coded as zero). Three

independent variables were considered in the analyses: infection by bot flies, vole abundance in

August, and latitude.

Third, factors related to the probability of infection of red-backed voles by bot flies were modeled

using GLMM with binomial distributions, and site as a random effect. The dataset included all

red-backed voles, dead or alive, captured in live traps. The binary dependent variable considered

the presence (coded as one) or absence (coded as zero) of at least one bot fly warble in host body.

Fourth, factors related to the abundance of bot fly larvae in red-backed voles were modeled using

GLMM with Poisson distributions and site as a random effect. The dataset included all red-backed

voles captured in pitfall traps, as well as those that had died in live traps. The third and fourth

analyses both accounted for seven independent variables: sex, body condition, vole abundance in

August, CWD volume, tree basal area, sapling basal area, and latitude.

Fifth, factors related to the growth rate of red-backed vole populations between mid-July and mid-

August were modeled using GLM with Gaussian distributions. The dataset included all red-backed

voles captured in live traps. Following Burns et al. (2005), the weekly growth rate of local

populations was calculated as vole abundance in August minus vole abundance in July, divided by

four weeks. Four independent variables were considered in the analyses: CWD volume, tree basal

area, sapling basal area, and the proportion of infected hosts in a population (i.e., the prevalence,

sensu Slansky 2007).

In all model comparisons, independent variables formed normal or uniform distributions, except

CWD volume that formed a skewed distribution and prevalence that presented a large number of

values near 0 and 1. Therefore, CWD volume was square-root transformed whereas prevalence

was arcsine of square-root transformed (Legendre & Legendre 1998, Zar 1999). Correlations

between pairs of independent variables were all <0.60.

62

Sets of candidate models were built for each of the five dependent variables. These models were

compared on the basis of fit and parsimony using the Bayesian information criteria, BIC (Burnham

& Anderson 2002). Weights of evidence (wi) were used to evaluate the probability that a given

model was the “best” predictor among the set of candidate models (Burnham & Anderson 2002).

The performance of the top models was evaluated using R² and pseudo-R2 for models estimated

based on Gaussian and Poisson distributions, respectively, and using receiver-operating

characteristic (ROC) curves for models based on binomial distributions. The ROC is a threshold-

independent method of validation, leading to an objective evaluation of statistical models. In

contrast, most of the other validation methods of logistic models are based on a confusion matrix,

which classifies cases as true positives, false positives, true negatives, and false negatives based on

an arbitrary threshold of 0.5 (Fielding & Bell 1997, Manel et al. 2001). The ROC evaluates the

performance of a model for all possible threshold values. The area under the ROC curve (AUC)

thus represents an accurate measure of model performance. A value of 1.0 would represent a

perfect model (e.g. perfect discrimination between dead and alive individuals) whereas a value of

0.5 would indicate no significant difference between the two states (Fielding & Bell 1997, Manel

et al. 2001, Guénette & Villard 2005). AUC < 0.70 indicates “poor” prediction accuracy, AUC ≥

0.70 indicates “fair” prediction accuracy, AUC ≥ 0.80 indicates “good” prediction accuracy, and

AUC ≥ 0.90 indicates “excellent” prediction accuracy (Hosmer & Lemeshow 2000). Analyses

were conducted with the R software, version 2.4.1 (R Development Core Team 2006), with

GLMM built using package lme4 (Bates & Sarkar 2006).

Results

We captured 197 red-backed voles over 1611 trap-nights in July 2004 and 341 red-backed voles

over 1585 trap-nights in August 2004. Red-backed voles were about 18 times more abundant than

any other rodent species captured at our study sites. Bot flies seemed to mainly infect red-backed

voles because only two other species were infected: a single heather vole (Phenacomys

intermedius) and a single southern bog lemming (Synaptomys cooperi).

63

Bot fly infection and red-backed vole’s survival and reproduction

Two of the 30 models predicting the survival probability of the 341 red-backed voles captured in

live traps had ∆BIC < 2 (Appendix 1), and thus were considered the best approximating models

for these data (Burnham & Anderson 2002). In the top-scoring model (S1), vole survival in live

traps decreased with bot fly infection, was greater in males than females, and increased with body

condition (Table 4). The second model (S2) also included a positive effect of air temperature on

survival. The influence of bot fly infection on survival was independent from air temperature

effects, as the consideration of the “Bot fly × Air temperature” interaction yielded a model with

relatively poor empirical support (Model S14: wi = 0.01). The two top-scoring models (S1 and S2)

were considered as “good” predictors of vole survival in live traps, with AUC ≥ 0.80 (Table 4).

We found that 33% of dead red-backed voles were infected, whereas only 15% of live red-backed

voles were infected (Fig. 6). The effect of sex was such that infected males had similar survival

probability as uninfected females (Fig. 7). We did not find evidence of density-dependent effects

on vole survival in live traps; the best model that included vole abundance in August (S9;

Appendix 1) had a weight of evidence of only 0.02.

We evaluated the reproductive activity of 160 females, 15 of which were captured in pitfall traps.

Reproductive females represented 58% of this sample. Model comparisons indicated that the null

model, the model including only the intercept, was among the top two scoring models (Appendix

1). The null model had a weight of evidence of 0.40. Also, the best model that included bot fly

effects (i.e., R4) had a weight of evidence of only 0.05. Therefore, the reproductive activity of red-

backed vole females did not seem to be influenced by bot fly infection.

Distance traveled by recaptured red-backed voles according to infection status

An analysis of the 39 red-backed voles that were captured twice revealed that infected individuals

were recaptured closer to the first trapping location (median distance [range] = 5 m [0 – 36 m]; n =

10) than uninfected individuals (10 m [0 – 67 m]; n = 29; Mann-Whitney test: p = 0.068).

Recaptured red-backed voles that had died during the second trapping event tended to have moved

further from the initial trapping location (20 m [0 – 67 m]; n = 19) than red-backed voles that were

recaptured and remained alive overnight (10 m [0 – 42 m]; n = 20; p = 0.11).

64

Link between infection risk, habitat attributes, life-history traits, and host abundance

Among the 341 individuals that were captured in live traps in August, 22% were infected. Only

one of the 11 candidate models predicting the probability of bot fly infection had ∆BIC < 2

(Appendix 1). The top model (PI1) showed increasing probability of infection with decreasing

vole abundance in August and decreasing latitude (weight of evidence = 0.90; Table 4; Fig 3a). An

average of 2.12 ± 1.80 larvae (range: 1−13 larvae) were found inside the 51 infected and dead

voles. Only a single individual had 13 larvae while all other individuals had between one and

seven larvae. Model comparisons conducted with and without the outlier (i.e., individual with 13

larvae) were similar. We thus present only the model without the outlier (Appendix 1). As for the

probability of infection, the only model with a ∆BIC < 2 showed that the abundance of bot fly

larvae inside individual red-backed voles decreased with vole abundance in August (weight of

evidence of 0.81; Table 4; Fig. 8b).

The top model (PI1) predicting the probability of infection of red-backed voles by bot flies was

considered as “good”, with AUC = 0.83 (Table 4). The top model (A1) predicting the abundance

of bot fly larvae in red-backed voles had a pseudo-R² of 0.19 (Table 4). The consideration of life

history traits (i.e. sex and body condition) and habitat attributes (i.e. CWD volume, basal area of

trees and snags, and basal area of saplings) did not help explain the probability of infection or the

abundance of bot flies in red-backed voles, as any models that included these variables had a

weight of evidence ≤ 0.01 (Appendix 1).

Bot fly prevalence and growth rate of red-backed vole populations over the summer

Local abundance varied broadly among the 36 populations of red-backed voles sampled over the

summer, with vole abundance ranging from zero (in five sites) to 41 individuals/100 trap-nights in

July and from five to 48 in August (Fig. 9). From mid-July to mid-August, individual vole

populations grew at an average rate of 2.3 ± 2.7 individuals/week (range: –3.5 to 7.6

individuals/week).

One of the 14 models predicting population growth rate between July and August had ∆BIC < 2

(Appendix 1). The top model (PGR1) showed increasing population growth rate with decreasing

bot fly prevalence in August and decreasing basal area of saplings. Model PGR1 also indicated

65

that population growth rate increased with the basal area of trees and snags (Table 4; Fig. 10). The

top scoring model (PGR1) was an efficient predictor of the growth rate of red-backed vole

populations, with an adjusted R² = 0.54 (Table 4).

Discussion

Our field experiment supports the growing idea that parasites can play a major role in population

dynamics of their hosts (Møller 2005, Lively 2006). We found that bot flies reduce survival of red-

backed voles in live traps. Bot flies seems to be able to infect large portions of vole populations

because bot flies are insensitive to the two life-history traits we studied, namely sex and body

condition, and to three habitats variables we studied, namely tree basal area, sapling basal area,

and CWD volume. Bot flies seem to have negative consequences on population growth rate of red-

backed voles over the summer because parasite prevalence was negatively linked to population

growth. We found little evidence of density dependence in parasite-vole interactions, indicating

that bot flies appear more likely to limit than regulate red-backed vole populations.

Consequences of bot fly infection on red-backed vole survival and reproduction

Model selection based on 341 red-backed voles captured from 36 populations indicated that the

probability of survival of red-backed voles in live traps was lower among individuals infected by

bot flies. Similarly, Boonstra et al. (1980) reported lower survival among infected Townsend’s

voles captured in four sites in British Columbia, Canada. Our finding contrasts, however, with

results reported for Peromyscus (e.g., Clark & Kaufman 1990, Burns et al. 2005, Jaffe et al. 2005,

Cramer & Cameron 2006). For example, Munger & Karasov (1991) found that survival was higher

for infected individuals in a field experiment on 125 white-footed mice in Wisconsin, USA.

In the past, the contrasting results on the survival of Peromyscus and voles were attributed to the

different host-parasite coevolution histories. Peromyscus was considered the primary host of bot

flies whereas voles were considered secondary hosts (Boonstra et al. 1980). According to the

coevolution hypothesis, reductions in host survival should be lowest when host-parasite

coevolution history is strongest and vice versa (Hunter & Webster 1973, Boonstra et al. 1980).

66

Parasites should not kill their hosts because their own survival depends on host persistence

(Slansky 2007). Warble site-specificity, an index of the coevolution strength represented by the

percentage of warbles located in the inguinal regions (Hunter & Webster 1973), was lower in

Townsend’s vole (37%; Boonstra et al. 1980) than in Peromyscus (96%; Hunter & Webster 1973).

Thus, the weaker effect of bot flies on Peromyscus survival than on vole survival reported in

previous studies were consistent with the predictions of the coevolution hypothesis (Hunter &

Webster 1973, Boonstra et al. 1980). In our study, we found that 96% of warbles were located in

the inguinal regions of red-backed voles. In addition, infection rate of bot flies was higher in the

red-backed vole than in any other rodent species, an observation previously reported in two other

studies conducted in eastern Canada (Bowman 2000, Pearce & Venier 2005a). These results

suggest that the red-backed vole represented the primary host of bot flies in our study area, and

potentially in northeastern Canada. Bot fly infection, however, was associated with a decreased

probability of survival of red-backed voles in live traps, which could indicate that the infection of

the red-backed voles by bot flies is fairly recent on an evolutionary scale, or that the coevolution

hypothesis does not hold in our study area.

Several authors have criticized the coevolution hypothesis, suggesting that the positive relationship

between bot fly infection and host survival is, in part, an artifact of mark-recapture studies (e.g.,

Wecker 1962; Hunter et al. 1972). Gravid bot flies lay their eggs near burrow entrances, increasing

the rate of infection among resident animals (Catts 1982, Slansky 2007). Resident animals also

have a higher probability of recapture than transient animals, inflating estimates of survival among

highly infected residents (Wecker 1962; Hunter et al. 1972). More recently, Burns et al. (2005)

showed that bot flies enhanced survival of white-footed mice captured in 12 sites in New-York

and Connecticut, USA. They suggested that bot flies would induce a shift in host life-history traits,

from energy allocated to reproduction, to energy allocated to survival. Our analysis of the distance

traveled by recaptured red-backed voles would tend to support the artifact hypothesis, where bias

in trapping probabilities among resident and transient voles would make it more likely to conclude

that infected individuals have relatively high survival probabilities simply because they travel less

and are thus more likely to be recaptured (Wecker 1962, Hunter et al. 1972). Consistently, we

found that, compared to infected voles, uninfected individuals were recaptured farther from the

trap where they were initially captured, while survival in live traps tended to be higher for

individuals recaptured closer to the first trapping site. Our experimental approach, however,

67

should have bypassed some of the potential problems associated with improper design (or faulty

assumptions) for the evaluation of survival from mark-recapture methods. We investigated the

probability of overnight survival during live trapping, and found that survival decreased with bot

fly infection.

Model selection based on the reproductive activity of 160 females indicated that the reproductive

activity of red-backed vole females was not influenced by bot fly infection. In contrast, Burns et

al. (2005) found that infected females had disproportionately high rates of reproductive activity

compared with uninfected females. In an experiment with nest boxes, they found, however, that

infected females produced fewer litters than uninfected ones. They concluded that external

examination of breeding condition was a misleading indicator of reproductive success.

Abiotic factors influencing vole survival

To obtain reliable information on the potential influence of parasitism on host survival by looking

at overnight mortality in live traps, infected hosts should not spend more time in traps than

uninfected hosts, and both groups should experience similar air temperature during confinement.

At six experimental sites, we contrasted the actual time spent in traps between infected and

uninfected red-backed voles. For this experiment, each trap was equipped with a timer triggered by

closing the trap’s door. Due to logistical difficulties in the field, we conducted the experiment with

only 56 traps during four nights, i.e. 224 trap-nights. We found that the period of confinement in

traps tended to be shorter for infected (mean ± S.D.: 7:04 ± 1:57 hours; n = 4) than uninfected red-

backed voles (11:53 ± 6:02 hours; n = 11). Thus infected red-backed voles did not seem to spend

more time in traps than uninfected red-backed voles. In addition, the effect of temperature was

additive to other variables influencing survival and did not interact with the presence of bot fly

(Appendix 1; Table 4). This result indicates that both groups experienced similar air temperature

during confinement.

The major cause of mortality in our live traps was probably hypothermia because our study area

had a cold maritime climate (annual precipitations ranged from 1000 to 1400 mm and annual mean

temperature ranged from -2.5 to 0.0 C, Boucher et al. 2003). Air temperature measured at

individual sites ranged from 9 to 28 °C in July and from 4 to 24 °C in August. Temperature have

68

been found to influence the survival of small mammals under natural conditions (Jackson et al.

2001), but most certainly to a lesser extent than in traps. The experiment of confinement, however,

shielded individuals from other sources of energy demand imposed to free individuals, i.e.

locomotion (Chappell et al. 2004), competition (Abramsky et al. 2000), or anti-predation behavior

(Ramos-Jiliberto et al. 2002).

Infection patterns of bot flies in red-backed voles

The abundance of conspecifics was the most important biotic factor predicting infection risk and

the abundance of bot flies in red-backed voles: both decreased with increasing vole abundance in

August. This result contrasts with previous studies that predicted an increase of prevalence of bot

flies with the abundance of red-backed voles (Bowman 2000, Pearce & Venier 2005a).

Nevertheless, our study is not the first to report a negative relationship between host abundance

and the probability of infection or abundance of parasites within hosts. For example, Stanko et al.

(2006) found a negative relationship for 23 associations of fleas and mammals.

At least three mechanisms could explain this negative relationship. First, it could be due to the one

year time-delay between population cycles of host and parasite (Burns et al. 2005). The pupation

of bot flies between larval and adult stages lasts about one year, from late summer or early autumn

to late spring or early summer (Catts 1982, Slansky 2007), most bot fly larvae may find a suitable

host during peak densities of small mammals, and emerge as adults the following year. The

density of adult bot flies would then peak during the following year when the density of small

mammals might be in the decline, resulting in a negative relationship between vole and bot fly

abundance (Boonstra & Krebs 2006). This hypothesis is supported by the yearly fluctuations in bot

fly infections. Infection rate was 22% in 2004, whereas it was 40% in 2005 (calculated from a

sample of 55 red-backed voles captured at 10 sites from 27 to 31 July), and 26% in 2006 (90 red-

backed voles captured at 35 sites from 29 July to 7 August) (K. Poitras, D. Fortin, D. W. Morris,

unpublished data). Second, transient voles might make up a larger proportion of the population

when vole densities are high and habitats are “saturated”; these individuals might have a lower

probability of infection because they do not use burrows and movement pathways (Stanko et al.

2006) where bot flies generally lay their eggs (Catts 1982, Slansky 2007). Third, individual

69

probability of infection could be reduced at high-population density because of the dilution of

infection risk among all hosts (Hart et al. 1992, Hart 1994, Sorci et al. 1997).

We also found that bot flies are able to infect large portions of host populations. Sex and body

condition of red-backed voles were not significant predictors of the probability of infection and the

abundance of bot flies in red-backed voles, consistent with observations in Peromyscus (Clark &

Kaufman 1990, Kollars 1995, Galindo-Leal 1997). Cramer and Cameron (2006) found, however,

that infected P. leucopus weighed significantly more when they were infected, regardless of body

length and after correction for parasite weight. Habitat complexity (i.e., tree basal area, sapling

basal area and CWD volume) was not a significant predictor of infection patterns when vole

abundance was taken into account. Although other studies found that infection patterns of bot flies

in small mammals appeared to be related to habitat attributes (Hensley 1976, Wolf & Batzli 2001,

Cockle & Richardson 2003), we identified host abundance in August as the main factor driving

both the probability of infection and the abundance of bot flies.

Consequences of bot fly infection on growth rate of red-backed vole populations over the summer

Effects of bot flies at the individual level seemed to have repercussions at the population level.

The growth rate of red-backed vole populations between July and August was negatively related to

the proportion of infected voles within the population in August. The impact of bot flies on growth

rates of host populations should be more important every two years, i.e. when population density

of bot flies increases (Burns et al. 2005). The negative impacts of bot flies that we identified on the

population growth rate of red-backed voles between mid-July and mid-August were estimated in

2004 when bot fly densities were low (prevalence 22% in 2004, 40% in 2005, and 26% in 2006)

and might underestimate the negative impacts in years of high bot fly density. Regardless, we

demonstrated that bot flies can impact the growth of host populations during the course of the

summer, which constitutes a critical part of the annual cycle of small mammals because the peak

of their reproductive activity occurs in summer (Falls et al. 2007). Bot fly parasitism appears more

likely to act as a limiting than regulating factor, and thus might contribute to the instability of red-

backed vole populations in the boreal forest of eastern Canada.

70

Table 4. Parameter estimates, standard errors, and p-values of factors contained in all models with

∆BIC < 2 for (1) the probability of survival in live traps of 341 red-backed voles captured at 36

sites in the boreal forest of the Côte-Nord region of Québec, Canada; (2) the probability of bot fly

infection for red-backed voles; (3) the abundance of bot fly larvae in 50 dead red-backed voles;

and (4) the growth of 36 red-backed vole populations between July and August. Model selections

1 to 3 were based on GLMM with sites as a random effect and model selection 4 was based on

GLM. Parameter estimates of the top-ranking models are presented, along with model fit (AUC,

R2 or pseudo-R2) and Akaike weight (wi). Complete model selections are presented in Appendix 1.

Model Variable ß se p Model fit wi

Probability of survival of red-backed voles in live traps

S1 Intercept 0.47 0.21 0.024 0.8b 0.35

Bot fly –1.04 0.34 0.002

Sexa 0.88 0.28 0.002

Body condition 0.39 0.05 <0.001

S2 Intercept -0.65 0.56 0.24 0.85c 0.21

Bot fly -1.01 0.34 0.003

Sexa 0.90 0.28 0.001

Body condition 0.41 0.005 <0.001

Air temperature 0.07 0.03 0.03

Probability of infection of individual red-backed voles by bot flies

PI1 Intercept 200.49 35.95 <0.001 0.83c 0.90

Vole abundance in August –0.08 0.02 <0.001

71

Latitude –3.95 0.71 <0.001

Abundance of bot flies in dead red-backed voles

A1 Intercept 1.6 0.22 <0.001 0.83c 0.81

Vole abundance in August –0.03 0.01 <0.001

Population growth rate of red-backed voles between mid-July and mid-August

PGR1 Intercept 5.09 1.55 0.002 0.54d 0.59

Prevalence –4.42 0.92 <0.001

Tree basal area 0.09 0.04 0.052

Sapling basal area –0.64 0.2 0.003

aFemale is the reference category

bAUC value

cPseudo-R2 value

dAdjusted-R2 value

72

Figure 6. Number of a. female and b. male red-backed voles (Myodes gapperi) that were captured

alive (white boxes) and dead (grey boxes) according to their infection status with bot fly

(Cuterebra spp.) parasites. The 341 individuals were sampled at 36 sites in the boreal forest of the

Côte-Nord region of Québec, Canada, from 12 to 22 July and from 12 to 22 August 2004.

Frequency

20

40

60

80

100

Infected Uninfected Infected Uninfected

a. Female b. Male

73

-10 0 10 20 30

0.0

0.2

0.4

0.6

0.8

1.0

Body condition

Probability of survival

uninfected malesuninifected fem.infected malesinfected fem.

Figure 7. Probability of survival in live traps of red-backed voles as a function of bot fly infection,

host sex, and body condition. Open circles represent raw data and curves represent the binomial fit

according to model S1 (see Table 4 for estimates and standard errors and Appendix 1 for an

explanation of model S1).

74

0 10 20 30 40 50

0.0

0.2

0.4

0.6

0.8

1.0Probability of infection

Latitude

LowMeanHigh

a

0 10 20 30 40 50

0

2

4

6

8

10

12

14

Vole abundance

Abundance of botflies per host

b

Figure 8. a) Probability of bot fly infections as a function of vole abundance (no per 100 trap

nights) in August based on low, mean or high latitude and b) abundance of bot flies per host. Open

circles represent raw data and the curve represent the fits according to model a. PI1 and b. A1 (see

Table 4 for estimates and standard errors and Appendix 1 for an explanation of model PI1 and

A1). In panel b the inclusion or exclusion of the outlier, i.e. 13 larvae in a host, yielded similar

results.

75

0 10 20 30 40 50

0

10

20

30

40

50

Vole abundance in July

Vole abundance in August

Figure 9. Relationship between the abundance of red-backed voles (number per 100 trap nights) in

August and in July at 36 individuals sites sampled in the boreal forest of the Côte-Nord region of

Québec, Canada. The equation of the linear regression is: y = 17 .8 + 0.5 x, p = 0.002.

76

0.0 0.2 0.4 0.6 0.8 1.0

-4

-2

0

2

4

6

8

Prevalence of infection

Short-term population growth rate

high tree, low sap.low tree, low sap.high tree, high sap.low tree, high sap.

Figure 10. Short-term growth rate of red-backed vole populations between July and August 2004.

Open circles represent raw data and lines represent the linear fit of the model PGR1 (see Table 4

for estimates and standard errors). Tree and sapling basal areas were categorized as low, i.e. first

quartile where first quartile was 17.9 m2/ha for trees and 2.3 m2/ha for saplings and high, i.e. third

quartile where third quartile was 31.0 m2/ha for trees and 5.5 m2/ha for saplings.

77

App

endi

ce 1

. Mod

el c

ompa

riso

ns b

ased

on

BIC

to

pred

ict

(1)

the

prob

abil

ity

of s

urvi

val

of r

ed-b

acke

d vo

les

in l

ive

trap

s, (

2) t

he r

epro

duct

ive

acti

vity

of

fem

ale

vole

s (i

.e.,

repr

oduc

tive

or

not)

, (3

) th

e pr

obab

ilit

y of

inf

ecti

on o

f vo

les

by b

ot f

lies

, (4

) th

e ab

unda

nce

of b

ot f

lies

ins

ide

indi

vidu

al v

oles

, an

d (5

) th

e po

pula

tion

gro

wth

rat

e of

vol

es b

etw

een

July

and

Aug

ust

2004

. M

odel

s w

ere

calc

ulat

ed u

sing

dat

a on

341

red

-

back

ed v

oles

sam

pled

at

36 s

ites

in

the

bore

al f

ores

t of

the

Côt

e-N

ord

regi

on o

f Q

uébe

c, C

anad

a. T

he n

umbe

r of

est

imat

ed p

aram

eter

s K

incl

uded

in

the

mod

el (

i.e.,

num

ber

of i

ndep

ende

nt v

aria

bles

+ t

he i

nter

cept

, fo

r al

l fi

ve m

odel

sel

ecti

ons,

+ t

he m

ixed

eff

ect,

for

mod

el

sele

ctio

ns o

ne t

o fo

ur),

the

∆B

IC (

i.e.,

the

diff

eren

ce b

etw

een

a gi

ven

mod

el's

BIC

and

the

low

est

BIC

am

ong

all

cand

idat

e m

odel

s),

and

the

wei

ght o

f ev

iden

ce (

wi)

are

prov

ided

for

eac

h m

odel

.

Mod

el S

truc

ture

M

odel

K

∆B

IC

wi

1. Probab

ility of survival of red-backed voles in

live traps (n = 341)

+ I

nter

cept

– B

ot f

ly +

Sex

* +

Bod

y co

ndit

ion

S1

5 0.

00

0.35

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n +

Air

tem

pera

ture

S

2 6

1.05

0.

21

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n –

Bot

fly

× S

ex

S3

6 2.

31

0.11

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n –

Bot

fly

× B

ody

cond

itio

n S

4 6

3.37

0.

06

+ I

nter

cept

+ S

ex +

Bod

y co

ndit

ion

S5

4 4.

07

0.05

+ I

nter

cept

+ S

ex +

Bod

y co

ndit

ion

+ A

ir te

mpe

ratu

re

S6

5 4.

31

0.04

+ I

nter

cept

– B

ot f

ly +

Bod

y co

ndit

ion

S7

4 4.

44

0.04

78

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n –

Lat

itud

e S

8 6

4.91

0.

03

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n –

Vol

e ab

unda

nce

in A

ugus

t S

9 6

5.28

0.

02

+ I

nter

cept

– B

ot f

ly +

Bod

y co

ndit

ion

+ A

ir te

mpe

ratu

re

S10

5

5.93

0.

02

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n +

Air

tem

pera

ture

– B

ody

cond

itio

n ×

Air

tem

pera

ture

S

11

7 6.

13

0.02

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n +

Air

tem

pera

ture

– V

ole

abun

danc

e in

Aug

ust

S12

7

6.49

0.

01

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n –

Vol

e ab

unda

nce

in A

ugus

t – B

ody

cond

itio

n ×

Vol

e ab

unda

nce

in A

ugus

t S

13

7 6.

54

0.01

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n +

Air

tem

pera

ture

+ B

ot f

ly ×

Air

tem

pera

ture

S

14

7 6.

82

0.01

+ I

nter

cept

– B

ot f

ly +

Bod

y co

ndit

ion

+ B

ot f

ly ×

Bod

y co

ndit

ion

S15

5

8.22

0.

01

+ I

nter

cept

+ B

ody

cond

itio

n S

16

3 9.

32

0.00

+ I

nter

cept

+ S

ex +

Bod

y co

ndit

ion

+ A

ir te

mpe

ratu

re –

Vol

e ab

unda

nce

in A

ugus

t S

17

6 10

.14

0.00

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n –

Vol

e ab

unda

nce

in A

ugus

t – S

ex ×

Vol

e ab

unda

nce

in A

ugus

t S

18

7 10

.23

0.00

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n –

Lat

itud

e+ I

nter

cept

– B

ot f

ly ×

Lat

itud

e S

19

7 10

.26

0.00

+ I

nter

cept

– B

ot f

ly +

Sex

+ B

ody

cond

itio

n –

Vol

e ab

unda

nce

in A

ugus

t+ I

nter

cept

– B

ot f

ly ×

Vol

e ab

unda

nce

in A

ugus

t S

20

7 10

.46

0.00

+ I

nter

cept

– B

ot f

ly

S21

3

106.

02 0

.00

79

+ I

nter

cept

– B

ot f

ly +

Sex

S

22

4 10

6.48

0.0

0

+ I

nter

cept

– B

ot f

ly +

Air

tem

pera

ture

S

23

4 11

0.32

0.0

0

+ I

nter

cept

– B

ot f

ly +

Sex

+ A

ir te

mpe

ratu

re

S24

5

110.

59 0

.00

+ I

nter

cept

– B

ot f

ly –

Vol

e ab

unda

nce

in A

ugus

t S

25

4 11

1.05

0.0

0

+ I

nter

cept

S

26

2 11

6.22

0.0

0

+ I

nter

cept

+ S

ex

S27

3

116.

29 0

.00

+ I

nter

cept

+ A

ir te

mpe

ratu

re

S28

3

120.

4 0.

00

+ I

nter

cept

– V

ole

abun

danc

e in

Aug

ust

S29

3

122.

06 0

.00

+ I

nter

cept

+ A

ir te

mpe

ratu

re –

Vol

e ab

unda

nce

in A

ugus

t S

30

4 12

6.23

0.0

0

2. Reproductive activity of females (n = 160)

– In

terc

ept +

Lat

itud

e R

1 3

0.00

0.

45

– In

terc

ept

R2

2 0.

25

0.39

– In

terc

ept –

Vol

e ab

unda

nce

in A

ugus

t R

3 3

4.36

0.

05

– In

terc

ept -

Bot

fly

R

4 3

4.43

0.

05

80

– In

terc

ept -

Bot

fly

+ L

atit

ude

R5

4 5.

08

0.04

– In

terc

ept -

Bot

fly

+ L

atit

ude

+ B

ot f

ly ×

Lat

itud

e R

6 5

7.61

0.

01

– In

terc

ept +

Bot

fly

– V

ole

abun

danc

e in

Aug

ust

R7

4 7.

94

0.01

– In

terc

ept +

Bot

fly

+ L

atit

ude

– V

ole

abun

danc

e in

Aug

ust

R8

5 9.

44

0.00

– In

terc

ept +

Bot

fly

– V

ole

abun

danc

e in

Aug

ust +

Bot

fly

× V

ole

abun

danc

e in

Aug

ust

R9

5 13

.02

0.00

3. Probab

ility of in

fection of individual red-backed voles by bot flies (n = 341)

– In

terc

ept –

Lat

itud

e –

Vol

e ab

unda

nce

in A

ugus

t P

I1

4 0.

00

0.90

– In

terc

ept –

Lat

itud

e –

Vol

e ab

unda

nce

in A

ugus

t – L

atit

ude

× V

ole

abun

danc

e in

Aug

ust

PI2

5

4.71

0.

09

– In

terc

ept –

Lat

itud

e –

Vol

e ab

unda

nce

in A

ugus

t + S

ex –

Bod

y co

ndit

ion

PI3

6

9.33

0.

01

– In

terc

ept –

Lat

itud

e P

I4

3 13

.55

0.00

– In

terc

ept –

Lat

itud

e –

Vol

e ab

unda

nce

in A

ugus

t + S

ex –

Bod

y co

ndit

ion

– C

WD

vol

ume

PI5

7

13.9

3 0.

00

– In

terc

ept –

Lat

itud

e –

Vol

e ab

unda

nce

in A

ugus

t + S

ex –

Bod

y co

ndit

ion

– S

apli

ng b

asal

are

a P

I6

7 14

.00

0.00

– In

terc

ept –

Lat

itud

e –

Vol

e ab

unda

nce

in A

ugus

t + S

ex –

Bod

y co

ndit

ion

– T

ree

basa

l are

a P

I7

7 15

.11

0.00

– In

terc

ept –

Lat

itud

e –

CW

D v

olum

e –

Sap

ling

bas

al a

rea

– T

ree

basa

l are

a P

I8

6 18

.32

0.00

81

– In

terc

ept –

Lat

itud

e +

Sex

– B

ody

cond

itio

n P

I9

5 24

.02

0.00

– In

terc

ept –

Vol

e ab

unda

nce

in A

ugus

t P

I10

3 24

.18

0.00

– In

terc

ept

PI1

1 2

25.8

1 0.

00

4. Abundan

ce of bot flies in

dead red-backed voles (n = 50)

+ I

nter

cept

– V

ole

abun

danc

e in

Aug

ust

A1

3 0.

00

0.81

+ I

nter

cept

+ L

atit

ude

– V

ole

abun

danc

e in

Aug

ust

A2

4 3.

64

0.13

+ I

nter

cept

– L

atit

ude

– V

ole

abun

danc

e in

Aug

ust –

Lat

itud

e ×

Vol

e ab

unda

nce

in A

ugus

t A

3 5

6.94

0.

03

+ I

nter

cept

A

4 2

7.82

0.

02

+ I

nter

cept

+ L

atit

ude

A5

3 10

.42

0.00

+ I

nter

cept

+ L

atit

ude

– V

ole

abun

danc

e in

Aug

ust +

Sex

– B

ody

cond

itio

n A

6 6

10.5

3 0.

00

+ I

nter

cept

– L

atit

ude

– V

ole

abun

danc

e in

Aug

ust +

Sex

– B

ody

cond

itio

n –

Tre

e ba

sal a

rea

A7

7 13

.00

0.00

+ I

nter

cept

– L

atit

ude

– V

ole

abun

danc

e in

Aug

ust +

Sex

– B

ody

cond

itio

n –

Sap

ling

bas

al a

rea

A8

7 13

.20

0.00

+ I

nter

cept

– L

atit

ude

– C

WD

vol

ume

– T

ree

basa

l are

a –

Sap

ling

bas

al a

rea

A9

6 13

.70

0.00

+ I

nter

cept

– L

atit

ude

– V

ole

abun

danc

e in

Aug

ust +

Sex

– B

ody

cond

itio

n –

CW

D v

olum

e A

10

7 13

.93

0.00

82

+ I

nter

cept

– L

atit

ude

+ S

ex –

Bod

y co

ndit

ion

A11

5

17.2

4 0.

00

5. Pop

ulation

growth rate of red-backed voles between m

id-July and m

id-A

ugu

st (n = 36)

+ I

nter

cept

– P

reva

lenc

e +

Tre

e ba

sal a

rea

– S

apli

ng b

asal

are

a P

GR

1 4

0.00

0.

59

+ I

nter

cept

– P

reva

lenc

e +

Tre

e ba

sal a

rea

– S

apli

ng b

asal

are

a –

CW

D v

olum

e P

GR

2 5

2.70

0.

15

+ I

nter

cept

– P

reva

lenc

e +

Tre

e ba

sal a

rea

– S

apli

ng b

asal

are

a –

Lat

itud

e P

GR

3 5

2.70

0.

15

+ I

nter

cept

– P

reva

lenc

e +

Tre

e ba

sal a

rea

– S

apli

ng b

asal

are

a –

CW

D v

olum

e –

Lat

itud

e P

GR

4 6

5.59

0.

04

+ I

nter

cept

– P

reva

lenc

e P

GR

5 2

6.87

0.

02

+ I

nter

cept

– P

reva

lenc

e +

Tre

e ba

sal a

rea

PG

R6

3 7.

25

0.02

+ I

nter

cept

– P

reva

lenc

e –

Lat

itud

e +

Pre

vale

nce

× L

atit

ude

PG

R7

4 8.

46

0.01

+ I

nter

cept

– P

reva

lenc

e –

CW

D v

olum

e P

GR

8 3

8.97

0.

01

+ I

nter

cept

– P

reva

lenc

e –

Tre

e ba

sal a

rea

+ P

reva

lenc

e ×

Tre

e ba

sal a

rea

PG

R9

4 9.

26

0.01

+ I

nter

cept

– P

reva

lenc

e –

CW

D v

olum

e –

Pre

vale

nce

× C

WD

vol

ume

PG

R10

4 9

.87

0.00

+ I

nter

cept

+ T

ree

basa

l are

a –

Sap

ling

bas

al a

rea

– C

WD

vol

ume

– L

atit

ude

PG

R11

5 1

3.84

0.

00

+ I

nter

cept

+ T

ree

basa

l are

a –

Sap

ling

bas

al a

rea

PG

R12

3 1

7.06

0.

00

83

+ I

nter

cept

+ T

ree

basa

l are

a –

Sap

ling

bas

al a

rea

– C

WD

vol

ume

PG

R13

4 1

9.55

0.

00

+ I

nter

cept

P

GR

14 1

24.

15

0.00

* F

emal

e is

the

refe

renc

e ca

tego

ry.

84

Chapitre 3. Deer mice mediate red-backed vole behavior and abundance along a gradient of habitat alteration

Jérôme Lemaître*, Daniel Fortin*, Douglas W. Morris† and Marcel Darveau‡

*NSERC-Université Laval industrial research chair in silviculture and wildlife, Département de Biologie, Université Laval, Québec, QC G1V 0A6, Canada.

†Lakehead Research Chair in Northern Studies, Department of Biology, Lakehead University, Thunder Bay, ON P7B 5E1, Canada.

‡Ducks Unlimited Canada, 710 rue Bouvier, bureau 260, Québec, QC G2J 1C2, Canada.

Article soumis à Oikos en Juillet 2009

85

Résumé

Nous avons étudié les effets de l’exploitation forestière sur les coûts de prédation et de

compétition du campagnol à dos roux (Myodes gapperi) à l’aide d’une expérience sur le

comportement d’approvisionnement de l’espèce. La densité de nourriture à l’abandon (GUD) dans

464 mangeoires réparties dans 29 sites déclinait davantage avec l’altération de l’habitat dans les

sites occupés par la souris sylvestre (Peromyscus maniculatus). Les coûts de prédation, i.e. ∆GUD

entre des mangeoires avec ou sans couvert anti-prédateur, diminuaient avec l’altération des

habitats mésiques (i.e., de plus haute qualité), comme prédit par le principe de protection des

acquis, et ils augmentaient avec l’intensité de l’altération des habitats xériques (i.e., de plus faible

élevée), à cause de la diminution du couvert anti-prédateur. L’abondance des campagnols

diminuait avec l’intensité de l’exploitation, un déclin accentué par la présence des souris. Nous

concluons que la compétition est le principal mécanisme responsable du déclin des populations de

campagnols à dos roux induit par l’exploitation forestière.

86

Abstract

Anthropogenic disturbances can alter habitat quality through changes in competition and

predation. Intra- and inter-specific competition should increase with forest harvesting because

logging reduces food availability for local populations. Predation risk should also increase with the

reduction of antipredatory cover resulting from logging activities. The response of prey to the

increase in predation risk should depend however on habitat quality. The asset-protection principle

states that foragers living in high-quality habitats should invest more in antipredatory behaviors

because the fitness cost of being injured or killed is greater than when they are in poor-quality

habitat. We used experiments on quitting-harvest rate, estimated with giving-up density (GUD), to

test these predictions for red-backed voles (Myodes gapperi) living with deer mice (Peromyscus

maniculatus) in a managed boreal forest. In 2006, we estimated rodent abundance following 3480

trap-nights, and GUDs using 464 artificial food patches, in 29 pairs of natural and logged habitats.

We used principal components analysis on 12 habitat variables to identify gradients of habitat

alteration and moisture. Multilevel statistical modeling revealed that quitting-harvest rate (mean

GUD) increased with the degree of habitat alteration, supporting our prediction. Mean GUD

declined more in sites with deer mice, indicating that deer mice exacerbated competition where

harvesting had been intensive. Mean GUD also declined more in xeric than mesic sites, and red-

backed voles were more abundant in mesic than in xeric sites. Mesic sites therefore provided

habitats of the highest quality for voles. As predicted, predation costs decreased with habitat

alteration in the high-quality mesic sites, and predation costs increased with forest harvesting

intensity in the low-quality xeric sites. Red-backed vole abundance decreased with logging

intensity, a decrease that was even steeper in the presence of deer mice. We conclude that intra-

and inter-specific competition, rather than predation, is the main mechanism responsible for the

decline of red-backed vole populations with forest harvesting.

87

Introduction

Conservation of biodiversity depends on our ability to understand the impacts of habitat alteration

on natural populations (Caro 2007, Morris et al. 2009). A major effect of habitat changes on

population dynamics comes through the modification of adaptive mechanisms of habitat selection

(Kotler et al. 2007, Morris et al. 2009). For example, competition may be affected directly by

changes in food resource availability or indirectly by the colonization of competitor species, while

predation may vary through the alteration of antipredatory cover or predator abundance (Caro

2007).

Foraging behavior can inform on the impact that natural and anthropogenic disturbances can have

on habitat quality, and more specifically on how these disturbances influence competition and

predation costs (e.g., Morris & Davidson 2000, Reed et al. 2005, Morris & Mukherjee 2007b,

Andruskiw et al. 2008). Foragers should exploit a given resource patch until the profit gained from

using the patch equals the foraging costs accumulated in that patch (Brown 1988). Accordingly,

the quitting-harvest rate of a patch (H) is expected to equal the sum of the metabolic (C) predation

(P) and missed opportunity costs of foraging (MOC) i.e. H = C + P + MOC (Brown 1988, Brown

et al. 1992). Quitting-harvest rate can be estimated by measuring the GUD in a patch for which the

initial amount of food is known (Brown 1988, Brown et al. 1992). With the appropriate design,

local differences in C, P or MOC can be assessed based on differences in GUDs. For example,

GUDs can be measured in pairs of experimental resource patches that are identical, with the

exception that one patch is placed in the open (risky) while the other is under protective cover

(Brown et al. 1992). Differences in GUDs should then be proportional to differences in predation

costs (Morris & Davidson 2000, Schmidt et al. 2005, Andruskiw et al. 2008).

Adaptive foraging behavior can inform on relative habitat quality because natural selection should

favor specific decisions, the nature of which varies in a predictive manner depending on the

foraging context. According to foraging theory, the higher is the quitting-harvest rate (i.e., mean

GUD) of a habitat, the higher is the profit in the habitat for the forager (Brown 1988, Brown et al.

1992). In other words, mean GUD is expected to be lower in poor than in rich habitats. Foraging

theory also predicts that predation costs should increase with the reduction of anti-predator cover

88

(Verdolin 2006, While & McArthur 2006, Eccard et al. 2008), a response that should depend

however on habitat quality. Indeed, the asset-protection principle of Clark (1994, with Ydenberg et

al. 1995, and Reed et al. 2005 providing empirical support) states that foragers living in a high-

quality habitats should invest more in antipredatory behaviors because they have more potential

fitness to lose if injured or killed than if they are living in a poor-quality habitat. Predation costs

should therefore decrease more dramatically following disturbance to high-quality (high fitness)

habitats, than they do following disturbance to low-quality habitats. In fact, predation costs could

even increase following the alteration of poor habitats (Clark 1994, Olsson & Molokwu 2007).

The greatest difference in GUDs between habitats is more likely caused by differences in food

availability rather than in predation risk (Olsson & Holmgren 1999, Olsson & Molokwu 2007).

In this study, we use observations on abundance and foraging behavior of red-backed voles

(Myodes gapperi) in a forest ecosystem also occupied by sympatric deer mice (Peromyscus

maniculatus) to determine how forest harvesting influences intra- and inter-specific competition as

well as predation costs. The red-backed vole is an omnivorous species feeding largely on

hypogeous fungi (Orrock & Pagels 2002), which are most abundant in mature and old-growth

forests (Johnson 1996). Also, mesic habitats are generally of higher quality than xeric habitats for

red-backed voles (Morris 1996), probably because of the higher level of soil moisture of mesic

sites that would allow the species satisfying its high metabolic demands of water (Getz 1968,

McShea et al. 2003). The main competitor of the red-backed vole, the omnivorous deer mouse

(Morris 1996), is an efficient forager in disturbed forests (Suzuki & Hayes 2003, Fuller et al.

2004). Quitting-harvest rates (mean GUD) should therefore decrease following forest harvesting

for red-backed voles because logging should reduce the abundance of their main food resources

and increase competition with deer mice.

The numerical response of the red-backed vole to habitat alteration has been studied intensively, in

part because it is the dominant species of the small-mammal community in North American boreal

forests (Pearce & Venier 2005b, Lemaître et al. 2009). Conclusions about the impact of forest

harvesting on red-backed vole populations are inconsistent. Several studies have documented

declines in abundance following logging (e.g., Mills 1995, Sullivan et al. 1999, Darveau et al.

2001, Moses & Boutin 2001), while others detected the opposite relationship (e.g., Hayward et al.

1999, Suzuki & Hayes 2003, Homyack et al. 2005). Behavioral indicators provided by foraging

89

experiments should help us resolve these conflicting results because they specifically address the

biological mechanisms of habitat quality, density, predation, and competition that underly habitat

use.

Material and methods

Study area and experimental design

The study took place in the boreal forest of northeastern Quebec, Canada (N 51°02’12’’, W

69°11’41’’; Fig. 11). Annual precipitation ranges from 1,000 to 1,400 mm, and annual mean

temperature varies from -2.5 to 0.0°C (Boucher et al. 2006). Xeric forest stands are dominated by

black spruce (Picea mariana) whereas mesic forest stands are generally dominated by a

combination of black spruce, balsam fir (Abies balsamea) and white birch (Betula papyrifera)

(Boucher et al. 2006). The study area is characterized by a long fire cycle (mean cycle length =

400 years) that produces uneven forest stands shaped by blowdown and senescence (Bouchard et

al. 2008). The main harvest regime is clearcutting although two partial harvest regimes are also

applied in the study area: cut with protection of high regeneration and soils (CPHRS), which

protects regeneration up to 2 m, and cut with protection of small merchantable stems (CPPTM),

which protects stems < 15 cm diameter at breast height (dbh).

We surveyed 29 paired habitats during summer 2006. Each pair included a harvested stand

adjacent to an uncut one. Stands were harvested with different intensity, which resulted in a tree

basal area ranging from 0 to 6 m²/ha. Tree basal area ranged from 5.5 to 46.5 m²/ha in uncut

stands. Each habitat (cut or uncut stand) covered an area of at least 200 × 300 m, and pairs shared

a boundary of at least 300-m. We pre-selected pairs of habitats using ArcGIS 9.1 (ESRI 2006)

according to three criteria: (1) the stand structure and composition within the forest habitat was

relatively uniform, (2) there were no streams and roads in either habitat, and (3) stand structure

and composition of the harvested habitat prior to harvesting was similar to the natural habitat.

These criteria yielded a total of 116 possible habitat pairs in the study area, and we randomly

selected 29 pairs.

90

Giving-up densities

Each pair of habitats was randomly assigned to one of four sampling sessions (4–6 July, 11–13

July, 29–31 July, and 5–7 August). A session included a “GUD experiment” followed by live-

trapping of small-mammals. We established two 100-m parallel transects positioned 100 m apart

in each habitat. Transects ran perpendicularly from the edge between the two habitats towards

their interior. We positioned two feeding stations per transect, one at 60 m and the other at 90 m

from the edge between the two habitats, i.e. four feeding stations per habitat. Each feeding station

was comprised of two food patches: one was covered with 4 m2 of freshly cut conifer branches to

represent a safe microhabitat, while the other was 2 m away and was not covered to represent a

riskier microhabitat. With the use of similar food patches at such close proximity, individuals

foraging in either patch should have identical metabolic and missed-opportunities costs (Brown

1988, Brown et al. 1992). Predation costs would thus be responsible for the difference in GUD

between pairs of patches (Brown et al. 1992, Schmidt et al. 2005).

We estimated quitting-harvest rates of red-backed voles by measuring GUD in the artificial food

patches. Each identical food patch was composed of a plastic tray (23 × 23 × 18 cm) in which we

added three liters of sifted dry sand and 50 unshelled sunflower seeds thoroughly mixed into the

sand (Kasparian & Millar 2004). A 2.5 cm entrance hole prevented access to larger animals such

as red squirrels (Tamiasciurus hudsonicus). Food patches were baited during nine days before

conducting the 3-day foraging experiment. We sieved the trays each day to collect the uneaten

seeds and recharged them with 50 new seeds. The number of uneaten seeds corresponded to the

GUD of the food patch (Reed et al. 2005).

Small-mammal abundance

We placed 20 Sherman collapsible live traps (7.7 × 8.8 × 23.0 cm; Sherman Traps, Tallahassee,

Fla.) in each habitat to estimate the abundance of small-mammals. Traps were located every 10 m

along the two 100-m parallel transects running into each habitat. Live-trapping lasted three days at

each habitat pair, and was conducted immediately after the GUD experiments. Traps were set and

inspected at dawn. Captured small-mammals were identified to species and marked with a unique

ear tag (style 1005-1 from National Band & Tag, Newport, Ky.) before being released. Red-

91

backed vole abundance was corrected for the number of night-traps at a given site (i.e. corrected

for sampling effort, which corresponded to the sum of the traps occupied by unmarked red-backed

voles plus traps where no individual was captured, Beauvais and Buskirk 1999). Sprung empty

traps counted as one half a trap-night (Beauvais & Buskirk 1999). Traps occupied by recaptured

individuals, or by species other than the red-backed vole, were not included because those traps

were not available to capture an unmarked red-backed vole (Beauvais & Buskirk 1999).

Habitat variables

We characterized vole habitat with 12 variables that represented anti-predator cover (Ucitel et al.

2003, Pearce & Venier 2005b) or reflected food availability for red-backed voles (Orrock &

Pagels 2002, Boonstra & Krebs 2006). Among the 12 variables, five variables also provided a

quantitative measure of habitat alteration (basal area of black spruce trees, ground cover of moss,

bare ground cover, ground cover of fungi, ground cover of fruits available for red-backed voles

[i.e., fruits of vaccinum spp and Cornus Canadensis]), while others distinguished xeric from mesic

sites (basal area of balsam fir trees, basal area of balsam fir saplings, coarse woody debris [CWD]

volume, ground cover lichens [epigeous and fallen arboricol], ground cover of Ledum

groenlandicum, ground cover of Kalmia angustifolia, ground cover of Vaccinum spp.).

We measured tree basal area at each feeding station using a 2× prism (Grosenbaugh 1952). Basal

area of saplings (m2/ha) was estimated from the dbh of saplings (i.e., ≤ 9 cm dbh and ≥ 1.30 m

height) in 10 2 × 2 m quadrats, at each feeding station. Quadrats were located every 2 m along a

randomly oriented transect that was centered on the feeding station. We visually estimated the

percent cover of bare ground, mosses, lichens, fruits, epigeous fungi, and ericaceous species within

two 1 × 1 m quadrats randomly located within 4 m of the feeding stations. CWD volume was

estimated for each habitat using line intercept sampling (Stahl 1997) along 100-m transect lines

running between the two feeding stations at 60 m and the two at 90 m. The diameter of each piece

of CWD (i.e., length ≥ 2 m and diameter at both ends > 9 cm) was recorded at the intersection

point with the line transect. CWD volume per habitat (m3/ha) was calculated using the equation

(De Vries 1973): CWD volume = π² / 8L × ∑(d²), where L = transect length, and d = diameter of

the CWD.

92

Statistical analyses

We used Principal Components Analysis (PCA) to reduce the 12 habitat variables to fewer

independent components. Habitat variables were normalized prior to the analysis with Box-Cox

transformations (Box & Cox 1964) and the loading significance was assessed by bootstrapping

(Peres-Neto et al. 2003). Principal components identified a gradient of habitat alteration, together

with a moisture gradient ranging from xeric to mesic sites (see Results). We included this moisture

gradient in our analyses because red-backed voles are known to prefer mesic over xeric sites

(Morris 1996). Using a gradient, rather than a bivariate comparison between harvested and natural

stands, allowed us to account for the broad habitat heterogeneity that characterized natural stands

in this irregular boreal forest (Bouchard et al. 2008) as well as that created by logging with various

levels of tree retention.

We tested predictions linking competition and predation costs to habitat disturbance while

accounting for the moisture gradient obtained with the PCA, and for the presence of deer mice in

habitat pairs (i.e., a dummy variable coded zero when deer mice were absent and one when they

were present in at least one habitat of the pair). The relationship between mean GUD and the three

independent variables (i.e., logging intensity [PC2], moisture gradient [PC1], and presence of deer

mouse) was evaluated with a Poisson distribution, using GUD per tray per day as the dependent

variable, because mean GUD (unit = number of seeds) was a discrete variable following a Poisson

distribution. ∆GUD, i.e. the difference between GUD of the covered food patch minus GUD of the

open food patch within a feeding station per day, followed a Gaussian distribution, and the

relationship between ∆GUD and the three independent variables was thus evaluated with a

Gaussian distribution. We accounted for the spatial design of the study using multilevel models

(Faraway 2006). Each pair of food patches (open and covered) was nested in a feeding station, the

feeding station was nested in a habitat (harvested or natural), and the habitat was part of a pair (n =

2 food patches × 4 feeding stations × 2 habitats × 29 pairs = 464 food patches). We took into

account the three experiment days using multilevel models with repeated measure (Faraway 2006).

Only pairs where red-backed voles occupied both habitats were included in GUD analyses. We

evaluated the relationship between the number of red-backed voles and habitat principal

components with generalized linear mixed-effect models (GLMM) with Poisson distribution,

using the pair as a random effect to take into account that each harvested stand was compared with

93

its adjacent natural forest stand. We used the Poisson distribution because the number of red-

backed voles was a discrete variable following this distribution, while accounting for the local

sampling effort (log transformed) through an offset.

We built the global model for each of the three dependent variables, i.e. mean GUD, ∆GUD, and

red-backed vole abundance. We then determined whether the least significant variable or

interaction in the global model was necessary to the overall model fit using the log-likelihood ratio

test (α = 0.05) between the global model (n variables) and the model with n-1 variables (Hosmer &

Lemeshow 2000). We continued this backward procedure until we found a significant difference

(p < 0.05) between models with n and n-1 variables. The model with n variables at this last step

was our final model (Hosmer & Lemeshow 2000). We determined model fit of our GLMM using

the pseudo-R2, i.e. the square of the correlation between fitted and observed values (Zar 1999).

Analyses were conducted with the R software, version 2.4.1 (R Development Core Team 2006).

Results

Habitat principal components

Tree basal area averaged 2.7 ± 1.1 m²/ha (mean ± S.D.) in harvested stands and 27.6 ± 14.4 m²/ha

in natural stands. Proportions of black spruce, balsam fir, and white birch in the canopy layer

averaged, respectively, 0.47 ± 0.40, 0.18 ± 0.27 and 0.34 ± 0.40 in harvested stands and 0.69 ±

0.27, 0.25 ± 0.22, 0.07 ± 0.08 in natural forests. The first two PCA axes together explained 54% of

the variance within the habitat dataset (Table 5). The first axis, PC1, mainly represented a moisture

gradient, from xeric sites where the ground was largely covered by lichens and ericaceous species

(namely, L. groenlandicum, K. angustifolia, and Vaccinum spp.) to mesic habitats with high basal

areas of balsam fir trees and saplings and high volume of CWD (Fig. 12).

The second PCA axis, PC2, represented a gradient of habitat alteration (Fig. 12). Basal area of

black spruce trees, which was highly correlated with basal area of all tree species (Spearman’s

rank correlation: rs = 0.91, p < 0.001), was negatively associated with PC2 (Table 5). Ground

cover of fungi, an indicator of mature and old-growth forests (Johnson 1996), was negatively

94

associated with PC2 (Table 5). Ground cover of mosses, which are primarily found in mature,

closed canopy, forests (Fenton & Bergeron 2007), was also negatively associated with PC2 (Table

5). Conversely, bare ground cover increased with PC2 (Table 5); bare ground is an indicator of

habitat alteration because soil in natural stands was covered by a rich floral carpet. Ground cover

of fruit-bearing shrubs, which are more abundant in disturbed, open canopy habitats, increased

with PC2 (Table 5).

Mean GUD and ∆ GUD

Red-backed voles were present in both the harvested and the uncut forest habitats in 17 of the 29

pairs; they occupied only the forest in eight pairs and only the harvested stand in the remaining

four pairs. Deer mice were present in 12 of the 29 pairs, including four pairs where they were

captured in both habitats, and eight pairs where they were found exclusively in the harvested

habitat. We pooled data from all the four sampling sessions because the inclusion of temporal

changes over the summer never improved model fit (p > 0.57) of mean GUD, ∆GUD, or red-

backed vole abundance.

Three of the 17 pairs where red-backed voles were present in both habitats had at least one habitat

with 75% to 100% of unused feeding stations, which provided little information about spatial

patterns in foraging effort. Therefore, we excluded these pairs from the analysis, and conducted

the analysis on six habitat pairs without deer mice and eight habitat pairs with deer mice, for a

total of 224 feedings stations. Mean GUD decreased with PC2, the gradient of habitat alteration

(Fig. 13a). Mean GUD declined more in sites with than without deer mice (significant PC2 × Deer

mouse interaction, Table 6). And mean GUD declined more in xeric (low PC1) than in mesic sites

(high PC1; significant PC1 × PC2 interaction, Table 6). Mean GUD did not vary with PC2 in

mesic sites lacking deer mice (Fig. 13a).

GUD was 34 ± 11 seeds (mean ± S.D.) in open food patches whereas it averaged 30 ± 11 seeds in

covered food patches. ∆GUD increased with PC2 in xeric sites whereas ∆GUD decreased with

PC2 in mesic sites (significant PC1 × PC2 interaction, Fig. 13c). ∆GUD was higher in xeric than

mesic habitats with deer mice (significant PC1 × Deer mouse interaction, Table 6) whereas GUD

was lower in xeric than mesic habitats without deer mice (Fig. 13b).

95

Small-mammal abundance

We captured a total of 278 different small-mammals over 3480 night-traps. The red-backed vole

was the most abundant rodent species (61% of all individuals, n = 169) followed by the deer

mouse (13%; n = 35). We also captured 10 red squirrels, eight rock voles (Microtus

chrotorrhinus), three heather voles (Phenacomys intermedius), and 53 shrews (Sorex spp.). Shrews

are insectivorous foragers that did not forage in our food patches. Also, the 2.5 cm hole at the

entrance of food patches prevented access by red-squirrels. Red-backed voles and deer mice thus

represented 95% (204/215) of the individuals that were able to forage in our food patches.

Red-backed vole abundance increased with PC1, implying that voles were more abundant in mesic

than in xeric sites (Table 6). Red-backed vole abundance also decreased with the level of habitat

alteration (Fig. 13c), an influence that varied with the presence of deer mice (significant PC2 ×

Deer mouse interaction, Table 6). At the undisturbed end of the gradient of habitat alteration

(PC2), red-backed voles were three times as abundant in sites with deer mice (10 to 16 red-backed

voles captured per habitat) than in sites without deer mice (3 to 4 red-backed voles captured per

habitat). In contrast, red-backed vole abundance was similar in sites with and without deer mice at

the upper end of PC2, with only 2 to 3 red-backed voles being captured per habitat (Fig. 13c).

Discussion

Our observations are consistent with previous studies reporting lower abundance of red-backed

voles in harvested than natural forest stands (e.g., Mills 1995, Sullivan et al. 1999, Darveau et al.

2001, Moses & Boutin 2001). Unlike these other studies, we considered the response of red-

backed vole to a broad gradient of habitat alteration made up of clear cuts, high-intensity partial

cuts (CPHRS), and low-intensity partial cuts (CPPTM). Also, we analyzed the influence of forest

harvesting on vole abundance while directly accounting for the presence or absence of deer mice,

the major competitor. We found that the effect of forest harvesting on vole populations depended

on the presence of mice. The abundance of voles was three times higher in uncut stands with than

without deer mice, suggesting that the presence of this competitor (along with mesic conditions)

provides an indication that the uncut stand is of particularly high-quality for both small-mammal

96

species. In addition, we detected a weak decrease in red-backed vole abundance with increasing

forest harvesting intensity among sites without deer mice, but found an exponential decrease

among sites with deer mice.

Competition

The quitting-harvest rate (mean GUD) decreased with logging intensity (i.e., from low to high

PC2-scores), indicating a decline in habitat quality with forest harvest. The presence of deer mice

exacerbated this effect, as found in other rodent species (Ziv & Kotler 2003, Eccard & Ylonen

2007). Mean GUD was lower in the presence of deer mice. There are two potential causes. First,

deer mice might have been the final foragers to visit the patches and could possess a lower GUD

than voles. Second, deer mice might simply have reduced resource abundance in the habitat as a

whole. We cannot be certain which effect was most pronounced in our study because we could not

differentiate vole and mouse foraging, although mice represented only 17% of the foragers.

Regardless, it is apparent that GUD was reduced in the presence of deer mice, suggesting that mice

are more efficient foragers than voles, and especially so in altered habitats (Suzuki & Hayes 2003,

Fuller et al. 2004). This result follows the R* rule that states that the winner between two

exploitative competitor species is the one that can depress mean resource abundance in the habitat

to the lowest level while ensuring its survival (Tilman 1990, Holt et al. 1994). Although red-

backed vole abundance was higher in sites with deer mice, mean GUD was similar between uncut

forest stands with and without deer mice. This result indicates either that uncut stands could

support more red-backed voles where deer mice were present than absent or that deer mice occupy

only the best undisturbed sites.

Predation costs

Red-backed voles also appeared to be risk-sensitive foragers. Red-backed voles should have more

to lose by foraging in the risky patches of mesic habitats because the overall quality was higher in

these sites than in xeric sites. According to the asset-protection principle (sensu Clark 1994), voles

should display the weakest antipredatory behavior in mesic habitats (Ekman & Ulliendahl 1993,

Ydenberg et al. 1995, Olsson & Molokwu 2007). Moreover, forest harvesting generally reduces

anti-predator cover (Ucitel et al. 2003, Fuller et al. 2004), which increases predation risk for

97

small-mammals (Morris & Davidson 2000, Verdolin 2006, Eccard et al. 2008). On the basis of

changes in habitat quality and availability of anti-predator cover, we predicted that predation costs,

i.e. ∆GUD between risky and safe food patches, would increase with logging intensity in xeric

habitats (Ucitel et al. 2003, Fuller et al. 2004), but would be less affected and could even decrease

with logging intensity in mesic habitats (Clark 1994, Olsson & Molokwu 2007). Our results are

consistent with these expectations. ∆GUD indicated that predation costs increased with logging in

low-quality xeric habitats, but decreased in high-quality mesic habitats.

Model fit was lower for ∆GUD than for mean GUD, indicating that the effect of greatest

difference in GUDs between environments was generated by a difference in food availability

rather than predation costs (Olsson & Holmgren 1999, Davidson & Morris 2001, Olsson &

Molokwu 2007). This conclusion is also consistent with Boonstra and Krebs’s (2006) contention

that competition, rather than predation, is the main mechanism controlling habitat selection by red-

backed voles. Red-backed vole populations thus seem to be primarily driven by competition

(bottom-up regulation) rather than by predation (top-down regulation).

Finally, the identification of intra- and inter-specific competition as the main mechanism

responsible for the decline of red-backed vole populations with forest harvesting is an important

cue for conservation and management. One cannot simply use the abundance of indicator species

to assess their habitat quality. Rather, studies must incorporate, explicitly, the effects of habitat

variation, habitat selection, and competing species on adaptive (e.g., foraging) behavior. Our

results suggest that expanding the use of behavioral indicators should benefit population ecology

and conservation biology by providing insight on adaptive mechanisms affected by habitat

alteration.

98

Table 5. Loadings and proportion of explained variance of the PCA conducted on 12 habitat

variables measured in 29 pairs of adjacent harvested and natural habitats (n = 58 habitats) in the

boreal forest of eastern Quebec. Stars indicate the level of significance of each loading according

to a bootstrap validation.

Variable PC1

PC2

Variables closely associated with habitat alteration

Basal area of black spruce trees -0.09 -0.48 **

Ground cover of Fungi 0.04 -0.34 **

Ground cover of Mosses 0.13 -0.48 **

Bare ground cover -0.04 0.33 *

Ground cover of fruits 0.08 0.37 **

Variables closely associated with moisture conditions

Basal area of balsam fir trees 0.37 *** -0.25 *

Basal area of balsam fir saplings 0.40 *** -0.18

Coarse woody debris volume 0.38 *** 0.09

Ground cover of Lichen -0.36 *** -0.08

Ground cover of Ledum groenlandicum -0.42 *** -0.03

Ground cover of Kalmia angustifolia -0.35 *** -0.09

Ground cover of Vaccinum spp. -0.30 *** -0.24 *

Proportion of explained variance 0.31

0.23

99

* p < 0.05

** p < 0.01

*** p < 0.001

10

0

Tab

le 6

. M

ulti

vari

able

GL

MM

of

mea

n G

UD

, ∆

GU

D (

i.e.

GU

D i

n op

en m

inus

GU

D i

n co

vere

d

food

pat

ch),

and

red

-bac

ked

vole

abu

ndan

ce a

s a

func

tion

of

habi

tat

alte

rati

on (

PC

2),

moi

stur

e

cond

itio

ns (

PC

1), a

nd t

he p

rese

nce

of d

eer

mic

e in

pai

rs, a

dum

my

vari

able

hav

ing

no d

eer

mou

se

as t

he r

efer

ence

val

ue.

Pse

udo-

R2 w

as 0

.56

for

vole

abu

ndan

ce,

0.58

for

mea

n G

UD

and

0.0

8 fo

r

∆G

UD

.

Var

iabl

e M

ean

GU

D

∆

GU

D

V

ole

abun

danc

e

β

SD

p

β

SD

p

β

SD

p

Inte

rcep

t -2

.61

0.13

<

0.00

1

4.10

0.

72

<0.

001

-3

.18

0.13

<

0.00

1

PC

1 -0

.01

0.01

0.

36

-1

.15

0.44

0.

04

0.

18

0.05

<

0.00

1

PC

2 -0

.02

0.02

0.

40

0.

48

0.42

0.

23

-0

.11

0.07

0.

10

Dee

r m

ouse

-0

.25

0.18

0.

17

0.

74

1.09

0.

35

0.

40

0.22

0.

07

PC

1 ×

PC

2 0.

01

0.01

0.

03

-0

.59

0.26

0.

04

PC

1 ×

Dee

r m

ouse

1.

65

0.72

0.

06

PC

2 ×

Dee

r m

ouse

-0

.06

0.03

0.

05

-0

.29

0.11

0.

01

101

Figure 11. Location of the study area in the Côte-Nord region of Quebec, Canada. Black dots in

the top-right panel indicate habitat pair location (n = 29 pairs).

102

-6 -4 -2 0 2 4 6

-6-4

-20

24

6

PC1

PC2

-6 -4 -2 0 2 4 6

-6-4

-20

24

6

PC1

PC2

Logged

Natural

-6 -4 -2 0 2 4 6-6

-4-2

02

46

PC1

CWD

Black spruce trees

Balsam fir trees

Bare ground

Balsam fir saplings

Mosses

Vaccinum spp

L. groenlendicus

K. angustifolia, Lichens

Fruits

Fungi

Figure 12. PCA conducted on 12 habitat variables in 29 pairs of adjacent harvested and natural

forests (n = 58 habitats). Habitats are represented in the PCA space on the left panel with black

dots representing natural habitats and white dots representing harvested habitats. The right panel

presents loadings of the PCA with arrow’s direction and length representing the strength of the

correlation between each habitat variable and principal components. PC1 represents a moisture

gradient, from xeric to mesic habitats, and PC2 represents a habitat alteration gradient, from

natural stands to stands harvested with high intensity.

103

a

c

-4 -2 0 2 4

25

30

35

40

PC2

GUD

Deer mice absent, low PC1 (Xeric)

Deer mice absent, high PC1 (Mesic)

Deer mice present, low PC1 (Xeric)

Deer mice present, high PC1 (Mesic)

-4 -2 0 2 4

0

2

4

6

8

10

PC2

Predation risk

Deer mice absent, low PC1 (Xeric)

Deer mice absent, high PC1 (Mesic)

Deer mice present, low PC1 (Xeric)

Deer mice present, high PC1 (Mesic)

∆∆ ∆∆ G

UD

b

Mean G

UD

-4 -2 0 2 4

0

5

10

15

20

PC2

Abundance

Deer mice absent, low PC1 (Xeric)

Deer mice absent, high PC1 (Mesic)

Deer mice present, low PC1 (Xeric)

Deer mice present, high PC1 (Mesic)

Red

-backed

vole ab

unda

nce

PC2

104

Figure 13. Interactions between red-backed voles and deer mice. (a) mean giving-up density

(GUD); (b) ∆GUD (i.e. GUD in open minus GUD in the paired covered food patch); and (c)

patterns of abundance in xeric and mesic habitats as a function of habitat alteration (PC2),

moisture conditions (PC1). Lines represent model fits (see Table 5 for estimates and p-values).

Low PC1, i.e. xeric habitats, refers to the 1st quartile of PC1 while high PC1, i.e. mesic habitats,

refers to the 3rd quartile of PC1.

105

Conclusion générale Cette thèse a permis de mieux comprendre les mécanismes de répartition des oiseaux nicheurs en

forêt boréale naturelle ainsi que les mécanismes de répartition des micromammifères en forêt

boréale naturelle et aménagée. Avec l’appui de mes directeurs de recherche, j’ai démontré que la

structure et la composition de l’habitat contribuaient à part égale à modeler l’assemblage de 25

espèces d’oiseaux boréaux à des échelles spatiales allant de 3 à 314 ha, soit des rayons allant de

100 à 1000 m autour des stations d’échantillonnage. Cependant, certaines espèces d’oiseaux

étaient exclusivement influencées soit par la structure de l’habitat, soit par sa composition. Avec

mes co-auteurs, j’ai ensuite découvert qu’un parasite, l’œstre (Cuterebra spp), avait le potentiel

de limiter les populations du campagnol à dos roux de Gapper (Myodes gapperi) en forêt boréale

naturelle, un phénomène jusqu’alors peu documenté. Cette découverte est d’autant plus

importante que le campagnol à dos roux est l’espèce de micromammifère la plus abondante en

forêt boréale naturelle. L’espèce devrait donc influencer de façon majeure la structure et le

fonctionnement des écosystèmes boréaux. Finalement, mes collaborateurs et moi avons

déterminé que l’abondance du campagnol à dos roux diminuait en fonction de l’intensité de

l’exploitation forestière. L’impact de l’aménagement forestier sur les populations de campagnols

était en grande partie une conséquence de l’augmentation de la compétition, particulièrement de

la compétition interspécifique avec la souris sylvestre (Peromyscus maniculatus).

Contribution relative de la structure et de la composition de l’habitat

Dans le chapitre 1, j’ai déterminé l’influence de l’hétérogénéité de l’habitat sur la répartition des

oiseaux nicheurs. J’ai trouvé que la structure et la composition de l’habitat expliquaient une

proportion semblable de variance dans l’assemblage des espèces d’oiseaux nicheurs de la forêt

boréale et ce, à toutes les échelles spatiales (étendue des rayons de 100 à 1000 m). En d’autre

mots, la structure et la composition de l’habitat seraient aussi importantes l’une que l’autre pour

la création des niches écologiques, probablement car ces attributs fourniraient des ressources

complémentaires en termes de couvert et de nourriture (Holmes & Robinson 1981, Robinson &

Holmes 1984, Deppe & Rotenberry 2008).

106

À l’intérieur des assemblages fauniques, j’ai observé que les espèces répondaient différemment à

des mêmes niveaux d’hétérogénéité, comme mentionné dans d’autres études de l’avifaune (e.g.,

Villard et al. 1999, Schmiegelow & Mönkkönen 2002). Ainsi, la probabilité d’occurrence de

plusieurs espèces était exclusivement reliée à la structure de l’habitat (i.e., bec-croisé bifascié,

Loxia leucoptera, grimpereau brun, Certhia americana, paruline à gorge noire, Dendroica virens,

pic à dos noir, Picoides arcticus et troglodyte mignon, Troglodytes troglodytes) ou à sa

composition (i.e., bruant familier, Spizella passerina, bruant de Lincoln, Melospiza lincolnii,

grive à dos olive, Catharus ustulatus, moucherolle des aulnes, Empidonax alnorum et paruline à

tête cendrée, Dendroica magnolia). Les patrons d’occurrence observés correspondent

généralement à l’autoécologie déjà reconnue des espèces.

Je n’ai détecté aucun effet de l’échelle spatiale sur la contribution relative de la structure et de la

composition de l’habitat, bien que j’aie considéré cinq échelles spatiales qui, pour les oiseaux

forestiers, englobaient des échelles locales associées aux peuplements et des échelles plus

étendues associées aux paysages (e.g., Lichstein et al. 2002, Guénette & Villard 2005, Boscolo &

Metzger 2009), i.e. rayons de 100 m, 250 m, 500 m, 750 m et 1000 m. Cependant, une étude

récente conduite dans le sud de l’Ontario indique que l’occurrence de plusieurs espèces d’oiseaux

pourraient aussi être influencée par des échelles régionales allant jusqu’à 24 km de rayon autour

des sites d’échantillonnage (Renaud 2005). Un des points forts de mon étude est que j’ai

caractérisé les mêmes variables d’habitat à toutes les échelles spatiales (i.e., proportion de

peuplements denses, ouverts et épars pour la structure ; proportion de peuplements feuillus,

mixtes et de conifères pour la composition) en m’inspirant de méthodes récentes (Anderson et al.

2005, Mowat 2006, Nams et al. 2006). Au contraire, les études visant à déterminer les effets de

l’échelle spatiale dans la contribution relative de la structure et de la composition sur la diversité

avaient jusqu’à présent généralement mesuré différents attributs de l’habitat à différentes échelles

spatiales (Wiens & Rotenberry 1981, Deppe & Rotenberry 2008), rendant l’interprétation des

résultats plus difficile.

J’ai aussi mis l’accent sur une caractérisation parcimonieuse de la structure et de la composition

de l’habitat afin de proposer des conclusions qui pourraient être facilement vérifiables dans

d’autres écosystèmes ou pour d’autres taxons (Tischendorf 2001, Tews et al. 2004, McElhinny et

al. 2005, Smith et al. 2009). Ainsi, avec seulement neuf variables indépendantes (i.e., six

107

variables d’habitat, en plus de la latitude, la longitude et l’altitude), j’ai expliqué une variance

dans l’assemblage des oiseaux (14.8%) qui était relativement similaire à Drapeau et al. (2000),

qui ont expliqué 16.4% de la variance avec 41 variables ou, dans une moindre mesure, que

Hobson et al. (2000), qui ont expliqué 22.8% de la variance avec 26 variables. En conclusion,

l’approche parcimonieuse et la considération des mêmes variables d’hétérogénéité de l’habitat à

plusieurs échelles spatiales a permis de déterminer que la structure et la composition de l’habitat

jouent toutes deux un rôle majeur dans la répartition des espèces d’oiseaux nicheurs de la forêt

boréale naturelle.

Potentiel régulateur des populations par un parasite

Les résultats du chapitre 2 ont permis d’illustrer qu’un parasite pouvait influencer la répartition

de son hôte et qu’il pouvait même jouer un rôle majeur dans la dynamique des populations. J’ai

montré que les œstres réduisaient la survie des campagnols à dos roux dans les pièges à capture

vivante (i.e., de type Sherman) alors que le confinement dans les pièges produit une réponse

physiologique similaire à la réponse physiologique induite par les risques naturels comme la

prédation chez les micromammifères (Harper & Austad 2001, Fletcher & Boonstra 2006). Les

œstres semblaient également capables d’infecter de grandes proportions des populations d’hôtes

car les patrons d’infection n’étaient pas influencés par les traits d’histoire de vie de l’hôte (sexe et

condition corporelle) ou par l’hétérogénéité de l’habitat (surface terrière des arbres, des gaules et

volume de débris ligneux). Les œstres semblaient avoir des conséquences négatives sur la

croissance estivale des populations de campagnols à dos roux puisque la prévalence des parasites

était négativement reliée à la croissance des populations d’hôtes.

J’ai déterminé que la probabilité de survie des campagnols à dos roux dans les pièges à capture

vivante était plus faible pour les individus infectés que pour les individus non infectés, un résultat

contraire aux observations sur les souris du genre Peromyscus (e.g., Clark & Kaufman 1990,

Burns et al. 2005, Jaffe et al. 2005, Cramer & Cameron 2006). Par le passé, la différence

d’impacts du parasite sur le campagnol de Townsend (Microtus townsendii) et les souris du genre

Peromyscus a été expliquée par le fait que les souris seraient les hôtes primaires du parasite

(Boonstra et al. 1980) et qu’un parasite n’a aucun avantage évolutif à tuer son hôte (Lively 2006).

108

Mes résultats permettraient de rejeter cette explication chez le campagnol à dos roux qui

semblerait être l’hôte principal de l’œstre dans l’Est du Canada (Chapitre 2, Bowman 2000,

Pearce & Venier 2005a). On pensait jusqu’alors que Peromyscus était l’hôte principal partout en

Amérique du Nord. Cependant, le campagnol à dos roux était le micromammifère qui avait le

taux de parasitisme le plus élevé dans mon aire d’étude. De plus, la proportion des larves de 3e

stade localisées dans la région inguinale chez nos campagnol à dos roux, un indicateur de la force

de la coévolution chez l’œstre et son hôte, était identique à celle qu’on retrouve chez Peromyscus

(96%, Hunter & Webster 1973). Mon étude suggère soit que l’infection du campagnol à dos roux

par l’œstre est récente sur une échelle évolutive, soit que l’œstre serait capable de réduire la

survie de son hôte primaire, un résultat qui se distinguerait de la majorité des études sur le sujet

(e.g., Clark & Kaufman 1990, Burns et al. 2005, Jaffe et al. 2005, Cramer & Cameron 2006).

De plus, mes analyses de déplacement ont dévoilé que les hôtes non infectés tendaient à être

recapturés plus loin du site de capture initial que les hôtes infectés, ce qui pourrait indiquer que le

parasite infecterait davantage les individus résidents que ceux en transit. Aussi, la survie dans les

pièges tendait à être plus élevée pour les individus recapturés plus près de l’endroit de capture

initial, ce qui pourrait indiquer que les hôtes résidents auraient une probabilité de survie plus

élevée que les hôtes en transit. Ces résultats supportent donc l’hypothèse de l’artefact (Wecker

1962, Hunter et al. 1972) et suggèrent que les études des effets de l’œstre sur la survie de son

hôte basées sur la méthode de capture-marquage-recapture pourraient être biaisées.

Enfin, l’effet négatif de l’œstre sur la survie de l’hôte était indépendant de la densité. De plus, la

probabilité d’infection et l’abondance des larves dans les hôtes n’étaient pas influencées par les

traits d’histoire de vie de l’hôte, ni pas l’hétérogénéité de l’habitat. Ainsi, le parasitisme par

l’œstre serait un mécanisme limitant plutôt que régulateur des populations de campagnols à dos

roux. En effet, le parasite est capable de réduire la survie d’une grande partie des individus de la

population d’une manière indépendante de la densité des hôtes (Anderson & May 1978, May &

Anderson 1978).

Le chapitre 2 se démarque par l’approche expérimentale que nous avons utilisée. L’étude de la

survie dans les pièges à capture vivante a permis de contourner le problème lié à l’hypothèse de

l’artefact pour découvrir des impacts négatifs des œstres sur la survie de leur hôte primaire. J’ai

obtenu ces résultats en échantillonnant des micromammifères au sein de 36 populations. En

109

comparaison, Boonstra et al. (1980) ont suivi quatre populations et Burns et al. (2005) en ont

suivi 12. En conclusion, le parasitisme est un mécanisme limitant qui pourrait contribuer à

l’instabilité des populations de campagnol à dos roux en forêt boréale, au même titre que la

compétition ou la prédation.

Réaction du campagnol à dos roux à l’aménagement forestier

La compétition et la prédation sont largement reconnues pour affecter la répartition des espèces et

la dynamique des populations (e.g., Stenseth et al. 1996, Gilg et al. 2003). Dans le chapitre 3, j’ai

utilisé des indicateurs du comportement d’approvisionnement du campagnol à dos roux pour

clarifier le lien de causalité entre la répartition des individus et l’altération de l’habitat, ce qui

n’avait jamais été fait auparavant. J’ai trouvé que l’abondance du campagnol à dos roux

diminuait avec l’intensité de la récolte ligneuse, un résultat qui supporte les études précédentes

(e.g., Mills 1995, Sullivan et al. 1999, Darveau et al. 2001, Moses & Boutin 2001). Mon étude se

démarque entre autres par le fait que j’ai échantillonné une grande variété d’approches sylvicoles

résultant en de larges variations du couvert résiduel. J’ai découvert que les effets négatifs de

l’exploitation forestière sur le campagnol à dos roux étaient étroitement associés à l’augmentation

de la compétition interspécifique avec la souris sylvestre.

Grâce à l’indicateur d’effort d’approvisionnement utilisé (GUD, Brown 1988, Brown et al.

1992), j’ai déterminé que la qualité générale de l’habitat diminuait avec l’intensité de

l’exploitation forestière. Cette relation était accentuée par la présence des souris. J’ai identifié

deux causes potentielles qui expliqueraient ce résultat. D’abord, il n’est pas exclut que l’on ait

mesuré l’effort d’approvisionnement des souris, qui s’approvisionnent de façon très efficace dans

les milieux perturbés (Suzuki & Hayes 2003, Fuller et al. 2004), plutôt que celle de campagnols

dans certaines mangeoires. Ensuite, la souris pourrait avoir réduit l’abondance générale des

ressources alimentaires dans la coupe à un niveau inférieur au minimum nécessaire à assurer la

survie du campagnol (règle R*, Tilman 1990, Holt et al. 1994). Les campagnols auraient donc

exploité davantage les parcelles de nourriture pour compenser la faible abondance de ressources

alimentaires et par conséquence, l’effort d’approvisionnement aurait augmenté dans les parcelles

expérimentales. On ne peut pas être certain duquel de ces effets était le plus important dans cette

110

étude car il ne m’était pas possible d’identifier l’espèce qui s’alimentait en dernier dans les

parcelles. Cependant, étant donné que les souris représentaient seulement 17% des individus, on

peut supposer que l’utilisation des mangeoires par les souris était généralement moindre que celle

des campagnols, un résultat appuyé par la comparaison de l’utilisation des mangeoires entre les

campagnols et les souris, à l’aide de caméras filmant l’entrée et la sortie des mangeoires par les

micromammifères (Mélanie-Louise Leblanc, Daniel Fortin et Marcel Darveau, données non

publiées). Au final, outre l’explication, il apparaît évident que les GUD étaient réduites en

présence de souris sylvestres, suggérant que les souris opposaient une forte compétition aux

campagnols à dos roux dans ces habitats.

J’ai détecté un effet relativement mineur de l’altération de l’habitat sur les coûts de prédation,

évalués par la différence (∆) de GUD entre une mangeoire placée sous un couvert anti-prédateur

et une mangeoire placée à proximité (2 m) sans couvert additionnel. Le pouvoir prédictif du

modèle ∆GUD était beaucoup plus faible que celui de la GUD moyenne, indiquant que la plus

grande différence dans les GUD entre les environnements était générée par une différence dans la

disponibilité de nourriture plutôt que dans les coûts de prédation, ce qui supporte d’autres études

(Olsson & Holmgren 1999, Davidson & Morris 2001, Olsson & Molokwu 2007). Mes résultats

supportent également ceux de Boonstra et Krebs (2006) qui mentionnent que la compétition

(régulation bottom-up), plutôt que la prédation (régulation top-down), contrôlerait la sélection de

l’habitat chez le campagnol à dos roux. En conclusion, ce chapitre est la première étude à

identifier, grâce à une approche comportementale, le lien de causalité entre l’altération de

l’habitat et la répartition du campagnol à dos roux. Ces résultats ont des implications importantes

pour la conservation.

Implications pour la conservation

Plusieurs implications pour la conservation de la biodiversité devraient découler de ma thèse, en

particulier en ce qui a trait aux chapitres 1 et 3. Dans le chapitre 1, j’ai trouvé que les espèces

d’oiseaux nicheurs des vieilles forêts boréales répondaient différemment à des caractéristiques

similaires de l’hétérogénéité de l’habitat. En conséquence, les indices de diversité comme la

richesse, qui sont utilisés par exemple pour identifier les points chauds de biodiversité (e.g.,

111

Prendergast et al. 1993, Virolainen et al. 2000), pourraient ne pas représenter la complexité de la

réponse interspécifique des assemblages (Hurlbert 1971, Gotelli & Colwell 2001). À la lumière

de ces résultats, il semble donc important d’utiliser d’autres indicateurs, centrés sur les besoins

des espèces, en complément de la richesse en espèces. On pourrait par exemple utiliser les

espèces parapluies, i.e. en protégeant de grandes superficies pour la conservation de ces espèces à

large domaine vital, on protégerait par ricochet les espèces utilisant une partie de ce grand

territoire (Ozaki et al. 2006); des espèces clés, i.e. en protégeant ces espèces qui modifient leur

environnement, on protégerait les espèces qui dépendantes des ces modifications (Martin & Eadie

1999) ; ou encore des espèces indicatrices, i.e. en protégeant l’habitat de ces espèces on assurerait

la protection des espèces occupant le même type d’habitat (Carignan & Villard 2002). Certaines

espèces occupent plusieurs rôle à la fois, comme par exemple le grand pic (Dryocopus pileatus)

dans les forêts mélangées et mixtes, qui possède un grand domaine vital, qui crée des cavités de

nidification ou de repos pour de nombreuses espèces et qui dépend étroitement des

caractéristiques des forêts âgées (Lemaître & Villard 2005).

Le chapitre 1 et implicitement le chapitre 3 fournissent également une référence écologique

concernant la diversité des oiseaux nicheurs et des micromammifères de la forêt boréale de l’est

de l’Amérique du Nord. En effet, la moitié de mon aire d’étude était vierge (Bouchard et al.

2008) et les peuplements échantillonnés étaient des vieux peuplements de plus de cinq hectares.

Grâce à cette référence, on devrait être en mesure d’orienter les stratégies de conservation. Par

exemple, les coupes partielles semblent limiter les impacts négatifs sur la faune par rapport aux

coupes plus intensives (Vanderwel et al. 2007). Les résultats du chapitre 1 permettent toutefois

de prédire que même les coupes partielles pourraient être néfastes pour les espèces fortement

associées à la structure de l’habitat, comme le grimpereau brun ou la paruline à gorge noire. De

plus, les coupes partielles pourraient également modifier la composition du peuplement et mes

résultats suggèrent que la grive solitaire ou la grive à dos olive pourraient être affectées par cette

pratique.

Le chapitre 2, quant à lui, est une démonstration du fort potentiel des indicateurs de

comportement pour améliorer les stratégies de conservation (Caro 2007, Morris et al. 2009). En

identifiant les mécanismes biologiques qui ont été modifiés de manière anthropique, il devient

plus aisé d’adopter des stratégies de conservation efficaces. Pour limiter les impacts de la coupe

112

sur le campagnol à dos roux, la priorité serait de limiter les effets négatifs de la compétition

interspécifique avec les souris, qui étaient beaucoup plus importants que les effets du risque de

prédation. Plusieurs méthodes auraient pu être envisagées comme l’ajout de nourriture dans les

habitats à protéger ou encore la limitation de la densité des souris dans ces habitats par des

aménagements forestiers réduisant la colonisation de cette espèce. J’encourage donc les études

futures à utiliser les indicateurs de comportement pour améliorer notre compréhension des

mécanismes qui influencent la répartition des espèces animales.

Avenues de recherche

Alors que mes travaux de doctorat ont permis d’améliorer les connaissances des mécanismes de

répartition des oiseaux et des micromammifères boréaux, plusieurs questions n’ont pas été

abordées et d’autres ont surgi au cours de mon étude. Par exemple, l’objectif du chapitre 1 était

de comprendre l’influence de l’hétérogénéité de l’habitat sur l’avifaune des vielles forêts boréales

à plusieurs échelles spatiales. Néanmoins, mon aire d’étude est parsemée d’autres types de

d’habitats que je n’ai pas échantillonné et qui contribuent à l’hétérogénéité naturelle de la matrice

de paysage et fort possiblement à la diversité aviaire en forêt boréale: peuplements de pins gris

(Venier & Pearce 2005), peuplements de feuillus (Barbaro et al. 2007), brulis (Imbeau et al.

1999) ou encore milieux humides (Aznar & Desrochers 2008). Il serait intéressant d’étudier la

contribution de ces habitats à la diversité de la faune des forêts boréales. Une telle étude pourrait

adresser spécifiquement les questions de dispersion que je n’ai pas abordées dans mon étude car

il se peut que ces habitats revêtent des caractéristiques uniques pour la faune. Ces habitats

pourraient représenter des îlots d’habitats propices pour certaines espèces dans une matrice

forestière (MacArthur & Wilson 1967, Brooks et al. 1999, Calmé & Desrochers 2000). De plus,

l’aménagement forestier gagne du terrain dans l’aire d’étude est amène de nouvelles formes

d’hétérogénéité de l’habitat. Il serait intéressant d’effectuer un suivi temporel des populations

fauniques pour comprendre les effets relatifs de l’hétérogénéité naturelle et anthropique de

l’habitat. Un suivi des peuplements que j’ai échantillonnés dans les vieilles forêts pourrait

permettre de répondre à cette question puisque plusieurs sites ont déjà été coupés alors que

d’autres sont situés dans des massifs forestiers protégés. Il serait alors possible de valider les

113

prédictions émises dans le chapitre 1 sur les effets potentiels de l’aménagement forestier sur

l’avifaune.

La démonstration, dans le chapitre 2, qu’un parasite a le potentiel de limiter la croissance des

populations du campagnol est très excitante pour les recherches futures puisque, par ricochet, ce

parasite pourrait être bénéfique aux autres espèces de l’assemblage de micromammifères (Holt et

al. 2003, Hatcher et al. 2006, Pedersen & Fenton 2007). Je prédis que la diversité de

l’assemblage de micromammifères devrait augmenter avec l’augmentation du taux de parasitisme

chez le campagnol à dos roux puisque le parasite réduit les populations de ce campagnol qui est

environ 10 fois plus abondant que n’importe qu’elle autre espèce de micromammifère dans les

forêts boréales naturelles. De plus, mon étude s’est déroulée lors d’une année de creux

d’infection et pourtant, j’ai trouvé des effets négatifs du parasite sur la croissance des populations

du campagnol à dos roux. Il serait intéressant de suivre les patrons d’infection dans ces mêmes

sites lors d’une année de pic pour comprendre encore davantage l’impact du cycle d’infection du

parasite sur les populations d’hôtes et sur l’assemblage des micromammifères (Holt et al. 2003,

Hatcher et al. 2006, Pedersen & Fenton 2007).

J’ai démontré l’utilité des indicateurs de comportement pour comprendre les mécanismes de

répartition des espèces dans le chapitre 3. Je reconnais par contre que ces méthodes sont parfois

difficiles à mettre en œuvre, notamment car elles nécessitent de valider plusieurs prémisses

(Brown 1988, Valone & Brown 1989). Le développement de l’approche que j’ai utilisée est le

fruit de nombreuses semaines d’expérimentations sur le terrain. Les études futures pourront

appliquer directement la méthode développée et ainsi se concentrer davantage sur les questions

de recherche plutôt que sur la validation des prémisses. Un objectif émergeant de mes recherches

est de tenter de comprendre l’influence conjointe des trois interactions biologiques étudiées

(compétition, prédation et parasitisme) sur la répartition du campagnol à dos roux et sur

l’assemblage des micromammifères en forêt boréale naturelle et aménagée. On pourrait utiliser

des enclos à ciel ouvert, y placer des individus parasités et non parasités et comparer leurs

densités de nourriture à l’abandon sous différents scénarios de compétition intra- et

interspécifique, de risque de prédation et d’aménagement forestier.

Bien qu’il reste de nombreuses avenues de recherche qui me semblent prometteuses, mes travaux

ont déjà permis de mieux comprendre comment l’hétérogénéité de l’habitat, la prédation, la

114

compétition intra- et interspécifique et le parasitisme influencent la répartition des organismes et

l’assemblage des espèces. Les nouvelles questions soulevées par mes recherches me semblent

d’autant plus fascinantes et y répondre devrait nous rapprocher davantage de la compréhension

des mécanismes qui régissent la biodiversité. Ultimement, ces recherches devraient contribuer à

l’établissement de stratégies d’aménagement durable des forêts.

115

Bibliographie

Abrams, P. A. (2000) The impact of habitat selection on the spatial heterogeneity of resources in

varying environments. Ecology 81:2902-2913.

Abramsky, Z., Rosenzweig, M. L. & Subach, A. (2000) The energetic cost of competition:

Gerbils as moneychangers. Evolutionary Ecology Research 2:279-292.

Albon, S. D., Stien, A., Irvine, R. J., Langvatn, R., Ropstad, E. & Halvorsen, O. (2002) The role

of parasites in the dynamics of a reindeer population. Proceedings of the Royal Society of

London B Biological Sciences 269:1625-1632.

Anderson, D. P., Forester, J. D., Turner, M. G., Frair, J. L., Merrill, E. H., Fortin, D., Mao, J. S.

& Boyce, M. S. (2005) Factors influencing female home range sizes in elk (Cervus

elaphus) in North American landscapes. Landscape Ecology 20:257-271.

Anderson, R. M. & May, R. M. (1978) Regulation and stability of host-parasite population

interactions: I. Regulatory processes. Journal of Animal Ecology 47:219-247.

Andruskiw, M., Fryxell, J. M., Thompson, I. D. & Baker, J. A. (2008) Habitat-mediated variation

in predation risk by the American marten. Ecology 89:2273-2280.

Azeria, E. T., Fortin, D., Hébert, C., Peres-Neto, P., Pothier, D. & Ruel, J.-C. (2009a) Using null

model analysis of species co-occurrences to deconstruct biodiversity patterns and select

indicator species. Diversity and Distributions (in press).

Azeria, E. T., Fortin, D., Lemaître, J., Janssen, P., Hebert, C., Darveau, M. & Cumming, S. G.

(2009b) Fine-scale structure and cross-taxon congruence of bird and beetle assemblages

in an old-growth boreal forest mosaic. Global Ecology and Biogeography 18:333-345.

Aznar, J.-C. & Desrochers, A. (2008) Building for the future: Abandoned beaver ponds promote

bird diversity. Ecoscience 15:250-257.

Barbaro, L., Rossi, J. P., Vetillard, F., Nezan, J. & Jactel, H. (2007) The spatial distribution of

birds and carabid beetles in pine plantation forests: the role of landscape composition and

structure. Journal of Biogeography 34:652-664.

Bates, D. & Sarkar, D. (2006) lme4: Linear mixed-effects models using S4 classes. R package

version 0.9975-7.

Beauvais, G. P. & Buskirk, S. W. (1999) Modifying estimates of sampling effort to account for

sprung traps. Wildlife Society Bulletin 27:32-38.

116

Begon, M., Harper, J. L. & Townsend, C. R. (2006) Ecology: Individuals, Populations and

Communities. 4th edition. Blackwell, Malden, MA, USA.

Bender, D. J., Tischendorf, L. & Fahrig, L. (2003) Using patch isolation metrics to predict animal

movement in binary landscapes. Landscape Ecology 18:17-39.

Bersier, L. F. & Meyer, D. R. (1994) Bird assemblages in mosaic forests - the relative importance

of vegetation structure and floristic composition along the successional gradient. Acta

Oecologica 15:561-576.

Bini, L. M., Diniz, J. A. F. & Hawkins, B. A. (2004) Macroecological explanations for

differences in species richness gradients: a canonical analysis of south American birds.

Journal of Biogeography 31:1819-1827.

Böhning-Gaese, K. (1997) Determinants of avian species richness at different spatial scales.

Journal of Biogeography 24:49-60.

Boonstra, R. & Krebs, C. J. (2006) Population limitation of the northern red-backed vole in the

boreal forests of northern Canada. Journal of Animal Ecology 75:1269-1284.

Boonstra, R., Krebs, C. J. & Beacham, T. D. (1980) Impact of botfly parasitism on Microtus

townsendii populations. Canadian Journal of Zoology 58:1683-1692.

Borcard, D., Legendre, P. & Drapeau, P. (1992) Partialling out the spatial component of

ecological variation. Ecology 73:1045-1055.

Boscolo, D. & Metzger, J. P. (2009) Is bird incidence in Atlantic forest fragments influenced by

landscape patterns at multiple scales? Landscape Ecology 24:907-918.

Bost, C. A., Thiebot, J. B., Pinaud, D., Cherel, Y. & Trathan, P. N. (2009) Where do penguins go

during the inter-breeding period? Using geolocation to track the winter dispersion of the

macaroni penguin. Biology Letters 5:473-476.

Bouchard, M., Pothier, D. & Gauthier, S. (2008) Fire return intervals and tree species succession

in the North Shore region of eastern Quebec. Canadian Journal of Forest Research

38:1621-1633.

Boucher, D., De Granpré, L. & Gauthier, S. (2003) Développement d'un outil de classification de

la structure des peuplements et comparaison de deux territoires de la pessière à mousses

du Québec. Forestry Chronicle 79:318-328.

117

Boucher, D., Gauthier, S. & De Granpré, L. (2006) Structural changes in coniferous stands along

a chronosequence and a productivity gradient in the northeastern boreal forest of Quebec.

Ecoscience 13:172-180.

Boulet, M., Darveau, M. & Belanger, L. (2000) A landscape perspective of bird nest predation in

a managed boreal black spruce forest. Ecoscience 7:281-289.

Bowman, J. (2000) Forest components associated with parasitism of small mammals by botflies

(Cuterebra). Mammalia 64:243-247.

Bowman, J., Forbes, G. & Dilworth, T. (2000) The spatial scale of variability in small-mammal

populations. Ecography 23:328-334.

Box, G. E. P. & Cox, D. R. (1964) An analysis of transformations. Journal of the Royal

Statistical Society B 26:211-252.

Boyce, M. S. (2006) Scale for resource selection functions. Diversity and Distributions 12:269-

276.

Brooks, T. M., Pimm, S. L. & Oyugi, J. O. (1999) Time lag between deforestation and bird

extinction in tropical forest fragments. Conservation Biology 13:1140-1150.

Brown, J. S. (1988) Patch use as an indicator of habitat preference, predation risk, and

competition. Behavioral Ecology and Sociobiology 22:37-47.

Brown, J. S. & Kotler, B. P. (2004) Hazardous duty pay and the foraging cost of predation.

Ecology Letters 7:999-1014.

Brown, J. S., Arel, Y., Abramsky, Z. & Kotler, B. P. (1992) Patch use by gerbils (Gerbillus

Allenbyi) in sandy and rocky habitats. Journal of Mammalogy 73:821-829.

Brown, L. N. (1965) Botfly parasitism in the brush mouse and white-footed mouse in the Ozarks.

Journal of Parasitology 51:302-304.

Burnham, K. P. & Anderson, D. R. (2002) Model Selection and Multimodel Inference: a

Practical Information - Theoretic Approach. 2nd edition. Springer-Verlag, New York,

NY.

Burns, C. E., Goodwin, B. J. & Ostfeld, R. S. (2005) A prescription for longer life? Bot fly

parasitism of the white-footed mouse. Ecology 86:753-761.

Burris, J. M. & Haney, A. W. (2005) Bird communities after blowdown in a late-successional

Great Lakes spruce-fir forest. Wilson Bulletin 117:341-352.

118

Bush, M. B., Silman, M. R. & Urrego, D. H. (2004) 48,000 years of climate and forest change in

a biodiversity hot spot. Science 303:827-829.

Calmé, S. & Desrochers, A. (2000) Biogeographic aspects of the distribution of bird species

breeding in Québec's peatlands. Journal of Biogeography 27:725-732.

Carignan, V. & Villard, M.-A. (2002) Selecting indicator species to monitor ecological integrity:

a review. Environmental Monitoring and Assesment 78:45-61.

Caro, T. (1999) The behaviour-conservation interface. Trends in Ecology & Evolution 14:366-

369.

Caro, T. (2007) Behavior and conservation: a bridge too far? Trends in Ecology & Evolution

22:394-400.

Catts, E. P. (1982) Biology of New World bot flies: Cuterebridae. Annual Review of Entomology

27:313-338.

Chappell, M. A., Garland, T., Rezende, E. L. & Gomes, F. R. (2004) Voluntary running in deer

mice: speed, distance, energy costs and temperature effects. Journal of Experimental

Biology 207:3839-3854.

Chevan, A. & Sutherland, M. (1991) Hierarchical partitioning. American Statistician 45:90-96.

Clark, B. K. & Kaufman, D. W. (1990) Prevalence of botfly (Cuterebra spp.) parasitism in

populations of small mammals in eastern Kansas. American Midland Naturalist 124:22-

30.

Clark, C. W. (1994) Antipredator behavior and the asset-protection principle. Behavioral Ecology

5:159-170.

Cockle, K. L. & Richardson, J. S. (2003) Do riparian buffer strips mitigate the impacts of

clearcutting on small mammals? Biological Conservation 113:133-140.

COSEWIC. (2009) The prioritized candidate species list. [Online]

http://www.cosewic.gc.ca/eng/sct3/index_e.cfm#2.

Côté, M., Ferron, J. & Gagnon, R. (2003) Impact of seed and seedling predation by small rodents

on early regeneration establishment of black spruce. Canadian Journal of Forest

Research 33:2362-2371.

Courbin, N., Fortin, D., Dussault, C. & Courtois, R. (2009) Landscape management for woodland

caribou: the protection of forest blocks influences wolf-caribou co-occurrence. Landscape

Ecology (in press).

119

Cramer, M. J. & Cameron, G. N. (2006) Effects of bot fly (Cuterebra fontinella) parasitism on a

population of white-footed mice (Peromyscus leucopus). Journal of Mammalogy

87:1103-1111.

Crête, M., Drolet, B., Huot, J., Fortin, M. J. & Doucet, G. J. (1995) Postfire sequence of

emerging diversity among mammals and birds in the north of the boreal forest in Quebec.

Canadian Journal of Forest Research 25:1509-1518.

Cronin, J. T. (2009) Habitat edges, within-patch dispersion of hosts, and parasitoid oviposition

behavior. Ecology 90:196-207.

Currie, D. J., Mittelbach, G. G., Cornell, H. V., Field, R., Guegan, J. F., Hawkins, B. A.,

Kaufman, D. M., Kerr, J. T., Oberdorff, T., O'brien, E. & Turner, J. R. G. (2004)

Predictions and tests of climate-based hypotheses of broad-scale variation in taxonomic

richness. Ecology Letters 7:1121-1134.

Cushman, S. A. & McGarigal, K. (2004) Hierarchical analysis of forest bird species-environment

relationships in the Oregon coast range. Ecological Applications 14:1090-1105.

Danielson, B. J. (1991) Communities in a landscape: the influence of habitat heterogeneity on the

interactions between species. American Naturalist 111:1119-1144.

Darveau, M., Labbé, P., Beauchesne, P., Bélanger, L. & Huot, J. (2001) The use of riparian forest

strips by small mammals in a boreal balsam fir forest. Forest Ecology and Management

143:96-104.

Darwin, C. (1859) On the Origin of Species. John Murray, London, UK.

Davidson, D. L. & Morris, D. W. (2001) Density-dependent foraging effort of deer mice

(Peromyscus maniculatus). Functional Ecology 15:575-583.

Deppe, J. L. & Rotenberry, J. T. (2008) Scale-dependent habitat use by fall migratory birds:

vegetation structure, floristics, and geography. Ecological Monographs 78:461-487.

Desrochers, A. & Hannon, S. J. (1997) Gap crossing decisions by forest songbirds during the

post-fledging period. Conservation Biology 11:1204-1210.

Dobson, A., Lafferty, K. D., Kuris, A. M., Hechinger, R. F. & Jetz, W. (2008) Homage to

Linnaeus: how many parasites? how many hosts? Proceedings of the National Academy

of Sciences of the United States of America 105:11482-11489.

120

Dobson, A. P. & Hudson, P. J. (1992) Regulation and stability of a free-living host-parasite

system - Trichostrongylus tenuis in red grouse. 2. population-models. Journal of Animal

Ecology 61:487-498.

Drapeau, P., Leduc, A., Giroux, J.-F., Savard, J.-P., Bergeron, Y. & Vickery, W. L. (2000)

Landscape-scale disturbances and changes in bird communities of boreal mixed-wood

forests. Ecological Monographs 70:423-444.

Dunaway, P. B., Payne, J. A., Lewis, L. L. & Story, J. D. (1967) Incidence and effects of

Cuterebra in Peromyscus. Journal of Mammalogy 48:38-51.

Ebenman, B., Law, R. & Borrvall, C. (2004) Community viability analysis: the response of

ecological communities to species loss. Ecology 85:2591-2600.

Eccard, J. A. & Ylonen, H. (2002) Direct interference or indirect exploitation? An experimental

study of fitness costs of interspecific competition in voles. Oikos 99:580-590.

Eccard, J. A. & Ylonen, H. (2003) Interspecific competition in small rodents: from populations to

individuals. Evolutionary Ecology 17:423-440.

Eccard, J. A. & Ylonen, H. (2007) Costs of coexistence along a gradient of competitor densities:

an experiment with arvicoline rodents. Journal of Animal Ecology 76:65-71.

Eccard, J. A., Pusenius, J., Sundell, J., Halle, S. & Ylonen, H. (2008) Foraging patterns of voles

at heterogeneous avian and uniform mustelid predation risk. Oecologia 157:725-734.

Edenius, L. & Sjöberg, K. (1997) Distribution of birds in natural landscape mosaics of old-

growth forests in northern Sweden: relations to habitat area and landscape context.

Ecography 20:425-431.

Ekman, J. & Ulliendahl, K. (1993) Using priority to food access: fattening strategics in

dominance-structured willow tit (Pants montanus) flocks. Behavioral Ecology 4:232-238.

Elias, S. P., Witham, J. W. & Hunter, M. L. (2006) A cyclic red-backed vole (Clethrionomys

gapperi) population and seedfall over 22 years in Maine. Journal of Mammalogy 87:440-

445.

ESRI. (2006) ArcGIS. ESRI, Redlands, CA, USA.

Falls, J. B., Falls, E. A. & Fryxell, J. M. (2007) Fluctuations of deer mice in Ontario in relation to

seed crops. Ecological Monographs 77:19-32.

Faraway, J. J. (2006) Extending the Linear Model with R: Generalized Linear, Mixed Effects and

Nonparametric Regression. Chapman & Hall, Boca Raton, FL, USA.

121

Fayt, P., Machmer, M. M. & Steeger, C. (2005) Regulation of spruce bark beetles by

woodpeckers - a literature review. Forest Ecology and Management 206:1-14.

Fenton, N. J. & Bergeron, Y. (2007) Sphagnum community change after partial harvest in of

black spruce boreal forests. Forest Ecology and Management 242:24-33.

Fielding, A. H. & Bell, J. F. (1997) A review for the assessment of prediction errors in

conservation presence/absence models. Environmental Conservation 24:38-49.

Fleishman, E., Noss, R. F. & Noon, B. R. (2006) Utility and limitations of species richness

metrics for conservation planning. Ecological Indicators 6:543-553.

Fleishman, E., McDonal, N., Mac Nally, R., Murphy, D. D., Walters, J. & Floyd, T. (2003)

Effects of floristics, physiognomy and non-native vegetation on riparian bird communities

in a Mojave Desert watershed. Journal of Animal Ecology 72:484-490.

Flemming, T. L. & Freedman, B. (1998) Conversion of natural, mixed-species forests to conifer

plantations: implications for dead organic matter and carbon storage. Ecoscience 5:23-

221.

Fletcher, Q. E. & Boonstra, R. (2006) Impact of live trapping on the stress response of the

meadow vole (Microtus pennsylvanicus). Journal of Zoology 270:473-478.

Fretwell, S. D. & Lucas, H. L. (1970) On territorial behaviour and other factors influencing

habitat distribution in birds. I. Theoretical development. Acta Biotheoretica 19:16-36.

Friedman, M. (1937) The use of ranks to avoid the assumption of normality implicit in the

analysis of variance. Journal of American Statistical Association 32:675-701.

Fuhlendorf, S. D., Harrell, W. C., Engle, D. M., Hamilton, R. G., Davis, C. A. & Leslie, D. M.

(2006) Should heterogeneity be the basis for conservation? Grassland bird response to fire

and grazing. Ecological Applications 16:1706-1716.

Fuller, A. K., Harrison, D. J. & Lachowski, H. J. (2004) Stand scale effects of partial harvesting

and clearcutting on small mammals and forest structure. Forest Ecology and Management

191:373-386.

Galindo-Leal, C. (1997) Infestation of rock mice (Peromyscus difficilis) by botflies: ecological

consequences of differences between sexes. Journal of Mammalogy 78:900-907.

Gaston, K. J. (2003) The how and why of biodiversity. Nature 421:900-901.

Gaston, K. J. & Spicer, J. I. (2004) Biodiversity: an Introduction. Blackwell, Malden, MA, USA.

122

Gaston, K. J., Blackburn, T. M. & Goldewijk, K. K. (2003) Habitat conversion and global avian

biodiversity loss. Proceedings of the Royal Society of London Series B-Biological

Sciences 270:1293-1300.

Gautestad, A. O. & Mysterud, I. (2005) Intrinsic scaling complexity in animal dispersion and

abundance. American Naturalist 165:44-55.

Getz, L. L. (1968) Influence of water balance and microclimate on local distribution of red-

backed vole and white-footed mouse. Ecology 49:276-286.

Gibbons, P., Briggs, S. V., Ayers, D. A., Doyle, S., Seddon, J., McElhinny, C., Jones, N., Sims,

R. & Doody, J. S. (2008) Rapidly quantifying reference conditions in modified

landscapes. Biological Conservation 141:2483-2493.

Gilg, O., Hanski, I. & Sittler, B. (2003) Cyclic dynamics in a simple vertebrate predator-prey

community. Science 302:866-868.

Gilg, O., Sittler, B., Sabard, B., Hurstel, A., Sane, R., Delattre, P. & Hanski, L. (2006) Functional

and numerical responses of four lemming predators in high arctic Greenland. Oikos

113:193-216.

Gingrich, R. E. (1981) Migratory kinetics of Cuterebra fontinella (Diptera: Cuterebridae) in the

white-footed mouse, Peromyscus leucopus. Journal of Parasitology 67:398-402.

Gotelli, N. J. & Colwell, R. K. (2001) Quantifying biodiversity: procedures and pitfalls in the

measurement and comparison of species richness. Ecology Letters 4:379-391.

Grand, J. & Cushman, S. A. (2003) A multi-scale analysis of species-environment relationships:

breeding birds in a pitch pine-scrub oak (Pinus rigida-Quercus ilicifolia) community.

Biological Conservation 112:307-317.

Grosenbaugh, L. R. (1952) Plotless timber estimates — new, fast, and easy. Journal of Forestry

50:32-37.

Guénette, J. S. & Villard, M.-A. (2005) Thresholds in forest bird response to habitat alteration as

quantitative targets for conservation. Conservation Biology 19:1168-1180.

Halvorson, C. H. (1982) Rodent occurrence, habitat disturbance, and seed fall in larch-fir forest.

Ecology 63:423-433.

Hanski, I., Henttonen, H., Korpimaki, E., Oksanen, L. & Turchin, P. (2001) Small-rodent

dynamics and predation. Ecology 82:1505-1520.

123

Harper, J. M. & Austad, S. N. (2001) Effect of capture and season on fecal glucocorticoid levels

in deer mice (Peromyscus Maniculatus) and red-backed voles (Clethrionomys Gapperi).

General and Comparative Endocrinology 123:337-344.

Hart, B. L. (1994) Behavioral defense against parasites - interaction with parasite invasiveness.

Parasitology 109:S139-S151.

Hart, B. L., Hart, L. A., Mooring, M. S. & Olubayo, R. (1992) Biological basis of grooming

behavior in antelope - the body-size, vigilance and habitat principles. Animal Behaviour

44:615-631.

Harvey, C. A., Medina, A., Sanchez, D. M., Vilchez, S., Hernandez, B., Saenz, J. C., Maes, J. M.,

Casanoves, F. & Sinclair, F. L. (2006) Patterns of animal diversity in different forms of

tree cover in agricultural landscapes. Ecological Applications 16:1986-1999.

Haslett, J. (1997) Insect communities and the spatial complexity of mountain habitat. Global

Ecology and Biogeography Letters 6:39-48.

Hatcher, M. J., Dick, J. T. A. & Dunn, A. M. (2006) How parasites affect interactions between

competitors and predators. Ecology Letters 9:1253-1271.

Hawkins, B. A. (2001) Ecology's oldest pattern? Trends in Ecology & Evolution 16:470.

Hawkins, C. P. (2006) Quantifying biological integrity by taxonomic completeness: Its utility in

regional and global assessments. Ecological Applications 16:1277-1294.

Hayward, G. D., Henry, S. H. & Ruggiero, L. F. (1999) Response of red-backed voles to recent

patch cutting in subalpine forest. Conservation Biology 13:168-176.

Hensley, M. (1976) Prevalence of cuterebrid parasitism among woodmice in Virginia. Journal of

Wildlife Diseases 12:172-179.

Hobson, K. A. & Schieck, J. (1999) Changes in bird communities in boreal mixedwood forest:

harvest and wildfire effects over 30 years. Ecological Applications 9:849-863.

Hobson, K. A., Kirk, D. A. & Smith, A. R. (2000) A multivariate analysis of breeding bird

species of western and central canadian boreal forests: stand and spatial effects.

Ecoscience 7:267-280.

Holling, C. S. (1959) Some characteristics of simple types of predation and parasitism. Canadian

Entomologist 91:385-398.

Holmes, R. T. & Robinson, S. K. (1981) Tree species preferences of foraging insectivorous birds

in a northern hardwoods forest. Oecologia 48:31-35.

124

Holt, R. D., Grover, J. & Tilman, D. (1994) Simple rules for interspecific dominance in systems

with exploitative and apparent competition. American Naturalist 114:741-771.

Holt, R. D., Dobson, A. P., Begon, M., Bowers, R. G. & Schauber, E. M. (2003) Parasite

establishment in host communities. Ecology Letters 6:837-842.

Homyack, J. A., Harrison, D. J. & Krohn, W. B. (2005) Long-term effects of precommercial

thinning on small mammals in northern Maine. Forest Ecology and Management 205:43-

57.

Hopper, S. D. & Gioia, P. (2004) The southwest Australian floristic region: evolution and

conservation of a global hot spot of biodiversity. Annual Review of Ecology Evolution and

Systematics 35:623-650.

Hosmer, D. W. & Lemeshow, S. (2000) Applied Logistic Regression. 2d edition. John Wiley &

Sons, New York, NY, USA.

Hubbell, S. P. (2001) The Unified Neutral Theory of Biodiversity and Biogeography. Princeton

University Press, Princeton, NJ, USA.

Hudson, P. J., Newborn, D. & Dobson, A. P. (1992) Regulation and stability of a free-living host-

parasite system - Trichostrongylus tenuis in red grouse .1. Monitoring and parasite

reduction experiments. Journal of Animal Ecology 61:477-486.

Huitu, O., Laaksonen, J., Norrdahl, K. & Korpimaki, E. (2005) Spatial synchrony in vole

population fluctuations - a field experiment. Oikos 109:583-593.

Hunter, D. M. & Webster, J. M. (1973) Determination of the migratory route of botfly larvae,

Cuterebra grisea (Diptera:Cuterebridae) in deermice. International Journal for

Parasitology 3:311-316.

Hunter, D. M., Sadleir, R. M. F. S. & Webster, J. M. (1972) Studies on the ecology of cuterebrid

parasitism in deermice. Canadian Journal of Zoology 50:25-29.

Hurlbert, A. H. (2004) Species-energy relationships and habitat complexity in bird communities.

Ecology Letters 7:714-720.

Hurlbert, S. H. (1971) The nonconcept of species diversity: a critique and alternative parameters.

Ecology 52:577-586.

Hutchinson, G. E. (1957) Concluding remarks. Cold Spring Harbor Symposium on Quantitative

Biology 22:415-427.

125

Iman, R. L. & Davenport, J. M. (1980) Approximations of the critical region of the Friedman

statistic. Communications in Statistics Part a-Theory and Methods 9:571-595.

Imbeau, L., Savard, J. P. L. & Gagnon, R. (1999) Comparing bird assemblages in successional

black spruce stands originating from fire and logging. Canadian Journal of Zoology

77:1850-1860.

Imbeau, L., Monkkonen, M. & Desrochers, A. (2001) Long-term effects of forestry on birds of

the eastern Canadian boreal forests: a comparison with Fennoscandia. Conservation

Biology 15:1151-1162.

Jackson, D. M., Trayhurn, P. & Speakman, J. R. (2001) Associations between energetics and

over-winter survival in the short-tailed field vole Microtus agrestis. Journal of Animal

Ecology 70:633-640.

Jaffe, G., Zegers, D. A., Steele, M. A. & Merritt, J. F. (2005) Long-term patterns of botfly

parasitism in Peromyscus maniculatus, P. leucopus, and Tamias striatus. Journal of

Mammalogy 86:39-45.

Janssen, P., Fortin, D. & Hebert, C. (2009) Beetle diversity in a matrix of old-growth boreal

forest: influence of habitat heterogeneity at multiple scales. Ecography 32:423-432.

Johnson, C. N. (1996) Interactions between mammals and ectomycorrhizal fungi. Trends in

Ecology & Evolution 11:503-507.

Johnson, D. H. (1980) The comparison of usage and availability measurements for evaluating

resource preference. Ecology 61:65-71.

Johnson, M. D. (2007) Measuring habitat quality: a review. Condor 109:489-504.

Jonsson, B. G. & Jonsell, M. (1999) Exploring potential biodiversity indicators in boreal forests.

Biodiversity and Conservation 8:1417-1433.

Kasparian, K. & Millar, J. S. (2004) Diet selection by red-backed voles (Clethrionomys Gapperi).

Acta Theriologica 49:289-300.

Kerr, J. T. & Packer, L. (1997) Habitat heterogeneity as a determinant of mammal species

richness in high-energy regions. Nature 385:252-254.

King, D. I., Lambert, J. D., Buonaccorsi, J. P. & Prout, L. S. (2008) Avian population trends in

the vulnerable montane forests of the Northern Appalachians, USA. Biodiversity and

Conservation 17:2691-2700.

126

Kollars, T. M. (1995) Factors affecting the distribution of bot flies (Diptera, Oestridae) on Islands

in Lake Barkley, Kentucky and Tennessee. Journal of Entomological Science 30:513-518.

Kotler, B. P., Morris, D. W. & Brown, J. S. (2007) Behavioral indicators and conservation:

wielding "the biologist's tricorder". Israel Journal of Ecology & Evolution 53:237-244.

Kotliar, N. B. & Wiens, J. A. (1990) Multiple scales of patchiness and patch structure - a

hierarchical framework for the study of heterogeneity. Oikos 59:253-260.

Krebs, C. J. (1985) Ecology: the Experimental Analysis of Distribution and Abundance. 3rd

edition. Harper & Row, New York, NY, USA.

Kremen, C. (2005) Managing ecosystem services: what do we need to know about their ecology?

Ecology Letters 8:468-479.

Kroll, J. C. & Fleet, R. R. (1979) Impact of woodpecker predation on over-wintering within-tree

populations of the southern pine beetle (Dendroctonus frontalis). Pages 269-281 in

Dickson, J. C., Conner, R. N., Fleet, R. R., Kroll, J. C. & Jackson, J. A., editors. The Role

of Insectivorous Birds in Forest Ecosystems. Academic Press, New York, NY, USA.

Lack, D. (1969) The numbers of bird species on islands. Bird Study 16:193-209.

Latimer, A. M., Silander, J. A. & Cowling, R. M. (2005) Neutral ecological theory reveals

isolation and rapid speciation in a biodiversity hot spot. Science 309:1722-1725.

Law, B. & Chidel, M. (2002) Tracks and riparian zones facilitate the use of Australian regrowth

forest by insectivorous bats. Journal of Applied Ecology 39:605-617.

Lawton, J. H. (1983) Plant architecture and the diversity of phytophagous insects. Annual Review

of Entomology 28:23-39.

Lee, P. Y. & Rotenberry, J. T. (2005) Relationships between bird species and tree species

assemblages in forested habitats of eastern North America. Journal of Biogeography

32:1139-1150.

Legendre, P. & Legendre, L. (1998) Numerical Ecology. Elsevier, Amsterdam, Netherlands.

Lemaître, J. & Villard, M.-A. (2005) Foraging patterns of Pileated Woodpeckers in a managed

Acadian forest: a resource selection function. Canadian Journal of Forest Research

35:2387-2393.

Lemaître, J., Fortin, D., Montiglio, P.-O. & Darveau, M. (2009) Bot fly parasitism of the red-

backed vole: host survival, infection risk, and population growth. Oecologia 159:283-294.

Levin, S. A. (1992) The problem of pattern and scale in ecology. Ecology 73:1943-1967.

127

Levins, R. (1968) Evolution in Changing Environments: Some Theoritical Explorations.

Princeton University Press, Princeton, NJ, USA.

Lichstein, J. W., Simons, T. R. & Franzreb, K. E. (2002) Landscape effects on breeding songbird

abundance in managed forests. Ecological Applications 12:836-857.

Lima, S. L. & Dill, L. M. (1990) Behavioral decisions made under the risk of predation - a review

and prospectus. Canadian Journal of Zoology 68:619-640.

Lively, C. M. (2006) The ecology of virulence. Ecology Letters 9:1089-1095.

Lopez, G. & Moro, M. J. (1997) Birds of Aleppo pine plantations in south-east Spain in relation

to vegetation composition and structure. Journal of Applied Ecology 34:1257-1272.

Loreau, M., Naeem, S., Inchausti, P., Bengtsson, J., Grime, J. P., Hector, A., Hooper, D. U.,

Huston, M. A., Raffaelli, D., Schmid, B., Tilman, D. & Wardle, D. A. (2001) Biodiversity

and ecosystem functioning: current knowledge and future challenges. Science 294:804-

808.

Mac Nally, R. C. (1990) The roles of floristics and physiognomy in avian community

composition. Australian Journal of Ecology 15:321-327.

MacArthur, R. H. (1964) Environmental factors affecting bird species diversity. American

Naturalist 98:387-397.

MacArthur, R. H. & Wilson, E. O. (1967) The Theory of Island Biogeography. Princeton

University Press, Princeton, NJ.

Macdonald, S. E., Eaton, B., Machtans, C. S., Paszkowski, C., Hannon, S. & Boutin, S. (2006) Is

forest close to lakes ecologically unique? Analysis of vegetation, small mammals,

amphibians, and songbirds. Forest Ecology and Management 223:1-17.

Manel, S., Ceri Willians, H. & Ormerod, S. J. (2001) Evaluating presence-absence models in

ecology: the need to account for prevalence. Journal of Applied Ecology 38:921-931.

Martin, K. & Eadie, J. M. (1999) Nest webs: A community-wide approach to the management

and conservation of cavity-nesting forest birds. Forest Ecology and Management

115:243-257.

May, R. M. (1988) How many species are there on Earth? Science 241:1441-1449.

May, R. M. & Anderson, R. M. (1978) Regulation and stability of host-parasite population

interactions: II. Destabilizing processes. Journal of Animal Ecology 47:249-267.

128

May, R. M., Hassell, M. P., Anderson, R. M. & Tonkyn, D. W. (1981) Density dependence in

host-parasitoid models. Journal of Animal Ecology 50:855-865.

McCollin, D. (1998) Forest edges and habitat selection in birds: a functional approach.

Ecography 21:247-260.

McElhinny, C., Gibbons, P., Brack, C. & Bauhus, J. (2005) Forest and woodland stand structural

complexity: Its definition and measurement. Forest Ecology and Management 218:1-24.

McShea, W. J., Pagels, J., Orrock, J., Harper, E. & Koy, K. (2003) Mesic deciduous forest as

patches of small-mammal richness within an Appalachian mountain forest. Journal of

Mammalogy 84:627-643.

Meadows, D. H. (1990) Biodiversity - the key to saving life on Earth. Land Stewardship

Newsletter (summer):4-5.

Mills, L. S. (1995) Edge effects and isolation: red-backed voles on forest remnants. Conservation

Biology 9:395-403.

Mittelbach, G. G., Steiner, C. F., Scheiner, S. M., Gross, K. L., Reynolds, H. L., Waide, R. B.,

Willig, M. R., Dodson, S. I. & Gough, L. (2001) What is the observed relationship

between species richness and productivity? Ecology 82:2381-2396.

Møller, A. P. (2005) Parasitism and the regulation of host populations. Pages 43-53 in Thomas,

F., Renaud, F. & Guéguan, J.-F., editors. Parasitism and Ecosystems. Oxford University

Press, New York, USA.

Morris, D. W. (1987) Tests of density-dependent habitat selection in a patchy environment.

Ecological Monographs 57:269-281.

Morris, D. W. (1996) Coexistence of specialist and generalist rodents via habitat selection.

Ecology 77:2352-2364.

Morris, D. W. (1999) Has the ghost of competition passed? Evolutionary Ecology Research 1:3-

20.

Morris, D. W. (2003) Toward an ecological synthesis: a case for habitat selection. Oecologia

136:1-13.

Morris, D. W. (2005) Paradoxical avoidance of enriched habitats: have we failed to appreciate

omnivores? Ecology 86:2568-2577.

Morris, D. W. & Davidson, D. L. (2000) Optimally foraging mice match patch use with habitat

differences in fitness. Ecology 81:2061-2066.

129

Morris, D. W. & Mukherjee, S. (2007a) Can we measure carrying capacity with foraging

behavior? Ecology 88:597-604.

Morris, D. W. & Mukherjee, S. (2007b) Is density-dependent resource harvest a reliable habitat

indicator for conservation and management? Israel Journal of Ecology & Evolution

53:371-387.

Morris, D. W., Kotler, B. P., Brown, J. S., Sundararaj, V. & Ale, S. B. (2009) Behavioral

indicators for conserving mammal diversity. Annals of the New York Academy of Sciences

1162:334-356.

Moses, R. A. & Boutin, S. (2001) The influence of clear-cut logging and residual leave material

on small mammal populations in aspen-dominated boreal mixedwoods. Canadian Journal

of Forest Research 31:483-495.

Mouillot, D., George-Nascimento, M. & Poulin, R. (2005) Richness, structure and functioning in

metazoan parasite communities. Oikos 109:447-460.

Mowat, G. (2006) Winter habitat associations of American martens Martes americana in interior

wet-belt forests. Wildlife Biology 12:51-61.

Munger, J. C. & Karasov, W. H. (1991) Sublethal parasites in white-footed mice: impact on

survival and reproduction. Canadian Journal of Zoology 69:398-404.

Myers, N. (1983) A Wealth of Wild Species. Westview Press, Boulder, CL, USA.

Myers, N. (2003) Conservation of biodiversity: how are we doing? Environmentalist 23:9-15.

Nams, V. O., Mowat, G. & Panian, M. A. (2006) Determining the spatial scale for conservation

purposes - an example with grizzly bears. Biological Conservation 128:109-119.

Niemi, G., Hanowski, J., Helle, P., Howe, R., Mönkkönen, M., Venier, L. & Welsh, D. (1998)

Ecological sustainability of birds in boreal forests. Conservation Ecology 2:17.

Norris, A. R. & Martin, K. (2008) Mountain pine beetle presence affects nest patch choice of red-

breasted nuthatches. Journal of Wildlife Management 72:733-737.

Olsson, O. & Holmgren, N. M. A. (1999) Gaining ecological information about Bayesian

foragers through their behaviour. I. Models with predictions. Oikos 87:251-263.

Olsson, O. & Molokwu, M. N. (2007) On the missed opportunity cost, GUD, and estimating

environmental quality. Israel Journal of Ecology & Evolution 53:263-278.

130

Orrock, J. L. & Pagels, J. F. (2002) Fungus consumption by the southern red-backed vole

(Clethrionomys gapperi) in the southern Appalachians. American Midland Naturalist

147:413-418.

Ozaki, K., Isono, M., Kawahara, T., Iida, S., Kudo, T. & Fukuyama, K. (2006) A mechanistic

approach to evaluation of umbrella species as conservation surrogates. Conservation

Biology 20:1507-1515.

Parrish, J. D., Braun, D. P. & Unnasch, R. S. (2003) Are we conserving what we say we are?

Measuring ecological integrity within protected areas. Bioscience 53:851-860.

Pearce, J. & Venier, L. (2005a) The incidence of botfly parasitism in small mammals near White

River, Ontario, Canada. Mammalia 69:97-101.

Pearce, J. & Venier, L. (2005b) Small mammals as bioindicators of sustainable boreal forest

management. Forest Ecology and Management 208:153-175.

Pedersen, A. B. & Fenton, A. (2007) Emphasizing the ecology in parasite community ecology.

Trends in Ecology & Evolution 22:133-139.

Peres-Neto, P. R., Jackson, D. A. & Somers, K. M. (2003) Giving meaningful interpretation to

ordination axes: assessing loading significance in principal component analysis. Ecology

84:2347-2363.

Poulin, J. F., Villard, M. A., Edman, M., Goulet, P. J. & Eriksson, A. M. (2008) Thresholds in

nesting habitat requirements of an old forest specialist, the Brown Creeper (Certhia

americana), as conservation targets. Biological Conservation 141:1129-1137.

Prendergast, J. R., Quinn, R. M., Lawton, J. H., Eversham, B. C. & Gibbons, D. W. (1993) Rare

species, the coincidence of diversity hotspots and conservation strategies. Nature

365:335-337.

Preston, M. I. & Harestad, A. S. (2007) Community and species responses by birds to group

retention in a coastal temperate forest on Vancouver Island, British Columbia. Forest

Ecology and Management 243:156-167.

Pulliam, H. R. (2000) On the relationship between niche and distribution. Ecology Letters 3:349-

361.

R Development Core Team. (2006) R: a language and environment for statistical computing. R

Fundation for Statistical Computing, Vienna, Austria.

131

Ramos-Jiliberto, R., Gonzalez-Olivares, E. & Bozinovic, F. (2002) Population-level

consequences of antipredator behavior: A metaphysiological model based on the

functional ecology of the leaf-eared mouse. Theoretical Population Biology 62:63-80.

Reed, A. W., Kaufman, G. A. & Kaufman, D. W. (2005) Rodent seed predation and guds: effect

of burning and topography. Canadian Journal of Zoology 83:1279-1285.

Renaud, C. (2005) Influence de l'échelle spatiale sur les relations oiseaux-habitat. M.Sc. Thesis.

Université Laval, Québec, QC.

Ressources naturelles Canada 2009. (2009) L'Atlas du Canada - La forêt boréale. [Online]

http://atlas.gc.ca/site/francais/learningresources/theme_modules/borealforest/index.html/d

ocument_view.

Ripple, W. J. & Beschta, R. L. (2004) Wolves and the ecology of fear: can predation risk

structure ecosystems? Bioscience 54:755-766.

Robinet, C., Lance, D. R., Thorpe, K. W., Onufrieva, K. S., Tobin, P. C. & Liebhold, A. M.

(2008) Dispersion in time and space affect mating success and Allee effects in invading

gypsy moth populations. Journal of Animal Ecology 77:966-973.

Robinson, S. K. & Holmes, R. T. (1984) Effects of plant-species and foliage structure on the

foraging behavior of forest birds. Auk 101:672-684.

Rosenzweig, M. L. (1981) A theory of habitat selection. Ecology 62:327-335.

Rosenzweig, M. L. (1991) Habitat selection and population interactions - the search for

mechanism. American Naturalist 137:S5-S28.

Rosenzweig, M. L. (1995) Species Diversity in Space and Time. Cambridge University Press,

New York, NY, USA.

Rotenberry, J. T. (1985) The role of habitat in avian community composition - physiognomy or

floristics. Oecologia 67:213-217.

Sabrosky, C. W. (1986) North American Species of Cuterebra, the Rabbit and Rodent Bot Flies

(Diptera : Cuterebridae). Entomological society of America, Lanham, MD, USA.

Sandlund, O. T., Hindar, K. & Brown, A. H. T. (1992) Conservation of Biodiversity for

Sustainable Development. Scandinavian University Press, Oslo, Norway.

Schmidt, K. A., Manson, R. & Lewis, D. (2005) Voles competing with mice: differentiating

exploitative, interference and apparent competition using patch use theory. Evolutionary

Ecology Research 7:273-286.

132

Schmiegelow, F. K. A. & Mönkkönen. (2002) Habitat loss and fragmentation in dynamic

landscapes: avian perspectives from the boreal forest. Ecological Applications 12:375-

389.

Schulte-Hostedde, A. I., Zinner, B., Millar, J. S. & Hickling, G. J. (2005) Restitution of mass-size

residuals: validating body condition indices. Ecology 86:155-163.

Sekercioglu, C. H., Daily, G. C. & Ehrlich, P. R. (2004) Ecosystem consequences of bird

declines. Proceedings of the National Academy of Sciences of the United States of

America 101:18042-18047.

Shannon, C. E. & Waever, W. (1949) The mathematical theory of communication. Urbana:

University of Illinois Press.

Simon, N. P. P., Diamond, A. W. & Schwab, F. E. (2003) Do northern forest bird communities

show more ecological plasticity than southern forest bird communities in eastern Canada?

Ecoscience 10:289-296.

Simpson, E. H. (1949) Measurement of diversity. Nature 163:688.

Slansky, F. (2007) Insect/mammal associations: effects of cuterebrid bot fly parasites on their

hosts. Annual Review of Entomology 52:17-36.

Smith, A. C., Koper, N., Francis, C. M. & Fahrig, L. (2009) Confronting collinearity: comparing

methods for disentangling the effects of habitat loss and fragmentation. Landscape

Ecology (in Press).

Smith, D. H. (1977) Natural-history and development of Cuterebra approximata (Diptera:

Cuterebridae) in its natural host, Peromyscus maniculatus (Rodentia: Cricetidae), in

Western Montana. Journal of Medical Entomology 14:137-145.

Sorci, G., deFraipont, M. & Clobert, J. (1997) Host density and ectoparasite avoidance in the

common lizard (Lacerta vivipara). Oecologia 111:183-188.

Soulé, M. E. (1986) Conservation biology: the science of scarcity and diversity. Sinauer

Associates, Sunderland, MA, USA.

Spies, T. A. & Franklin, J. F. (1991) The structure of natural young, mature, and old-growth

Douglas-Fir forests in Oregon and Washington. Pages 91-109 in Aubry, K. B., Brookes,

M. H., Agee, J. K., Anthony, R. G. & Franklin, J. F., editors. Wildlife and Vegetation of

Unmanaged Douglas-Fir Forests. USDA Forest Service, Portland, OR, USA.

133

Stahl, G. (1997) Transect relascope sampling for assessing coarse woody debris: the case of a

pi/2 relascope angle. Scandanavian Journal of Forest Research 12:375-381.

Stanko, M., Krasnov, B. R. & Morand, S. (2006) Relationship between host abundance and

parasite distribution: inferring regulating mechanisms from census data. Journal of

Animal Ecology 75:575-583.

Stenseth, N. C., Bjornstad, O. N. & Falck, W. (1996) Is spacing behaviour coupled with

predation causing the microtine density cycle? a synthesis of current process-oriented and

pattern-oriented studies. Proceedings of the Royal Society of London Series B-Biological

Sciences 263:1423-1435.

Sullivan, T. P. & Boateng, J. O. (1996) Comparison of small mammal community responses to

broadcast burning and herbicide application in cutover forest habitats. Canadian Journal

of Forest Research 26:462-473.

Sullivan, T. P., Lautenschlager, R. A. & Wagner, R. G. (1999) Clearcutting and burning of

northern spruce-fir forests: implications for small mammal communities. Journal of

Applied Ecology 36:327-344.

Sutherland, G. D., Harestad, A. S., Price, K. & Lertzman, K. P. (2000) Scaling of natal dispersal

distances in terrestrial birds and mammals. Conservation Ecology 4:16.

Suzuki, N. & Hayes, J. P. (2003) Effects of thinning on small mammals in Oregon coastal forests.

Journal of Wildlife Management 67:352-371.

ter Braak, C. J. F. (1986) Canonical correspondence analysis: a new eigenvector technique for

multivariate direct gradient analysis. Ecology 67:1167-1179.

Tews, J., Brose, U., Grimm, V., Tielborger, K., Wichmann, M. C., Schwager, M. & Jeltsch, F.

(2004) Animal species diversity driven by habitat heterogeneity/diversity: the importance

of keystone structures. Journal of Biogeography 31:79-92.

Tilman, D. (1990) Mechanisms of plant competition for nutrients: the elements of a predictive

theory of competition. Pages 117-142 in Grace, J. B. & Tilman, D., editors. Perspective of

Plant Competition. Academic Press, San Diego, CA, USA.

Tilman, D. (2000) Global environmental impacts of agriculture expansions: the need for

sustainable and efficient practices. Proceedings of the National Academy of Sciences of

the United States of America 96:5995-2000.

134

Timm, R. M. & Cook, E. F. (1979) Effect of bot fly larvae on reproduction in white-footed mice,

Peromyscus leucopus. American Midland Naturalist 101:211-217.

Tischendorf, L. (2001) Can landscape indices predict ecological processes consistently?

Landscape Ecology 16:235-254.

Turchin, P. & Hanski, I. (2001) Contrasting alternative hypotheses about rodent cycles by

translating them into parameterized models. Ecology Letters 4:267-276.

Ucitel, D., Christian, D. P. & Graham, J. M. (2003) Vole use of coarse woody debris and

implications for habitat and fuel management. Journal of Wildlife Management 67:65-72.

Valone, T. J. & Brown, J. S. (1989) Measuring patch assessment abilities of desert granivores.

Ecology 70:1800-1810.

Van Horne, B. (1983) Density as a misleading indicator of habitat quality. Journal of Wildlife

Management 47:893-901.

Vanderwel, M. C., Maicolm, J. R. & Mills, S. C. (2007) A meta-analysis of bird responses to

uniform partial harvesting across North America. Conservation Biology 21:1230-1240.

Venier, L. A. & Pearce, J. L. (2005) Boreal bird community response to jack pine forest

succession. Forest Ecology and Management 217:19-36.

Verdolin, J. L. (2006) Meta-analysis of foraging and predation risk trade-offs in terrestrial

systems. Behavioral Ecology and Sociobiology 60:457-464.

Verner, J. (1988) Optimizing the duration of point counts for monitoring trends in bird

populations. Research Note PSW-395, USDA Forest Service.

Villard, M.-A., Trzcinski, M. K. & Merriam, G. (1999) Fragmentation effects on forest birds:

relative influence of woodland cover and configuration on landscape occupancy.

Conservation Biology 13:774-783.

Virkkala, R. & Rajasärkkä, A. (2006) Spatial variation of bird species in landscapes dominated

by old-growth forests in northern boreal finland. Biodiversity and Conservation 15:2143-

2162.

Virolainen, K. M., Ahlroth, P., Hyvarinen, E., Korkeamaki, E., Mattila, J., Paivinen, J., Rintala,

T., Suomi, T. & Suhonen, J. (2000) Hot spots, indicator taxa, complementarity and

optimal networks of taiga. Proceedings of the Royal Society of London Series B-

Biological Sciences 267:1143-1147.

135

Vitousek, P. M., Mooney, H. A., Lubchenco, J. & Melillo, J. M. (1997) Human domination of

Earth's ecosystems. Science 277:494-499.

Wecker, S. C. (1962) The effects of bot fly parasitism on a local population of the white-footed

mouse. Ecology 43:561-565.

Westerberg, L. & Wennergren, U. (2003) Predicting the spatial distribution of a population in a

heterogeneous landscape. Ecological Modelling 166:53-65.

While, G. M. & McArthur, C. (2006) Distance from cover affects artificial food-patch depletion

by macropod herbivores. Wildlife Research 33:565-570.

Wiens, J. A. (1989) Spatial scaling in ecology. Functional Ecology 3:385-397.

Wiens, J. A. & Rotenberry, J. T. (1981) Habitat associations and community structure of birds in

shrubsteppe environments. Ecological Monographs 51:21-41.

Wigley, T. B. & Roberts, T. H. (1997) Landscape-level effects of forest management on faunal

diversity in bottomland hardwoods. Forest Ecology and Management 90:141-154.

Wilson, E. O. & Peter, F. M. (1989) Biodiversity. National Academy Press, Washington DC,

USA.

Witting, L. (2000) Population cycles caused by selection by density dependent competitive

interactions. Bulletin of Mathematical Biology 62:1109-1136.

Wolf, M. & Batzli, G. O. (2001) Increased prevalence of bot flies (Cuterebra fontinella) on

white-footed mice (Peromyscus leucopus) near forest edges. Canadian Journal of

Zoology 79:106-109.

Worm, B. & Duffy, J. E. (2003) Biodiversity, productivity and stability in real food webs. Trends

in Ecology & Evolution 18:628-632.

Ydenberg, R. C., Clark, C. W. & Harfenist, A. (1995) Interspecific fledging mass variation in the

Alcidae, with special reference to the seasonal fledging mass decline. American Naturalist

145:412-433.

Zar, J. H. (1999) Biostatistical Analysis. 4th edition. Prentice-Hall, Upper Saddle River, NJ.

Ziv, Y. & Kotler, B. P. (2003) Giving-up densities of foraging gerbils: the effect of interspecific

competition on patch use. Evolutionary Ecology 17:333-347.