Bayesian spatio‐temporal reconstruction reveals rapid ... · 3Real Jardín Botánico, RJB-CSIC,...

R E S E A R CH P A P E R

Bayesian spatio‐temporal reconstruction reveals rapiddiversification and Pleistocene range expansion in thewidespread columnar cactus Pilosocereus

Pacircmela Lavor1 | Alice Calvente2 | Leonardo M Versieux2 | Isabel Sanmartin3

1Departamento de Botacircnica Aplicada

Centro de Ciecircncias Bioloacutegicas Universidade

Regional do Cariri Crato CE Brazil

2Departamento de Botacircnica e Zoologia

Centro de Biociecircncias Universidade Federal

do Rio Grande do Norte Natal RN Brazil

3Real Jardiacuten Botaacutenico RJB-CSIC Madrid

Spain

Correspondence

Isabel Sanmartin Real Jardiacuten Botaacutenico

CSIC Madrid Spain

Email isanmartinrjbcsices

Funding information

Conselho Nacional de Desenvolvimento

Cientiacutefico e Tecnoloacutegico Coordenaccedilatildeo de

Aperfeiccediloamento de Pessoal de Niacutevel

Superior CAPES CNPq MINECOFEDER

GrantAward Number CGL2015-67849-P

Proap (CAPESPPGSE UFRN) PPBio

Semi-Aacuterido GrantAward Number 457427

2012-4

Editor Lars Chatrou

ABSTRACT

Aim Pilosocereus is one of the richest and most widespread genera of columnar

cacti extending from south‐west USA to southern Brazil Most species occur in the

seasonally dry tropical forest biome but can also be found in xeric microhabitats

inside woody savannas (Cerrado) and moist forests (Brazilian Atlantic forest) The

genus exhibits a highly disjunct distribution across the Neotropics Using a 90

complete species‐level phylogeny we reconstructed the spatio‐temporal evolution

of Pilosocereus to explore the historical factors behind the species richness of

Neotropical dry formations

Location South America Mesoamerica Caribbean south‐western North America

Taxon Genus Pilosocereus (Cactaceae Cactoideae Cereeae)

Methods We used plastid and nuclear DNA sequences and Bayesian inference to

estimate phylogenetic relationships and lineage divergence times Ancestral ranges

were inferred within the Pilosocereus subgenus Pilosocereus s s clade using the

DispersalndashExtinctionndashCladogenesis model in a Bayesian framework to account for

parameter estimation uncertainty and the effect of geographical distance on disper-

sal rates

Results Pilosocereus was recovered as polyphyletic with representatives of other

Cereeae nested within The Pilosocereus subgenus Pilosocereus s s clade originated

around the PliocenendashPleistocene transition (27 Ma) probably within the Caatinga

seasonally dry tropical forest (SDTF) formation Species divergences were dated in

the Middle and Upper Pleistocene often constrained to the same geographic region

but also associated to migration events to other xeric habitats in Mesoamerica and

northern South America dispersal rates were not dependent on distance

Main conclusions Diversification dynamics in the Pilosocereus subgenus Pilosocereus

s s clade agree with other infrageneric studies in cacti Species divergence was

rapid driven by in situ diversification and migration events between SDTF dry for-

mations and xeric microhabitats within other biomes and probably linked to Pleis-

tocene climatic changes This dynamic history differs from that found in woody

SDTF lineages which are older in age and characterized by low‐dispersal rates and

long‐term isolation

Received 4 July 2017 | Revised 23 September 2018 | Accepted 30 September 2018

DOI 101111jbi13481

Journal of Biogeography 20181ndash13 wileyonlinelibrarycomjournaljbi copy 2018 John Wiley amp Sons Ltd | 1

K E YWORD S

biogeography cacti diversification Neotropical region Northeastern Brazil Pleistocene

climatic changes

1 | INTRODUCTION

The Neotropics are among the regions with the greatest floristic

diversity in the world (Kier et al 2005) Throughout the region a

variety of plant formations are arranged in alternating patches of

vegetation of different size creating a mosaic‐like landscape (Olson

et al 2001) These include moist (eg tropical rain forests and wet-

lands) and drier habitat types such as seasonally dry tropical forests

(SDTFs) savannas and rocky fields high‐elevation Andean grasslands

and deserts all embedded within a complex physiography with a

variety of edaphic climatic and topographic conditions (Antonelli amp

Sanmartiacuten 2011 Burnham amp Graham 1999 Hughes Pennington amp

Antonelli 2013) Among these the extremely rich tropical lowland

forests of the Amazonian region have attracted the attention of bio-

geographic studies (Hoorn et al 2010 Rull 2008) but there is

increasing interest in other biomes especially SDTF dry formations

characterized by high levels of beta diversity (Banda et al 2016

Pennington Lavin amp Oliveira‐Filho 2009)The timing and potential diversification drivers behind the Neo-

tropics hyperdiversity are still under debate (Antonelli et al 2018)

Geological events during the Neogene (23ndash26 million years Ma)

including the uplift of the northern Andes and the emergence of the

Isthmus of Panama have been suggested as responsible for major

vegetation changes (Antonelli amp Sanmartiacuten 2011 Bacon et al

2015) The rapid climatic fluctuations that characterized the Pleis-

tocene (26ndash001 Ma) have also been suggested as a driver shaping

geographic distributions at interspecies and species levels (Garzoacuten‐Orduntildea Benetti‐Longhini amp Brower 2014 Rull 2008) Palynological

evidence supports vegetation changes in this period (Behling Bush

amp Hooghiemstra 2010 Van der Hammen amp Hooghiemstra 2000)

but whether Pleistocene glacialndashinterglacial cycles affected only spe-

cies distribution ranges or were linked to an increase in speciation

rates is a matter of debate likely dependent on the age and phyto-

geographical adscription of the Neotropical lineage (De‐Nova et al

2012 Hoorn et al 2010 Koenen Clarkson Pennington amp Chatrou

2015) For example molecular time estimates of Amazonian lowland

rain forest taxa support pre‐Pleistocene diversification (Antonelli et

al 2018 Hoorn et al 2010) though the age of extant species may

be younger (Garzoacuten‐Orduntildea et al 2014) For drought‐adaptedNeotropical taxa (Table 1) the age of diversification ranges between

the early Cenozoic and the Pleistocene Most SDTF‐centred lineages

originated during the Late Miocene and Pliocene when arid environ-

ments became dominant in the Americas (De‐Nova et al 2012 Pen-

nington et al 2009) These vegetation types differ also in their

biogeographic patterns moist forests (Amazonian) are dominated by

a few taxa with large distributions suggestive of high migration

rates In contrast woody SDTF species are often confined to small

scattered patches or nuclei and exhibit a phylogenetic geographic

structure compatible with low‐dispersal rates and long‐term isolation

driven by niche conservatism (Pennington et al 2009)

The cacti family Cactaceae are one the most conspicuous ele-

ments of SDTFs and other Neotropical arid and semi‐arid formations

(deserts savannas rocky outcrops highland rocky fields) (Nyffeler amp

Eggli 2010) Comprising 1850 species and 400 genera the family

extends from southwestern USA and Mexico through Central Amer-

ica and the Caribbean to Peru and south‐west Brazil (Nyffeler amp

Eggli 2010) Initial diversification within Cactaceae was dated around

the Late EocenendashOligocene Cooling Event (c 34 Ma Zachos Dick-

ens amp Zeebe 2008) while major cacti lineages (ie subfamilies and

tribes) did not start to diversify until the Mid‐Late Miocene (c 15ndash11 Ma) after a global drop in temperatures led to the expansion of

arid and semi‐arid habitats worldwide (Arakaki et al 2011 Hernaacuten-

dez‐Hernaacutendez Brown Schlumpberger Eguiarte amp Magalloacuten 2014)

Although there are still few generic studies in cacti (Franck

Cochrane amp Garey 2013 Franco et al 2017 Majure et al 2012

Silva Antonelli Lendel Moraes amp Manfrin 2018) estimates from

these and more inclusive family‐level phylogenies (Arakaki et al

2011 Hernaacutendez‐Hernaacutendez et al 2014) indicate an age of origin

of many genera around the PliocenendashPleistocene transition (258 Ma

wwwstratigraphyorg) with species divergences dated well within

the Pleistocene (Bonatelli et al 2014 Franco et al 2017 Silva et

al 2018) These studies suggest that events of rapid climate change

since the Paleogene have been important drivers in the evolution of

cacti As with other SDTF elements widespread cacti genera exhibit

disjunct geographic distributions across the Neotropics with centres

of diversity in Mesoamerica (Mexico and Central America) and north-

eastern South America (Brazil) (Hernaacutendez‐Hernaacutendez et al 2014)

In this work we reconstruct a nearly complete phylogeny (90

species diversity) of the widespread genus Pilosocereus Byles amp Row-

ley (subfamily Cactoideae tribe Cereeae) to infer the patterns of

spatio‐temporal diversification at the infrageneric level in cacti Pilo-

socereus is one of the largest genera of columnar cacti with 42 spe-

cies traditionally dividedmdashbased on morphological charactersmdashinto

two subgenera Pilosocereus subgen Gounellea (with three species

Pilosocereus gounellei Pilosocereus tuberculatus and Pilosocereus fre-

wenii) and Pilosocereus subgen Pilosocereus (with 39 species) (Cal-

vente et al 2017 Zappi 1994) The genus is a characteristic

element of Neotropical arid environments abundant in SDTF habi-

tats as well as xeric microhabitats (rocky outcrops and highland

rocky fields in the Cerrado savanna and rocky sandy soils in the

Brazilian Atlantic Forest) (Zappi 1994) The genus stands out for its

exceptionally widespread distribution compared to other genera in

the family extending from south‐west USA to Mexico and the Carib-

bean towards northern Paraguay Bolivia Peru and south‐west

2 | LAVOR ET AL

Brazil Most species are distributed in two core regions eastern and

central Brazil which harbour the highest species diversity and Cen-

tral America and the Caribbean islands (Taylor amp Zappi 2004 Zappi

1994) As in other cacti genera phylogeographic work has been lim-

ited to the study of patterns within species or closely related species

complexes (Bonatelli et al 2014 Figueredo Nassar Garciacutea‐Rivas ampGonzaacutelez‐Carcaciacutea 2010) Recently Calvente et al (2017) recon-

structed species relationships within the genus based on a data set

of four plastid and one nuclear DNA marker for 33 species of Piloso-

cereus They recovered a polyphyletic Pilosocereus with P gounellei

(subgen Gounellea) and P bohlei (subgen Pilosocereus) more closely

related to representatives of other Cereeae genera than to the

remaining Pilosocereus species However limited taxon sampling

among outgroups outside Cereeae and poor resolution at deeper

nodes within Pilosocereus make it difficult to reconstruct the evolu-

tionary history of the genus

Here we increase the number of molecular markers (adding the

plastid coding gene ycf1) and extend taxon sampling among out-

groups outside Cereeae and within Pilosocereus (from 33 to 38 spe-

cies 90 of total diversity) to generate a robust phylogeny which

is used as a template to estimate lineage divergence times diversifi-

cation rates and historical migration events in the genus We aimed

to answer the following questions (a) Did Quaternary climatic

changes play a central role in driving species divergence (b) Which

was the ancestral range of Pilosocereus and how did it achieve such

a widespread and disjunct distribution (c) Is there support for a pat-

tern of niche conservatism and dispersal limitation in which closely

related species share the same geographic area and habitat as

observed in other SDTF elements (De‐Nova et al 2012 Pennington

et al 2009) (d) Are migration events directional from SDTFs to

other vegetation habitats (De‐Nova et al 2012)

2 | MATERIALS AND METHODS

21 | Taxon sampling DNA amplification andsequencing

Nyffeler and Eggli (2010) ascribed genus Pilosocereus to tribe Cer-

eeae (subtribe Cereinae) which comprises columnar and globular

taxa centred in South America within subfamily Cactoideae the

TABLE 1 Crown ages of drought‐adapted Neotropical taxa with distribution in different xeric environments

Family Genus Clade nameNumber ofspecies

Crownage (Ma) Distribution Environment Source

Anacardiaceae Loxopterygium Loxopterygium 3 spp 8 South America SDTF and moist

forest

(Guianas and

Venezuelan

Guayana)

Pennington

et al (2004)

Boraginaceae Tiquilia Tiquilia 30 spp 331 North and South

America

Desert Moore and

Jansen (2006)

Cactaceae Cereus Cereus c 30 spp 233 South America Typically xeric

environment

Franco

et al (2017)

Cactaceae Harrisia Harrisia 20 spp 175ndash330 South America

and Caribbean

SDTFs Franck

et al (2013)

Cactaceae Opuntia Opuntia ss 150ndash180 spp 56 From North to

South America

Deserts and SDTFs Majure

et al (2012)

Leguminosae Astragalus Clade F 90 spp 189 South America Deserts Scherson Vidal

and Sanderson

(2008)

Leguminosae Chaetocalyx Chaetocalyx subclade 13 spp 59 South America SDTFs Pennington

et al (2004)

Leguminosae Centrolobium Centrolobium 7 spp 99 Northern South

America

Typically SDTFs Pirie et al

(2009)

Leguminosae Nissolia Nissolia 13 spp 16 Mesoamerica SDTFs Pennington

et al (2004)

Leguminosae Prosopis ASP + Xerocladia 43 spp 284 From North to

South America

Arid and semi‐aridenvironments

Catalano Vilardi

Tosto and

Saidman (2008)

Heliotropiaceae Heliotropium Heliotropium Sect

Cochranea

19 spp 14 Peru and Chile Deserts Luebert and

Wen (2008)

Solanaceae Nolana Nolana 89 spp 402 Peru and Chile Lomas formations Dillon Tu Xie

Quipuscoa

Silvestre and

Wen (2009)

LAVOR ET AL | 3

richest subfamily in Cactaceae (see Supporting Information

Appendix S1 for more details on the Study Group) To provide a

phylogenetic backbone for the position of Pilosocereusmdashand to sam-

ple relevant calibration nodes for molecular dating (see below)mdashour

data set included 10 outgroup taxa representing different genera in

tribes Rhipsalideae and Cereeae as well as several species from sub-

tribe Cereinae (sensu Nyffeler amp Eggli 2010) all outgroup taxa (ob-

tained from herbaria or fieldwork) were carefully examined and

identified by the first author or acknowledged experts on Cactaceae

We did not include any representative of subfamily Opuntioideae

sister to Cactoideae due to difficulties in DNA sequencing (presence

of mucus) GenBank sequences were not used because several mark-

ers were missing Pereskia grandifolia was used as the most external

outgroup to root the trees in agreement with other family‐levelphylogenetic studies (Hernaacutendez‐Hernaacutendez et al 2014) Within

Pilosocereus we increased taxon sampling by adding new species

P tuberculatus (subgen Gounellea) P lanuginosus P oligolepis

P piauhyensis and P polygonus (subgen Pilosocereus) and increased

the infrasampling for species P arrabidae P chrysostele P collinsii

P flavipulvinatus P gounellei P magnificus P multicostatus and

P parvus relative to Calvente et al (2017) Supporting Information

Appendix S1 Table S11 lists species names voucher information

and geographic location for all samples

We sequenced six DNA regions (a) four non‐coding intergenic

spacers of plastid DNA (cpDNA) sequenced also by Calvente et al

(2017) trnS‐trnG psbD‐trnT trnL‐trnT petL‐psbE (b) the coding ycf1

gene (new for this studymdasha giant open reading frame that presents

potential to amplification in many Cactaceae (Franck Cochrane amp

Garey 2012) and (c) one low‐copy nuclear gene phytochrome C

(PHYC) Genomic DNA was either extracted from silica‐dried stems

or roots or from herbarium specimens protocols for extraction and

PCR amplification followed Calvente et al (2017) and are listed in

Supporting Information Appendix S1 Table S12 Some regions could

not be sequenced for several species and were coded as missing data

in the matrix (Supporting Information Appendix S1 Table S11) In

total we added 141 new sequences relative to Calvente et al (2017)

molecular data set (comprising 134 sequences) so our final data set

included 275 sequences divided as trnS‐trnG (17 sequences) psbD‐trnT (23 sequences) trnL‐trnT (18 sequences) petL‐psbE (22

sequences) ycf1 (43 sequences) and PHYC (18 sequences) Indels

were coded using Simmons and Ochoterenas (2000) simple coding

method as presenceabsence data (01)

22 | Phylogenetic inference

Phylogenetic relationships were reconstructed using Bayesian infer-

ence implemented in MRBAYES 322 (Ronquist et al 2012) hosted

on the CIPRES Science Gateway (Miller Pfeiffer amp Schwartz

2010) Substitution models were selected using the Akaike infor-

mation criterion (AIC) implemented in MRMODELTEST 22 (Nylander

2004) Bayesian analyses were performed on individual genes

using the selected molecular substitution models GTR GTR+I

GTR+G GTR+I+G F81+I and HKY+G for psbD-trnT petL- psbE

ycf1 trnS-trnG trnL-trnT and PHYC respectively Two analyses of

four chains each were run for 10 times 107 generations sampling

every 1000th Convergence was assessed with the potential scale

reduction factor and by monitoring the standard deviation of split

frequencies (lt001) A 50 majority rule consensus tree was con-

structed after discarding the first 25 samples as burnin Clade

posterior probability values were considered as ldquoweakrdquo support if

PP lt 070 ldquomoderaterdquo support if 070 lt PP gt 095 and high sup-

port if PP gt 095 following Alfaro Zoller and Lutzoni (2003) No

incongruent clades receiving high support were found among the

individual gene trees (Supporting Information Appendix S2 Figures

S21andashf) so we concatenated all markers into a combined plastid

nuclear data set which was used for further analyses Markers

were partitioned by genome (GTR+I+G was used for the plastid

regions HKY+G for PHYC) with the overall substitution rate

unlinked between partitions since plastid DNA generally exhibits

slower evolutionary rates than nuclear DNA (Wolfe Li amp Sharp

1987)

23 | Divergence time estimation

Lineage divergence times were estimated on the partitioned plastid

nuclear data set using Bayesian relaxed clock models implemented in

BEAST 183 (Drummond Suchard Xie amp Rambaut 2012) hosted on

CIPRES Analyses were run with the birthndashdeath (BD) model as tree

prior and the uncorrelated lognormal clock as the clock model We

ran two MCMC chains for 10 times 107 generations sampling every

1000th and used TRACER 16 to monitor convergence and adequate

mixing (Effective Sample Size ESS gt200 Rambaut Suchard Xie amp

Drummond 2014) A maximum clade credibility (MCC) tree (burnin

10) was constructed in TREEANNOTATOR 182 (Drummond amp Ram-

baut 2016)

There are not known fossils of cacti so we used two higher‐levelphylogenetic studies on Cactaceae (Arakaki et al 2011 Hernaacutendez‐Hernaacutendez et al 2014) to obtain secondary age estimates as calibra-

tion points in our phylogeny In these studies dating of Cactaceae

relied also on secondary calibration derived from more inclusive

fossil‐rich analyses of seed plants Hence estimates for major lineage

divergences within Cactaceae differed slightly between Arakaki et al

(2011) and Hernaacutendez‐Hernaacutendez et al (2014) so we decided to

carry out a preliminary sensitivity analysis to assess the influence of

calibration constraints on our age estimates (see Supporting Informa-

tion Appendix S1 for more details)

We set normal distribution priors to accommodate the uncer-

tainty in secondary calibration with the mean of the prior distribu-

tion equal to the mean age estimate in the original study and the SD

spanning the upper and lower bounds of the 95 highest posterior

density (HPD) credibility intervals Schenk (2016) criticized the use

of normal priors in molecular dating analysis because the posterior

distribution of age estimates may differ from that inferred in the

original study To address this we ran preliminary analyses under

our priors and next corroborated in Tracer that there was no signifi-

cant departure (in the mean and 95 HPD interval) between the

4 | LAVOR ET AL

prior and the estimated posterior distribution see Supporting Infor-

mation Appendix S1 for more details

We performed three different analyses (a) Analysis ldquoAE 1rdquo

included two calibration points from Arakaki et al (2011) the crown

age of Cactaceae (the root node in our data set) with M = 286 Ma

SD = 19 (95 HPD 2547ndash3173 Ma) and the crown age of subfam-

ily Cactoideae (the node splitting P grandifolia from all other taxa)

using M = 218 Ma SD = 17 95 HPD 19ndash246 Ma (b) Analysis

ldquoAE 2rdquo included the Hernaacutendez‐Hernaacutendez et al (2014) estimates for

the same calibration points Cactaceae crown age (M = 2688 Ma

SD = 62 95 HPD 1668ndash3708 Ma) and Cactoideae crown age

(M = 1715 Ma SD = 3 95 HPD 1222ndash2208 Ma) (c) Analysis ldquoAE

3rdquo added a third calibration point to AE2 from Hernaacutendez‐Hernaacutendez

et al (2014) the crown node of Cereeae (M = 528 Ma SD = 13

95 HPD 314ndash741 Ma) which corresponds to the node splitting P

grandifolia Copiapoa cinerea and Rhipsalis baccifera from the remain-

ing taxa

The three calibration analyses generated divergence time esti-

mates with overlapping credibility intervals for all major clades

(Table 2) We selected the AE2‐MCC tree for further analyses and

discussion because (a) it generated the narrowest 95 HPD inter-

vals (b) is based on secondary age estimates from a well‐sampled

Cactaceae phylogeny (Hernaacutendez‐Hernaacutendez et al 2014) (c) the lat-

ter study used a Bayesian relaxed clock whereas Arakaki et al

(2011) assumed autocorrelated rates in molecular dating (see Sup-

porting Information Appendix S1)

24 | Diversification and biogeographic analyses

Since Pilosocereus was recovered as non‐monophyletic in our phy-

logeny (see below) diversification analyses were restricted to the

monophyletic clade containing the majority of species Pilosocereus

subgen Pilosocereus sensu stricto (s s) with the exclusion of P boh-

lei We first plotted the pattern of lineage accumulation through time

(LTT) using the R program language (R Core Team 2016) package

ape (Paradis Bolker amp Strimmer 2004) to visually inspect the diver-

sification trajectory We then used whole‐tree episodic BD models

implemented in the R program language (R Core Team 2016) pack-

age lsquoTreeParrsquo (Stadler 2015) to compare the fit of this trajectory to

a BD process with a constant diversification rate (r = speciation

minus extinction) and turnover rate (ε = extinctionspeciation) against

timendashvariable models in which these two parameters change at dis-

crete points in time We estimated the magnitude and times of rate

and turnover rate changes using a grid with 02 Ma discrete time

intervals to detect potential rate shifts We assumed that all lineages

were sampled at shift times (ρ = 1) except at present (ρ = 090) to

account for incomplete taxon sampling and used likelihood ratio

tests to compare models with an increasing number of rate shifts

Biogeographic analyses used the DispersalndashExtinctionndashCladogen-esis model (Ree amp Smith 2008) implemented in a Bayesian frame-

work in RevBayes (Houmlhna et al 2016) This allowed us to estimate

the mean and 95 credibility intervals for the rates of range expan-

sion and local extinction and to compute posterior probability values

TABLE 2 Results from three molecular dating analyses in BEAST using different calibration priors derived from secondary age estimates (seetext for further explanation) only values for clades with high support (PP gt 095) are given Values are given in millions of years (Ma) meandivergence (95 high posterior density [HPD] credibility intervals) numbered nodes are those plotted in Figure 3

Nodes Analysis AE1 Analysis AE2 Analysis AE3

1 2759 (2400ndash3125) 2052 (1308ndash2885) 1803 (1137ndash2601)



















39 136 (ndash) 102 (ndash) 074 (ndash)

LAVOR ET AL | 5

(PP) for ancestral range inheritance scenarios Nine areas were

delimited based on the distribution patterns of species geological

and biome history (Antonelli Nylander Persson amp Sanmartiacuten 2009

Olson et al 2001) and areas of endemism used in previous works

to maximize biogeographic congruence (Givnish et al 2014 Hernaacuten-

dez‐Hernaacutendez et al 2014 Zappi 1994) (A) northeastern Brazil

(equivalent to the Caatinga ecoregion sensu Olson et al 2001) (B)

Central Brazil (equivalent to the Cerrado ecoregion) (C) Coastal Bra-

zil (equivalent to the Brazilian Atlantic Forest ecoregion) (D) North

Brazil (State of Roraima Guiana region) (E) northwestern South

America (F) Central America (G) western Mexico (H) eastern Mex-

ico and (I) the Caribbean Geographic occurrence data were com-

piled from online databases SpeciesLink (CRIA httpsplinkcria

orgbr) Global Biodiversity Information Facility (GBIF httpwww

gbiforg) the Virtual Herbarium REFLORA (Flora do Brasil 2020

httpfloradobrasiljbrjgovbr) totalling 2617 records (Figure 1)

Supporting Information Appendix S1 provides more details on area

definition and corroboration of accuracy of geographic records

Analyses were run for 10000 generations with default values as

defined in the RevBayes tutorial (Michael Landis httprevbayes

githubiotutorialshtml) except for the migration rate parameter

ldquorate_bgrdquo (representing the number of migration events per unit of

time) which was assigned a more restricted (informative) hyperprior

bounding the migration rate between 0001 and 10 events per Ma

Ancestral ranges were constrained to include a maximum of three dis-

crete areas the range size of the most widespread species in the phy-

logeny To explore the influence of geographic distance in dispersal

rates we ran a second analysis in which the migration rate parameter

(ldquorate_bgrdquo) was scaled by the inverse of the geographic distance

between areas according to a factor the ldquodistance_scalerdquo parameter

(a) which was estimated from the data If (a) is ~0 (no scaling) then

the dispersal rate is equal for all areas Pairwise area distances were

estimated in QUANTUM GIS 2140 (QGis ndash Quantum GIS Development

Team 2011) using the geographic centroids of the areas as defined

above Supporting Information Appendix S1 gives more details on

these analyses and provides the corresponding RevBayes scripts

3 | RESULTS

31 | Phylogenetic and divergence time estimates

Phylogenetic relationships among outgroup taxa and within Pilosocer-

eus were largely congruent between the MrBayes tree (Figure 2) and

the BEAST AE2‐MCC tree (Figure 3) with backbone nodes receiving

F IGURE 1 Map showing the geographic distribution of the cacti genus Pilosocereus (Cactoideae Cereeae) with pictures of someemblematic species

6 | LAVOR ET AL

mostly high support Stephanocereus leucostele was recovered as sis-

ter to a polyphyletic Pilosocereus (PP = 1) Within the latter a clade

formed by P gounellei and P tuberculatus (PP = 1) splits first fol-

lowed by P bohlei which is sister to a clade composed of represen-

tatives of other Cereeae genera (Arrojadoa rhodantha Cereus

jamacaru and Melocactus zehntneri PP = 1) and a second clade

grouping the remaining species in Pilosocereus subgen Pilosocereus s

s (PP = 1) Within this clade Pilosocereus aureispinus branches basally

to a clade divided into two well‐supported (PP = 1) subclades Clades

A and B (Figures 2 and 3) Clade A further splits into two smaller

clades each receiving high support (PP = 1) Clade AI comprising

Brazilian species Pilosocereus glaucochrous Pilosocereus pentae-

drophorus and Pilosocereus piauhyensis (Figures 2 and 3 PP = 1) and

Clade AII grouping all extra‐Brazilian species (Figures 2 and 3 PP =

1) Clade B clusters together the remaining Brazilian species (PP = 1)

and is divided into three subclades (Clades BI BII BIII) Relationships

among these are unresolved in the MrBayes tree (Figure 2) and

receive weak to moderate support in the BEAST tree (Figure 3) Other

differences concern the clade Pilosocereus azulensisndashP floccosus

which MrBayes includes in Clade BIII with Pilosocereus fulvilanatus

(Figure 2 PP = 086) and BEAST in Clade BII with Pilosocereus jauruen-

sis (Figure 3 PP = 052)

Table 2 and Figure 3 provide mean and 95 HPD intervals for

major divergences within Pilosocereus in the AE2‐MCC tree The ori-

gin of genus Pilosocereus was dated in the Early Pliocene (c 5 Ma

Figure 3 Table 2) and that of Pilosocereus subgen Pilosocereus s s

around the PliocenendashPleistocene transition (270 Ma) The split

between Clades A and B was dated in the Early Pleistocene Gela-

sian (228 Ma) and that of Clades AI and AII in the Calabrian

(172 Ma) Diversification in Clade B began also in this period

(154 Ma 95 HPD 063ndash281 node 27) Most extant species are

dated as originating in the Middle (078ndash012 Ma) and Upper Pleis-

tocene (012ndash001 Ma) with a few species diverging slightly earlier

(Figure 3)

F IGURE 2 Majority‐rule consensus tree derived from the MrBayes analysis of the concatenated plastidndashnuclear data set (trnS‐trnG psbD‐trnT trnL‐trnT petL‐psbE ycf1 and PHYC) posterior probability (PP) values for clade support are shown above branches numbers for cladesmentioned in the text are given at nodes

LAVOR ET AL | 7

8 | LAVOR ET AL

32 | Diversification and Biogeographic analyses

The LTT plot of Pilosocereus subgen Pilosocereus s s shows an

upshift in the cumulative number of lineages after 2 Ma (Supporting

Information Appendix S2 Figure S22) TreePar however did not

reject a model of constant net diversification (r = 126 Maminus1) and

background extinction (ε = 012) over the one‐ and two‐rate shift

models (p = 073 and 043 respectively) suggesting that the LTT

upshift is an artefact of the ldquopull‐of‐the‐presentrdquo (Stadler 2011) To

incorporate the uncertainty in age estimates we further estimated

net diversification rates for the clade Pilosocereus subgen Pilosocer-

eus s s using the methods‐of‐moments estimator of Magalloacuten and

Sanderson (2001) We used a total clade size of N = 38 (ie assum-

ing that three non‐sampled species of Pilosocereus belonged to this

clade based on morphology) the 95 HPD interval for estimated

clade age 270 (471ndash117) Ma and the value of ε = 012 estimated

by TreePar This gave us a net rate of diversification r = 126 (062ndash251) Maminus1

The unconstrained Bayesian DEC analysis (lnLik = 107189)

estimated rate_bg as 0104 Maminus1 (95 HPD 0053ndash0162) and the

extirpation_rate as 1331 (0272ndash2824) Maminus1 In the distance‐modelled DEC analysis (minus92642) these values were 2230 (0393ndash5253) Maminus1 for rate_bg and 0979 (0322ndash1710) Maminus1 for extirpa-

tion A low value was estimated for the distance_scale parameter

00011 (00006ndash00016) suggesting that the baseline dispersal or

migration rate is independent of geographic distance that is equal

among all areas Bayes factor comparison of the marginal likelihood

estimates of the two models (using the harmonic mean of posterior

likelihoods) favoured the distance‐modelled DEC analysis over the

unconstrained analysis (2lnBF = 291) However exploration in Tra-

cer of the output showed poorer mixing and ESS values for all

parameters in the distance‐modelled analysis This model provided

similar reconstructions to the unconstrained analysis except that

area A was slightly favoured over other ranges for the root node

Results below followed the unconstrained analysis (Figure 4)

The inferred ancestral area of Pilosocereus subgen Pilosocereus s s

(node 1) was ambiguous in the unconstrained analysis with ranges

ACG (northeastern BrazilndashCoastal Brazilndashwestern Mexico) and A

(northeastern Brazil = Caatinga) receiving the highest posterior prob-

abilities (PP lt 020 Figure 4 Supporting Information Appendix S1

Table S11) The most recent common ancestor of Clades A and B

(node 2) was reconstructed as Caatinga with low probability (PP =

010) The first migration event within Clade A was from northeast-

ern Brazil to western Mexico (node 3 Figure 4) Within Clade AI

migration from Caatinga to Coastal Brazil was reconstructed in the

clade Pilosocereus glaucochorusndashP pentaedrophorus (node 5) within

Clade AII there were dispersal events to the Caribbean (Pilosocereus

royeniindashP polygonus node 12 Figure 4) and northwestern South

America (P lanuginosus node 11) Diversification in Clade B (node

14 Figure 4) was preceded by migration from northeastern Brazil to

central (B) and Coastal Brazil (C) followed by back migration events

to the Caatinga and northwards dispersal to area D in P oligolepis

4 | DISCUSSION

41 | PliocenendashPleistocene origin and rapidPleistocene diversification in widespread cacti genera

Our temporal estimates for the origin of the Pilosocereus subgen Pilo-

socereus s s clade (270 Ma Figure 3) and the MiddlendashUpper Pleis-

tocene (078ndash002 Ma) age of divergence for extant species agree well

with those obtained by Franco et al (2017) for the widespread cacti

genus Cereus Comparable divergence time estimates in Cereus hild-

mannianus Mill (Silva et al 2018) and the Pilosocereus aurisetus spe-

cies complex (Bonatelli et al 2014) are probably upwardly biased due

to the use of distant fossil calibrations and improper tree priors when

there is ongoing gene flow (Ho et al 2015) If our age estimates are

right species divergence was rapid with a net diversification rate

resembling those exhibited by some island radiations (Vitales et al

2014) This might explain the weak support for internal nodes recov-

ered in our study (Figure 2) and in other infrageneric studies of cacti

(Bonatelli et al 2014 Franco et al 2017)

The PliocenendashPleistocene transition was a period of major climate

change globally which marked the onset of Quaternary glaciations

(Zachos et al 2008) Phylogenetic studies have supported a role for

Pleistocene climatic changes in structuring geographic patterns among

and within species in Cactaceae (Bonatelli et al 2014 Franco et al

2017 Silva et al 2018) According to the hypothesis of ldquointerglacial

refugiardquo (Bonatelli et al 2014) cacti species would have extended

their geographic ranges during Pleistocene glacial cycles probably via

stepping stone dispersal but became restricted to refugia during the

warmer interglacial periods Large confidence intervals in divergence

time estimates in our phylogeny (Figure 4) prevent establishing a

direct correlation between migration events and specific glacial cycles

Yet it is interesting to note that mean ages for nodes associated to

migration and subsequent allopatric speciation are dated around two

especially dry and severe glacial periods the Anglian Kansas glaciation

c 455000ndash300000 years (nodes 4 11ndash13 20 22ndash24) and the LGM

(21000ndash7000 years nodes 5 32) (Figure 4) This pattern stands in

contrast to that found in other SDTF‐centred genera where species

divergence largely predates the Pleistocene and is therefore inconsis-

tent with a Pleistocene diversification scenario (De‐Nova et al 2012

Pennington et al 2009 Pirie Klitgaard amp Pennington 2009) This

does not discard that intraspecies divergence within widespread SDTF

F IGURE 3 Maximum clade credibility (MCC) tree of genus Pilosocereus and related outgroups showing 95 HPD credibility intervals forphylogenetic relationships and lineage divergence times inferred by BEAST using different calibration constraints (a) Analysis AE2 (b) AnalysisAE1 (c) Analysis AE3 Red asterisks indicate the calibration points used in the different analyses Numbers above branches indicate mean agesthose below branches correspond to posterior probability (PP) values Clades with numbers close to nodes are referenced in Table 2 thosewith names are referred in the text

LAVOR ET AL | 9

taxa could be driven by Quaternary climatic changes (Pennington et

al 2004)

42 | Inter‐region colonization of similar habitats asa driver of diversification in cacti

Evolutionary studies in Cactaceae point to morphological key innova-

tions and the colonization of novel ecological nichesmdasharid and semi‐arid environments that became dominant across the America in the

Neogenemdashas the main drivers behind initial diversification in the

family (Arakaki et al 2011 Hernaacutendez‐Hernaacutendez et al 2014) Sub-

sequent diversification of cacti lineages however was associated to

range expansion and the colonization of geographic regions with

similar xeric conditions and involved reduced morphological variation

(Hernaacutendez‐Hernaacutendez et al 2014) Genus Pilosocereus is one of

few genera of Cereeae that are distributed outside South America

extending across arid habitats from southern Brazil and Peru to

southern USA but exhibits a surprisingly homogeneous morphology

(Zappi 1994 see Figure 1) That morphology alone is not a good

systematic criterion in cacti is suggested by the frequent recovery in

phylogenetic studies of non‐monophyletic genera and species (Bona-

telli et al 2014 Franco et al 2017 this study) This differs from

the pattern found in other Neotropical families (Carvalho amp Renner

2012 Lohmann Bell Calioacute amp Winkworth 2013) where diversifica-

tion has been linked to the colonization of novel habitat types and

concomitant morphological change

Cactaceae is one of the most species‐rich families and a charac-

teristic element of SDTFs (Pennington et al 2009) Woody SDTF‐centred lineages in South America often exhibit a high degree of

niche conservatism with sister‐species living in the same nuclei and

generally a stronger phylogenetic geographic structure and lower

immigration rates than savanna and rain forest taxa When there are

habitat shifts these are predominantly from SDTF patches to other

vegetation types (De‐Nova et al 2012 Pennington et al 2009)

Our reconstruction of the biogeographic history of Pilosocereus sup-

ports a pattern of niche conservatism with diversification

F IGURE 4 (a) Geographic areas used in the biogeographic analysis (b) Bayesian DispersalndashExtinctionndashCladogenesis (DEC) reconstruction ofthe spatio‐temporal evolution of the Pilosocereus subgen Pilosocereus s s clade The phylogeny is the AE2 MCC tree obtained in BEASTnumbers below branches correspond to posterior probability values numbered nodes refer to those listed in Supporting InformationAppendix S2 Table S21 Coloured rectangle close to taxon names indicates the present distributions Range inheritance scenarios arepresented at each node pie graph represents the posterior probability values (PP) for the alternative ancestral ranges squares represent theinherited descendant ranges immediately after speciation for the scenario with the highest posterior probability

10 | LAVOR ET AL

constrained to the same geographic area and habitat type (67 of

the total number of nodes involve in situ diversification) or when

driven by migration events (33 of nodes) these involve other

Neotropical dry formations for example SDTFs in Mesoamerica

(Clade A) or xeric microhabitats embedded within woody savannas

and moist forest biomes for example rocky outcrops and sandy soils

in the Cerrado and the Atlantic Forest in Clade B

The earliest migration events have their origin in the Caatinga (north-

eastern Brazil) the largest SDTF nuclei (Cardoso amp Queiroz 2011) which

was reconstructed in our study as the possible ancestral area of the Pilo-

socereus subgen Pilosocereus s s clade (albeit with large uncertainty Fig-

ure 4) The Caatinga dry formation was climatically more stable than

other regions during the Quaternary (Werneck 2011) and in our study

it is the source area of dispersal events into xeric microhabitats within

nearby regions such as the central Brazilian Cerrado (eg Pilosocereus

splendidusndashPilosocereus albisumums) and the Coastal Brazilian Atlantic

Forest (P glaucochrusndashP peantaedrophorus Figure 4) Although cacti are

sensitive to fire and therefore relatively rare in the Cerrado (where natu-

ral fires occur seasonally) Pilosocereus is an exception because species

are restricted to rocky outcrops where fires cannot penetrate and are

very abundant in the Brazilian Central plateau (Taylor amp Zappi 2004)

Migrations from the Cerrado and Caatinga to Coastal Brazil (Atlantic For-

est) could have involved a coastal plain formation ldquoPortal de Torresrdquo

which acted as a corridor during the Middle Pleistocene (c 04 Ma) link-

ing the central Brazilian savanna with rocky and sandy substrate micro-

habitats in the Brazilian coastal plains (Silva et al 2018) When moist

forests expanded these populations became isolated in ldquoisland‐likerdquo refu-gia inside the Atlantic forest giving rise to endemic sister‐species such as

Pilosocereus azulensis and P brasiliensis (Figure 4)

Some sister‐species in our phylogeny occur in similar habitats but

geographically distant regions for example SDTF patches in north-

western South America and Mesoamerica and the Caribbean (eg P

lanuginosus and P polygonusndashPilosocereus chrysacanthus) or highland

rocky fields and rocky outcrops in the Guiana Shield and central‐east-ern Brazil (eg P oligolepis and P chrysostelendashP flavipulvinatus Fig-ure 4) Indeed DEC suggests that geographic distance is not an

important constraint for the migration rate of Pilosocereus These long‐distance migration events probably involved biotic‐aided dispersal

Chiropterochorymdashdispersal by batsmdashhas been reported in Pilosocer-

eus (Zappi 1994) while Salles et al (2014) describe several fossil bat

species from the Late Miocene to the Holocene with a disjunct distri-

bution in eastern Brazil (mainly rocky caves in the Caatinga) and Cen-

tral America and Mexico similar to the one observed in Pilosocereus

5 | CONCLUSIONS

We found a PliocenendashPleistocene origin for the widespread columnar

cactus Pilosocereus with species divergence occurring as late as the

Middle and Upper Pleistocene in agreement with other infrageneric

studies in cacti Diversification was driven by in situ speciation and

migration events to other SDTF patches and xeric microhabitats

embedded within woody savannas and moist forest biomes

Although our study supports the pattern of phylogenetic niche con-

servatism observed in SDTF‐centred lineages (De‐Nova et al 2012

Pennington et al 2009) it also implies that cacti genera are younger

in age and exhibit a more dynamic migration history probably linked

to vegetation changes during the Pleistocene glacial cycles as well as

long‐distance dispersal events

ACKNOWLEDGEMENTS

We thank CAPES for funding the PhD ldquosandwichrdquo scholarship of PL

with IS CNPq for funding the project ldquoPhylogeny of Pilosocereusrdquo

under AC supervision and project CGL2015‐67849‐P (MINECO

FEDER) for funding IS LMV thanks CNPq for his productivity and

research grants Part of the fieldwork was funded by Proap (CAPES

PPGSE UFRN) and PPBio Semi‐Aacuterido (4574272012‐4) Thanks are

extended to the herbaria MA UFRN and QCNE for access to speci-

mens J Lavor for help during fieldwork M Pace S Arias and T Ter-

razas for samples of Mexican and Central America taxa and S Arias

and M F Freitas for authorizing the publication of their photos

ORCID

Pacircmela Lavor httpsorcidorg0000-0002-2791-6532

Isabel Sanmartin httpsorcidorg0000-0001-6104-9658

REFERENCES

Alfaro M E Zoller S amp Lutzoni F (2003) Bayes or bootstrap A simu-

lation study comparing the performance of Bayesian Markov chain

Monte Carlo sampling and bootstrapping in assessing phylogenetic

confidence Molecular Biology and Evolution 20 255ndash266 httpsdoiorg101093molbevmsg028

Antonelli A Ariza M Albert J Andermann T Azevedo J Bacon C

hellip Edwards S V (2018) Conceptual and empirical advances in

Neotropical biodiversity research Peer 6 e5644 httpsdoiorg

107717peerj5644

Antonelli A Nylander J A A Persson C amp Sanmartiacuten I (2009) Tracing

the impact of the Andean uplift on Neotropical plant evolution Pro-

ceedings of the National Academy of Sciences of the United States of

America 106 9749ndash9754 httpsdoiorg101073pnas0811421106Antonelli A amp Sanmartiacuten I (2011) Why are there so many plant species

in the Neotropics Taxon 60 403ndash414Arakaki M Christin P A Nyffeler R Lendel A Eggli U Ogburn R

M hellip Edwards E J (2011) Contemporaneous and recent radiations

of the worlds major succulent plant lineages Proceedings of the

National Academy of Sciences of the United States of America 108

8379ndash8384 httpsdoiorg101073pnas1100628108Bacon C D Silvestro D Jaramillo C Smith B T Chakrabarty P amp

Antonelli A (2015) Biological evidence supports an early and com-

plex emergence of the Isthmus of Panama Proceedings of the National

Academy of Sciences of the United States of America 112 6110ndash6115httpsdoiorg101073pnas1423853112

Banda K Delgado-Salinas A Dexter K G Linares-Palomino R Oli-

veira-Filho A Prado D hellip Pennington R T (2016) Plant diversity

patterns in neotropical dry forests and their conservation implica-

tions Science 353 1383ndash1387Behling H Bush M amp Hooghiemstra H (2010) Biotic development of

quaternary amazonia A palynological perspective In C Hoorn amp F

P Wesselingh (Eds) Amazonia Landscape and species evolution A

look into the past Oxford UK Wiley-Blackwell

LAVOR ET AL | 11

Bonatelli I A S Perez M F Peterson T Taylor N P Zappi D C

Machado M C hellip Moraes E M (2014) Interglacial microrefugia

and diversification of a cactus species complex Phylogeography and

palaeodistributional reconstructions for Pilosocereus aurisetus and

allies Molecular Ecology 23 3044ndash3063 httpsdoiorg101111

mec12780

Burnham R J amp Graham A (1999) The history of Neotropical vegeta-

tion New developments and status Annals of the Missouri Botanical

Garden 86 546ndash589 httpsdoiorg1023072666185Calvente A Moraes E M Lavor P Bonatelli I A S Nacaguma P

Versieux L M hellip Zappi D C (2017) Phylogenetic analyses of Pilo-

socereus (Cactaceae) inferred from plastid and nuclear sequences

Botanical Journal of the Linnean Society 183 25ndash38Cardoso D B O S amp deQueiroz L P (2011) Caatinga no contexto de

uma metacomunidade evidecircncias da biogeografia padrotildees filogeneacuteti-

cos e abundacircncia de espeacutecies em Leguminosas In C J B Carvalho amp

E A B Almeida (Org) Biogeografia da Ameacuterica do Sul padrotildees e pro-

cessos (pp 241ndash260) Satildeo Paulo Brazil Roca

Carvalho F A amp Renner S S (2012) A dated phylogeny of the papaya

family (Caricaceae) reveals the crops closest relatives and the familysbiogeographic history Molecular Phylogenetics and Evolution 65 46ndash53 httpsdoiorg101016jympev201205019

Catalano S A Vilardi J C Tosto D amp Saidman B O (2008) Molecu-

lar phylogeny and diversification history of Prosopis (Fabaceae Mimo-

soideae) Biological Journal of the Linnean Society 93 621ndash640httpsdoiorg101111j1095-8312200700907x

De-Nova J A Medina R Montero J C Weeks A Rosell J A Olson

M E hellip Magalloacuten S (2012) Insights into the historical construction

of species‐rich Mesoamerican seasonally dry tropical forests The

diversification of Bursera (Burseraceae Sapindales) New Phytologist

193 276ndash287 httpsdoiorg101111j1469-8137201103909xDillon M O Tu T Xie L Quipuscoa Silvestre V amp Wen J (2009)

Biogeographic diversification in Nolana (Solanaceae) a ubiquitous

member of the Atacama and Peruvian Deserts along the western

coast of South America Journal of Systematics and Evolution 47

457ndash476 httpsdoiorg101111j1759-6831200900040xDrummond A J amp Rambaut A (2016) TreeAnnotator v182 [computer

program] Retrieved from httpbeastbioedacuk

Drummond A J Suchard M A Xie D amp Rambaut A (2012) Bayesian

phylogenetics with BEAUti and the BEAST 17 [computer program]

Molecular Biology and Evolution 29 1969ndash1973 httpsdoiorg10

1093molbevmss075

Figueredo C J Nassar J M Garciacutea-Rivas A E amp Gonzaacutelez-Carcaciacutea J

A (2010) Population genetic diversity and structure of Pilosocereus

tillianus (Cactaceae Cereeae) a columnar cactus endemic to the

Venezuelan Andes Journal of Arid Environments 74 1392ndash1398httpsdoiorg101016jjaridenv201005020

Franck A R Cochrane B J amp Garey J R (2012) Low‐copy nuclear pri-

mers and ycf1 primers in Cactaceae American Journal of Botany 99

e405ndashe407 httpsdoiorg103732ajb1200128Franck A R Cochrane B J amp Garey J R (2013) Phylogeny biogeogra-

phy and infrageneric classification of Harrisia (Cactaceae) Systematic

Botany 38 210ndash223 httpsdoiorg101600036364413X662105Franco F F Silva G A R Moraes E M Taylor N Zappi D C

Jojima C L amp Machado M C (2017) Plio‐Pleistocene diversifica-

tion of Cereus Mill (Cactaceae Cereeae) and closely allied genera

Botanical Journal of Linnean Society 183 199ndash210 httpsdoiorg

101093botlinneanbow010

Garzoacuten-Orduntildea I J Benetti-Longhini J E amp Brower A V (2014) Tim-

ing the diversification of the Amazonian biota Butterfly divergences

are consistent with Pleistocene refugia Journal of Biogeography 41

1631ndash1638 httpsdoiorg101111jbi12330Givnish T J Barfuss M H J Ee B V Riina R Schulte K Horres R

hellip Sytsma K J (2014) Adaptive radiation correlated and contingent

evolution and net species diversification in Bromeliaceae Molecular

Phylogenetic and Evolution 71 55ndash78 httpsdoiorg101016jympe

v201310010

Hernaacutendez-Hernaacutendez T Brown J W Schlumpberger B O Eguiarte

L E amp Magalloacuten S (2014) Beyond aridification Multiple explana-

tions for the elevated diversification of cacti in the New World Suc-

culent Biome New Phytologist 202 1382ndash1397 httpsdoiorg101111nph12752

Ho S Y W Tong K J Foster C S P Ritchie A M Lo N amp Crisp

M D (2015) Biogeographic calibrations for the molecular clock

Biology Letters 11 20150194 httpsdoiorg101098rsbl2015

0194

Houmlhna S Landis M J Heath T A Boussau B Lartillot N Moore B

R hellip Ronquist F (2016) RevBayes Bayesian phylogenetic inference

using graphical models and an interactive model‐specification lan-

guage Systematic Biology 65 726ndash736 httpsdoiorg101093sysbiosyw021

Hoorn C Wesselingh F P ter Steege H Bermudez M A Mora A

Sevink J hellip Antonelli A (2010) Amazonia through time Andean

uplift climate change landscape evolution and biodiversity Science

330 927ndash931 httpsdoiorg101126science1194585Hughes C E Pennington R T amp Antonelli A (2013) Neotropical plant

evolution Assembling the big picture Botanical Journal of the Linnean

Society 171 1ndash18 httpsdoiorg101111boj12006Kier G Mutke J Dinerstein E Ricketts T H Kuper W Kreft H amp

Barthlott W (2005) Global patterns of plant diversity and floristic

knowledge Journal of Biogeography 32 1107ndash1116 httpsdoiorg101111j1365-2699200501272x

Koenen E J M Clarkson J J Pennington T D amp Chatrou L W

(2015) Recently evolved diversity and convergent radiations of rain-

forest mahoganies (Meliaceae) shed new light on the origins of rain-

forest hyperdiversity New Phytologist 207 327ndash339 httpsdoiorg101111nph13490

Lohmann L G Bell C D Calioacute M F amp Winkworth R C (2013) Pat-

tern and timing of biogeographical history in the Neotropical tribe

Bignonieae (Bignoniaceae) Botanical Journal of the Linnean Society

171 154ndash170 httpsdoiorg101111j1095-8339201201311xLuebert F amp Wen J (2008) Phylogenetic analysis and evolutionary

diversification of Heliotropium sect Cochranea (Heliotropiaceae) in the

Atacama Desert Systematic Botany 33 390ndash402 httpsdoiorg101600036364408784571635

Magalloacuten S amp Sanderson M J (2001) Absolute diversification rates in

angiosperm clades Evolution 55 1762ndash1780 httpsdoiorg10

1111j0014-38202001tb00826x

Majure L C Puente R Griffith M P Judd W S Soltis P S amp Soltis

D E (2012) Phylogeny of Opuntia ss (Cactaceae) Clade delineation

geographic origins and reticulate evolution American Journal of Bot-

any 99 847ndash864 httpsdoiorg103732ajb1100375Miller M A Pfeiffer W amp Schwartz T (2010) Creating the CIPRES

science gateway for inference of large phylogenetic trees In 2010

Gateway Computing Environments Workshop (GCE) (pp 1ndash8) Retrievedfrom httpsieeexploreieeeorgdocument5676129

Moore M J amp Jansen R K (2006) Molecular evidence for the age ori-

gin and evolutionary history of the American desert plant genus

Tiquilia (Boraginaceae) Molecular Phylogenetics and Evolution 39

668ndash687 httpsdoiorg101016jympev200601020

Nyffeler R amp Eggli U (2010) A farewell to dated ideas and concepts ndashMolecular phylogenetics and a revised suprageneric classification of

the family Cactaceae Schumannia 6 109ndash149Nylander J A A (2004) MrModeltest v2 [computer program] Uppsala

Sweden Evolutionary Biology Centre Uppsala University

Olson D M Dinerstein E Wikramanayake E D Burgess N D Pow-

ell G V N Underwood E C hellip Kassem K R (2001) Terrestrial

ecoregions of the world A new map of life on Earth BioScience 51

933ndash938httpsdoiorg1016410006-3568(2001)051[0933TEOTWA]20CO2

12 | LAVOR ET AL

Paradis E Bolker B amp Strimmer K (2004) APE Analysis of phyloge-

netics and evolution in R language Bioinformatics 20 289ndash290 R

package version 40 Retrieved from httpscranr-projectorgweb

packagesape httpsdoiorg101093bioinformaticsbtg412

Pennington R T Lavin M amp Oliveira-Filho A (2009) Woody plant

diversity evolution and ecology in the tropics Perspectives from

seasonally dry tropical forests Annual Review of Ecology Evolution

and Systematics 40 437ndash457 httpsdoiorg101146annurevecolsys110308120327

Pennington R T Lavin M Prado D E Pendry C A Pell S K amp But-

terworth C A (2004) Historical climate change and speciation

Neotropical seasonally dry forest plants show patterns of both Ter-

tiary and Quaternary diversification Philosophical Transactions of the

Royal Society B Biological Sciences 359 515ndash538 httpsdoiorg101098rstb20031435

Pirie M D Klitgaard B B amp Pennington R T (2009) Revision and bio-

geography of Centrolobium (Leguminosae ndash Papilionoideae) Systematic

Botany 34 345ndash359 httpsdoiorg101600036364409788606262QGis ndash Quantum GIS Development Team (2011) Quantum GIS geo-

graphic information system Open Source Geospatial Foundation Pro-

ject [computer program]

R Core Team (2016) R A language and environment for statistical com-

puting Vienna Austria R Foundation for Statistical Computing

Retrieved from httpswwwR-projectorg

Rambaut A Suchard M A Xie D amp Drummond A J (2014) Tracer

v16 Retrieved from httpsgithubcombeast-devtracerreleases

Ree R H amp Smith S A (2008) Maximum‐likelihood inference of

geographic range evolution by dispersal local extinction and

cladogenesis Systematic Biology 57 4ndash14 httpsdoiorg10108010635150701883881

Ronquist F Teslenko M van der Mark P Ayres D Darling A Houmlhna

S hellip Huelsenbeck J P (2012) MrBayes 32 Efficient Bayesian phylo-

genetic inference and model choice across a large model space System-

atic Biology 61 539ndash542 httpsdoiorg101093sysbiosys029Rull V (2008) Speciation timing and neotropical biodiversity The Ter-

tiary‐Quaternary debate in the light of molecular phylogenetic evi-

dence Molecular Ecology 17 2722ndash2729 httpsdoiorg101111j1365-294X200803789x

Salles L O Arroyo-Cabrales J Lima A C D M Lanzelotti W Perini

F A Velazco P M amp Simmons N B (2014) Quaternary bats from

the Impossiacutevel‐Ioiocirc Cave System (Chapada Diamantina Brazil) Hum-

eral remains and the first fossil record of Noctilio leporinus (Chi-

roptera Noctilionidae) from South America American Museum

Novitates 3798 1ndash32 httpsdoiorg10120637981Schenk J J (2016) Consequences of secondary calibrations on diver-

gence time estimates PLoS ONE 11 e0148228 httpsdoiorg10

1371journalpone0148228

Scherson R A Vidal R amp Sanderson M J (2008) Phylogeny biogeog-

raphy and rates of diversification of New World Astragalus (Legumi-

nosae) with an emphasis on South American radiations American

Journal of Botany 95 1030ndash1039 httpsdoiorg103732ajb

0800017

Silva G A R Antonelli A Lendel A Moraes E M amp Manfrin M H

(2018) The impact of early Quaternary climate change on the diversifi-

cation and population dynamics of a South American cactus species

Journal of Biogeography 45 76ndash88 httpsdoiorg101111jbi13107Simmons M P amp Ochoterena H (2000) Gaps as characters in

sequence‐based phylogenetic analyses Systematic Biology 49 369ndash381 httpsdoiorg101093sysbio492369

Stadler T (2011) Mammalian phylogeny reveals recent diversification

rate shifts Proceedings of the National Academy of Sciences of the Uni-

ted States of America 108 6187ndash6192 httpsdoiorg101073pnas1016876108

Stadler T (2015) TreePar Estimating birth and death rates based on

phylogenies R package version 33 Retrieved from httpscranr-pro

jectorgwebpackagesTreePar

Taylor N P amp Zappi D C (2004) Cacti of Eastern Brazil London UK

Royal Botanic Gardens Kew

Van der Hammen T amp Hooghiemstra H (2000) Neogene and quater-

nary history of vegetation climate and plant diversity in Amazonia

Quaternary Science Reviews 19 725ndash742 httpsdoiorg101016

S0277-3791(99)00024-4

Vitales D Garnatje T Pellicer J Vallegraves J Santos-Guerra A amp San-

martiacuten I (2014) The explosive radiation of Cheirolophus (Asteraceae

Cardueae) in Macaronesia BMC Evolutionary Biology 14 118

httpsdoiorg1011861471-2148-14-118

Werneck F P (2011) The diversification of eastern South American

open vegetation biomes Historical biogeography and perspectives

Quaternary Science Reviews 30 1630ndash1648 httpsdoiorg101016jquascirev201103009

Wolfe K Li H W H amp Sharp P M (1987) Rates of nucleotide substi-

tution vary greatly among plant mitochondrial chloroplast and

nuclear DNAs Proceedings of the National Academy of Sciences of the

United States of America 84 9054ndash9058 httpsdoiorg101073

pnas84249054

Zachos J C Dickens G R amp Zeebe R E (2008) An early Cenozoic

perspective on greenhouse warming and carbon‐cycle dynamics Nat-

ure 451 279ndash283 httpsdoiorg101038nature06588Zappi D C (1994) Pilosocereus (Cactaceae) The genus in Brazil Succu-

lent Plant Research 3 1ndash160

BIOSKETCH

Pamela Lavor is a researcher working on the systematics and

evolution of plants The other authors are specialists on system-

atics of Cactaceae (AC) and analytical biogeographic methods

(IS) collaborating in disentangling the origins of the Neotropical

xeric flora

Author contributions PL AC and IS conceived the study PL

performed molecular sequencing and phylogenetic analyses with

help from AC and IS IS performed the biogeographic analy-

ses PL and IS wrote the manuscript with contributions from

AC and LMV

SUPPORTING INFORMATION

Additional supporting information may be found online in the

Supporting Information section at the end of the article

How to cite this article Lavor P Calvente A Versieux LM

Sanmartin I Bayesian spatio‐temporal reconstruction reveals

rapid diversification and Pleistocene range expansion in the

widespread columnar cactus Pilosocereus J Biogeogr

2018001ndash13 httpsdoiorg101111jbi13481

LAVOR ET AL | 13

K E YWORD S

biogeography cacti diversification Neotropical region Northeastern Brazil Pleistocene

climatic changes

1 | INTRODUCTION

The Neotropics are among the regions with the greatest floristic

diversity in the world (Kier et al 2005) Throughout the region a

variety of plant formations are arranged in alternating patches of

vegetation of different size creating a mosaic‐like landscape (Olson

et al 2001) These include moist (eg tropical rain forests and wet-

lands) and drier habitat types such as seasonally dry tropical forests

(SDTFs) savannas and rocky fields high‐elevation Andean grasslands

and deserts all embedded within a complex physiography with a

variety of edaphic climatic and topographic conditions (Antonelli amp

Sanmartiacuten 2011 Burnham amp Graham 1999 Hughes Pennington amp

Antonelli 2013) Among these the extremely rich tropical lowland

forests of the Amazonian region have attracted the attention of bio-

geographic studies (Hoorn et al 2010 Rull 2008) but there is

increasing interest in other biomes especially SDTF dry formations

characterized by high levels of beta diversity (Banda et al 2016

Pennington Lavin amp Oliveira‐Filho 2009)The timing and potential diversification drivers behind the Neo-

tropics hyperdiversity are still under debate (Antonelli et al 2018)

Geological events during the Neogene (23ndash26 million years Ma)

including the uplift of the northern Andes and the emergence of the

Isthmus of Panama have been suggested as responsible for major

vegetation changes (Antonelli amp Sanmartiacuten 2011 Bacon et al

2015) The rapid climatic fluctuations that characterized the Pleis-

tocene (26ndash001 Ma) have also been suggested as a driver shaping

geographic distributions at interspecies and species levels (Garzoacuten‐Orduntildea Benetti‐Longhini amp Brower 2014 Rull 2008) Palynological

evidence supports vegetation changes in this period (Behling Bush

amp Hooghiemstra 2010 Van der Hammen amp Hooghiemstra 2000)

but whether Pleistocene glacialndashinterglacial cycles affected only spe-

cies distribution ranges or were linked to an increase in speciation

rates is a matter of debate likely dependent on the age and phyto-

geographical adscription of the Neotropical lineage (De‐Nova et al

2012 Hoorn et al 2010 Koenen Clarkson Pennington amp Chatrou

2015) For example molecular time estimates of Amazonian lowland

rain forest taxa support pre‐Pleistocene diversification (Antonelli et

al 2018 Hoorn et al 2010) though the age of extant species may

be younger (Garzoacuten‐Orduntildea et al 2014) For drought‐adaptedNeotropical taxa (Table 1) the age of diversification ranges between

the early Cenozoic and the Pleistocene Most SDTF‐centred lineages

originated during the Late Miocene and Pliocene when arid environ-

ments became dominant in the Americas (De‐Nova et al 2012 Pen-

nington et al 2009) These vegetation types differ also in their

biogeographic patterns moist forests (Amazonian) are dominated by

a few taxa with large distributions suggestive of high migration

rates In contrast woody SDTF species are often confined to small

scattered patches or nuclei and exhibit a phylogenetic geographic

structure compatible with low‐dispersal rates and long‐term isolation

driven by niche conservatism (Pennington et al 2009)

The cacti family Cactaceae are one the most conspicuous ele-

ments of SDTFs and other Neotropical arid and semi‐arid formations

(deserts savannas rocky outcrops highland rocky fields) (Nyffeler amp

Eggli 2010) Comprising 1850 species and 400 genera the family

extends from southwestern USA and Mexico through Central Amer-

ica and the Caribbean to Peru and south‐west Brazil (Nyffeler amp

Eggli 2010) Initial diversification within Cactaceae was dated around

the Late EocenendashOligocene Cooling Event (c 34 Ma Zachos Dick-

ens amp Zeebe 2008) while major cacti lineages (ie subfamilies and

tribes) did not start to diversify until the Mid‐Late Miocene (c 15ndash11 Ma) after a global drop in temperatures led to the expansion of

arid and semi‐arid habitats worldwide (Arakaki et al 2011 Hernaacuten-

dez‐Hernaacutendez Brown Schlumpberger Eguiarte amp Magalloacuten 2014)

Although there are still few generic studies in cacti (Franck

Cochrane amp Garey 2013 Franco et al 2017 Majure et al 2012

Silva Antonelli Lendel Moraes amp Manfrin 2018) estimates from

these and more inclusive family‐level phylogenies (Arakaki et al

2011 Hernaacutendez‐Hernaacutendez et al 2014) indicate an age of origin

of many genera around the PliocenendashPleistocene transition (258 Ma

wwwstratigraphyorg) with species divergences dated well within

the Pleistocene (Bonatelli et al 2014 Franco et al 2017 Silva et

al 2018) These studies suggest that events of rapid climate change

since the Paleogene have been important drivers in the evolution of

cacti As with other SDTF elements widespread cacti genera exhibit

disjunct geographic distributions across the Neotropics with centres

of diversity in Mesoamerica (Mexico and Central America) and north-

eastern South America (Brazil) (Hernaacutendez‐Hernaacutendez et al 2014)

In this work we reconstruct a nearly complete phylogeny (90

species diversity) of the widespread genus Pilosocereus Byles amp Row-

ley (subfamily Cactoideae tribe Cereeae) to infer the patterns of

spatio‐temporal diversification at the infrageneric level in cacti Pilo-

socereus is one of the largest genera of columnar cacti with 42 spe-

cies traditionally dividedmdashbased on morphological charactersmdashinto

two subgenera Pilosocereus subgen Gounellea (with three species

Pilosocereus gounellei Pilosocereus tuberculatus and Pilosocereus fre-

wenii) and Pilosocereus subgen Pilosocereus (with 39 species) (Cal-

vente et al 2017 Zappi 1994) The genus is a characteristic

element of Neotropical arid environments abundant in SDTF habi-

tats as well as xeric microhabitats (rocky outcrops and highland

rocky fields in the Cerrado savanna and rocky sandy soils in the

Brazilian Atlantic Forest) (Zappi 1994) The genus stands out for its

exceptionally widespread distribution compared to other genera in

the family extending from south‐west USA to Mexico and the Carib-

bean towards northern Paraguay Bolivia Peru and south‐west

2 | LAVOR ET AL








































forest

(Guianas and

Venezuelan

Guayana)

Pennington

et al (2004)


America

Desert Moore and

Jansen (2006)


environment

Franco

et al (2017)


and Caribbean

SDTFs Franck

et al (2013)


South America


et al (2012)


and Sanderson

(2008)


et al (2004)


America


(2009)


et al (2004)


South America


Catalano Vilardi

Tosto and

Saidman (2008)


Cochranea


Wen (2008)


Quipuscoa

Silvestre and

Wen (2009)

LAVOR ET AL | 3



































































1987)












baut 2016)




















4 | LAVOR ET AL

















ing taxa
























































LAVOR ET AL | 5





































3 | RESULTS






6 | LAVOR ET AL


































(Figure 3)


LAVOR ET AL | 7

8 | LAVOR ET AL

















































4 | DISCUSSION







































LAVOR ET AL | 9


al 2004)































10 | LAVOR ET AL








































5 | CONCLUSIONS













ACKNOWLEDGEMENTS











ORCID



REFERENCES








107717peerj5644




















LAVOR ET AL | 11






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13








































forest

(Guianas and

Venezuelan

Guayana)

Pennington

et al (2004)


America

Desert Moore and

Jansen (2006)


environment

Franco

et al (2017)


and Caribbean

SDTFs Franck

et al (2013)


South America


et al (2012)


and Sanderson

(2008)


et al (2004)


America


(2009)


et al (2004)


South America


Catalano Vilardi

Tosto and

Saidman (2008)


Cochranea


Wen (2008)


Quipuscoa

Silvestre and

Wen (2009)

LAVOR ET AL | 3



































































1987)












baut 2016)




















4 | LAVOR ET AL

















ing taxa
























































LAVOR ET AL | 5





































3 | RESULTS






6 | LAVOR ET AL


































(Figure 3)


LAVOR ET AL | 7

8 | LAVOR ET AL

















































4 | DISCUSSION







































LAVOR ET AL | 9


al 2004)































10 | LAVOR ET AL








































5 | CONCLUSIONS













ACKNOWLEDGEMENTS











ORCID



REFERENCES








107717peerj5644




















LAVOR ET AL | 11






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13



































































1987)












baut 2016)




















4 | LAVOR ET AL

















ing taxa
























































LAVOR ET AL | 5





































3 | RESULTS






6 | LAVOR ET AL


































(Figure 3)


LAVOR ET AL | 7

8 | LAVOR ET AL

















































4 | DISCUSSION







































LAVOR ET AL | 9


al 2004)































10 | LAVOR ET AL








































5 | CONCLUSIONS













ACKNOWLEDGEMENTS











ORCID



REFERENCES








107717peerj5644




















LAVOR ET AL | 11






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13

















ing taxa
























































LAVOR ET AL | 5





































3 | RESULTS






6 | LAVOR ET AL


































(Figure 3)


LAVOR ET AL | 7

8 | LAVOR ET AL

















































4 | DISCUSSION







































LAVOR ET AL | 9


al 2004)































10 | LAVOR ET AL








































5 | CONCLUSIONS













ACKNOWLEDGEMENTS











ORCID



REFERENCES








107717peerj5644




















LAVOR ET AL | 11






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13





































3 | RESULTS






6 | LAVOR ET AL


































(Figure 3)


LAVOR ET AL | 7

8 | LAVOR ET AL

















































4 | DISCUSSION







































LAVOR ET AL | 9


al 2004)































10 | LAVOR ET AL








































5 | CONCLUSIONS













ACKNOWLEDGEMENTS











ORCID



REFERENCES








107717peerj5644




















LAVOR ET AL | 11






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13


































(Figure 3)


LAVOR ET AL | 7

8 | LAVOR ET AL

















































4 | DISCUSSION







































LAVOR ET AL | 9


al 2004)































10 | LAVOR ET AL








































5 | CONCLUSIONS













ACKNOWLEDGEMENTS











ORCID



REFERENCES








107717peerj5644




















LAVOR ET AL | 11






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13

8 | LAVOR ET AL

















































4 | DISCUSSION







































LAVOR ET AL | 9


al 2004)































10 | LAVOR ET AL








































5 | CONCLUSIONS













ACKNOWLEDGEMENTS











ORCID



REFERENCES








107717peerj5644




















LAVOR ET AL | 11






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13

















































4 | DISCUSSION







































LAVOR ET AL | 9


al 2004)































10 | LAVOR ET AL








































5 | CONCLUSIONS













ACKNOWLEDGEMENTS











ORCID



REFERENCES








107717peerj5644




















LAVOR ET AL | 11






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13


al 2004)































10 | LAVOR ET AL








































5 | CONCLUSIONS













ACKNOWLEDGEMENTS











ORCID



REFERENCES








107717peerj5644




















LAVOR ET AL | 11






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13








































5 | CONCLUSIONS













ACKNOWLEDGEMENTS











ORCID



REFERENCES








107717peerj5644




















LAVOR ET AL | 11






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13






mec12780





























1093molbevmss075





















v201310010








0194

























1111j0014-38202001tb00826x


















12 | LAVOR ET AL













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13













































0800017

















S0277-3791(99)00024-4




httpsdoiorg1011861471-2148-14-118








pnas84249054





BIOSKETCH





xeric flora





AC and LMV









LAVOR ET AL | 13

Bayesian spatio‐temporal reconstruction reveals rapid ... · 3Real Jardín Botánico, RJB-CSIC,...

Documents

Transcript of Bayesian spatio‐temporal reconstruction reveals rapid ... · 3Real Jardín Botánico, RJB-CSIC,...