The Contribution of Temperature and Continental Fragmentation to Amphibian Diversification

Received: 25 May 2018 | Revised: 30 March 2019 | Accepted: 2 April 2019 DOI: 10.1111/jbi.13592

RESEARCH PAPER

The contribution of temperature and continental fragmentation to amphibian diversification

Jonathan Rolland1 | Fabien L. Condamine2

1Department of Zoology, University of British Columbia, Vancouver, British Abstract Columbia, Canada Aim: Abiotic factors such as global temperature or continental fragmentation may 2 CNRS, UMR 5554, Institut des Sciences favour speciation through the ecological and geographical isolation of lineages, but de l'Evolution de Montpellier (Université de Montpellier), Montpellier, France macroevolutionary quantifications of such effect with both fossil and phylogenetic data are rarely performed. Here, we propose to use biogeographical estimations and Correspondence Jonathan Rolland, Department of Zoology, palaeo‐environmental diversification models to estimate whether and how palae‐ University of British Columbia, #4200‐6270 otemperature and the sequential break‐ups of Pangaea, Gondwana and Laurasia University Blvd., Vancouver, BC, Canada. Email: [email protected] have affected the diversification of amphibians through time. Location: Global. Funding information University of British Columbia, Grant/Award Methods: Using a time‐calibrated phylogeny for 3,309 amphibian species and a Number: 151042; Agence Nationale de la genus‐level fossil record, we estimated the diversification rates of the group with Recherche, Grant/Award Number: ANR‐10‐ LABX‐25‐01 birth–death models allowing rates to depend on the temporal variations of the envi‐ ronment. We used estimates of global palaeotemperature and an index of continental Editor: Richard Ree fragmentation through time to test the association between speciation and/or ex‐ tinction rates and past temperature and fragmentation. We also estimated the bio‐ geographical history based on a time‐stratified parametric model informed by the global palaeogeography. We inferred whether vicariance or dispersal events ex‐ plained the ancient and current geographical distribution of amphibians. Results: The diversification analyses on the whole amphibians showed that tempera‐ ture‐dependent models are better supported than tectonic‐dependent, time‐de‐ pendent and constant‐rate models for both the fossil and phylogenetic data. The best‐fitting temperature‐dependent model indicated a positive dependence of both speciation and extinction rates with the temperature through time. Biogeographical analyses indicated a Pangaean origin for amphibians and also showed that allopatric speciation (vicariance) explained important phases of the evolution of geographical ranges in the Mesozoic. Main conclusions: Our results support that palaeotemperatures have positively im‐ pacted amphibian diversification. Our study provides additional insights into how to quantify the effect of the landmass fragmentation on the diversification processes and shows with biogeographical reconstruction that continental fragmentation is linked to allopatric speciation in the early history of the clade.

Jonathan Rolland and Fabien Condamine contributed equally to the study.

KEYWORDS Anura, birth–death, Caudata, extinction, Gymnophiona, landmass fragmentation, macroevolution, plate tectonics, speciation

1 | INTRODUCTION linking explicitly the speciation rate with the number of landmasses through time has been addressed by some studies with fossil data Current species diversity has originated through the diversification and found mixed support for a link between diversification and con‐ process of ancestral lineages (Benton, 2015). The rates at which spe‐ tinental fragmentation through time (Jordan et al., 2016; Lehtonen cies originate and vanish have drastically varied through time and et al., 2017; Leprieur et al., 2016; Vavrek, 2016; Zaffos et al., 2017). across clades (Alfaro et al., 2009; Morlon, Parsons, & Plotkin, 2011; For example, Zaffos et al. (2017) found a positive correlation be‐ Rabosky, Slater, & Alfaro, 2012; Stadler, 2011). Explaining why di‐ tween global marine invertebrate generic richness and an inde‐ versification rates (speciation minus extinction) vary through time pendently derived quantitative index describing the fragmentation is a challenging task because many factors can alter the pace of of continental crust. On the contrary, Jordan et al. (2016) concluded speciation and extinction (Benton, 2009, 2015). Factors affecting that continental drift is not sufficient to account for the increase in the diversification of lineages are often placed in two categories: (a) terrestrial species richness observed in the fossil record. Tectonic the abiotic factors (Barnosky, 2001) and (b) the biotic factors (Van changes have also been examined recently by studies evaluating the Valen, 1973). Thanks to recent methodological developments, the effect of mountain orogeny on diversification using phylogenies of relative contribution of abiotic and biotic factors to the variation of plants (Lagomarsino et al., 2016; Pérez‐Escobar et al., 2017) and an‐ speciation and extinction rates has been assessed in various groups imals (Condamine et al., 2018; Xu, Kuntner, Liu, Chen, & Li, 2018). using fossil data (Ezard, Aze, Pearson, & Purvis, 2011; Lehtonen A second important abiotic factor likely playing an important role et al., 2017; Liow, Reitan, & Harnik, 2015; Roalson & Roberts, in species diversification is temperature. Previous palaeontological 2016; Silvestro, Antonelli, Salamin, & Quental, 2015) or phyloge‐ studies have already indicated that palaeotemperature had a posi‐ netic data (Condamine, Rolland, Höhna, Sperling, & Sanmartín, tive effect on biodiversity (Erwin, 2009; Ezard et al., 2011; Jaramillo, 2018; Condamine, Sperling, Wahlberg, Rasplus, & Kergoat, 2012; Rueda, & Mora, 2006; Mayhew, Bell, Benton, & McGowan, 2012). Lagomarsino, Condamine, Antonelli, Mulch, & Davis, 2016; Pérez‐ Temperature has been related to higher rate of speciation and lower Escobar et al., 2017). rate of extinction in plants (Jaramillo et al., 2006) and in macroper‐ Among the abiotic factors potentially impacting diversification, forate planktonic foraminifera (Ezard et al., 2011). The reasons why one can distinguish factors related to geography with the fragmenta‐ temperature should be related to diversification rates are still de‐ tion and the movements of landmasses (Valentine & Moores, 1970; bated. The metabolic theory proposes that temperature is associated Zaffos, Finnegan, & Peters, 2017), and factors related to climate and positively with faster rate of mutation, smaller generation time and palaeotemperature (Erwin, 2009; Hannisdal & Peters, 2011; Peters, ultimately higher speciation rate (Allen, Gillooly, Savage, & Brown, 2008). First, plate tectonics and continental fragmentation are likely 2006; Brown, Gillooly, Allen, Savage, & West, 2004). Hypotheses instrumental for understanding species diversification since evolu‐ related to the latitudinal diversity gradient have also proposed that tionary biologists often evidence the role of allopatric speciation warmer tropical regions have more stable climate through geolog‐ at small scales (Baselga, Recuero, Parra‐Olea, & Garcia‐Paris, 2011; ical scales, which might have considerably decreased the rates of Coyne & Orr, 2004; Funk, 1998), while biogeographers stress the extinction, comparatively to temperate regions at higher latitudes role of vicariance at large scales (Leprieur et al., 2016; Lomolino, (Dynesius & Jansson, 2000; Pianka, 1966; Rohde, 1992). Riddle, Whittaker, & Brown, 2010; Losos & Glor, 2003; Mao et al., Although the role of temperature has been already tested in 2012; Orr & Smith, 1998). The break‐up and isolation of landmasses several clades (Condamine, Rolland, & Morlon, 2013; Condamine might trigger the divergence of spatially isolated populations. A et al., 2018; Pyron & Wiens, 2013; Roalson & Roberts, 2016), the positive association between geographical fragmentation and net link between diversification and continental fragmentation remains diversification rates has already been proposed for global marine seldom tested empirically with birth–death models. Recent develop‐ biodiversity (Belanger et al., 2012; Jablonski & Bottjer, 1991; Peters, ments in the field of diversification models permit to assess the role 2005; Peters, Kelly, & Fraass, 2013; Valentine & Moores, 1970), for of variables that vary through time on the diversification processes mammalian lineages in the Late Cretaceous (100.5–66 million years (Condamine et al., 2013; Morlon et al., 2016). Given this progress, ago, Ma; Hedges, Parker, Sibley, & Kumar, 1996), and in Cupressaceae it is now possible to test whether the palaeotemperature and the (gymnosperms) during the break‐up of Pangaea (183–124 Ma; Mao number of landmasses have affected speciation and/or extinction et al., 2012). Despite a large number of biogeographical studies sug‐ through time in a given clade. To address this question, amphibi‐ gesting that continental fragmentation might impact diversification ans (frogs, salamanders and caecilians) are a relevant clade because (Cermeño, Benton, Paz, & Vérard, 2017; Hedges et al., 1996; Jordan, their relatively long evolutionary history (ca. 300–270 Ma; Feller & Barraclough, & Rosindell, 2016; Mao et al., 2012), the hypothesis Hedges, 1998; Feng et al., 2017; Pyron, 2014) has allowed long‐term ROLLAND and CONDAMINE | 3 fluctuations of temperature and number of landmasses to possibly model of range evolution (DEC; Ree & Smith, 2008) implemented affect their diversification. Indeed, amphibians have likely experi‐ in C++ (Beeravolu & Condamine, 2016; Smith, 2009). We also fit‐ enced the effect of continental fragmentation through time with ted the DEC+J model (Matzke, 2014) but did not compare to DEC the successive break‐up of Pangaea, Gondwana and Laurasia (San because of concerns with statistical validity and model choice in Mauro, Vences, Alcobendas, Zardoya, & Meyer, 2005). Moreover, BioGeoBEARS (Ree & Sanmartín, 2018). We also preferred DEC given that less than 5% of amphibians can tolerate saline habitats to DEC+J because the latter often infers null or extremely low (Hopkins & Brodie, 2015; Pyron, 2014), they are likely to be isolated extinction rates (Sanmartín & Meseguer, 2016), and because the by continental fragmentation. DEC+J model favours direct dispersal over widespread ranges, Inspired by the work of von Humboldt (1807), we attempt to which might not be adequate for reconstructing the history of old test the evolutionary links between geology, climate and amphib‐ groups. ian diversification. To do so, we rely on both phylogenetic and fossil The species distribution ranges were categorized based on the data of amphibians to estimate the relationship between tempera‐ IUCN Global Amphibian Assessment (http://www.iucnredlist.org/ ture, geography and diversification rates using a combination of bio‐ initiatives/amphibians). Definition of biogeographical units was geographical analyses and diversification models. We address the based on the present‐day distribution of species but also informed by following question: How and to which extent was amphibian biodi‐ plate tectonics (Sanmartín, Enghoff, & Ronquist, 2001; Sanmartín & versity affected by fluctuations of palaeotemperature and the frag‐ Ronquist, 2004). Species ranges were defined by presence–absence mentation of the Earth's landmasses? We expect that vicariance has in each of the eight regions defined: NT = Neotropics (South America fostered amphibian diversification because studies have reported and Central America); NEA = Nearctic (North America); AF = Africa allopatric speciation at different stages of the evolutionary history (sub‐Saharan Africa and southern Arabia); PA = Palaearctic (Europe, of the clade (Duellman, Marion, & Hedges, 2016; Frazão, da Silva, & Asia, North Africa and northern Arabia); MG = Madagascar (includ‐ de Moraes Russo, 2015; Kok et al., 2018; Portik & Papenfuss, 2015; ing surrounding islands); OA = Oceania (Australia, New Zealand, Xavier, Guedes, & Napoli, 2015). Finally, we also expect amphibian New Caledonia and surrounding islands); SEA = Southeast Asia diversification to be positively associated with temperature, given (from Myanmar to the Lesser Sunda Islands, including south‐east that previous studies have found that their diversification rate was China, Taiwan and the Philippines, separated from OA by Weber's higher in the tropics (Pyron & Wiens, 2013). Line); and SA = South Asia (India and adjacent areas). We defined the model M0 in which the biogeographical history of species was reconstructed without any constraint. We followed 2 | MATERIALS AND METHODS the approach of Pyron (2014), such that the inferred ancestral ranges can only be composed of a maximum of two areas. First, we 2.1 | Fossil record and dated phylogeny ran the DEC and DEC+J models using the files provided by Pyron For estimating past amphibian diversity dynamics, we com‐ (2014, Dryad: https://doi.org/10.5061/dryad.jm453). Second, we piled information on fossil occurrences of amphibians from the imposed a time‐stratified biogeographical model covering the lin‐ Paleobiology Database (PBDB: http://paleobiodb.org). Using the eage diversification history of amphibians from the present to the word “Lissamphibia”, we included only fossil occurrences identified past 270 Ma, which we refer to as model M1. Based on palaeo‐ at the genus level and we adopted the conservative approach of geographical reconstructions (Blakey, 2008; Scotese, 2004; Seton excluding all occurrences that were not explicitly assigned to a de‐ et al., 2012), we constructed a time‐stratified model (Appendix S1) scribed genus in the PBDB. To visualize the raw data, we computed with five time slices that specified constraints on area connections the lineages‐through‐time plots, that is the number of living species and spanned the amphibian evolutionary history; each interval rep‐ through time, with range‐through‐diversity trajectories. We then resents a period bounded by major changes in tectonic and climatic identified whether the periods of higher diversity in the fossil record conditions thought to have affected the distribution of amphibians. were also corresponding to periods of continental fragmentation. The five time intervals are as follows: (a) 0–34 Ma, (b) 34–90 Ma, We used the molecular time‐calibrated phylogeny of amphibians (c) 90–120 Ma, (d) 120–170 Ma and (e) 170–270 Ma. Note that the including 3,309 species (Pyron, 2014; Pyron & Wiens, 2011, 2013). “maximum of two areas” constraint applies to the M1 model as These authors performed maximum‐likelihood dating analyses with well. As the latter constraint might seem unreasonable, especially a supermatrix of 12,809 nucleotides (9 nuclear and 3 mitochondrial for ancient clades which may have had a geographical spread over genes), and divergence times were estimated using multiple fossils wide continental landmasses (e.g., Pangaea, Gondwana, Laurasia), that provided minimum ages for the deeper nodes (i.e., crown of su‐ we extend the M1 model to have a maximum number of areas per perfamilies; Appendix S12). ancestral range to four, which will be referred to as the M2 model. Biogeographical ranges larger than two or four areas were disal‐ lowed as valid biogeographical states if they were not subsets of the 2.2 | Biogeographical analyses terminal species ranges; widespread ranges comprising areas that Biogeographical analyses were performed on the consensus tree have never been geographically connected (Sanmartín et al., 2001) (outgroups removed), using the Dispersal–Extinction–Cladogenesis were also removed. 4 | ROLLAND and CONDAMINE

Biogeographical estimates based on extant data only may con‐ time with no extinction (BCST) and one with speciation and ex‐ flict with the fossil record of the group. To overcome this bias, we tinction rates constant through time (BCSTDCST). Second, we fit‐ also performed DEC analyses including “fossil constraints” (DEC+FC) ted six time‐dependent models: two models with only speciation by introducing information on the distribution of fossil taxa, in the rate varying exponentially or linearly through time and no extinc‐ form of “soft” geographical constraints, placed at specific nodes ac‐ tion (BtimeVar_EXPO/LIN), two models with speciation rate vary‐ cording to the range of distribution of fossil occurrences assigned to ing exponentially or linearly through time and constant extinction a particular taxon during the relevant time frame (Appendix S2). We (BtimeVarDCST_EXPO/LIN), and two models with speciation rate consider this as a “soft” constraint because other areas different to and extinction rate both varying exponentially or linearly through that of the fossil could be included in the ancestral states. We per‐ time (BtimeVarDtimeVar_EXPO/LIN). We then fitted 16 models with formed DEC+FC analyses with the C++ version using the command speciation and extinction rates varying according to an external en‐ “fossil”. vironmental variable, of which eight models had the temperature curve and eight had the continental fragmentation index curve as follows: four models with speciation varying exponentially/linearly 2.3 | Global palaeotemperature and as a function of the environmental variable (BtemperatureVar_ palaeofragmentation EXPO/LIN and BcontinentVar_EXPO/LIN), four models with spe‐ We tested the effect of two abiotic variables on diversification: the ciation varying exponentially/linearly as a function of the variable past global variations of temperature and the number of landmasses with constant extinction rate (BtemperatureVarDCST_EXPO/LIN through time. The palaeotemperature curve was obtained from a and BcontinentVarDCST_EXPO/LIN), four models with extinction 18 compilation of studies measuring oxygen isotopes (δ O) as meas‐ rate varying exponentially/linearly as a function of the variable and ured from the benthic foraminiferal shells preserved in oceanic sedi‐ constant speciation rate (BCSTDtemperatureVar_EXPO/LIN and ments (Prokoph, Shields, & Veizer, 2008; Zachos, Dickens, & Zeebe, BCSTDcontinentVar_EXPO/LIN), and four models with both specia‐ 18 2008). δ O were transformed into deep‐sea temperature esti‐ tion and extinction rates varying exponentially/linearly as a function mates using the approach of Epstein, Buchsbaum, Lowenstam, and of the variable (BtemperatureVarDtemperatureVar_EXPO/LIN and Urey, (1953). The curve from Zachos et al. (2008) was used for the BcontinentVarDcontinentVar_EXPO/LIN). Cenozoic (0–66 Myr), and the curve from Prokoph et al. (2008) was We chose either a linear or an exponential dependence with used for the Mesozoic. The number of landmasses through time was time, continental fragmentation or temperature and diversification obtained as a fragmentation index estimated from palaeogeographi‐ rates because these functions can take a very broad range of shapes cal maps while assuming a plate tectonics model (see Zaffos et al., depending on the strength and direction of the dependence to the 2017, for details). The data were used to build penalized smoothing fitted variable. Speciation rate (λ) and extinction rate (μ) are param‐ splines for each environmental variable as an input in the diversi‐ eterized by the set of following parameters: when λ and μ can be fication analyses using the R package “pspline 1.0‐18” (Heckman & drawn as linear functions of the number of landmasses through time

Ramsay, 1996) with a default degree of freedom for each environ‐ denoted C(t), λ(C(t)) = λ0 + αC(t) and μ(C(t)) = μ0 + βC(t), where λ0 and μ0 mental variable. are respectively the speciation rate and the extinction rate expected when fragmentation is at its minimum (e.g., supercontinent), and α and β are the coefficients that measure the strength and the sign of 2.4 | Diversification analyses with the phylogeny the relationship with continental fragmentation (e.g. α > 0 and β > 0 To test the hypothesis that speciation was linked to continental indicate that speciation and extinction rates increase with continen‐ break‐up, we designed and applied four types of diversification tal fragmentation). When speciation and extinction rates are expo‐ models: constant‐rate models, time‐dependent models, tempera‐ nential functions of the number of landmasses through time, the

C(t) C(t) ture‐dependent models and tectonic‐dependent models. These equations are the following: (C(t)) = 0 ×e and (C(t)) = 0 ×e models rely on the maximum‐likelihood framework originally de‐ . Similar equations can be written for a linear and exponential re‐ veloped by Morlon et al. (2011) and implemented in the R package lationship between temperature through time and speciation and T (T ) = ×e (t) “rpanda 1.3” (Morlon et al., 2016). We accounted for the missing extinction rates: λ(T(t)) = λ0 + αT(t) or (t) 0 , where T(t) is the species in the phylogeny in the form of a global sampling fraction temperature at time t and λ0 is the speciation rate at the temperature that is the ratio of sampled species diversity (3,309) over the total of 0°C. All speciation and extinction rates were constrained to be described species diversity (6,990) based on Frost (2018). Given that between 0 and +∞. The models were compared with the corrected the heterogeneity of rates between clades may bias our results, we Akaike information criterion (AICc) and the Akaike weight (AICω). ran all diversification analyses with the whole dataset and indepen‐ The model with the lowest AICc and highest AICω was considered as dently for the three main orders: Anura (2,785 of 6,990 species), the best‐fitting model for the phylogeny. Caudata (469 of 722 species) and Gymnophiona (55 of 209 spe‐ We assessed whether the K‐Pg transition has affected amphib‐ cies). We designed and fitted a total of 26 diversification models. ian diversification with two other models: (a) a two‐parameter model We first fitted two constant‐rate models, considered as references including two constant speciation rates separated with a shift at the for model comparison: one with speciation rate constant through K‐Pg transition (66 Ma) and no extinction, and (b) a four‐parameter ROLLAND and CONDAMINE | 5 model including two constant speciation rates and two constant ex‐ fragmentation through time. We ran a temperature‐dependent and tinction rates separated with a shift at 66 Ma. Finally, we assessed a tectonic‐dependent diversification model using default gamma the robustness of the best palaeo‐environment‐dependent model priors on the baseline speciation and extinction rates and normal by fitting this model using a smoothed environmental curve (with a priors on the correlation parameters α and β, for speciation and degree of freedom of 3) to test whether and to which extent speci‐ extinction, respectively (Silvestro et al., 2015). For each model, we ficities of the environmental curve matter in explaining support for ran 10 million MCMC iterations with a sampling frequency of 1,000 an environmental effect on diversification when features of the en‐ and combined the posterior samples of the parameters from the 10 vironmental curve are removed. replicates after excluding the first 20% of the samples as burn‐in. We monitored chain mixing and effective sample sizes by examin‐

ing the log files in tracer. Posterior samples of the parameters were 2.5 | Diversification analyses with the fossil record summarized over all replicates as mean values and 95% credibility To assess the diversification dynamics of amphibian fossils, we interval. We considered the correlation to be statistically significant used a Bayesian approach (Silvestro, Schnitzler, Liow, Antonelli, & when 0 was not included within the 95% credibility interval of α and Salamin, 2014) in which fossil occurrences are modelled as the result β. To summarize the results on diversification dynamics, we gener‐ of two processes: preservation and diversification. This method, im‐ ated rates‐through‐time (origination and extinction) plots to identify plemented in pyrate (Silvestro, Salamin, & Schnitzler, 2014), uses all the periods of higher net diversification in the fossil record corre‐ fossil occurrences of a given taxon to estimate the parameters of the sponding to periods of continental fragmentation. To rank compet‐ preservation process, the times of speciation and extinction, the spe‐ ing diversification models, we estimated the marginal likelihoods of ciation and extinction rates and rate shifts through time. Assuming the different models using thermodynamic integration (Lartillot & that preservation is a stochastic process, it can be modelled as a Philippe, 2006) to compute Bayes factors and relative probabilities. non‐homogeneous Poisson process in which the rate parameter is a function of time and the expected number of fossil occurrences per 3 | RESULTS lineage per million years is given by the preservation rate q, which is estimated from the data (Silvestro, Schnitzler, et al., 2014). Thus, 3.1 | Raw genus‐level palaeodiversity times of speciation and extinction are estimated while taking the preservation process into account, instead of assuming that first In total, we obtained 1,885 fossil occurrences for 195 genera (mean and last appearances in the fossil record represent the true specia‐ of 9.6 occurrences per genus) including 135 extinct and 60 extant tion and extinction times. A birth–death process (with constant or genera. The raw counts of amphibian genera through time indicated variable rates) is set as a prior on the times of speciation and extinc‐ that amphibian diversity has not been constant through time, with tion of all taxa and is itself estimated from the data. pyrate has been the number of genera increasing rapidly during the Jurassic and tested with simulations and shown to provide accurate estimates of peaking in the mid‐Cretaceous and also during the Neogene, which preservation, speciation and extinction rates under a range of diver‐ is a recovery after the decrease in the Late Cretaceous (results not sification scenarios and preservation regimes (Silvestro, Schnitzler, shown). et al., 2014). It is robust against incomplete taxon sampling and vari‐ ations of the preservation process across lineages and through time 3.2 | Historical biogeography and geographical (Silvestro, Schnitzler, et al., 2014). mode of speciation We used a birth–death Markov chain Monte Carlo algorithm (BDMCMC) to estimate diversification rates (Silvestro, Schnitzler, We performed biogeographical analyses with DEC, DEC+J (jump dis‐ et al., 2014). Fossil occurrences in the raw data are often assigned persal) and DEC+FC (fossil constraints) models (Table 1). Dispersal to a temporal range, which depicts the uncertainty on their age. To rates were similar in the three models, but extinction rates were account for this uncertainty, we randomly selected fossil ages within different, with an extinction rate equal to zero in DEC+J. Despite the range of each occurrence, which generated 10 randomized data‐ these differences, the biogeographical reconstructions were similar sets for the clade, and we replicated the analyses on the random‐ (Figure 1; Appendices S3, S4 and S10). Figure 1 displays the most ized datasets (Silvestro, Schnitzler, et al., 2014). For each replicate, likely ancestral ranges at each node for the DEC analysis (DEC M2 we ran 10 million BDMCMC iterations, discarded the first 20% as model). We chose this model because it includes a stratification tak‐ burn‐in and sampled every 1,000 iterations to obtain posterior pa‐ ing into account the palaeogeographical changes though time and is rameter estimates. We used tracer 1.7.1 to monitor chain mixing and less constrained for the estimates of ancestral ranges (with four pos‐ effective sample sizes (Rambaut, Drummond, Xie, Baele, & Suchard, sible areas at each node). A discussion of the other DEC inferences is 2018). presented in Appendix S13. We then used the estimated times of speciation and extinction The DEC M2 analysis estimated that the most likely ancestral of all taxa as estimated in the BDMCMC analyses to carry out ad‐ area of origin of amphibians is a range encompassing Africa, the ditional analyses and test whether origination and/or extinction Neotropics, the Nearctic and the Palaearctic. Two hundred sev‐ rate dynamics correlate with either temperature or the continental enty million years ago, such a distribution range was covering a 6 | ROLLAND and CONDAMINE

TABLE 1 Results of DEC, DEC+J and DEC+FC models for the amphibian phylogeny

Model Analysis Description lnL NP d e j Inferred origin

DEC Pyron (2014) M0 unconstrained, −1721.2 2 0.00075 0.00075 0 NEA + NT max. 2 areas DEC This study M0 unconstrained, −1617.95 2 0.00036 4.726e‐6 0 NEA + NT max. 2 areas DEC+FC This study M0 unconstrained, −1707.55 2 0.00035 0.00029 0 NT + AF + MG + SA max. 2 areas DEC+J Pyron (2014) M0 unconstrained, −1804.3 3 0.00099 0.00098 0.000997 NEA + NT max. 2 areas DEC+J This study M0 unconstrained, −1550.96 3 0.00024 1e‐12 0.0015 NEA + NT max. 2 areas DEC This study M1 strat. adjacency, –1670.22 2 0.00175 0.00057 0 NEA + AF max. 2 areas DEC+J This study M1 strat. adjacency, −1571.13 3 0.00022 1e‐12 0.001961 NEA + NT max. 2 areas DEC This study M2 strat. adjacency, –1587.86 2 0.00148 0.00051 0 NEA + NT + AF + PA max. 4 areas DEC+FC This study M2 strat. adjacency, −1709.25 2 0.00149 0.00062 0 NEA + NT + OA max. 4 areas DEC+J This study M2 strat. adjacency, −1572.84 3 0.00021 1e‐12 0.0019297 NEA + AF + PA max. 4 areas

Note. For each model, the log‐likelihood (lnL) value, the number of free parameters (NP), and the parameter estimates for d (dispersal), e (extinction) and j (founder‐event speciation) are reported. The ancestral origin of the clade is also indicated. Abbreviations: AF, Africa; MG, Madagascar; NEA, Nearctic; NT, Neotropics; OA, Australasia; PA, Palaearctic; SA, South Asia (India).

large part of the Pangaea. The second most likely range included tree, ancestral ranges reconstructed with DEC and DEC+FC were the Neotropics, the Nearctic, the Palaearctic and Oceania, and extremely similar given that only 44 nodes out of 3,308 were dif‐ the third was composed by the Neotropics, the Nearctic, Oceania ferent (1.3%; Appendix S5). We found that the ancestral origin for and India. From this origin, we inferred vicariance events between amphibians was an area encompassing the Nearctic, the Neotropics Laurasia and Gondwana in the Jurassic and Cretaceous leading to and Australasia for the DEC+FC M2 model (including an adjacency the divergence of Gymnophiona with the rest of amphibian clades matrix) (Table 1, Figure 1). This same model estimated that Anura (Figure 1; Appendices S3 and S4). Our biogeographical estimations originated in the Neotropics, Caudata originated in Nearctic, and also supported that speciation events in the Jurassic and Early Gymnophiona had a widespread area of origin comprising the Cretaceous were congruent with the subsequent separation of the Neotropics, Africa, Australasia and India. main lineages of amphibians following the break‐up of Laurasia into North America, Europe/Asia (that then split again into European and 3.3 | Diversification dynamics with the Asian landmasses) and Gondwana, which also got fragmented and phylogeny and the fossil record led to the formation of South America, Africa, India and Australia– Antarctica (Figure 1; Appendices S3 and S4). The best‐fitting model among constant, temperature‐dependent When comparing the ancestral ranges estimated with the DEC and tectonic‐dependent models includes both speciation and ex‐ M2 model and the fossil record, we found that seven out of 26 tinction rates varying linearly according to temperature through nodes (27%; corresponding to major orders or families) had conflict‐ time (∆AICc = 1.92 with the second best‐fitting model, AICω = 0.72; ing results (Appendix S2). These nodes were then used to constrain Table 2). This temperature‐dependent model indicates that tem‐ the DEC+FC model. Figure 1 also displays the most likely ancestral perature is positively associated with speciation (α = 0.006) and ranges at each node for the DEC+FC analysis (estimated ranges are negatively associated with extinction rates, that is although we es‐ depicted just above DEC inferences when different). On the whole timate a low extinction rate for the most part of amphibian history

FIGURE 1 Historical biogeography of Amphibia. Comparison of the biogeographical histories as inferred by the best DEC and DEC+FC models for the amphibians based on the dated phylogeny of Pyron (2014). The most likely ancestral ranges are shown at each node (DEC estimates are depicted at nodes, and DEC+FC results—when different from DEC—are depicted just above). The biogeographical model includes a time‐stratified palaeogeographical model and restricts the ancestral range size to maximum four areas. Important vicariance events are highlighted by grey triangles, which are also shown on palaeogeographical maps (sensu Blakey, 2008). The detailed results are available in Appendices S3 and S4. A comparison between DEC and DEC+FC is provided in Appendix S5 ROLLAND and CONDAMINE | 7 8 | ROLLAND and CONDAMINE

(β = −0.009; Figure 2). Past temperature has impacted the diversi‐ is currently too scarce (only three occurrences for three extinct fication dynamics of amphibians with periods of elevated diversifi‐ genera). cation associated with periods of warming. Interestingly, speciation rate shows a peak during the Cretaceous when the number of landmasses was maximal (Figure 2). The second best‐fitting model 4 | DISCUSSION is a time‐dependent model for which speciation and extinction are negatively associated with time (α = 0.005 and β = 0.005). The first Identifying the relative effect of different environmental factors tectonic‐dependent model is the third best model and is not sup‐ to clades’ diversification is an emerging goal in macroevolution ported (∆AIC = 47.2 with the best‐fitting model, and AICω = 0), even (Benton, 2015; Condamine et al., 2012, 2018; Ezard et al., 2011; if it shows that continental fragmentation is strongly positively as‐ Roalson & Roberts, 2016). In this context, it is important to combine sociated with speciation and negatively associated with extinction palaeontological and phylogenetic approaches and to test a wide (α = 0.419, β = −0.277). At the order level, we find that the tem‐ range of diversification models. Here, we show that temperature is perature‐dependent model is supported for Anura (∆AIC = 5.5 with likely the most important factor associated with the diversification the second model). Consistent with the results for all amphibians, of amphibians compared to other factors, such as time or landmass temperature is positively associated with temperature and specia‐ fragmentation. tion (α = 0.005, β = −0.008; Appendix S6). A time‐dependent model We found an association between temperature and diversifi‐ is supported for Caudata, with a decrease in speciation and extinc‐ cation rates in amphibians with both the whole phylogeny and the tion rates thought time (α = 0.0149 and β = 0.01543; Appendix S7). fossil record. Similar results were found at the order level with the Identifying the best‐fitting model is unclear for Gymnophiona given Anuran phylogeny and with the fossil record of Anura and Caudata. that a time‐dependent (first), a tectonic‐dependent (second) and a We detected a positive association between temperature and spe‐ constant‐rate (third) model had close AICc within a range of ∆AIC = 2 ciation in all of these clades: speciation increased when tempera‐ (Appendix S8). AICω is 0.23 for the time‐dependent model with a ture increased. This result is consistent with a previous study of decrease in both speciation and extinction through time (α = 0.006 amphibians showing higher speciation in tropical regions, where and β = 0.006), 0.16 for a model with a positive effect of tectonic temperature is warmer (Pyron & Wiens, 2013). It is also consistent dependence on speciation rate (α = 0.396) and 0.11 for a model with with palaeontological studies predicting that temperature is pos‐ constant speciation and no extinction through time. itively associated with species diversity (Erwin, 2009; Ezard et al., The inclusion of a shift in speciation and extinction rates at the 2011; Jaramillo et al., 2006; Mayhew et al., 2012). This result can be K‐Pg boundary (66 Ma) did not improve the likelihood of the mod‐ also potentially explained by the metabolic theory of ecology, which els (Appendix S11), which suggests that we did not detect any sig‐ stipulates that rates of mutations and speciation are accelerated in nal for a shift in diversification rates at the K/T boundary. We also warmer climate (Allen et al., 2006; Brown et al., 2004), and there is found that smoothing the temperature curve strongly decreased the evidence that substitution rates are accelerated in amphibians at low support for temperature‐dependent model (∆AIC = 99.6 between a latitude and low elevation (Wright, Gillman, Ross, & Keeling, 2010). model with a smoothed curve and the original curve), which suggests The dependence between temperature and extinction rate that the overall trend of the temperature curve (linear decline) was was less clear given that temperature was negatively associated not sufficient to explain the support for temperature‐dependent with extinction in phylogenies (although extinction was low during model and that fine‐scale fluctuations of temperature had an impact the entire history of the group; Figure 2), but positively associ‐ on the likelihood (Appendix S11). ated with extinction in the fossil record for the whole clade or at

The pyrate analyses of the fossil record with constant, tem‐ the order level. Most theories propose that lineages distributed in perature‐dependent and tectonic‐dependent models also indicate warmer (tropical) regions, with more stable climate, have a lower that a temperature‐dependent model best explains the diver‐ extinction rate (Dynesius & Jansson, 2000; Pianka, 1966; Rohde, sification of amphibians (relative probability = 1; Table 3). This 1992). Very few examples of high extinction rate in warm regions temperature‐dependent model estimates that both origination have been documented, even if one case has been described in the and extinction rates are positively associated with temperature Early Triassic (Sun et al., 2012). We cannot assert that temperature (γλ = 0.039 and γμ = 0.054; Table 3). Increasing temperature led to was the leading factor explaining the diversification of Caudata higher origination and extinction rates in amphibians reaching a and Gymnophiona given that we only found support for tempera‐ peak in the Late Cretaceous (Figure 3). Time‐dependent and tec‐ ture‐dependent model with the fossil record but not with the phy‐ tonic‐dependent models are not supported by our data. We find logeny. The alternative hypothesis (the best‐fitting model in both consistently that the temperature‐dependent model is selected as groups with phylogenies) is a decreasing speciation and extinction the best‐fit model in Anura and Caudata (Table 3). Temperature is rate through time. Given that we only found this result in the two also positively associated with origination and extinction in these poorest sampled groups, it is also possible that the results are less two groups (for Anura: γλ = 0.024 and γμ = 0.040; and for Caudata: robust. Phylogenetic and age uncertainties and the estimation of γλ = 0.057 and γμ = 0.057; Table 3). We did not run diversification extinction process are potential biases that could affect the infer‐ analyses on Gymnophiona because the quality of the fossil record ences and results of the study. ROLLAND and CONDAMINE | 9

TABLE 2 Diversification analysis of Amphibia based on the phylogeny

Models NP logL AICc ∆AICc AICω λ α μ β BtemperatureVarDtemperatureVar_LIN 4 −13661.9 27331.8 0.0 0.72 0.035 0.006 0.044 −0.009 BTimeVarDTimeVar_LIN 4 −13662.8 27333.7 1.9 0.28 0.039 0.005 0.005 0.005 BcontinentVarDcontinentVar_LIN 4 −13685.5 27379.0 47.2 0 0.122 0.419 0.102 −0.277 BCSTDtemperatureVar_LIN 3 −13687.5 27381.1 49.3 0 0.067 — 0.021 −0.003 BCSTDTimeVar_LIN 3 −13693.1 27392.2 60.5 0 0.065 — 0.004 0.000 BCSTDtemperatureVar_EXPO 3 −13698.7 27403.4 71.7 0 0.066 — 0.002 0.130 BTimeVar_LIN 2 −13702.8 27409.7 77.9 0 0.068 0.000 — — BTimeVarDCST_LIN 3 −13702.8 27411.7 79.9 0 0.068 0.000 0.000 — BcontinentVarDCST_LIN 3 −13705.4 27416.9 85.1 0 0.070 0.303 0.026 — BTimeVar_EXPO 2 −13707.3 27418.6 86.8 0 0.069 −0.005 — — BTimeVarDCST_EXPO 3 −13707.3 27420.6 88.8 0 0.069 −0.005 0.000 — BTimeVarDTimeVar_EXPO 4 −13707.3 27422.6 90.8 0 0.069 −0.005 0.000 0.001 BcontinentVarDCST_EXPO 3 −13709.0 27424.0 92.2 0 0.010 4.345 0.029 — BCSTDTimeVar_EXPO 3 −13715.0 27436.1 104.3 0 0.067 — 0.007 0.012 BcontinentVarDcontinentVar_EXPO 4 −13718.9 27445.9 114.1 0 0.019 2.773 0.125 −4.187 BtemperatureVar_LIN 2 −13721.8 27447.7 115.9 0 0.073 −0.001 — — BtemperatureVarDCST_LIN 3 −13721.8 27449.7 117.9 0 0.073 −0.001 0.000 — BcontinentVar_LIN 2 −13725.0 27454.0 122.3 0 0.034 0.196 — — BtemperatureVar_EXPO 2 −13726.5 27457.0 125.3 0 0.073 −0.022 — — BtemperatureVarDCST_EXPO 3 −13726.5 27459.0 127.3 0 0.072 −0.021 0.000 — BtemperatureVarDtemperatureVar_EXPO 4 −13725.9 27459.8 128.0 0 0.073 −0.019 0.002 0.036 BcontinentVar_EXPO 2 −13729.4 27462.9 131.1 0 0.016 2.754 — — BCST 1 −13746.6 27495.3 163.5 0 0.061 — — — BCSTDCST 2 −13745.8 27495.7 163.9 0 0.064 — 0.005 — BCSTDcontinentVar_LIN 3 −13744.9 27495.9 164.1 0 0.066 — 0.051 −0.088 BCSTDcontinentVar_EXPO 3 −13746.1 27498.2 166.4 0 0.064 — 0.004 0.008

Note. Models are ranked based on AICc and AICω. The best‐fitting model is in bold. This model shows a positive dependence (α > 0 and β > 0) between past temperature and speciation and extinction. Abbreviations: AICc, corrected Akaike's information criterion; ΔAICc, the difference in AICc between the model with the lowest AICc and the others; AICω, the Akaike weight of each model; B, birth (speciation); continent, rate varies as a function of the landmass fragmentation index through time; CST, constant rate; D, death (extinction); EXPO, exponential relationship; LIN, linear relationship; logL, log‐likelihood; NP, number of parameters; temperature, rate varies as a function of the palaeotemperature through time; Time, time‐variable rate (see Section 2 for more details on the models).

Our second main result was that the landmass fragmentation analyses with time‐stratified DEC models suggest that vicariance did not appear to be a major factor affecting the diversification of events related to tectonic plates dynamics are associated with vi‐ amphibians. This result was surprising given that the raw amphib‐ cariance (allopatric speciation) in the early evolutionary history of ian palaeodiversity substantially increased during the Jurassic and amphibians (Figure 1). Our ancestral area estimations indicate that Cretaceous (results not shown), two major periods of continental the Western Gondwana break‐up triggered vicariance events sep‐ fragmentation (Blakey, 2008; Seton et al., 2012). The tectonic‐de‐ arating the main lineages of Neobatrachia and Pipidae (150 Ma), pendent model was never the best‐fitting model in our analyses, both having a lineage in Africa sister to a lineage in South America suggesting that if continental fragmentation had an effect on diversi‐ (Figures 1 and 4). We also found several vicariance events within fication, this effect was weaker than the effect of temperature. One Gymnophiona (109 Ma) due to the successive break‐up of Western possible reason to explain this result is that continental fragmenta‐ and Eastern Gondwana in the Cretaceous, and even a vicariance be‐ tion was low when the group began to diversify in the Triassic and tween India and Madagascar in the Palaeocene (Figure 4). Another Jurassic, and landmass fragmentation might thus have fostered the example involves a vicariance within Salamandridae (Pleurodelinae) diversification of amphibians in the Cretaceous but was not found as between Nearctic and Palaearctic lineages in the early Eocene the most important factor through the whole history of the clade. (Figure 4). Our results are consistent with previous studies on am‐ Although continental fragmentation was not as important as phibians proposing that the divergences of the main amphibian and temperature in our diversification analyses, the biogeographical anuran lineages may be related to vicariance events (Feng et al., 10 | ROLLAND and CONDAMINE

(a) FIGURE 2 Effect of temperature changes on the amphibian diversification estimated from the phylogeny. (a) The index of continental fragmentation through time as calculated with palaeogeographical maps (Zaffos et al., 2017), and the temperature profile through time as estimated from oxygen isotopes (Prokoph et al., 2008; Zachos et al., 2008). (b) Rates‐through‐time plot for the best diversification model. We found support for a positive dependence of both speciation and extinction rates with temperature through time. Amphibians diversified faster when temperature increased. Neog., Neogene; Palaeog., Palaeogene

(b)

2017; Frazão et al., 2015; Pyron, 2014). The biogeographical anal‐ 2007). Amphibian diversification might be particularly prone to be yses further suggest that vicariance was frequent and widespread associated with continental fragmentation as it has been shown that across the amphibian phylogeny when the continental fragmenta‐ amphibian species have smaller distribution ranges, low dispersal tion increased during the Jurassic and throughout the Cretaceous abilities across salt water (Hopkins & Brodie, 2015; Pyron, 2014) and (Figure 4), and the frequency of these events tends to decrease after narrower climatic niches than other groups, for example mammals continental fragmentation slowed down in the Cenozoic (Appendix and birds (Rolland & Salamin, 2016; Rolland et al., 2018). An alter‐ S9). We found in our biogeographical estimations a number of allo‐ native explanation for a positive correlation between speciation and patric speciation events temporally associated with the successive continental fragmentation is that continental fragmentation and the fragmentation of Pangaea, Gondwana and Laurasia in the Mesozoic. subsequent creation of epicontinental seas likely increase hygrome‐ These results are consistent with several studies proposing a role try and pluviometry in large areas and may increase the size of suit‐ of geography, topography and vicariance in the dynamics of diver‐ able habitats for amphibians. sification and divergence between populations (Baselga et al., 2011; Studying the impact of temperature and tectonic change on spe‐ Hedges et al., 1996; Jordan et al., 2016; Mao et al., 2012; Vavrek, cies diversification is an active area of research (Condamine et al., 2016; Zaffos et al., 2017), and also consistent with previous studies 2013; Jordan et al., 2016; Lewitus & Morlon, 2018; Zaffos et al., on amphibians reporting that geographical fragmentation explains 2017). We currently use deep‐sea temperatures from a compilation 18 the divergence of lineages (Fouquet et al., 2012; Vences & Wake, of studies measuring δ O from the oceanic sediments (Prokoph ROLLAND and CONDAMINE | 11 and 0.092) −0.740) 0.068) 3.907) 0.076) −0.085) 0.057 (0.023, (0.023, 0.057 — (−10.020, −5.470 0.040 (0.013, (0.013, 0.040 — −1.290 (−6.217, 0.054 (0.031, (0.031, 0.054 γμ — −3.301 (−6.616, 0.087) 0.892) 0.047) −0.746) 0.059) −1.078) 0.057 (0.026, (0.026, 0.057 — −3.233 (−7.390, 0.024 (6.201e‐4, — −4.543 (−8.258, 0.039 (0.020, (0.020, 0.039 Correlation parameters CI) (95% γλ — −3.825 (−6.471, 1 1.337e‐05 1.675e‐05 5.541e‐09 0.999 4.769e‐09 0.906 0.003 0.091 Relative Relative probability (positive)

strong) 0 22.4 (very strong) (very21.9 strong) 0 (strong)11.7 4.6 38.02 (very 0 38.3 (very strong) Bayes factors −523.607 −534.829 −534.604 −783.51 −789.366 −785.808 −1365.117 −1346.106 −1365.267 Marginal likelihood Marginal

variations fragmentation variations fragmentation variations fragmentation Constant model Continental Temperature Constant model Continental Temperature Constant model Continental Temperature Model fitted Model

. Bayes factors and relative probabilities are used to select the best model for each dataset (grey). Posterior estimates of the parameters γλ 189.6 195.3 171.8 (167.6–181.7) (184.5–195.4) (190.4–199.0) Clade age (Ma) pyrate

) 0.272 0.487 analyses for amphibians α (0.223–0.65) (0.430–0.564) 0.5283 (0.422–0.607) Heterogeneity Heterogeneity ( pyrate

) 1.21 q 1.065 (0.921–1.455) 1.239 (1.105–1.363) (1.101–1.335) Preservation ( . Posterior estimates of the preservation rate (expected number of occurrences per lineage per Myr) and heterogeneity parameter (shape parameter of the gamma distribution) estimated under the quantifying the correlation between either continental fragmentation or global temperature variations and origination and extinction rates CIs are (95% given in parentheses). The value in bold Caudata Anura Amphibians Clade Note model birth–death Markov chain Monte Carlo algorithm. Relative probability of palaeo‐environmental models for twoas estimated drivers of diversification. with the thermodynamic Model comparisons integration are of based on the marginal likelihoods TABLE 3 SummaryTABLE of the γμ highlights the parameter displaying a significant correlation. 12 | ROLLAND and CONDAMINE

(a) FIGURE 3 Effect of temperature changes on the amphibian diversification estimated from the fossil record. Plots of origination and extinction rates through time for the best pyrate model (Table 3), which supports a positive dependence of both origination and extinction rates with the temperature variations through time. Amphibians diversified faster during warming periods. The shaded areas represent the 95% credibility intervals of the parameters, estimated with pyrate. Pictured is a specimen of †Eopelobates (50 Ma), belonging to Pelobatidae (© Wikimedia Commons). Neog., Neogene; (b) Palaeog., Palaeogene

et al., 2008; Zachos et al., 2008). However, sea‐level fluctuations a dynamic continuous process over time. Probabilistic biogeographi‐ and ice volumes might bias temperature estimates during cold pe‐ cal models based on discrete state spaces relied on continuous‐time riods when ice sheets formed at the poles (Cramer, Miller, Barrett, Markov chains, implying that the geographical mode of speciation is & Wright, 2011; Hansen, Sato, Russell, & Kharecha, 2013). There is not modelled as a continuous‐time stochastic process. For all these currently no palaeotemperature curve that accurately accounts for reasons, DEC‐based inferences focusing on these scenarios should these biases during the Phanerozoic (but see Hansen et al., 2013, for be treated with utmost caution (Ree & Sanmartín, 2018). the Cenozoic). We thus acknowledge that another palaeotempera‐ Our study, nonetheless, brings new insights into the spatiotem‐ ture curve could potentially improve our diversification estimates. poral roles of temperature and geological changes in shaping the We encourage future studies, especially those focusing on clades diversity of one of the major vertebrate clades, suggesting that the diversifying in the Cenozoic, to account for measurement and con‐ impact of temperature and geographical factors on species diversifi‐ 18 version errors between δ O and temperature by taking into account cation may have been overlooked compared to biotic interactions and the variations of sea level and atmospheric CO2 in the deep‐sea tem‐ adaptive divergence (Losos et al., 2003; Lamichhaney et al., 2015, but perature estimates (Hansen et al., 2013). There are also methodolog‐ see Wang, Glor, & Losos, 2013). Our study suggests that the temporal ical limitations when reconstructing ancestral ranges and estimating variations of temperature and continental fragmentation should be rates of speciation and extinction. For example, our analyses nei‐ included more often in studies looking for the causes of clades’ diver‐ ther accounted for the distribution of entire extinct clades, nor ac‐ sification at the global scale, as they may be major factors affecting counted for uncertainty in the phylogenetic tree given that they were macroevolutionary dynamics. In particular, we suggest that the role based on a single maximum‐likelihood tree (Pyron, 2014; Pyron & of vicariant speciation should be tested in other groups such as mam‐ Wiens, 2011, 2013). Ancestral range analyses also currently assume mals and birds that also experienced a Cretaceous interordinal di‐ static set of predefined regions, while continental fragmentation is versification when the continental fragmentation was maximal (Jetz,

FIGURE 4 Evidence of vicariance events in three amphibian clades. The biogeographical histories of Gymnophiona (a), Pleurodelinae (b) and Pipidae (c) are inferred with the DEC M2 and DEC+FC models, and the most likely ancestral ranges are shown at each node (estimates of DEC M2 and DEC+FC M2 were the same for those clades). Grey triangles highlight vicariance events, which are also shown on palaeogeographical maps (sensu Blakey, 2008). The detailed results of the DEC and DEC+FC M2 models are available in Appendices S3 and S4. K, Cretaceous; Q, Quaternary; Pl, Pliocene ROLLAND and CONDAMINE | 13

(a)

(b)

Thomas, Joy, Hartmann, & Mooers, 2012; Meredith et al., 2011). REFERENCES Detecting the effect of environmental factors on diversification is Alfaro, M. E., Santini, F., Brock, C., Alamillo, H., Dornburg, A., Rabosky, nonetheless a difficult task, given that the many lineages that may D. L., … Harmon, L. J. (2009). Nine exceptional radiations plus high have kept the traces of Cretaceous diversification have gone extinct turnover explain species diversity in jawed vertebrates. Proceedings during the Cretaceous–Palaeogene mass extinction. Recent diversifi‐ of the National Academy of Sciences USA, 106, 13410–13414. https:// cation of lineages, not related to temperature or continental break‐up, doi.org/10.1073/pnas.0811087106 Allen, A. P., Gillooly, J. F., Savage, V. M., & Brown, J. H. (2006). Kinetic may have also blurred the phylogenetic signal of speciation and ex‐ effects of temperature on rates of genetic divergence and specia‐ tinction. Consistent with the increase of the proportion of sympatric tion. Proceedings of the National Academy of Sciences USA, 103, 9130– speciation events detected in the late history of the clade (Appendix 9135. https://doi.org/10.1073/pnas.0603587103 S9), recent speciation events are likely less associated with continen‐ Barnosky, A. D. (2001). Distinguishing the effects of the Red Queen and Court Jester on Miocene mammal evolution in the northern Rocky tal fragmentation because the number of landmasses has decreased Mountains. Journal of Vertebrate Paleontology, 21, 172–185. https:// during the Cenozoic but instead within‐continent speciation events doi.org/10.1671/0272-4634(2001)021[0172:DTEOTR]2.0.CO;2 were paramount, like in the Neotropics (Hutter, Lambert, & Wiens, Baselga, A., Recuero, E., Parra‐Olea, G., & Garcia‐Paris, M. (2011). 2017; Santos et al., 2009). Although this study aimed at testing the Phylogenetic patterns in zopherine beetles are related to ecological niche width and dispersal limitation. Molecular Ecology, 20, 5060– role of temperature and geological changes on the diversification 5073. https://doi.org/10.1111/j.1365-294X.2011.05342.x processes of the amphibians, we do not rule out the possibility that Beeravolu, C. R., & Condamine, F. L. (2016). An extended maximum like‐ the diversification of the group was also influenced by other factors lihood inference of geographic range evolution by dispersal, local ex‐ not studied here, such as intrinsic species’ traits or the increase of tinction and cladogenesis. bioRxiv. https://doi.org/10.1101/038695 Belanger, C. L., Jablonski, D., Roy, K., Berke, S. K., Krug, A. Z., & Valentine, wet environments over a large latitudinal range in the Cretaceous J. W. (2012). Global environmental predictors of benthic marine bio‐ (Ziegler et al., 2003). geographic structure. Proceedings of the National Academy of Sciences USA, 109, 14046–14051. https://doi.org/10.1073/pnas.1212381109 Benton, M. J. (2009). The Red Queen and the Court Jester: Species diver‐ 5 | CONCLUSION sity and the role of biotic and abiotic factors through time. Science, 323, 728–732. https://doi.org/10.1126/science.1157719 Benton, M. J. (2015). Exploring macroevolution using modern and fos‐ von Humboldt (1807) envisioned a link between geological, cli‐ sil data. Proceedings of the Royal Society B: Biological Sciences, 282, matic and biodiversity changes over time and space. In this study, 20150569. https://doi.org/10.1098/rspb.2015.0569 we show that temperature has played an important role on the di‐ Blakey, R. C. (2008). Gondwana paleogeography from assembly to breakup – A 500 million year odyssey. In C. R. Fielding, T. D. Frank & versification of amphibian diversity through time. We think that J. L. Isbell (Eds.), Resolving the Late Paleozoic Ice Age in time and space combining neontological and palaeontological data provides an op‐ (pp. 1–28). Special Paper, 441. McLean, VA: The Geological Society portunity to detect and analyse the effect of temperature on the of America. diversification of a given clade. We also think that analysing these Brown, J. H., Gillooly, J. F., Allen, A. P., Savage, V. M., & West, G. B. (2004). Toward a metabolic theory of ecology. Ecology, 85, 1771– data in the light of the ancestral biogeography of the group might 1789. https://doi.org/10.1890/03-9000 ultimately lead to understand the emergence of global biodiver‐ Cermeño, P., Benton, M. J., Paz, Ó., & Vérard, C. (2017). Trophic and sity patterns, proposing here the major role of temperature and tectonic limits to the global increase of marine invertebrate di‐ landmass fragmentation as predictors of species diversification. versity. Scientific Reports, 7, 15969. https://doi.org/10.1038/ s41598-017-16257-w Further studies may consider analysing the effect of continental Condamine, F. L., Rolland, J., Höhna, S., Sperling, F. A. H., & Sanmartín, I. fragmentation and temperature on more clades at the global scale, (2018). Testing the role of the Red Queen and Court Jester as drivers and to ultimately better understand the combined role of abiotic of the macroevolution of Apollo butterflies. Systematic Biology, 67, factors on biological evolution, such as the relative contribution of 940–964. https://doi.org/10.1093/sysbio/syy009 Condamine, F. L., Rolland, J., & Morlon, H. (2013). Macroevolutionary geographical speciation event and ecological speciation on current perspectives to environmental change. Ecology Letters, 16, 72–85. diversity. https://doi.org/10.1111/ele.12062 Condamine, F. L., Sperling, F. A. H., Wahlberg, N., Rasplus, J. Y., & Kergoat, G. J. (2012). What causes latitudinal gradients in species ACKNOWLEDGEMENTS diversity? Evolutionary processes and ecological constraints on swallowtail biodiversity. Ecology Letters, 15, 267–277. https://doi. We thank Dolph Schluter for helpful comments on the revised ver‐ org/10.1111/j.1461-0248.2011.01737.x sion of the manuscript. R.J. received funding from a Banting postdoc‐ Coyne, J. A., & Orr, H. A. (2004). Speciation. Sunderland, MA: Sinauer. toral fellowship, Grant/Award Number: 151042; Agence Nationale Cramer, B. S., Miller, K. G., Barrett, P. J., & Wright, J. D. (2011). Late de la Recherche, Grant/Award Number: ANR‐10‐ LABX‐25‐01. Cretaceous–Neogene trends in deep ocean temperature and continental ice volume: Reconciling records of benthic foramin‐ 18 iferal geochemistry (δ O and Mg/Ca) with sea level history. ORCID Journal of Geophysical Research: Oceans, 116, C12023. https://doi. org/10.1029/2011JC007255 Jonathan Rolland https://orcid.org/0000-0002-7141-0870 Duellman, W. E., Marion, A. B., & Hedges, S. B. (2016). Phylogenetics, classification, and biogeography of the treefrogs (Amphibia: Anura: Fabien L. Condamine https://orcid.org/0000-0003-1673-9910 ROLLAND and CONDAMINE | 15

Arboranae). Zootaxa, 4104, 1–109. https://doi.org/10.11646/ Jaramillo, C., Rueda, M. J., & Mora, G. (2006). Cenozoic plant diversity zootaxa.4104.1.1 in the Neotropics. Science, 311, 1893–1896. https://doi.org/10.1126/ Dynesius, M., & Jansson, R. (2000). Evolutionary consequences of science.1121380 changes in species' geographical distributions driven by Milankovitch Jetz, W., Thomas, G. H., Joy, J. B., Hartmann, K., & Mooers, A. O. (2012). climate oscillations. Proceedings of the National Academy of Sciences of The global diversity of birds in space and time. Nature, 491, 444–448. the USA, 97, 9115–9120. https://doi.org/10.1038/nature11631 Epstein, S., Buchsbaum, R., Lowenstam, H. A., & Urey, H. C. (1953). Jordan, S. M., Barraclough, T. G., & Rosindell, J. (2016). Quantifying the Revised carbonate-water isotopic temperature scale. Geological effects of the break up of Pangaea on global terrestrial diversifica‐ Society of America Bulletin, 64, 1315–1326. tion with neutral theory. Philosophical Transactions of the Royal Society Erwin, D. H. (2009). Climate as a driver of evolutionary change. Current B: Biological Sciences, 371, 20150221. https://doi.org/10.1098/ Biology, 19, R575–R583. https://doi.org/10.1016/j.cub.2009.05.047 rstb.2015.0221 Ezard, T. H., Aze, T., Pearson, P. N., & Purvis, A. (2011). Interplay be‐ Kok, P. J., Ratz, S., MacCulloch, R. D., Lathrop, A., Dezfoulian, R., Aubret, tween changing climate and species’ ecology drives macroevolu‐ F., & Means, D. B. (2018). Historical biogeography of the palaeoen‐ tionary dynamics. Science, 332, 349–351. https://doi.org/10.1126/ demic toad genus Oreophrynella (Amphibia: Bufonidae) sheds a new science.1203060 light on the origin of the Pantepui endemic terrestrial biota. Journal of Feller, A. E., & Hedges, S. B. (1998). Molecular evidence for the early Biogeography, 45, 26–36. https://doi.org/10.1111/jbi.13093 history of living amphibians. Molecular Phylogenetics and Evolution, 9, Lagomarsino, L. P., Condamine, F. L., Antonelli, A., Mulch, A., & Davis, 509–516. https://doi.org/10.1006/mpev.1998.0500 C. C. (2016). The abiotic and biotic drivers of rapid diversification Feng, Y. J., Blackburn, D. C., Liang, D., Hillis, D. M., Wake, D. B., in Andean bellflowers (Campanulaceae). New Phytologist, 210, 1430– Cannatella, D. C., & Zhang, P. (2017). Phylogenomics reveals rapid, 1442. https://doi.org/10.1111/nph.13920 simultaneous diversification of three major clades of Gondwanan Lamichhaney, S., Berglund, J., Almén, M. S., Maqbool, K., Grabherr, M., frogs at the Cretaceous–Paleogene boundary. Proceedings of the Martinez‐Barrio, A., … Andersson, L. (2015). Evolution of Darwin's National Academy of Sciences USA, 114, E5864–E5870. https://doi. finches and their beaks revealed by genome sequencing. Nature, 518, org/10.1073/pnas.1704632114 371–375. https://doi.org/10.1038/nature14181 Fouquet, A., Noonan, B. P., Rodrigues, M. T., Pech, N., Gilles, A., & Lartillot, N., & Philippe, H. (2006). Computing Bayes factors using ther‐ Gemmell, N. J. (2012). Multiple quaternary refugia in the eastern modynamic integration. Systematic Biology, 55, 195–207. https://doi. Guiana Shield revealed by comparative phylogeography of 12 frog org/10.1080/10635150500433722 species. Systematic Biology, 61, 461–489. https://doi.org/10.1093/ Lehtonen, S., Silvestro, D., Karger, D. N., Scotese, C., Tuomisto, H., Kessler, sysbio/syr130 M., … Antonelli, A. (2017). Environmentally driven extinction and op‐ Frazão, A., da Silva, H. R., & de Moraes Russo, C. A. (2015). The Gondwana portunistic origination explain fern diversification patterns. Scientific breakup and the history of the Atlantic and Indian oceans unveils Reports, 7, 4831. https://doi.org/10.1038/s41598-017-05263-7 two new clades for early neobatrachian diversification. PLoS ONE, 10, Leprieur, F., Descombes, P., Gaboriau, T., Cowman, P. F., Parravicini, V., e0143926. https://doi.org/10.1371/journal.pone.0143926 Kulbicki, M., … Pellissier, L. (2016). Plate tectonics drive tropical ma‐ Frost, D. R. (2018). Amphibian species of the world: An online reference. rine biodiversity dynamics. Nature Communications, 7, 11461. https:// Version 6.0. New York, NY: American Museum of Natural History. doi.org/10.1038/ncomms11461 Retrieved from http://research.amnh.org/herpetology/amphibia/ Lewitus, E., & Morlon, H. (2018). Detecting environment‐dependent index.html diversification from phylogenies: A simulation study and some em‐ Funk, D. J. (1998). Isolating a role for natural selection in specia‐ pirical illustrations. Systematic Biology, 67, 576–593. https://doi. tion: Host adaptation and sexual isolation in Neochlamisus org/10.1093/sysbio/syx095 bebbianae leaf beetles. Evolution, 52, 1744–1759. https://doi. Liow, L. H., Reitan, T., & Harnik, P. G. (2015). Ecological interactions on org/10.1111/j.1558-5646.1998.tb02254.x macroevolutionary time scales: Clams and brachiopods are more Hannisdal, B., & Peters, S. E. (2011). Phanerozoic Earth system evolu‐ than ships that pass in the night. Ecology Letters, 18, 1030–1039. tion and marine biodiversity. Science, 334, 1121–1124. https://doi. https://doi.org/10.1111/ele.12485 org/10.1126/science.1210695 Lomolino, M. V., Riddle, B. R., Whittaker, R. J., & Brown, J. H. (2010). Hansen, J., Sato, M., Russell, G., & Kharecha, P. (2013). Climate sensitivity, Biogeography (4th ed.). Sunderland, MA: Sinauer Associates, Inc. sea level and atmospheric carbon dioxide. Philosophical Transactions Losos, J. B., & Glor, R. E. (2003). Phylogenetic comparative methods and of the Royal Society A: Mathematical, Physical and Engineering Sciences, the geography of speciation. Trends in Ecology and Evolution, 18, 220– 371, 20120294. https://doi.org/10.1098/rsta.2012.0294 227. https://doi.org/10.1016/S0169-5347(03)00037-5 Heckman, N., & Ramsay, J. O. (1996). Spline smoothing with model based Losos, J. B., Leal, M., Glor, R. E., de Queiroz, K., Hertz, P. E., Schettino, L. penalties. McGill University. Retrieved from https://cran.r-project. R., … Larson, A. (2003). Niche lability in the evolution of a Caribbean org/web/packages/pspline/pspline.pdf lizard community. Nature, 424, 542–545. https://doi.org/10.1038/ Hedges, S. B., Parker, P. H., Sibley, C. G., & Kumar, S. (1996). Continental nature01814 breakup and the ordinal diversification of birds and mammals. Nature, Mao, K., Milne, R. I., Zhang, L., Peng, Y., Liu, J., Thomas, P., … Renner, 381, 226–229. https://doi.org/10.1038/381226a0 S. (2012). Distribution of living Cupressaceae reflects the breakup Hopkins, G. R., & Brodie, E. D. (2015). Occurrence of amphibi‐ of Pangea. Proceedings of the National Academy of Sciences USA, 109, ans in saline habitats: A review and evolutionary perspective. 7793–7798. https://doi.org/10.1073/pnas.1114319109 Herpetological Monographs, 29, 1–27. https://doi.org/10.1655/ Matzke, N. J. (2014). Model selection in historical biogeography reveals that HERPMONOGRAPHS-D-14-00006 founder‐event speciation is a crucial process in island clades. Systematic Hutter, C. R., Lambert, S. M., & Wiens, J. J. (2017). Rapid diversification Biology, 63, 951–970. https://doi.org/10.1093/sysbio/syu056 and time explain amphibian richness at different scales in the tropi‐ Mayhew, P. J., Bell, M. A., Benton, T. G., & McGowan, A. J. (2012). cal Andes, Earth's most biodiverse hotspot. American Naturalist, 190, Biodiversity tracks temperature over time. Proceedings of the 828–843. https://doi.org/10.1086/694319 National Academy of Sciences USA, 109, 15141–15145. https://doi. Jablonski, D., & Bottjer, D. J. (1991). Environmental patterns in the or‐ org/10.1073/pnas.1200844109 igins of higher taxa: The post‐Paleozoic fossil record. Science, 252, Meredith, R. W., Janecka, J. E., Gatesy, J., Ryder, O. A., Fisher, C. A., 1831–1833. https://doi.org/10.1126/science.252.5014.1831 Teeling, E. C., … Murphy, W. J. (2011). Impacts of the Cretaceous 16 | ROLLAND and CONDAMINE

Terrestrial Revolution and KPg extinction on mammal diver‐ Roalson, E. H., & Roberts, W. R. (2016). Distinct processes drive diver‐ sification. Science, 334, 521–524. https://doi.org/10.1126/ sification in different clades of Gesneriaceae. Systematic Biology, 65, science.1211028 662–684. https://doi.org/10.1093/sysbio/syw012 Morlon, H., Lewitus, E., Condamine, F. L., Manceau, M., Clavel, J., & Rohde, K. (1992). Latitudinal gradients in species diversity: The search Drury, J. (2016). RPANDA: An R package for macroevolutionary for the primary cause. Oikos, 514–527. analyses on phylogenetic trees. Methods in Ecology and Evolution, 7, Rolland, J., & Salamin, N. (2016). Niche width impacts vertebrate diversi‐ 589–597. https://doi.org/10.1111/2041-210X.12526 fication. Global Ecology and Biogeography, 25, 1252–1263. https://doi. Morlon, H., Parsons, T. L., & Plotkin, J. B. (2011). Reconciling mo‐ org/10.1111/geb.12482 lecular phylogenies with the fossil record. Proceedings of the Rolland, J., Silvestro, D., Schluter, D., Guisan, A., Broennimann, O., National Academy of Sciences USA, 108, 16327–16332. https://doi. & Salamin, N. (2018). The impact of endothermy on the climatic org/10.1073/pnas.1102543108 niche evolution and the distribution of vertebrate diversity. Nature Orr, M. R., & Smith, T. B. (1998). Ecology and speciation. Trends in Ecology and Evolution, 2, 459–464. https://doi.org/10.1038/ Ecology and Evolution, 13, 502–506. https://doi.org/10.1016/ s41559-017-0451-9 S0169-5347(98)01511-0 San Mauro, D., Vences, M., Alcobendas, M., Zardoya, R., & Meyer, A. Pérez‐Escobar, O. A., Chomicki, G., Condamine, F. L., Karremans, A. P., (2005). Initial diversification of living amphibians predated the Bogarín, D., Matzke, N. J., … Antonelli, A. (2017). Recent origin and breakup of Pangaea. American Naturalist, 165, 590–599. https://doi. rapid speciation of Neotropical orchids in the world's richest plant org/10.1086/429523 biodiversity hotspot. New Phytologist, 215, 891–905. https://doi. Sanmartín, I., Enghoff, H., & Ronquist, F. (2001). Patterns of ani‐ org/10.1111/nph.14629 mal dispersal, vicariance and diversification in the Holarctic. Peters, S. E. (2005). Geologic constraints on the macroevolutionary his‐ Biological Journal of the Linnean Society, 73, 345–390. https://doi. tory of marine animals. Proceedings of the National Academy of Sciences org/10.1111/j.1095-8312.2001.tb01368.x USA, 102, 12326–12331. https://doi.org/10.1073/pnas.0502616102 Sanmartín, I., & Meseguer, A. S. (2016). Extinction in phylogenetics and Peters, S. E. (2008). Environmental determinants of extinction selectivity biogeography: From timetrees to patterns of biotic assemblage. in the fossil record. Nature, 454, 626–629. https://doi.org/10.1038/ Frontiers in Genetics, 7, 35. nature07032 Sanmartín, I., & Ronquist, F. (2004). Southern Hemisphere biogeography in‐ Peters, S. E., Kelly, D. C., & Fraass, A. J. (2013). Oceanographic controls ferred by event‐based models: Plant versus animal Patterns. Systematic on the diversity and extinction of planktonic foraminifera. Nature, Biology, 53, 216–243. https://doi.org/10.1080/10635150490423430 493, 398–401. https://doi.org/10.1038/nature11815 Santos, J. C., Coloma, L. A., Summers, K., Caldwell, J. P., Ree, R., & Pianka, E. R. (1966). Latitudinal gradients in species diversity: A review of Cannatella, D. C. (2009). Amazonian amphibian diversity is pri‐ concepts. The American Naturalist, 100, 33–46. marily derived from late Miocene Andean lineages. PLoS Biology, 7, Portik, D. M., & Papenfuss, T. J. (2015). Historical biogeography resolves e1000056. the origins of endemic Arabian toad lineages (Anura: Bufonidae): Scotese, C. R. (2004). A continental drift flipbook. Journal of Geology, 112, Evidence for ancient vicariance and dispersal events with the Horn 729–741. https://doi.org/10.1086/424867 of Africa and South Asia. BMC Evolutionary Biology, 15, 152. https:// Seton, M., Müller, R. D., Zahirovic, S., Gaina, C., Torsvik, T., Shephard, doi.org/10.1186/s12862-015-0417-y G., … Chandler, M. (2012). Global continental and ocean basin re‐ Prokoph, A., Shields, G. A., & Veizer, J. (2008). Compilation and time‐ constructions since 200 Ma. Earth‐Science Reviews, 113, 212–270. 18 13 87 86 34 series analysis of a marine carbonate δ O, δ C, Sr/ Sr and δ S https://doi.org/10.1016/j.earscirev.2012.03.002 database through Earth history. Earth‐Science Reviews, 87, 113–133. Silvestro, D., Antonelli, A., Salamin, N., & Quental, T. B. (2015). The role https://doi.org/10.1016/j.earscirev.2007.12.003 of clade competition in the diversification of North American canids. Pyron, R. A. (2014). Biogeographic analysis reveals ancient continental Proceedings of the National Academy of Sciences USA, 112, 8684– vicariance and recent oceanic dispersal in amphibians. Systematic 8689. https://doi.org/10.1073/pnas.1502803112 Biology, 63, 779–797. https://doi.org/10.1093/sysbio/syu042 Silvestro, D., Salamin, N., & Schnitzler, J. (2014). PyRate: A new program Pyron, R. A., & Wiens, J. J. (2011). A large‐scale phylogeny of Amphibia in‐ to estimate speciation and extinction rates from incomplete fossil cluding over 2800 species, and a revised classification of extant frogs, data. Methods in Ecology and Evolution, 5, 1126–1131. https://doi. salamanders, and caecilians. Molecular Phylogenetics and Evolution, org/10.1111/2041-210X.12263 61, 543–583. https://doi.org/10.1016/j.ympev.2011.06.012 Silvestro, D., Schnitzler, J., Liow, L.‐H., Antonelli, A., & Salamin, N. (2014). Pyron, R. A., & Wiens, J. J. (2013). Large‐scale phylogenetic analyses re‐ Bayesian estimation of speciation and extinction from incomplete veal the causes of high tropical amphibian diversity. Proceedings of fossil occurrence data. Systematic Biology, 63, 349–367. https://doi. the Royal Society B: Biological Sciences, 280, 20131622. https://doi. org/10.1093/sysbio/syu006 org/10.1098/rspb.2013.1622 Smith, S. A. (2009). Taking into account phylogenetic and divergence‐time Rabosky, D. L., Slater, G. J., & Alfaro, M. E. (2012). Clade age and species uncertainty in a parametric biogeographical analysis of the Northern richness are decoupled across the eukaryotic tree of life. PLoS Biology, Hemisphere plant clade Caprifolieae. Journal of Biogeography, 36, 10, e1001381. https://doi.org/10.1371/journal.pbio.1001381 2324–2337. https://doi.org/10.1111/j.1365-2699.2009.02160.x Rambaut, A., Drummond, A. J., Xie, D., Baele, G., & Suchard, M. A. (2018). Stadler, T. (2011). Mammalian phylogeny reveals recent diversification Posterior summarisation in Bayesian phylogenetics using Tracer 1.7. rate shifts. Proceedings of the National Academy of Sciences USA, 108, Systematic Biology, 67, 901–904. https://doi.org/10.1093/sysbio/ 6187–6192. https://doi.org/10.1073/pnas.1016876108 syy032 Sun, Y., Joachimski, M. M., Wignall, P. B., Yan, C., Chen, Y., Jiang, H., … Ree, R. H., & Sanmartín, I. (2018). Conceptual and statistical problems Lai, X. (2012). Lethally hot temperatures during the Early Triassic with the DEC+ J model of founder‐event speciation and its com‐ greenhouse. Science, 338, 366–370. https://doi.org/10.1126/ parison with DEC via model selection. Journal of Biogeography, 45, science.1224126 741–749. https://doi.org/10.1111/jbi.13173 Valentine, J. W., & Moores, E. M. (1970). Plate‐tectonic regulation of fau‐ Ree, R. H., & Smith, S. A. (2008). Maximum likelihood inference nal diversity and sea level: A model. Nature, 228, 657–659. https:// of geographic range evolution by dispersal, local extinction, doi.org/10.1038/228657a0 and cladogenesis. Systematic Biology, 57, 4–14. https://doi. Van Valen, L. (1973). A new evolutionary law. Evolutionary Theory, 1, org/10.1080/10635150701883881 1–30. ROLLAND and CONDAMINE | 17

Vavrek, M. J. (2016). The fragmentation of Pangaea and Mesozoic terres‐ Ziegler, A., Eshel, G., Rees, P. M., Rothfus, T., Rowley, D., & Sunderlin, D. trial vertebrate biodiversity. Biology Letters, 12, 20160528. https:// (2003). Tracing the tropics across land and sea: Permian to present. doi.org/10.1098/rsbl.2016.0528 Lethaia, 36, 227–254. https://doi.org/10.1080/00241160310004657 Vences, M., & Wake, D. B. (2007). Speciation, species boundaries and phylogeography of amphibians. Amphibian Biology, 7, 2613–2671. von, Humboldt, A. (1807). Essai sur la géographie des plantes: BIOSKETCH Accompagné d'un tableau physique des régions équinoxiales, fondé sur des mesures exécutées, depuis le dixième degré de lat‐ Jonathan Rolland and Fabien L. Condamine are interested in itude boréale jusqu'au dixième degré de latitude australe, pendant the diversification processes of clades and the emergence of les années 1799, 1800, 1801, 1802 et 1803. Paris, France: Levrault, biodiversity patterns. They worked together on different pro‐ Schoell et Compagnie, Libraires. Wang, I. J., Glor, R. E., & Losos, J. B. (2013). Quantifying the roles of ecol‐ jects in the last 5 years. Their area of expertise ranges from ogy and geography in spatial genetic divergence. Ecology Letters, 16, genomics, phylogenetics, biogeography to evolutionary biology 175–182. https://doi.org/10.1111/ele.12025 in general. Jonathan Rolland is currently a Banting postdoctoral Wright, S. D., Gillman, L. N., Ross, H. A., & Keeling, J. D. (2010). Energy researcher in Dolph Schluter Lab in Vancouver (Canada), and and the tempo of evolution in amphibians. Global Ecology and Fabien Condamine is a research scientist at CNRS in Montpellier Biogeography, 19, 733–740. Xavier, A. L., Guedes, T. B., & Napoli, M. F. (2015). Biogeography of an‐ (France). urans from the poorly known and threatened coastal sandplains of eastern Brazil. PLoS ONE, 10, e0128268. https://doi.org/10.1371/ journal.pone.0128268 SUPPORTING INFORMATION Xu, X., Kuntner, M., Liu, F., Chen, J., & Li, D. (2018). Formation of riv‐ ers and mountains drives diversification of primitively segmented Additional supporting information may be found online in the spiders in continental East Asia. Journal of Biogeography, 45, 2080– Supporting Information section at the end of the article. 2091. https://doi.org/10.1111/jbi.13403 Zachos, J. C., Dickens, G. R., & Zeebe, R. E. (2008). An early Cenozoic perspective on greenhouse warming and carbon‐cycle dynamics. How to cite this article: Rolland J, Condamine FL. The Nature, 451, 279–283. https://doi.org/10.1038/nature06588 contribution of temperature and continental fragmentation Zaffos, A., Finnegan, S., & Peters, S. E. (2017). Plate tectonic regula‐ tion of global marine animal diversity. Proceedings of the National to amphibian diversification. J Biogeogr. 2019;00:1–17. Academy of Sciences USA, 114, 5653–5658. https://doi.org/10.1073/ https://doi.org/10.1111/jbi.13592 pnas.1702297114