Determinantes Ecológicos De Processos Macro E Micro Evolutivos Em Regiões Complexas

ADRIÁN GARCÍA-RODRÍGUEZ

DETERMINANTES ECOLÓGICOS DE PROCESSOS MACRO E MICRO EVOLUTIVOS EM REGIÕES COMPLEXAS

Natal, Rio Grande do Norte - Brasil 2018

ADRIÁN GARCÍA-RODRÍGUEZ

DETERMINANTES ECOLÓGICOS DE PROCESSOS MACRO E MICRO EVOLUTIVOS EM REGIÕES COMPLEXAS

Tese apresentada à Universidade Federal do Rio Grande do Norte, como parte das exigências do Programa de Pós-Graduação em Ecologia, para obtenção do título de Doutor.

Orientador Dr. Gabriel Corrêa Costa

Co-orientador Dr. Adrian Antonio Garda

ADRIÁN GARCÍA-RODRÍGUEZ

DETERMINANTES ECOLÓGICOS DE PROCESSOS MACRO E MICRO EVOLUTIVOS EM REGIÕES COMPLEXAS

Tese apresentada à Universidade Federal do Rio Grande do Norte, como parte das exigências do Programa de Pós-Graduação em Ecologia, para obtenção do título de Doutor.

Dr. Fabricio Villalobos Dr. Diogo Borges Provete Membro titular externo Membro titular externo Instituto de Ecología, A.C. UFMS (INECOL). México

Dr. Adrian Antonio Garda Dr. Sergio Maia Queiroz Lima Membro titular interno Membro titular interno UFRN UFRN

______Dr. Gabriel Corrêa Costa Orientador Auburn University, Alabama, EU

Universidade Federal do Rio Grande do Norte - UFRN Sistema de Bibliotecas - SISBI Catalogação de Publicação na Fonte. UFRN -

Biblioteca Setorial Prof. Leopoldo Nelson -Centro de Biociências - CB

García-Rodríguez, Adrián. Determinantes ecológicos de processos macro e microevolutivos em regiões complexas / Carlos Adrián García Rodríguez. - Natal, 2018. 160 f.: il.

Tese (Doutorado) - Universidade Federal do Rio Grande do Norte. Centro de Biociências. Departamento de Ecologia. Programa de Pós-Graduacão em Ecologia. Orientador: Prof. Dr. Gabriel Correa Costa.

1. Bioacústica - Tese. 2. Heterogeneidad climática - Tese. 3. Complexidade topográfica - Tese. 4. Divergência genética - Tese. 5. Especiação - Tese. 6. Macroevolução - Tese. I. Costa, Gabriel Correa. II. Universidade Federal do Rio Grande do Norte. III. Título.

Elaborado por KATIA REJANE DA SILVA - CRB-15/351

Vive como si fueses a morir mañana. Aprende como si fueses a vivir para siempre.

Mahatma Gandhi

La ciencia se compone de errores, que a su vez, son los pasos hacia la verdad Julio Verne

AGRADECIMIENTOS

Tras cuatro años de mucho aprendizaje y crecimiento personal, cierro aquí la que puedo decir -sin temor a equivocarme- ha sido la experiencia más enriquecedora de mi vida. Quedan plasmadas en estas páginas, las ideas que poco a poco fui madurando durante este tiempo, pero más que un documento que pretende hacer una pequeña contribución al conocimiento, este trabajo es el vivo reflejo de un esfuerzo (y sacrifício) conjunto, y a la vez mi humilde homenaje a todas las personas que me acompañaron en el camino.

Agradezco sobre todo a mi família, que es el centro de mi vida. A mis padres, Carlos Alberto y Carmen María, por ser mi luz, mi ejemplo y mi más grande apoyo en todo momento. Por su amor incondicional, por siempre motivarme a perseguir lo que me apasiona, por acuerpar mis decisiones y por enseñarme, desde que tengo memoria, a valorar la educación como el tesoro más preciado que me podían dar. A mis hermanas Silvi, Lauri y Marce y mis sobrinos Sofi, Ale y Amandita por su cariño, su comprensión, su complicidad, su admiración y por ser mi fuerza y motivación aún en la distancia.

A mi orientador Gabriel Costa, por ser mi principal guía en este proceso.

Por su integridad profesional y su sincera amistad, por la admirable dedicación y paciencia que tiene con sus alumnos. Por su inspiradora capacidad para sacar

lo mejor de cada uno de nosotros y estar presente y disponible aún a 7000 km de distancia. A mis co-autores y amigos Marcelo Araya, Andrew Crawford,

Adrian Garda, Carlos Guarnizo, Pablo Martinez, Brunno Oliveira y Alex Pyron por toda su colaboración y críticas constructivas durante el desarrollo de estos trabajos.

A mis mentores en Costa Rica, Cachí, Fede y GB por su influencia y consejos durante mi formación como estudiante y después como profesional, por su apoyo contínuo hasta el día de hoy. A la Escuela de Biología de la

Universidad de Costa Rica, especialmente a Gustavo Gutiérrez, Viviana Lang,

Elsa de la O y demás funcionarios que siempre me han apoyado a lo largo de esta etapa, facilitando todos los procesos administrativos que permitieron mantener un vínculo profesional con mi querida Álma Mater.

A todos mis amigos en Costa Rica (mis “Brothers” del cole, mis queridos

“Peleles” de pretil, y toda la chusma de Biolo) por siempre tener una sonrisa para recibirme y un abrazo para despedirme, por desearme lo mejor y ante todo por no dejar que la distancia nos separe. A Juanca y Eu, mis hermanitos ticos que se embarcaron conmigo en esta aventura brasilera y fueron siempre mi pedacito de Tiquicia en el exilio. A los ticos con los que en algún momento coincidí estando Brasil: Sarita, Kabeto, Boris y Hellen, gracias por la ayuda, la solidaridad y los buenos momentos. A mis grandes amigos y colegas, Pitillo,

Erick, Victicor, Sofi Rodríguez, Sofi Granados y Bety por su ayuda en el campo y su amistad sincera.

A mis amigos en Brasil, quienes hoy son mi família lejos de casa. Mis colegas del lab Juampi, Bruninho, André, Brunnão, Tales y Felipe, por la compañia, las buenas energias y la ayuda siempre desinteresada. A mis roomies, con quiénes compartí mil historias y cuya compañia hizo todo más fácil desde el inicio, principalmente a Camura y Anita mis hermanitos y cómplices, gracias por tanto cariño y sonrisas compartidas, tamo junto sempre. A Vekinha, que le tocó aguantarme en la recta final de esta tesis, gracias por el apoyo, la paciencia y los chineos durante esta “labor de parto”. A quienes a ojos cerrados literalmente me entregaron sus casas, sus carros y sus mascotas: mis hermanos y consejeros Hélder y Carolzinha, mis amados Duka e Helo, mi otro hermano

Juampi y mis queridas Tamy e Isa, gracias no sólo por todo lo que facilitaron mi vida, sino por esas grandes muestras de confianza, somos familia. A mis grandes amigos y colegas Castiele, Eliana, Vinicius y Francisco, que haciendo un esfuerzo gigantesco me visitaron en Costa Rica y también me acompañaron en mis giras de campo, recuerdos para siempre mis queridos. A mis amigas, consejeras, confidentes y hasta enfermeras Andressa y Nadia por cuidarme cuando no estuve bien y escucharme siempre que lo necesité. A Gus y Serginho

por su sincera amistad y por siempre tener esa energía leve y una palabra adecuada para compartir.

Agradezco a todos los profesores, amigos y alumnos con quiénes coincidí en el

Programa de Posgraduacao em Ecologia de la UFRN. Ha sido un placer y un honor ser parte del programa y un gusto inmenso haberlos encontrado en este camino. Todos y cada uno de ustedes forman parte de mi historia, son el más puro reflejo de la solidaridad y una muestra clara de que la amistad trasciende idiomas y fronteras... saudades galera!

Agradezco infinitamente a este Brasil querido, por recibirme de brazos abiertos, por presentarme algunos de los lugares más lindos y personas más especiales que conocí en mi vida. Por desbordarme con su diversidad cultural y sus bellezas naturales, por contagiarme de esa alegría que nunca acaba y hacerme sentir en casa. Nada de esto hubiera sido posible sin el apoyo de la Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) que financió durante 48 meses mi vida en Brasil y National Geographic Society que apoyó mi trabajo de campo en Costa Rica.

Gracias a todos, gracias por todo!

SUMÁRIO Agradecimientos...... 6 Resumo...... 12

Abstract...... 14

Introdução Geral...... 16

Capítulo 1. Faster amphibian speciation supports the role of mountains as biodiversity pumps

Abstract……………………………………………….. 26

Introdução...... 27

Material e métodos...... 30

Resultados...... 36

Discussão...... 44

Referências...... 50

Material suplementar...... 61

Capítulo 2. Idiosyncratic responses to drivers of genetic differentiation in the complex landscapes of Isthmian Central America

Abstract………………………………………………... 65

Introdução...... 67

Material e métodos...... 72

Resultados...... 82

Discussão...... 88

Referências...... 98

Material suplementar...... 104

Capítulo 3. The role of geography, topography and climate in the acoustic divergence of Neotropical Diasporus frogs

Abstract...... 116

Introdução...... 118

Material e métodos...... 121

Resultados...... 128

Discussão...... 136

Referências...... 144

Considerações finais...... 158

Resumo

As áreas de montanha do mundo cobrem menos de 15% da superfície terrestre; no entanto, elas concentram cerca de 90% dos hotspots de diversidade de espécies e 40% dos hotspots de endemismo. As evidências sugerem que fatores como a complexidade topográfica, a heterogeneidade climática e sua dinâmica histórica nas montanhas podem desempenhar um papel importante na evolução e manutenção de suas ricas biotas. Nesta tese, pretendi avaliar o papel de tais fatores tanto em escala macro (ou seja, nos padrões globais de especiação) quanto em escalas microevolutivas (ou seja, intraespecíficas de divergência genética e de traits) usando anfíbios como sistema de estudo. No primeiro capítulo, contrastei as taxas de especiação entre regiões de alta e baixa complexidade topográfica. Para este fim, usei uma filogenia quase completa de anfíbios contendo 7238 espécies (>90% da diversidade existente) para rodar uma Análise Bayesiana de Misturas Macroevolutivas (BAMM) que permite estimar as taxas de especiação. Posteriormente, projetei na geografia essa informação usando os mapas de distribuição disponíveis, para explorar padrões geográficos de especiação em anfíbios e avaliei sua associação com terrenos complexos, estimando um índice global de complexidade topográfica. Encontrei que, globalmente, as taxas de especiação são mais rápidas em regiões de alta complexidade topográfica independentemente da latitude. Desconstruí esse padrão repetindo as análises nas regiões Zoogeográficas de Wallace - levando em consideração as histórias evolutivas regionais independentes - e encontrei a mesma tendência em oito dos 11 reinos zoogeográficos. No segundo capítulo, avalio o papel relativo de diferentes componentes da paisagem na promoção da diversificação da linhagem na complexa topografia da América Central Ístmica (ACI: Costa Rica e Panamá), uma região geologicamente jovem, mas altamente biodiversa. Aqui usei DNA mitocondrial para estimar a divergência genética dentro de 11 espécies de anfíbios (9 anuros e 2 salamandras) com diferentes atributos ecológicos que ocorrem conjuntamente na região. Então, utilizei análises de Matriz Múltipla de Regressão com Randomização e Modelagem de Dissimilaridade Generalizada para quantificar o papel relativo do isolamento por distância, ambiente e resistência (topografia e adequação) na modelagem de padrões geográficos de estrutura genética dentro de cada espécie. Encontrei respostas idiossincráticas que podem refletir aspectos específicos de suas histórias de vida e poderiam dar uma visão sobre o papel da ACI como motor da especiação. No terceiro capítulo, testei se as barreiras climáticas e topográficas podem influenciar a variação dos sinais acústicos de duas espécies de sapos do gênero Diasporus. Este é um traço comportamental importante que possui características particulares que

permitem o reconhecimento intra-específico e podem desempenhar um papel importante como mecanismo de isolamento reprodutivo. Para este capítulo, gravei vocalizações de anúncio de 170 machos de duas espécies de sapos do gênero Diasporus distribuídos na Costa Rica. Eu realizei gravações em 21 locais em todo o país, desde o nível do mar até 2800 metros de altitude. Com essa informação realizei análises bioacústicas para documentar a variação geográfica e análises correlativas de matrizes múltiplas para testar se a distância geográfica, as barreiras físicas ou climaticas entre populações, ou adaptação às condições locais podem moldar tais padrões. Para esse fim, eu incorporei análises espaciais (modelos de nicho, estimativas de rugosidade do terreno e teoria dos circuitos) para estimar níveis de isolamento das populações e ajustar um modelo de dissimilaridade generalizada para abordar esta questão. Nas duas espécies, encontrei altos níveis de variação acústica, assim como de isolamento entre populações, gerado pelos fatores testados. No entanto, somente as barreiras topográficas explicaram significativamente a variação acústica em D. diastema. Entretanto, a dissimilaridade climática e distância geográfica só possui associação marginal com os padrões de variação acústica encontrados. Em conclusão, consideramos forças que operam em uma escala local e de forma independente (por exemplo a seleção sexual, o deslocamento de caracteres ou mesmo deriva genética) poderiam então ser mais determinantes na evolução desses sinais nas espécies de estudo.

Palavras chave: Bioacústica, Complexidade Topográfica, Divergência Genética, Especiação, Genética da Paisagem, Heterogeneidade Climática, Isolamento, Macroecologia, Macroevolução, Montanhas

Abstract

Mountain areas around the world cover less than 15% of global land surface; nevertheless, they concentrate around 90% of the hotspots of species diversity and 40% of the hotspots of endemism. Available evidence suggest that ecological factors such as landscape features (i.e topographic complexity, climatic heterogeneity and their historical dynamics) of mountains may play an important role in the evolution and maintenance of rich biotas at such regions. In my dissertation I aim to evaluate the role of such factors in both macro (i.e global speciation patterns) and microevolutionary (i.e intra-specific genetic and trait divergence) processes using amphibians as study system. In the first chapter, we tested in a global scale the Montane Pumps hypothesis, which proposes that speciation rates are faster in mountains explaining higher diversities in those regions. To this end we used a near complete Amphibian phylogeny containing 7238 species (>90% of the group’s extant diversity) and conducted a Bayesian Analysis in Macroevolutionary Admixtures (BAMM) to estimate speciation rates. Then we combined this information with available range maps to explore Amphibian geographic patterns of speciation and evaluated its association with complex terrains (mountains) by estimating a global index of topographic complexity. We found that globally, speciation rates are faster in regions of high topographic complexity independently of latitude. We repeated our analyses using the Wallace’s Zoogeographic regions, taking into account regional independent evolutionary histories, and found the same pattern in eight out of the total 11 zoogeographical realms. In a second chapter, we assess the relative role of different components of the landscape in promoting lineage diversification across the roughed topography of Isthmian Central America (Costa Rica & Panama), a geologically young but highly biodiverse region. Here we use available mitochondrial DNA to estimate genetic divergence within 10 amphibian species (8 anurans and 2 salamanders) with different biologies that co-occur in the region. Then, we use a Multiple Matrix of Regression with Randomization to assess the relative role of isolation by distance, by environment and by resistance (topography, current climate, and LGM paleoclimate) in shaping the geographic patterns of genetic structuration within each species. So far, we have not found a general force that explains genetic divergence in all studied species. Instead, we have found idiosyncratic responses that may reflect specific aspects of their life histories, such as dispersal capabilities, range size or reproductive potential. In the third chapter,

we test how climatic and topographic barriers may influence variation in an important behavioral trait such as are advertisement calls. In anurans, such calls has species-specific features that play an important role in recognition. Then, variation in spectro-temporal features between populations has been proposed as a mechanism of reproductive isolation that may promote speciation in the long term. For this chapter I recorded advertisement calls of 170 males from 2 species of Diasporus frogs distributed in Costa Rica. I made recordings at 21 sites in all the country ranging from sea level to 2800 meters elevation. We use such information we conduct bioacoustics analyses to first document geographic variation and then test if the geographic distance, physical or ecological barriers between populations, or adaptation to local conditions could shape such patterns. To this end, we incorporate spatial analyses (niche models, terrain roughness estimations and circuit theory) to generate levels of population isolation and apply Generalized Dissimilarity Matrix test to address this question. In both species, I found high levels of acoustic variation and among population isolation derived by the tested factors. However, only topography significantly explained acoustic divergence in D. diastema while climatic dissimilarity and geographic distance are only marginally associated with the patters of acoustic variation in D. hylaeformis. In conclusion, other forces operating independently in the local scale -such as sexual selection, character displacement or genetic drift- may be more determinant in the evolution of acoustic signals in these species.

Keywords: Bioacoustics, Climatic Heterogeneity, Genetic Divergence, Isolation, Landscape Genetics, Macroecology, Macroevolution, Mountains, Speciation, Topographic Complexity

INTRODUÇÃO GERAL

Um dos fenômenos naturais mais amplamente documentados é a distribuição desigual que tem a diversidade em múltiplas dimensões (Menge and Sutherland

1976). Os padrões de riqueza de espécies, variam através do espaço, do tempo e dos clados: algumas regiões são mais diversas que outras (Hillebrand 2004), a composição da diversidade hoje não é a mesma que no passado (Johnson

2009) ainda, alguns grupos taxonômicos são muito diversos, outros contem poucos representantes (Wiens 2011). Entender o porquê dessa variação tem se tornado um dos maiores objetivos de pesquisa na intersecção da ecologia e evolução, gerando hipóteses derivadas dessas duas áreas da ciência.

Na escala espacial, uma das mais extremas variações na distribuição da diversidade acontece em áreas de topografias irregulares. Globalmente os sistemas montanhosos tem uma distribuição desigual que abrange somente uma oitava parte da superfície da terra (Antonelli 2015, Körner et al. 2017). No entanto, essas regiões concentram altas riquezas de espécies, sendo que 90% dos hotspots de diversidade e 40% dos hotspots de endemismo ocorrem em

áreas de montanha (Myers et al. 2000, Orme et al. 2005). A tendência que tem as regiões de alta complexidade topográfica para suportar altos números de espécies é um padrão bem documentado em diversos grupos animais e vegetais

(Ruggiero and Hawkins 2008). Porém, os determinantes ecológicos e os

mecanismos macro e micro evolutivos que geram essa diversidade biótica ainda são pouco conhecidos.

Tem-se sugerido que as regiões montanhosas poderiam agir como motores de especiação, pelo efeito duplo que as topografias complexas e os fortes gradientes ambientais contidos nelas podem ter nos processos de divergência genética (Funk et al. 2016). As configurações irregulares de topos de montanha e vales alternados representam mosaicos de habitats favoráveis e desfavoráveis (Kozak and Wiens 2006), que aumentam o isolamento entre populações e em consequência a probabilidade de especiação alopátrica (Orr and Smith 1998, Moritz et al. 2000, Rull 2005, Guarnizo et al. 2009).

Complementariamente, os amplos espectros ambientais representados em curtas distâncias ao longo dos gradientes altitudinais (Graham et al. 2014,

Merckx et al. 2015), oferecem condições ideais em que a especiação ecológica em parapatria pode ocorrer (Rundle and Nosil 2005). Nessas circunstâncias, as pressões locais poder promover divergência entre populações, levando ao surgimento de novas espécies, mesmo na ausência de barreiras físicas maiores

(Knox and Palmer 1995, Graham et al. 2004, Caro et al. 2013, Chapman et al.

2013).

Dentro de uma perspectiva macro ecológica, a montagem de comunidades e riqueza de espécies numa região especifica, num dado momento,

é determinada pelos processos de especiação, extinção e dispersão (Hutter et al.

2013). Portanto, a identificação dos fatores que potencialmente influenciam nestes processos é crucial para entender a origem dos amplos padrões de diversidade atual. Em escalas mais locais, as abordagens desenvolvidas nas

áreas da filogeografia e genética da paisagem tem sido úteis para abordar essa questão com maior resolução espacial mas menor alcance taxonômico e geográfico.

Nesta tese avalio em diferentes escalas geográficas de que forma as paisagens complexas determinadas por regiões montanhosas possuindo perfis climáticos heterogêneos influenciam em processos evolutivos que contribuem para a formação dos padrões biológicos que observamos. O meu interesse foi primeiramente abordar essa questão tentando obter o ‘big picture’ da generalidade de certos padrões macro evolutivos em escala global; ao mesmo tempo, que procurei aprofundar numa maior resolução, testando o rol que tem certos atributos físicos e ecológicos da paisagem na geração de pressões locais que influenciam os processos micro evolutivos de diferenciação genética e divergência acústica no espaço. Para atingir esses objetivos eu incorporei diversas análises evolutivas, conceitos de genética de populações, abordagens da ecologia do comportamento e ferramentas de machine learning para projetar

no espaço múltiplos padrões de variação e avaliar quais são as forças que lós explicam melhor.

No primeiro capítulo, testei em escala global se existe uma relação entre taxas de especiação mais rápidas e regiões topograficamente complexas, que potencialmente poderia explicar maiores diversidades nessas regiões. Para este fim, usei uma filogenia quase completa de anfíbios para estimar dinâmicas evolutivas. Posteriormente, espacializei essa informação para explorar padrões geográficos de especiação em anfíbios e avaliei sua associação com terrenos complexos, estimando um índice global de complexidade topográfica. No segundo capítulo, avalio o papel relativo de diferentes componentes da paisagem na promoção da diversificação da linhagem na complexa topografia da América Central Ístmica (ACI: Costa Rica e Panamá). Aqui usei DNA mitocondrial para estimar a divergência genética dentro de 11 espécies de anfíbios que ocorrem conjuntamente na região. Posteriormente quantifiquei o papel relativo do isolamento por distância, ambiente e resistência (topografia e adequação bioclimatica) na modelagem de padrões geográficos de estrutura genética dentro de cada espécie. No terceiro capítulo, testei como as barreiras climáticas e topográficas podem influenciar a variação nas chamadas de anuncio de duas espécies de sapos do gênero Diasporus. Para este capítulo, gravei vocalizações de anúncio de 170 machos em 21 locais na Costa Rica,

desde o nível do mar até 2800 metros de altitude. Com essa informação eu documentei a variação acústica intraespecifica e testei se a distância geográfica, o isolamento gerado pela topografia e o clima, ou a adaptação às condições locais podem moldar tais padrões.

Referências

Antonelli, A. 2015. Biodiversity: multiple origins of mountain life. - Nature 524: 300–301.

Caro, L. M. et al. 2013. Ecological speciation along an elevational gradient in a tropical passerine bird? - J. Evol. Biol. 26: 357–374.

Chapman, M. A. et al. 2013. Genomic divergence during speciation driven by adaptation to altitude. - Mol. Biol. Evol. 30: 2553–2567.

Funk, W. C. et al. 2016. Elevational speciation in action? Restricted gene flow associated with adaptive divergence across an altitudinal gradient. - J. Evol. Biol. 29: 241–252.

Graham, C. H. et al. 2004. Integrating phylogenetics and environmental niche models to explore speciation mechanisms in dendrobatid frogs. - Evolution. 58: 1781–93.

Graham, C. H. et al. 2014. The origin and maintenance of montane diversity: integrating evolutionary and ecological processes. - Ecography. 37: 711– 719.

Guarnizo, C. E. et al. 2009. The relative roles of vicariance versus elevational gradients in the genetic differentiation of the high Andean tree frog, Dendropsophus labialis. - Mol. Phylogenet. Evol. 50: 84–92.

Hillebrand, H. 2004. On the generality of the latitudinal diversity gradient. - Am. Nat. 163: 192–211.

Johnson, C. N. 2009. Ecological consequences of Late Quaternary extinctions of megafauna. - Proc. R. Soc. B 276: 2509–2519.

Knox, E. B. and Palmer, J. D. 1995. Chloroplast DNA variation and the recent radiation of the giant senecios (Asteraceae) on the tall mountains of eastern Africa. - Proc. Natl. Acad. Sci. U. S. A. 92: 10349–10353.

Körner, C. et al. 2017. A global inventory of mountains for bio-geographical applications. - Alp. Bot. 127: 1–15.

Kozak, K. H. and Wiens, J. J. 2006. Does niche conservatism promote speciation? A case study in North American salamanders. - Evolution. 60: 2604–21.

Mayr, E. 1963. Animal species and evolution. - Eugen. Rev. 55: 226–228.

Menge, B. A. and Sutherland, J. P. 1976. Species Diversity Gradients: Synthesis of the Roles of Predation, Competition, and Temporal Heterogeneity. - Am. Nat. 110: 351.

Merckx, V. S. F. T. et al. 2015. Evolution of endemism on a young tropical mountain. - Nature 524: 347–350.

Moritz, C. et al. 2000. Diversification of rainforest faunas: an integrated molecular approach. - Annu. Rev. Ecol. Syst. 31: 533–563.

Myers, N. et al. 2000. Biodiversity hotspots for conservation priorities. - Nature 403: 853–8.

Orme, C. D. L. et al. 2005. Global hotspots of species richness are not congruent with endemism or threat. - Nature 436: 1016–1019.

Orr, M. R. and Smith, T. B. 1998. Ecology and speciation. - Trends Ecol. Evol. 13: 502–506.

Ruggiero, A. and Hawkins, B. A. 2008. Why do mountains support so many species of birds? - Ecography (Cop.). 31: 306–315.

Rull, V. 2005. Biotic diversification in the Guayana Highlands: a proposal. - J. Biogeogr. 32: 921–927.

Rundle, H. D. and Nosil, P. 2005. Ecological speciation. - Ecol. Lett. 8: 336– 352.

Wiens, J. J. 2011. The causes of species richness patterns across space, time, and clades and the role of “ecological limits”. - Q. Rev. Biol. 86: 75–96.

Wiens, J. J. and Graham, C. H. 2005. Niche Conservatism: integrating evolution, ecology, and conservation biology. - Annu. Rev. Ecol. Evol. Syst. 36: 519–539.

CAPÍTULO I*

Faster amphibian speciation supports the role of mountains as biodiversity pumps

1,2 3 1,4 Adrián García-Rodríguez , Pablo A. Martínez , Brunno F. Oliveira , R. Alexander Pyron5 & Gabriel C. Costa6

1 Departamento de Ecologia, Universidade Federal do Rio Grande do Norte, Natal

- RN, Brasil, 59078-900

2 Escuela de Biología, Universidad de Costa Rica, San Pedro, 11501-2060 San José,

Costa Rica.

3 PIBi Lab. (Laboratorio de Pesquisas Integrativas em Biodiversidade), Programa de Pós-Graduação em Ecologia e Conservação, Universidade Federal do Sergipe,

São Cristóvão, Brasil

4 Department of Wildlife Ecology and Conservation, University of Florida,

Gainesville, FL 32611-0430, USA

5 Department of Biological Sciences, The George Washington University, 2023 G

Street NW, Washington, DC 20052, USA

6 Department of Biology, Auburn University at Montgomery, Montgomery, AL

36124, United States of America.

*Corresponding author; email: [email protected]

ABSTRACT

Continental mountain areas cover less than 15% of global land surface; nevertheless, around 90% of the hotspots of species diversity and 40% of the hotspots of endemism are concentrated in these regions. Such high diversities could be explained by higher diversification rates in regions of high topographic complexity, giving mountains the character of speciation pumps. We specifically focused on testing whether speciation is faster in mountains by conducting macro evolutionary analyses on a near complete Amphibian phylogeny and evaluating geographic patterns of this evolutionary rate. We accounted for the role of topographic complexity on speciation patterns across the globe and within zoogeographic realms. We found that globally, speciation rates are higher in mountainous areas. At a regional scale, we found the same pattern for most zoogeographical realms. Moreover, clades showing the fastest speciation rates are groups with predominantly montane distributions. Our study bolsters the importance of mountains as engines of speciation at different geographical scales. Due to their remoteness, the real contribution of such areas to the origin and maintenance of global biodiversity is probably still underestimated. These facts and the risk these regions face from global change suggests that mountains around the globe should be conservation priorities in local and regional agendas.

Keywords: Amphibians, BAMM, Macroecology, Macroevolution, Topographic

Complexity

BACKGROUND

Nearly one-third of the world’s terrestrial species diversity is concentrated in regions of high topographic complexity [1]. High diversity in mountain regions is a well- documented pattern [2–4], reported for numerous taxa and regions [5]. In Central and North America for example, mammal diversity is greater in regions dominated by mountains and complex reliefs [6]. Likewise, peaks of species richness and endemism of Afrotropical avifauna occurs within mountains and mountain-lowland complexes [7] . Worldwide, most global centres of vascular plant richness (>5000 species per 10,000 km2) are located in regions dominated by mountainous areas such as Costa Rica-Chocó, Tropical Eastern Andes, Atlantic Brazil, Northern Borneo and

New Guinea [8]. As a consequence, despite continental mountain areas covering less than 15% of global land surface [9], around 90% of the hotspots of species diversity and 40% of the hotspots of endemism [10,11] are concentrated in these regions.

Although this pattern has been reported for several taxa and across different regions [5], we still lack a comprehensive understanding of the mechanisms that drive higher diversity in mountains [12]. From an evolutionary perspective, montane systems have been hypothesized to be engines of diversification, because of their potential to drive speciation, both in allopatry and parapatry [13]. Evidence of allopatric speciation [14] promoted by the vicariant settings implicit in complex topographies have been widely documented in a variety of taxa [15–17]. For many

groups, the irregular configuration of alternate mountaintops and valleys represent mosaics of favourable and unfavourable habitats [18] that increases isolation among populations, thereby increasing opportunities for allopatric speciation [19].

Moreover, the distribution of such suitable regions has varied in response to historical climatic oscillations, increasing allopatric diversification in the mountains

[20]. Other features of mountains are the wide environmental spectrums they cover in short distances along their elevational gradients [12,21]. These transitions offer ideal conditions where ecological speciation in parapatry can take place [22]. In these circumstances, local pressures can drive adaptive divergence between populations, leading to the formation of new species, in the absence of hard geographic barriers [23–26].

Whether by allopatric or parapatric speciation, the idea that mountains act as cradles of biodiversity has been supported in several studies that linked the chronology of orogenic events to radiations of clades. For instance, the rise of the

Tibetan Plateau seems to have triggered the rapid radiation of glyptosternoid catfishes [27]; ranid frogs [28] and plants of the families Asteraceae and Fabaceae

[29,30]. Similarly, accumulating evidences suggests that the Andes uplift impacted evolutionary dynamics of Neotropical taxa such as hummingbirds from the genus

Adelomyia [31], butterflies from the subtribe Oleriina [32], and a variety of angiosperm clades [33–36].

Since the processes of speciation, extinction and dispersal are the ultimate determinants of diversity occurring in a given geographic region [37], identifying potential factors that drives these processes is crucial to understand the origin and distribution of past, present, and future biodiversity. Recently, several hypotheses based on this evolutionary framework have been proposed to explain the rich biotas in montane regions [12]. One of them is the Montane species pump hypothesis, which predicts that clades occurring at mountains have higher rates of net diversification [38] likely as consequence of their higher rates of speciation.

Evidence supporting this model has been reported for Mesoamerican hylid frogs as well as for tanagers and butterflies from the Andes; in these cases, montane clades showed higher speciation rates than those whose ranges are restricted to lowlands

[38–40].

However, the few studies testing whether complex topographic regions are speciation pumps were too restrictive in terms of their phylogenetic scope (i.e. few specific clades were analysed) and geographical extent (i.e. explored only local to regional scales), which limits our ability to determine the generality of topographic complex regions as speciation pumps. Here, we assess the prediction that complex topographies promote faster speciation rates. To do this, we use amphibians as a study system, and integrate global information on species distributions, terrain complexity and novel analyses on evolutionary dynamics across a nearly complete

phylogeny of the group. Amphibians are a particularly suitable study system to test this hypothesis because they represent an ancient radiation (~7700 species, www.amphibiaweb.org), with widespread latitudinal and altitudinal distribution across the globe [41] and a growing availability of phylogenetic information [42,43].

In addition, their high philopatry [44], restricted dispersal capabilities [45], limited osmotic tolerance [46], high sensitivity to temperature in early developmental stages

[47], and adaptations to particular elevations [48,49] bond their evolutionary fate strongly to their geographic settings, providing a valuable opportunity to investigate the forces shaping speciation patterns in montane regions.

METHODS

Amphibian Phylogeny

When inferring diversification dynamics through time, inclusion of all lineages in focal clades or regions has been proven to be of special importance [50,51].

Considering the known sampling bias towards particular clades and specific geographic areas as well as the global character of our approach, we attempted to improve the performance of our analysis by using a tree containing as many species as possible, even those lacking molecular data. Recent practice enables the incorporation of lineages lacking genetic data on tree inference using a given set of priors on branching times [52]. Then, we based our macroevolutionary analyses on

recently published trees which to date represent the most complete amphibian phylogenetic inference [53]. These trees were constructed using the Phylogenetic

Assembly with Soft Taxonomic Inferences (PASTIS) approach [52] updating an existing molecular supermatrix [43] that contains sequence data (5 mitochondrial and 10 nuclear genes) for ~56% of extant amphibian species. A Maximum-

Likelihood (ML) topology for these species then served as backbone for a set of 10,

000 trees containing 7,238 species, which represent ~94% of the known extant amphibian diversity and includes most families, subfamilies and genera. For detailed description on dating and tree construction, see [53].

Amphibian Evolutionary Dynamics

In order to estimate evolutionary rates, we modelled macroevolutionary dynamics across the amphibian phylogeny using Bayesian Analysis of Macroevolutionary

Mixtures (BAMM) [54]. BAMM models complex dynamics of speciation, extinction and trait evolution on phylogenetic trees, by detecting and quantifying heterogeneity on those rates while exploring a vast parameter-space of diversification models via reversible Markov Chain Monte Carlo (MCMC) [55].

This approach is useful since it does not assume that rates of speciation and extinction are constant, and can account for rate variation through time and among lineages [56]. The performance and theoretical foundations of BAMM has recently

received criticism, mainly dealing with the algorithm’s likelihood function, the posterior distribution on the number of rate shifts and the reliability of its diversification rate estimates [57]. However, BAMM’s authors have provided detailed evidence to clarify those concerns and demonstrated satisfactory and consistent performance of the method [58]. BAMM analysis provide speciation and extinction rates per species as direct output, and is possible to estimate net diversification rates by subtracting extinction rates from speciation rates [59].

However, we decided to focused our analyses on speciation rates, because they can be estimated with much more confidence than extinction rates, for which confidence intervals tend to be large, even when all assumptions of the inference model are satisfied [60]. Details of this analysis are provided in the supplementary material

(electronic supplementary material, text S1).

Spatial patterns of Amphibian speciation

We used geographical range maps for 6311 amphibian species obtained from the

IUCN (www.iucnredlist.org). These maps represent approximately 85% of the known extant amphibian species (~7500 species, www.amphibiaweb.org). Although we estimated macro evolutionary dynamics using ~94% of amphibian diversity represented in our phylogenetic tree, available range maps limited our analyses to a smaller number of species projected in the geographical space. We overlaid species

range maps in a 1x1 degree global grid and extracted species presence-absence within each grid cell, creating a presence-absence matrix for the 6311 species in the phylogeny that had range maps available. These analyses were conducted in the R package LetsR [61]. We further estimated speciation rates based on species composition within each grid cell.

Some authors have argued that species ranges may be too dynamic and this would mask any potential relationship between current distributions and the geography of speciation [62]. However, strong evidence supporting range stasis is available in the literature for a variety of organisms, from fossil molluscs to living insects and mammals [63–65]. We considered that it is unlikely that all species have altered their ranges enough to remove geographical signal from their past distribution. Most amphibian species have low dispersal ability [66] and are highly sensitivity to environmental conditions, resulting in a high proportion of species of small range sizes [67,68]. Therefore, the effects of range dynamics on the geographical signal we are investigating should be a minor concern in this study, especially at the scales we are working.

Topographic complexity

In order to have an informative proxy of geomorphologic heterogeneity, we generated a global index of topographic complexity (TC). Using a global layer of

elevation at 30-second resolution (~1km at the equator, http://www.worldclim.org/) we calculated the standard deviation of differences between 100x100 adjacent elevations. This procedure has been demonstrated to more accurately represent topographic roughness than elevation range, which only indicates the strength of a gradient within a cell [69]. We projected our TC layer to match the 1x1 degree resolution of our species distribution dataset.

Amphibian speciation in topographic complex regions

TC is not evenly distributed around the world [70]. This pattern reflects in our metric of TC, for which the number of cells with low values widely exceeds the number of cells with high values across the globe (Fig. 1c). To account for this, we created two categories: low topographic complexity (LTC) and high topographic complexity

(HTC). We considered as HTC cells that have a complexity index value higher than

300. Our complexity index is correlated with altitude and a value of 300 assures that we are selecting regions that are at least 600 meters elevation. This approach is conservative considering that Körner et al. (2017) defined montains as those areas above 200m elevation.

To test the montane pump hypothesis, we compared speciation rates between

LTC and HTC regions. According to this hypothesis, HTC areas should show higher speciation rates than LTC areas. HTC cells represent only a small fraction of the

total number of cells across the globe (2196 cells or 14.5% of all cells analysed).

Therefore, to test for significant differences between LTC and HTC speciation rates we used a rarefaction procedure [71]. We calculated the average speciation rate for all HTC cells and next, randomly sampled 2196 cells from the LTC and calculate the average speciation rate. Finally, we repeated this procedure 10,000 times, generating a distribution of average speciation rates for LTC. The observed average for HTC was then compared with the LTC generated distribution to assess significance. In order to test how speciation rates vary between LTC and HTC regions at different latitudes, we conducted this same analysis within the updated zoogeographic realms, a classification that defines robust biogeographic units based on global distributions and phylogenetic relations from over 20,000 world´s vertebrate species [71]. Therefore, using this delimitation also allows us to consider the evolutionary histories of the different zoological Realms.

Finally, we provided some examples to illustrate general patterns, where we compared mean speciation trajectories between predominantly montane groups and groups mainly distributed in adjacent lowlands. For this, we gathered species- specific information on elevation ranges for species belonging to several montane or lowland genera available at http://www.iucnredlist.org. We used this information to plot elevational distribution pattern for each genus and extract their respective

speciation rates to visualize how they vary through time and how different they are between lowland and montane clades.

RESULTS

Evolutionary Dynamics

When checking for convergence of BAMM runs, we obtained values of 210.66 and

418.99 for the effective sample sizes of the log-likelihood and the number of shift events present in each sample respectively. These values have been shown to be reasonable for very large datasets confirming convergence of our analyses [72]. We found strong evidence for heterogeneous diversification dynamics in amphibians.

Based on the values of posterior quasi-probability across all bootstrap replicates from post burn-in BAMM, we found support for 45 evolutionary rate shifts (mean =

48.35; median = 48) (electronic supplementary material, Fig S1). We focus our discussion on speciation rates dynamics, however since we found a high positive linear correlation between speciation and net diversification rates (Pearson's r = 0.97, p < 0.001) we consider that speciation might provide good insights on the diversification of amphibians.

Figure 1. Species richness based on the distribution of 6311 species (A); mean speciation rate (B) and topographic complexity (C) per 1° grid cell in a global scale.

Geographic patterns of Amphibian speciation

Mean amphibian speciation-rates are unevenly distributed across the world.

Speciation rates show an inverse latitudinal gradient in the New World, with faster speciation rates towards the poles. In the Old World speciation rates increase only towards northern latitudes, while regions such as Africa, Madagascar and Western

Australia are characterized by low speciation rates. A major portion of Southeast

Asia and the Neotropical Region show low to intermediate mean speciation rates

(Fig. 1).

We detected that speciation rates vary widely within regions. Such variability peaks in Mesoamerica, Patagonia and North America (Fig. 2), where there is a mixture of groups with both fast and slow speciation rates (Fig. 2). Rapidly diversifying groups are concentrated in the Neotropical, Panamanian, Nearctic and

Australian regions. In contrast, we found, lowest values of speciation rates in western Africa and most of the Palearctic region (Fig. 2).

Figure 2. Variability on speciation rates across the world

Topographic complexity as driver of speciation

At the global scale, we found faster speciation rates in HTC regions than in LTC regions (HTCmean=0.0679, LTCmean=0.0651, p-value <0.0001). Considering independent evolutionary histories, we applied the same approach across global biogeographic realms. At this regional scale, we found the same pattern of faster speciation in HTC in eight out 11 realms (Fig. 3, Table 1). In addition, speciation rates also tended to be higher in HTC in the realms were statistical difference were not significant (i.e., Afrotropical, Madagascan and Nearctic realms) (Table 1, Fig 3).

Table 1. Differences in mean speciation rates between LTC and HTC areas in global scale and within the 11 Zoogeographical regions of the world.

Region Mean Speciation Mean Speciation SD of Speciation P-value Rate in HTC Rate in LTC Rates in LTC Global 0.0679 0.0651 0.0004 <0.001 Neotropical 0.0605 0.0551 0.0002 <0.001 Afrotropical 0.0532 0.0523 0.0005 0.969 Madagascan 0.0500 0.0483 0.0008 <0.001 Australian 0.0648 0.0546 0.0005 <0.001 Nearctic 0.0724 0.0717 0.0002 0.996 Oceania 0.0605 0.0560 0.0002 <0.001 Oriental 0.0639 0.0592 0.0002 <0.001 Panamanian 0.0642 0.0604 0.0004 <0.001 Saharo-Arabian 0.0685 0.0639 0.0006 <0.001 Sinojapanese 0.0746 0.0713 0.0004 <0.001 Palearctic 0.0682 0.0651 0.0002 <0.001

Figure 3. Mean speciation rates for LTC and HTC areas in a global scale and within the different zoogeographic realms. Histograms represent the distribution of values obtained after resampling 10 000 times the number of cells in HTC from the pool of LTC cells. Dashed lines represent the mean values of speciation rate for LTC (blue) and HTC (red) regions.

Faster speciation rates are generally associated with clades that predominantly inhabit HTC areas. In the New World for example, those clades occur in several

Andean, Mesoamerican and North American mountain chains, and in a series of islands dominated by steeped topographies such as Jamaica and Dominican

Republic. In the Old World, speciation rates peak at the Himalayans and other major mountainous systems in China, Philippines and Papua New Guinea. In Australia, we found speciation rates maxima similar to those of the other regions although they were not exclusively associated to mountainous areas.

When comparing speciation rates across the phylogeny, we found faster rates in salamanders (Caudata = 0.0781±0.034; Anura = 0.053±0.016;

Gymnophiona=0.028±0.001; p = 0.0054, Df = 2). Differences are also significant among Amphibian families (p<0.0001, Df = 75) as well as among families within the orders Anura (p<0.0001, Df = 57) and Caudata (p<0.0001, Df = 8). Mean speciation rates among Gymnophiona families did not differ significantly (p = 0.077,

Df = 8). At the genus level, the fastest speciation rates occur in the Patagonian spiny frogs Alsodes (mean = 0.1934±0.006, n = 18). Other anuran genera showing high rates of speciation are the bufonid genera Rhinella and Atelopus, ranids of the genera

Rana, Odorrana, Babina and Amolops, as well as the New World direct developing frogs of the genus Brachycephalus. Salamanders of the family Plethodonthidae, which includes genera such as Bolitoglossa, Eurycea, Pseudoeurycea,

Batrachoseps, Thorius, Nototriton and Oedipina (electronic supplementary material,

Fig S2) showed the highest speciation rates at the family level (mean =

0.0982±0.0383, n = 450).

Figure 4. Comparative patterns of altitudinal distributions and speciation trajectories for contrasting montane and lowland anuran genera. A-B. The bufonid highland genus Atelopus from Northern Andes, a closely related genus (Rhaebo)

and a highly diverse hylid genus (Scinax), both distributed in lower elevations mainly the adjacent Amazon Basin; C-D. Three centrolenid genera from the Andean and Mesoamerican Region presenting different altitudinal distributions: Hyalinobatrachium with the lower elevation range and most species occurring below 1000 m.a.s.l and Nymphargus and Centrolene with mean altitudes around 2000 m.a.s.l; E-F. Three Ranid genera from the Old World with different patterns of altitudinal distribution: Pelophylax and Meristogenys with most of their representatives occurring below 500 m.a.sl and Odorrana with a peak of diversity above 1000 m.a.s.l and several species reaching 3000 m.a.s.l

As predicted, most of the rapidly diversifying clades showed predominantly montane distributions. To exemplify this trend, we compared speciation through time plots between some of these montane genera and lowland genera. To make it comparable, we contrast genera with similar richness. In all cases, speciation rates were higher in clades that are mostly montane, and the differences were constant through the evolutionary history of these groups, depicting historical differences in their speciation trajectories (Fig 4).

DISCUSSION

We found that speciation rates are generally higher in HTC regions than in LTC regions at a global scale, in concordance with the montane-pump hypothesis. In addition, our results provide evidence showing that maximum speciation rates are generally associated with clades that predominantly inhabit HTC regions. These includes several Andean ranges, Mesoamerican mountain chains, various Sierras in

North America, and a series of islands dominated by steeped topographies such as

Jamaica and Hispaniola in the Western Hemisphere. In the Old World, the

Himalayans and other major mountainous systems in China, Philippines and Papua

New Guinea exhibit similar dynamics. This suggests that highly complex reliefs around the globe, independently of their latitude, have an important role as engines of speciation. It also suggests that these dynamics are specific to the geographical setting of montane regions generally, and not specific geographic areas or traits possessed by specific lineages that confer increased diversification.

A growing body of literature provide evidence supporting the role of mountains as species cradles for numerous taxa. A few examples are the

Australasian Sky Islands [73,74], the Hendguan Mountains [75,76], and the

Anatolian Mountains in the eastern hemisphere [75,76]. In the New World, evidence of such tendency has been documented in regions such as the Andes [32,75–80] and the North American Sky Islands [81,82]. Such studies have often focused on few clades and specific geographic regions that exhibit high diversity. Our study is the first to our knowledge to contrast speciation rates between HTC and LTC regions at global scale. We provide evidence of the general importance of mountain ranges as speciation pumps. Importantly, our results suggest that mountains affect speciation rates independently of region, diversity, or specific lineage in amphibians.

Across the phylogeny, we found that salamanders have the highest mean speciation rates among Amphibians, followed by anurans and caecilians.

Salamanders are abundant in North temperate regions where seven of the 10 families in the order are distributed (www.amphibiaweb.org). Within the order, we found the highest speciation rates in Plethodontidae, a family whose representatives reach the tropics of the western hemisphere [83]. The major radiation of this family is the

Neotropical tribe Bolitoglossini, which occurs throughout complex topographies within Mesoamerica, and contains nearly 300 species, that accounts for over 65% of the species in the family and 43% of the diversity in the order [84].

Among anurans, speciation rates also peak in montane-associate clades.

Fastest speciation rates occur in the genera Alsodes [85] and Eupsophus from the

Patagonian Andes (despite the low diversity of this region) and bufonids such as the

Harlequin toads of the genus Atelopus which have mainly radiated in the highlands of the northern Andes [17]. In mountain ranges of south eastern and eastern Asia, ranids of the genus Odorrana [86–88] also rank among the anuran clades with the highest means of speciation rate. As examples of these evolutionary contrasts, we compared altitudinal distributions and speciation trajectories within these genera with those of closely relatives or similarly diverse clades occurring in adjacent lowlands. In all cases, it is evident that montane clades have higher speciation rates and these differences have been constant through time (Fig 4).

Rates of speciation can be influenced by both intrinsic biological attributes and extrinsic environmental factors [13,89]. Some of the latter factors may be magnified in topographically complex landscapes. For example, characteristic rugged reliefs in mountainous regions are more likely to impose physical barriers, fragmenting species ranges and promoting geographical isolation [14,16,17,21,90].

Furthermore, altitudinal gradients in these complex landscapes, provide heterogeneous environmental conditions that could promote ecological specialization and niche divergence based on trait differences [91,92]. Both scenarios restrict gene flow, augmenting founder effects and driving speciation whether in allopatric or parapatric conditions [43,85]. For groups with low dispersal rates such as amphibians, these conditions appear to have a major impact on the processes of incipient population differentiation, and ultimately, speciation [43,92].

Our results also provide insights on the latitudinal and zoogeographic patterns of amphibian speciation. We found high latitudinal variance in amphibian speciation rates. Such variability is strikingly decoupled from the well-documented latitudinal diversity gradient (LDG) present in amphibians and many other groups [94] . For example, mean speciation rates for all amphibians are higher in temperate zones of both the New and the Old world, while lower mean rates were concentrated in more speciose regions such as Africa, Madagascar, and Western Australia. Other hotspots of diversity, including a major portion of Southeast Asia and the Amazon Basin [60],

showed intermediate mean speciation rates. We suggest that the great variability of speciation rates in speciose areas with heterogeneous species compositions may obscure the latitudinal patterns of speciation. However, future studies should explore the relation between latitude and speciation rates more deeply, in order to understand the main evolutionary forces shaping the LDG in amphibians.

CONCLUSIONS

Our findings bolster the general importance of mountains as engines of speciation at different geographical scales and independently of latitude. However, due to their remote conditions, many mountain ranges remain unexplored and their real contribution to the origin and maintenance of global biodiversity is still underestimated. For these reasons and the risk these regions face during ongoing global changes [71], mountains around the world must be considered conservation priorities in local and regional agendas. The evidence presented here highlights the role of such areas in the evolutionary history of modern patterns of diversity; further efforts must be oriented to increase the knowledge of these areas to inform future decisions for the conservation of their particular biotas.

DATA ACCESSIBILITY

The phylogeny used here is a random tree extracted from the topologies made available by Jetz & Pyron [53] at https://vertlife.org/files_20170703/#amphibians.

TCI was calculated using an elevation layer available at www.worldclim.org at 30 secs resolution. R code and associated files are available as electronic supplementary material

AUTHORS' CONTRIBUTIONS

AGR conceived of the study, discussed design, conducted analyses and drafted the manuscript. PAM conceived of the study and participated in data analyses. BFO participated in data analyses. RAP participated in data analyses. GCC conceived of the study, discussed design of analyses and drafted the manuscript. All authors improved the draft of the manuscript and gave final approval for publication.

COMPETING INTERESTS

We have no competing interests.

FUNDING

AGR was supported by Coordenação de Aperfeiçoamento de Pessoal de Nível

Superior, Brazil. GCC thanks CNPq produtivity grant 302297/2015-4. RAP was

supported by US NSF grant DEB-1441719 and DEB-1655737. BFO thanks

University of Florida for providing generous support.

ACKNOWLEDGEMENTS

We thank Marcelo Araya and Juan Pablo Zurano for valuable discussion and suggestions during the development of this study. To our colleagues Jodi Rowley,

Luis Coloma, Santiago Ron, Alexander Haas, Andreas Nöllert and Brian Gratwicke that kindly provide permission to use the photos included in figure 3, and Paula

Acosta who helped with the edition and improving of one of those images.

REFERENCES

1. Körner C. 2002 Mountain biodiversity, its causes and function. In Mountain biodiversity: a global assessment (eds C Körner, EM Spehn), pp. 3–20. London, UK: Parthenon Publishing. (doi:10.2307/25094582) 2. Ruggiero A, Hawkins BA. 2008 Why do mountains support so many species of birds? Ecography. 31, 306–315. (doi:10.1111/j.0906-7590.2008.05333.x) 3. Körner C. 2000 Why are there global gradients in species richness? Mountains might hold the answer. Trends Ecol. Evol. 15, 513–514. 4. Spehn EM, Messerli B, Körner C. 2002 A global assessment of mountain biodiversity: synthesis. In Mountain biodiversity: a global assessment., London, UK: The Parthenon Publishing Group. 5. Lomolino M V. 2001 Elevation gradients of species-density: historical and prospective views. Glob. Ecol. Biogeogr. 10, 3–13. (doi:10.1046/j.1466- 822x.2001.00229.x) 6. Simpson GG. 1964 Species density of North American recent mammals. Syst.

Zool. 13, 57–73. (doi:10.2307/2411825) 7. De Klerk HM, Crowe TM, Fjelds?? J, Burgess ND. 2002 Biogeographical patterns of endemic terrestrial Afrotropical birds. Divers. Distrib. 8, 147–162. (doi:10.1046/j.1472-4642.2002.00142.x) 8. Barthlott W, Mutke J, Rafiqpoor D, Kier G, Kreft H. 2005 Global centers of vascular plant diversity. Nov. Acta Leopoldina 92, 61–83. 9. Antonelli A. 2015 Biodiversity: multiple origins of mountain life. Nature 524, 300–301. (doi:10.1038/nature14645) 10. Myers N, Mittermeier RA, Mittermeier CG, Fonseca GAB da, Kent E. 2000 Biodiversity hotspots for conservation priorities. Nature. 403, 853–858. (doi:10.1038/35002501) 11. Orme CDL et al. 2005 Global hotspots of species richness are not congruent with endemism or threat. Nature 436, 1016–1019. (doi:10.1038/nature03850) 12. Graham CH et al. 2014 The origin and maintenance of montane diversity: integrating evolutionary and ecological processes. Ecography. 37, 711–719. (doi:10.1111/ecog.00578) 13. Funk WC, Murphy M., Hoke KL, Muths SM, Amburgey, Lemmon EM, Lemmon A. 2016 Elevational speciation in action? Restricted gene flow associated with adaptive divergence across an altitudinal gradient. J. Evol. Biol. 29, 241–252. 14. Mayr E. 1963 Animal species and evolution. Eugen. Rev. 55, 226–228. (doi:10.1016/0169-5347(94)90187-2) 15. Guarnizo CE, Amézquita A, Bermingham E. 2009 The relative roles of vicariance versus elevational gradients in the genetic differentiation of the high Andean tree frog, Dendropsophus labialis. Mol. Phylogenet. Evol. 50, 84–92. (doi:10.1016/j.ympev.2008.10.005) 16. Rull V. 2005 Biotic diversification in the Guayana Highlands: a proposal. J. Biogeogr. 32, 921–927. (doi:10.1111/j.1365-2699.2005.01252.x) 17. Orr MR, Smith TB. 1998 Ecology and speciation. Trends Ecol. Evol. 13, 502– 506. (doi:10.1016/S0169-5347(98)01511-0) 18. Kozak KH, Wiens JJ. 2006 Does niche conservatism promote speciation? A case study in North American salamanders. Evolution. 60, 2604–21. 19. Moritz C, Patton JL, Schneider CJ, Smith TB. 2000 Diversification of

rainforest faunas: an integrated molecular approach. Annu. Rev. Ecol. Syst. 31, 533–563. (doi:10.1146/annurev.ecolsys.31.1.533) 20. Wiens JJ, Graham CH. 2005 Niche Conservatism: integrating evolution, ecology, and conservation biology. Annu. Rev. Ecol. Evol. Syst. 36, 519–539. (doi:10.1146/annurev.ecolsys.36.102803.095431) 21. Merckx VSFT et al. 2015 Evolution of endemism on a young tropical mountain. Nature 524, 347–350. (doi:10.1038/nature14949) 22. Rundle HD, Nosil P. 2005 Ecological speciation. Ecol. Lett. 8, 336–352. (doi:10.1111/j.1461-0248.2004.00715.x) 23. Graham CH, Ron SR, Santos JC, Schneider CJ, Moritz C. 2004 Integrating phylogenetics and environmental niche models to explore speciation mechanisms in dendrobatid frogs. Evolution. 58, 1781–93. 24. Caro LM, Caycedo-Rosales PC, Bowie RCK, Slabbekoorn H, Cadena CD. 2013 Ecological speciation along an elevational gradient in a tropical passerine bird? J. Evol. Biol. 26, 357–374. (doi:10.1111/jeb.12055) 25. Chapman MA, Hiscock SJ, Filatov DA. 2013 Genomic divergence during speciation driven by adaptation to altitude. Mol. Biol. Evol. 30, 2553–2567. (doi:10.1093/molbev/mst168) 26. Knox EB, Palmer JD. 1995 Chloroplast DNA variation and the recent radiation of the giant senecios (Asteraceae) on the tall mountains of eastern Africa. Proc. Natl. Acad. Sci. U. S. A. 92, 10349–10353. (doi:10.1073/pnas.92.22.10349) 27. Peng Z, Ho SYW, Zhang Y, He S. 2006 Uplift of the Tibetan plateau: evidence from divergence times of glyptosternoid catfishes. Mol. Phylogenet. Evol. 39, 568–572. (doi:10.1016/j.ympev.2005.10.016) 28. Zhou WW, Wen Y, Fu J, Xu YB, Jin JQ, Ding L, Min MS, Che J, Zhang YP. 2012 Speciation in the Rana chensinensis species complex and its relationship to the uplift of the Qinghai-Tibetan Plateau. Mol. Ecol. 21, 960–973. (doi:10.1111/j.1365-294X.2011.05411.x) 29. Liu JQ, Wang YJ, Wang AL, Hideaki O, Abbott RJ. 2006 Radiation and diversification within the Ligularia-Cremanthodium- Parasenecio complex (Asteraceae) triggered by uplift of the Qinghai-Tibetan Plateau. Mol. Phylogenet. Evol. 38, 31–49. (doi:10.1016/j.ympev.2005.09.010) 30. Zhang ML, Fritsch PW. 2010 Evolutionary response of Caragana (Fabaceae) to Qinghai-Tibetan Plateau uplift and Asian interior aridification. Plant Syst.

Evol. 288, 191–199. (doi:10.1007/s00606-010-0324-z) 31. Chaves JA, Weir JT, Smith TB. 2011 Diversification in Adelomyia hummingbirds follows Andean uplift. Mol. Ecol. 20, 4564–4576. (doi:10.1111/j.1365-294X.2011.05304.x) 32. De-Silva DL, Elias M, Willmott K, Mallet J, Day JJ. 2016 Diversification of clearwing butterflies with the rise of the Andes. J. Biogeogr. 43, 44–58. (doi:10.1111/jbi.12611) 33. Antonelli A, Nylander JAA, Persson C, Sanmartín I. 2009 Tracing the impact of the Andean uplift on Neotropical plant evolution. Proc. Natl. Acad. Sci. USA 106, 9749–9754. (doi:10.1073/pnas.0811421106) 34. Sanín MJ et al. 2016 The Neogene rise of the tropical Andes facilitated diversification of wax palms (Ceroxylon: Arecaceae) through geographical colonization and climatic niche separation. Bot. J. Linn. Soc. 182, 303–317. (doi:10.1111/boj.12419) 35. Hughes C, Eastwood R. 2006 Island radiation on a continental scale: exceptional rates of plant diversification after uplift of the Andes. Proc. Natl. Acad. Sci. USA 103, 10334–10339. (doi:10.1073/pnas.0601928103) 36. Lagomarsino LP, Condamine FL, Antonelli A, Mulch A, Davis CC. 2016 The abiotic and biotic drivers of rapid diversification in Andean bellflowers (Campanulaceae). New Phytol. 210, 1430–1442. (doi:10.1111/nph.13920) 37. Hutter CR, Guayasamin JM, Wiens JJ. 2013 Explaining Andean megadiversity: the evolutionary and ecological causes of glassfrog elevational richness patterns. Ecol. Lett. 16, 1135–1144. (doi:10.1111/ele.12148) 38. Smith SA, De Oca ANM, Reeder TW, Wiens JJ. 2007 A phylogenetic perspective on elevational species richness patterns in middle American treefrogs: why so few species in lowland tropical rainforests? Evolution. 61, 1188–1207. (doi:10.1111/j.1558-5646.2007.00085.x) 39. Fjeldsa J, Rahbek C. 2006 Diversification of tanagers, a species rich bird group, from lowlands to montane regions of South America. Integr. Comp. Biol. 46, 72–81. (doi:10.1093/icb/icj009) 40. Hall JPW. 2005 Montane speciation patterns in Ithomiola butterflies (Lepidoptera: Riodinidae): are they consistently moving up in the world? Proc. R. Soc. B 272, 2457–2466. (doi:10.1098/rspb.2005.3254) 41. Duellman WE. 1999 Patterns of distribution of amphibians: a global

perspective. Baltimore, Maryland.: John Hopkins University Press. 42. Pyron RA, Wiens JJ. 2011 A large-scale phylogeny of Amphibia including over 2800 species, and a revised classification of extant frogs, salamanders, and caecilians. Mol. Phylogenet. Evol. 61, 543–83. (doi:10.1016/j.ympev.2011.06.012) 43. Pyron RA. 2014 Biogeographic analysis reveals ancient continental vicariance and recent oceanic dispersal in amphibians. Syst. Biol. 63, 779–797. (doi:10.1093/sysbio/syu042) 44. Blaustein AR, Wake DB, Sousa WP. 1994 Amphibian declines: judging stability, persistence, and susceptibility of populations to local and global extinctions. Conserv. Biol. 8, 60–71. (doi:10.1046/j.1523- 1739.1994.08010060.x) 45. Beebee TJC. 2005 Conservation genetics of amphibians. Heredity. 95, 423–7. (doi:10.1038/sj.hdy.6800736) 46. Balinsky JB. 1981 Adaptation of nitrogen metabolism to hyperosmotic environment in Amphibia. J. Exp. Zool. 215, 335–350. (doi:10.1002/jez.1402150311) 47. Berven KA, Grudzien TA. 1990 Dispersal in the wood frog (Rana sylvatica): implications for genetic population structure. Evolution. 44, 2047–2056. (doi:10.2307/2409614) 48. Lüddecke H, Sánchez OR. 2002 Are tropical highland frog calls cold-adapted? The case of the Andean frog Hyla labialis. Biotropica 34, 281–288. (doi:10.1111/j.1744-7429.2002.tb00539.x) 49. Bonin A, Taberlet P, Miaud C, Pompanon F. 2006 Explorative genome scan to detect candidate loci for adaptation along a gradient of altitude in the common frog (Rana temporaria). Mol. Biol. Evol. 23, 773–783. (doi:10.1093/molbev/msj087) 50. Jetz W, Thomas GH, Joy JB, Hartmann K, Mooers AO. 2012 The global diversity of birds in space and time. Nature 491, 444–448. (doi:10.1038/nature11631) 51. Morlon H, Parsons TL, Plotkin JB. 2011 Reconciling molecular phylogenies with the fossil record. Proc. Natl. Acad. Sci. USA 108, 16327–32. (doi:10.1073/pnas.1102543108) 52. Thomas GH, Hartmann K, Jetz W, Joy JB, Mimoto A, Mooers AO. 2013

PASTIS: an R package to facilitate phylogenetic assembly with soft taxonomic inferences. Methods Ecol. Evol. 4, 1011–1017. 53. Jetz W, Pyron RA. 2018 The interplay of past diversification and evolutionary isolation with present imperilment across the amphibian tree of life. Nat. Ecol. Evol. In Press. 54. Rabosky DL, Santini F, Eastman J, Smith S a, Sidlauskas B, Chang J, Alfaro ME. 2013 Rates of speciation and morphological evolution are correlated across the largest vertebrate radiation. Nat. Commun. 4, 1958. (doi:10.1038/ncomms2958) 55. Rabosky DL. 2014 Automatic detection of key innovations, rate shifts, and diversity-dependence on phylogenetic trees. PLoS One 9, e89543. (doi:10.1371/journal.pone.0089543) 56. McGuire JA, Witt CC, Remsen J V., Corl A, Rabosky DL, Altshuler DL, Dudley R. 2014 Molecular phylogenetics and the diversification of hummingbirds. Curr. Biol. 24, 910–916. (doi:10.1016/j.cub.2014.03.016) 57. Moore BR, Höhna S, May MR, Rannala B, Huelsenbeck JP. 2016 Critically evaluating the theory and performance of Bayesian analysis of macroevolutionary mixtures. Proc. Natl. Acad. Sci. 113, 9569–9574. (doi:10.1073/pnas.1518659113) 58. Rabosky DL, Mitchell JS, Chang J. 2017 Is BAMM flawed? Theoretical and practical concerns in the analysis of multi-rate diversification models. Syst. Biol. 66, 477–498. (doi:10.1093/sysbio/syx037) 59. Morlon H. 2014 Phylogenetic approaches for studying diversification. Ecol. Lett. 17, 508–525. (doi:10.1111/ele.12251) 60. Rabosky DL, Title PO, Huang H. 2015 Minimal effects of latitude on present- day speciation rates in New World birds. Proc. R. Soc. B 282, 20142889. (doi:10.1098/rspb.2014.2889) 61. Vilela B, Villalobos F. 2015 LetsR: A new R package for data handling and analysis in macroecology. Methods Ecol. Evol. 6, 1229–1234. (doi:10.1111/2041-210X.12401) 62. Fitzpatrick BM, Turelli M. 2006 The geography of mammalian speciation: mixed signals from phylogenies and range maps. Evolution. 60, 601–615. (doi:10.1111/j.0014-3820.2006.tb01140.x) 63. Ribera I, Castro A, Díaz JA, Garrido J, Izquierdo A, Jäch MA, Valladares LF.

2011 The geography of speciation in narrow-range endemics of the ‘Haenydra’ lineage (Coleoptera, Hydraenidae, Hydraena). J. Biogeogr. 38, 502–516. (doi:10.1111/j.1365-2699.2010.02417.x) 64. Cardillo M. 2015 Geographic range shifts do not erase the historic signal of speciation in mammals. Am. Nat. 185, 343–353. (doi:10.1086/679663) 65. Martínez-Solano I, Gonçalves HA, Arntzen JW, García-París M. 2004 Phylogenetic relationships and biogeography of midwife toads (Discoglossidae: Alytes). J. Biogeogr. 31, 603–618. (doi:10.1046/j.1365- 2699.2003.01033.x) 66. Munguía M, Rahbek C, Rangel TF, Diniz-Filho JAF, Araújo MB. 2012 Equilibrium of global amphibian species distributions with climate. PLoS One 7, e34420. (doi:10.1371/journal.pone.0034420) 67. Buckley LB, Jetz W. 2008 Linking global turnover of species and environments. Proc. Natl. Acad. Sci. USA. 105, 17836–17841. (doi:10.1073/pnas.0803524105) 68. Grenyer R et al. 2006 Global distribution and conservation of rare and threatened vertebrates. Nature 444, 93–96. (doi:10.1038/nature05237) 69. McCarroll D, Nesje A. 1996 Rock surface roughness as an indicator of degree of rock surface weathering. Earth Surf. Process. Landforms 21, 963–977. (doi:10.1002/(SICI)1096-9837(199610)21:10<963::AID-ESP643>3.0.CO;2- J) 70. Körner C, Jetz W, Paulsen J, Payne D, Rudmann-Maurer K, M. Spehn E. 2017 A global inventory of mountains for bio-geographical applications. Alp. Bot. 127, 1–15. (doi:10.1007/s00035-016-0182-6) 71. Holt BG, Al E. 2013 An updated of Wallace’s zoogeographic regions of the World. Science. 339, 74–79. (doi:10.1126/science.1228282) 72. López-Pujol J, Zhang F-M, Sun H-Q, Ying T-S, Ge S. 2011 Mountains of Southern China as ‘Plant Museums’ and ‘Plant Cradles’: evolutionary and conservation insights. Mt. Res. Dev. 31, 261–269. (doi:10.1659/MRD- JOURNAL-D-11-00058.1) 73. Ansell SW et al. 2017 The importance of Anatolian mountains as the cradle of global diversity in Arabis alpina, a key arctic-alpine species. Ann. Bot. 108, 241–252. (doi:10.1093/aob/mcrl34) 74. Atalay I. 2006 The effects of mountainous areas on biodiversity: a case study

from the northern Anatolian Mountains and the Taurus Mountains. Proc. 8th Int. Symp. High Mt. Remote Sens. Cartogr. 41, 17–26. 75. Kattan GH, Franco P, Rojas V, Morales G. 2004 Biological diversification in a complex region: a spatial analysis of faunistic diversity and biogeography of the Andes of Colombia. J. Biogeogr. 31, 1829–1839. (doi:10.1111/j.1365- 2699.2004.01109.x) 76. Santos JC, Coloma LA, Summers K, Caldwell JP, Ree R, Cannatella DC. 2009 Amazonian amphibian diversity is primarily derived from late Miocene Andean lineages. PLoS Biol. 7, 0448–0461. (doi:10.1371/journal.pbio.1000056) 77. Mendoza ÁM, Ospina OE, Cárdenas-Henao H, García-R JC. 2015 A likelihood inference of historical biogeography in the world’s most diverse terrestrial vertebrate genus: Diversification of direct-developing frogs (Craugastoridae: Pristimantis) across the Neotropics. Mol. Phylogenet. Evol. 85, 50–58. (doi:10.1016/j.ympev.2015.02.001) 78. Sedano RE, Burns KJ. 2010 Are the Northern Andes a species pump for Neotropical birds? Phylogenetics and biogeography of a clade of Neotropical tanagers (Aves: Thraupini). J. Biogeogr. 37, 325–343. (doi:10.1111/j.1365- 2699.2009.02200.x) 79. Fjeldsa J, Bowie RCK, Rahbek C. 2012 The role of mountain ranges in the diversification of birds. Annu. Rev. Ecol. Evol. Syst. 43, 249–265. (doi:10.1146/annurev-ecolsys-102710-145113) 80. McGuire JA, Witt CC, Remsen J V., Corl A, Rabosky DL, Altshuler DL, Dudley R. 2014 Molecular phylogenetics and the diversification of hummingbirds. Curr. Biol. 24, 910–916. (doi:10.1016/j.cub.2014.03.016) 81. Shepard DB, Burbrink FT. 2008 Lineage diversification and historical demography of a sky island salamander, Plethodon ouachitae, from the Interior Highlands. Mol. Ecol. 17, 5315–35. (doi:10.1111/j.1365- 294X.2008.03998.x) 82. Knowles LL. 2000 Tests of pleistocene speciation in montane grasshoppers (genus Melanoplus) from the sky islands of western north America. Evolution. 54, 1337–1348. (doi:10.1111/j.0014-3820.2000.tb00566.x) 83. Bell T. 1843 Reptiles. In The Zoology of the Voyage of the H.M.S. Beagle, Under the Command of Captain Fitzroy, R.N., During the Years 1832 to 1836. (ed C ed. Darwin), p. 1–51. London, UK: Smith, Elder and Co.

84. Noonan BP, Gaucher P. 2005 Phylogeography and demography of Guianan harlequin toads (Atelopus): diversification within a refuge. Mol. Ecol. 14, 3017–31. (doi:10.1111/j.1365-294X.2005.02624.x) 85. Frost D. 2017 Amphibian Species of the World: an Online Reference. Am. Museum Nat. Hist. Version 6., 6–8. (doi:http://research.amnh.org/herpetology/amphibia/index.php.) 86. Koscinski D, Yates AG, Handford P, Lougheed SC. 2009 Effects of landscape and history on diversification of a montane, stream-breeding amphibian. J. Biogeogr. 36, 255–265. (doi:10.1111/j.1365-2699.2008.02009.x) 87. Wollenberg KC, Vieites DR, van der Meijden A, Glaw F, Cannatella DC, Vences M. 2008 Patterns of endemism and species richness in Malagasy cophyline frogs support a key role of mountainous areas for speciation. Evolution. 62, 1890–907. (doi:10.1111/j.1558-5646.2008.00420.x) 88. Smith BT et al. 2014 The drivers of tropical speciation. Nature 515, 1–8. (doi:10.1038/nature13687) 89. Kozak KH, Wiens JJ. 2007 Climatic zonation drives latitudinal variation in speciation mechanisms. Proc. R. Soc. B 274, 2995–3003. (doi:10.1098/rspb.2007.1106) 90. Rundle HD, Nosil P. 2005 Ecological speciation. Ecol. Lett. 8, 336–352. (doi:10.1111/j.1461-0248.2004.00715.x) 91. Guarnizo CE, Cannatella DC. 2013 Genetic divergence within frog species is greater in topographically more complex regions. J. Zool. Syst. Evol. Res. 51, 333–340. (doi:10.1111/jzs.12027) 92. Rodríguez A, Börner M, Pabijan M, Gehara M, Haddad CFB, Vences M. 2015 Genetic divergence in tropical anurans: deeper phylogeographic structure in forest specialists and in topographically complex regions. Evol. Ecol. 29, 765– 785. (doi:10.1007/s10682-015-9774-7) 93. Wiens JJ. 2007 Global patterns of diversification and species richness in amphibians. Am. Nat. 170, 86–106. (doi:10.1086/519396) 94. Weir JT, Schluter D. 2007 The latitudinal gradient in recent speciation and extinction rates of birds and mammals. Science. 315, 1574–1576. (doi:10.1126/science.1135590)

-Electronic Supplementary Material-

Faster amphibian speciation in mountains support their role as biodiversity pumps

Supplementary material, text S1.

BAMM analysis

To perform the BAMM analysis we first used the 'setBAMMpriors' function to generate a prior block in accordance with the scale of our tree. We ran 60,000,000 generations of reversible-jump MCMC sampling on our phylogeny with samples drawn from the posterior every 5,000 generations. Considering the massive number of tips in our tree we used a poisson prior of 0.033, which correspond to 30 expected shifts of diversification. The selection of low values for this parameter allows the algorithm to explore a broader range of shift configurations, facilitating convergence. We used the BAMMtools R package -version 2.0-[1] in R, to analyze consistency among BAMM outputs. We visually checked for convergence of the

MCMC algorithm by inspecting the relation between likelihood scores and sampled generations. Then, we reviewed adequate mixing of chains, examined for effective sample sizes above at least 10% of our sampled generations, discarded the first 20% of samples as burn-in and used the 80% left to estimate evolutionary rate values for each tip and resulting shifts of diversification across the tree. We used the R package

‘coda’ [2] to diagnose convergence, considering as satisfactory values of effective

sample size of the log-likelihood and the number of shift events present in each sample above 200. BAMM analyses provide speciation and extinction rates per species as direct output and is possible to estimate net diversification rates by subtracting extinction rates from speciation rates [3]. However, we focused our analyses on speciation rates, because they can be estimated with much more confidence than extinction rates, for which confidence intervals tend to be large, even when all assumptions of the inference model are satisfied [4]

Supplementary References

1. Rabosky DL, Grundler M, Anderson C, Title P, Shi JJ, Brown JW, Huang H,

Larson JG. 2014 BAMMtools: An R package for the analysis of evolutionary

dynamics on phylogenetic trees. Methods Ecol. Evol. 5, 701–707.

(doi:10.1111/2041-210X.12199)

2. Plummer M, Best N, Cowles K, Vines K. 2006 CODA: Convergence

Diagnostics and Output Analysis for MCMC. R News 6, 7–10.

3. Morlon H. 2014 Phylogenetic approaches for studying diversification. Ecol.

Lett. 17, 508–525. (doi:10.1111/ele.12251)

4. Rabosky DL, Title PO, Huang H. 2015 Minimal effects of latitude on present-

day speciation rates in New World birds. Proc. R. Soc. B 282, 20142889.

(doi:10.1098/rspb.2014.2889)

SUPPLEMENTARY FIGURES

Figure S1. Tree showing best configuration of speciation rate shifts among groups in extant Amphibians. Colours in branches represent speciation rate dynamics, blue represent slower while red represent faster rates. Red dots highlights the regions of the phylogeny where major shifts were detected .

Figure S2. Clade-specific speciation trajectories for some of the groups with the highest values of speciation rates detected. A. Patagonian Spiny frogs (Alsodes, 18 species); B. South American Harlequin toads (Atelopus,97 spp); C. Neotropical true toads (Rhinella, 35 spp); D, E and F. Bolitoglossine salamanders (Bolitoglossa, 137 spp; Oedipina, 36 spp and Pseudoeurycea, 50 spp).

CAPÍTULO II**

Idiosyncratic responses to drivers of genetic differentiation in the complex landscapes of Isthmian Central America

Adrián García-Rodríguez1,2, Carlos E. Guarnizo3, Andrew J. Crawford3,4, Adrian A.

Garda1 & Gabriel C. Costa5

1Departamento de Ecologia, Universidade Federal do Rio Grande do Norte, Natal,

59078-900 RN, Brasil.

2Escuela de Biología, Universidad de Costa Rica, San Pedro, 11501-2060 San José,

Costa Rica.

3Departamento de Ciencias Biológicas, Universidad de los Andes, A.A. 4976,

Bogotá, Colombia.

3Smithsonian Tropical Research Institute, Apartado, 0843–03092, Panamá,

Republic of Panama.

5Department of Biology, Auburn University at Montgomery, Montgomery AL

36124.

Corresponding author: Adrián García-Rodríguez [email protected]

ABSTRACT The isthmian portion of Central America (ICA) is one of the most biodiverse regions in the world, hosting the highest number of species per unit area, for many taxa. Due to the geological history of this region, the assembly of this biota is relatively recent and results from dispersal events and in situ diversification processes across a complex landscape. Here, we combined information on mitochondrial DNA sequence variation with climatic and physical environmental features to understand abiotic forces that might have promoted diversification. To this end, we evaluated the role of isolation by distance, topography, habitat suitability, and environment in shaping patterns of genetic differentiation in eleven amphibian species with disparate life histories co-distributed in the region. In seven of the species studied, we found that at least one the factors tested, significantly explains genetic divergence patterns. Instead of finding a major force responsible for the intraspecific genetic divergence in ICA, our results reveal idiosyncratic responses of each species, suggesting that intrinsic characteristics of each species play an important role in determining responses to different drivers of isolation. We show that confluence of several determinants of isolation with a heterogeneous biota having different life histories, geographic origins, and arrival times to ICA maximizes the chances of genetic differentiation. We conclude that evolutionary dynamics of ICA’s biota is far more complex than simply vicariance between Caribbean and Pacific clades as a main form of speciation in the region. Drivers of diversification likely act even in short distances in complex landscapes, contributing to high levels of endemism as is the case LCA. More research is needed, not only to understand the causal relation between environment and genetic differentiation, but also to better document a diversity that is still remains underestimated.

Keywords: Gene flow, Isolation by distance, Isolation by resistance, Isolation by environment, Generalized dissimilarity Modeling, Multiple matrix regression with randomization

INTRODUCTION

Isthmian Central America (ICA), centered at Costa Rica and Panama, is one of the most diverse regions in the globe (Myers et al. 2000). Inserted in the Mesoamerican hotspot of biodiversity, this region roughly covers 0.1% of earth's land surface, nonetheless harbors an immense number of species; reaching estimates for some taxonomic groups of 4-10% of the global biodiversity residing in Costa Rica alone and Panama may be more diverse (Bagley and Johnson, 2014). Some examples of highly diverse taxa in the region includes birds (>1,000 spp), amphibians (~300 spp), reptiles (~500 spp), insects (>300,000 spp) and vascular plants (> 20,000 spp)

(Anger and Dean, 2010; Garrigues and Dean, 2014; Frost, 2017).

ICA has a relatively recent geological origin, nevertheless, the precise timing of the formation of the region´s major landscape features is still under debate

(Montes et al., 2012; O’Dea et al., 2016). Historically, the most accepted hypothesis was that the Panama Isthmus closed relatively recently, around 3-4 million years ago

(Ma) (Coates et al., 1992). Multiple independent sources of evidence supported this hypothesis, including divergence times in marine organisms separated by the

Isthmus, and the fossil record found at the Panama Canal area (Keigwin, 1978;

O’Dea et al., 2016). However, recent studies using sources of evidence such as petrographic, geochronological, and termo-chronological data suggests that the

Panama Isthmus closed much earlier, around 15 Ma (Montes et al., 2012). This

earlier date is also supported by recent molecular-based studies showing early dispersal between North and South America around 10 Ma (Pinto-Sánchez et al.,

2012; Bacon et al., 2015).

Independently of the debate regarding the precise date of the Isthmus closing, this geological formation precipitated one of the greatest biogeographic events of the Cenozoic, the Great American Biotic Interchange (GABI), a bidirectional dispersal of terrestrial mammals between the previously isolated North and South

American landmasses, around 2 Ma (Marshall, 1988; Webb, 2006). Evidence available for other groups such as frogs, birds and invertebrates (Webb, 2006; Weir et al., 2009; Pinto-Sánchez et al., 2012; Wilson et al., 2014) also highlights the important influence of the rising of the isthmus in the conformation of the regional biota. In consequence, biodiversity in ICA is largely constituted by northern and southern lineages that arrived after the completion of the land bridge (Janzen, 1991;

Savage, 2002). Surprisingly, the Isthmian fauna also has a striking high number of endemics (e.g. Rodriguez-Herrera et al. 2004; Kluge & Kessler 2006; Savage &

Bolaños 2009; Bogarín et al. 2013; Garrigues & Dean 2014), including lineages of considerably old age (Wang et al., 2008) . This high prevalence of endemism highlights the importance of in situ processes of diversification contributing to this region´s biota.

The complex tectonic and geological history of LCA resulted in a steep topography where a variety of habitats and climatic regimes converges in a small surface (Weyl, 1980; Gabb et al., 2007). For example, in less than 1000 km along the Pacific coast of Costa Rica, mean annual precipitation varies from ~1800 mm in some areas of the northern dry forest of Guanacaste to nearly 5000 mm in the very humid rainforests of Peninsula de Osa (Savage, 1966; Coen, 1991). The physiography of the region is characterized by NW-trending volcanic cordilleras that can reach altitudes over 3000 m in several peaks (Luteyn, 1999). Such cordilleras are bisected by valleys which result in a mosaic of sky-islands along with steep altitudinal gradients, microclimates, and diverse vegetation zones (Marshall, 2007;

Bagley and Johnson, 2014). In addition, climatic and sea level fluctuations during the Quaternary (Horn, 1990; Islebe et al., 1995, 1996) likely promoted species range contractions and expansions throughout this complex landscape, imposing or removing barriers to gene flow and playing a central role in diversification processes

(Hewitt, 2004). Given all these conditions, allopatric speciation (Mayr, 1942) through vicariant events promoted by physical barriers and parapatric speciation

(Doebeli and Dieckmann, 2003) across environmental gradients may have occurred in the region, although strong evidence of the latter is still scarce around the globe.

Considering its recent geologic history and high levels of richness and endemism, ICA represent an ideal natural laboratory to study the role of landscape

factors in driving early phases of genetic differentiation. One way to understand the role that such factors have had on the in-situ diversification of the ICA is to study the relationship between geographic or environmental variables and intraspecific genetic divergence. In the simplest of these relationships, called isolation by distance (IBD), genetic differentiation is expected to increase with geographic distance due to restricted gene flow by individuals (Wright, 1943; Slatkin, 1993).

However, in highly heterogeneous regions, dispersal across the landscape is more restricted due to the effect of barriers imposed by complex topographies and their associated climatic gradients (Manel and Holderegger, 2013). Consequently, genetic isolation is expected to be more associated with landscape heterogeneity than with geographic distances alone. Isolation by resistance (IBR) accounts for the reduced dispersal among populations caused by the relative unsuitability or ‘friction’ presented by heterogeneous landscape components located in between two populations (McRae, 2006). In contrast, Isolation by Environment (IBE) has been recently proposed to describe a pattern in which genetic isolation between two populations increases with environmental differences at the respective localities, independently of the resistance imposed by the landscape found between the two populations (Wang and Bradburd, 2014).

Discerning between IBD, IBR, and IBE can help discriminate among different forces promoting genetic divergence within species and its relationship with the

prevalent endemism of ICA. An ideal study system with which to test this idea are amphibians: their low dispersal capabilities (Blaustein et al., 1994) prevent them from moving across landscapes as much as other vertebrates (Beebee 2005), promoting the accumulation of genetic differences among populations (Funk et al.,

2005; Wollenberg et al., 2011). Because of their permeable skin and poikliothermic physiology, amphibians are also thought to be relatively sensitive to environmental variation (Navas & Otani 2007; Cruz-Piedrahita et al. 2018). Thus, amphibians should show a marked imprint of the potential effects of historical, ecological and geographic factors in driving genetic divergence among populations.

In this study, we explore how ICA´s landscape features predict patterns of genetic variation within eleven nominal species of amphibian belonging to 7 taxonomic families. We quantified the relative role of geographic distance (IBD), topography (IBRt) and climate (IBRsuitability and IBE) in shaping genetic divergence in each species. Using this information, we tested the following hypotheses and related predictions applied to amphibian species with diverse life histories.

1) Linear geographic distance itself is not a good predictor of genetic divergence in the complex landscapes of ICA. Variation in genetic divergence is better explained by metrics that acknowledge the topographic and climatic heterogeneity among populations.

2) Prominent mountain chains in the region represent major vicariant barriers for the local fauna. Due to the massive dimensions of these orogenic formations, we expect major genetic structure in species occurring on both the Pacific and Caribbean lowlands, regardless of their biology.

3) Climatic suitability increases the probability of a population to persist in the environment. We predict that genetic structure may be explain by patterns of isolation due to patchy distribution of climatically suitable habitats in the complex landscapes of the region.

4) High climatic heterogeneity, promotes divergence among populations of the same species occurring in distinct environments. Different climatic regimes within the region promoted adaptation to local conditions and such process may reflect the patterns of genetic divergence.

METHODS

Study species

We used as study system a set of 11 amphibian species, including nine anurans and two salamanders for which we had mitochondrial DNA (mtDNA) sequence data.

Such species represent a variety of reproductive strategies, body sizes, dispersal capabilities, and distribution ranges (see details in Table 1).

1 Table 1. Amphibian species studied with details on several biological attributes and genetic data analyzed.

2 Family Species SVL Reproductive Mode Clutch Elevation Genes Localities (mm) Size (m) Plethodontidae Bolitoglossa lignicolor 46-81 Direct Development Unknown 0-900 11 (16S) 9

Plethodontidae Oedipina alleni 40-58 Direct Development Unknown 0-900 5 (16S) 4

Eleutherodactylidae Diasporus diastema 16-24 Direct Development 7 to 19 1- 1500 11 (COI) 8 Arboreal eggs Strabomantidae Pristimantis ridens 16-25 Direct Development Unknown 15-1600 14 (16S) 11 Terrestrial eggs Centrolenidae Sachatamia albomaculata 20 -32 Arboreal eggs Unknown 0-850 14 (16S), 7,7 tadpoles into streams 14 (COI) Centrolenidae Espadarana prosoblepon 21 -31 Arboreal eggs, ~20 20-1900 26 (16S) 12 tadpoles into streams Hylidae Dendropsophus ebraccatus 23-35 Arboreal eggs,tadpoles into 15 to 296 0-1300 24 (16S) 10 puddles, ponds, streams Hylidae Smilisca phaeota 40-78 Eggs and tadpoles in small 1.5K -> 2K 0-1000 9 (16S) 8 ponds or shallow streamlets Hylidae Agalychnis callidryas 30-71 Arboreal eggs, tadpoles into 11 to 104 1 -1000 51 (16S) 14 puddles, ponds, streams Ranidae Lithobates warszewitschii 37-52 Eggs /tadpoles in lotic water Unknown 1 - 1750 29 (16S) 11

Bufonidae Rhinella marina 85-175 Eggs /tadpoles in lentic water 5K - 25K 1 - 2100 9 (16S) 8

We selected these species from a bigger data set of mtDNA from which we discarded cryptic species or species with less than 5 sampled localities. The final set analyzed here included sequences from 350 individual amphibians, corresponding to 95 nominal species sampled from localities across Costa Rica (Fig. 1).

Figure 1. Maps with the distribution of available genetic data for the 11 amphibian species included in this study

Sampling and sequencing methods

Most tissues used in this study were obtained from samples deposited at the

Herpetology Collection of the Museo de Zoología de la Universidad de Costa Rica

(UCR). These new data were supplemented with sequence data available in

GenBank obtained from Costa Rica as well as from Panama. The final dataset analyzed here contained 136 sequences of the cytochrome oxidase I (COI-5’, also known as the Barcode of Life fragment [Hebert et al. 2003]) and 321 sequences of a fragment of the 16S ribosomal RNA gene. Species names, field collection numbers, museum numbers, localities, Barcode of Life Data Systems (BoLD;

Ratnasingham and Hebert 2007) Process ID for each specimen, and GenBank accession numbers for each sequence used in the present study are provided in Table

S1.

DNA extraction was carried out on a BioSprint 96 (Qiagen) robotic extractor based on magnetic beads, including digestion with proteinase K (0.4 mg/mL) at 55 °C. We amplified two mitochondrial gene fragments, the fast-evolving COI, and the more slowly-evolving 16S gene, following Crawford et al. (2010). The primer pairs used to amplify and Sanger-sequence the COI-5’ gene were: dgHCO2198 (5′-TAA ACT

TCA GGG TGA CCA AAR AAY CA-3′) and dgLCO1490 (5′-GGT CAA CAA

ATC ATA AAG AYA TYG G-3′) (Folmer et al. 1994; Meyer et al. 2005) and 0.25

µg/µL of bovine serum albumin. The 16S gene fragment was amplified and

sequenced with primers 16SB-H (aka, 16Sbr-H) (5′-CCG GTC TGA ACT CAG

ATC ACG T-3′) and 16SA-L (aka, 16Sar-L) (5′-CGC CTG TTT ATC AAA AAC

AT-3′) (Kessing et al. 2004). PCR products were cleaned using ExoI and SAP enzymes (Werle et al. 1994), with Sanger sequencing reactions run on ABI 3130 automated sequencers. All enzymatic and sequencing reactions were performed in a high-throughput 96-well format. Both genes were sequenced bi-directionally to confirm base calls. The sequences of each gene were aligned independently using default parameters in MUSCLE (Edgar, 2004) -available at https://www.ebi.ac.uk/Tools/msa/muscle/ - and reviewed by eye.

Potential cryptic species handling

Preliminary genetic analyses showed that some museum voucher specimens were misidentified, a common mistake in mega-diverse regions such as ICA, where a high number of sister species maintain morphological stasis after speciation (Padial and

De La Riva, 2009; Funk et al., 2012). To prevent potential errors assigning DNA sequences to species: First, we used the Automated Barcode Gap Discovery (ABGD) algorithm (http://wwwabi.snv.jussieu.fr/public/abgd/), which identifies clusters of sequences that may correspond to more than one species based on the distribution of pairwise genetic distances between the aligned DNA sequences. This method statistically infers multiple potential barcode gaps or thresholds, and partitions the sequences such that the distance between two sequences taken from distinct clusters

is larger than the barcode gap (Puillandre et al., 2012). This method performs well in terms of efficiency and success in species identification compared with other

DNA barcode algorithms (Paz and Crawford, 2012). The COI, 16S, and concatenated alignments were processed in ABGD assuming a Kimura two- parameter (K2P) nucleotide substitution model (Kimura, 1980) and the following settings: prior for the maximum value of intraspecific divergence between 0.001 and

0.1, with 10 recursive steps within the primary partitions defined by the first estimated gap, and a gap width of 1.0. K2P is the standard model of DNA substitution for barcode studies, performing as well as other more complex models in identifying specimens (Collins and Cruickshank, 2012). Even though there are not many DNA substitution models available in ABGD, a recent study suggests that species identification success rate is not affected by the model (Collins et al., 2012)

We corroborated the ABGD results by checking if there was evidence of highly divergent DNA sequences in sympatry, which may suggest potential misidentifications. To do this, we contrasted geographic versus genetic distances for each nominal species and searched for cases where genetic divergence at short geographic distances (in sympatry) were above 5% for COI and 2.5% for 16S. We selected these thresholds considering those proposed for candidate species in frogs,

10% for COI and 5% (Vences et al., 2005) or 3% (Fouquet et al. 2007) 16S. The

total number of sequences and nominal species that we ended up after the ABDG analyses are in Table S

Estimation of genetic distances

Pairwise genetic p-distances were estimated between 16S and COI sequences available for each species (See Table 1). We chose the best-fit model of nucleotide substitution for each set of sequences using jModelTest (Posada, 2008) based on the

Akaike information criterion (AIC). Then, we estimated genetic distances with

MEGA 4.0 (Tamura et al., 2013) using the pairwise-deletion option, therefore excluding inferred gaps in each pair of sequences for the genetic distance calculations. We used these genetic distances as a proxy for gene flow (Rousset,

1997; Selonen et al., 2010). We estimated separately genetic distances for COI or

16S (we did not concatenated both genes) since in many nominal species there were no sequences available for both genes. We used the landscape genetics approach, where the individual sequence is the unit of analysis, to avoid biases in the identification of populations (step needed to estimate other genetic divergence measurements, such as Fst).

Estimation of geographic, resistance and environmental distances

To understand the relationship between geography and the intraspecific genetic divergence we estimated landscape derived distances, including geographic distance

(IBD), and resistance distances that account for topography (IBRt) and climatic suitability (IBRcs) as well as local environmental dissimilarity (IBE) between all individual pairwise combinations within each species. This method allowed us to compare the relative importance of different landscape variables in explaining genetic divergence within each species.

To test for IBD, we estimated linear geographic distances as pairwise

Euclidean distances (in km) between the geographic position of each pair of individuals of each species using the R package raster (Hijmans and van Etten,

2010). To test for IBR we estimated resistance distances using topography and environmental suitability as friction layers based on a circuit-theory approach conducted in Circuitscape V.4.0 (Shah and McRae, 2008). For topography, we created a layer that quantifies local terrain complexity, using a raster grid of elevation with a resolution of 30 arcseconds (~0.8 km at the equator) and estimated the standard deviation of a set of adjacent cells to obtain a value of topographic complexity for cells of ~5 by 5 km resolution. In the case of environmental suitability, we used the inverse of species distribution model suitability (SDMs) for each species, assuming that areas with low suitability have higher resistance for dispersal (Wang, Yang, et al., 2008).

To construct SDMs, we gathered additional occurrence data from the Global

Biodiversity Information Facility and the Collection of Herpetology of Universidad

de Costa Rica. We generated SDMs using 11 out of the 19 bioclimatic variables available at worldclim.org (2.5 arcmin resolution ~ 4.5 km at the equator) for current conditions. We excluded variables using a Variance Inflation Factor Analysis to avoid predictor redundancy. We created all models with a Maximum Entropy algorithm using the R package dismo (Hijmans et al., 2012) after a process of parameter tuning and evaluation conducted in the R package ENMeval (Muscarella et al., 2014). We provide details on modeling procedures in the supplementary material. We also quantified historical isolation by projecting these models to Mid

Holoce (~ 6 Kya) and Last Glacial Maximum -LGM- (~ 22 Kya) conditions and using the inverse of these projections as friction layers.

To test for IBE, we estimated environmental dissimilarity between locality pairs. For each sampled locality, we extracted the values for the 19 bioclimatic variables available in the worldclim data set (worldclim.org) at 30 arcseconds resolution (~ 1km). Then we estimated Euclidean pairwise distances in the multidimensional space using the function dist in R which computes specific distances between the rows of a multivariate matrix.

Quantifying the relative effects of IBD, IBR and IBE

We tested for the independent association between geographic, resistance (i.e. topographic and climatic suitability), and climatic distances against genetic distances

using Multiple Matrix Regression with Randomization Analysis (MMRR) and

Generalized Dissimilarity Matrix. MMRR uses randomized permutation to account for the fact that distances are not independent from each other (Wang et al., 2013).

We preferred this method to partial Mantel tests because of the well-known high type-I error rate and low power characteristic of these kind of tests (Raufaste and

Rousset 2001; Harmon and Glor 2010; Guillot and Rousset 2013). In contrast to partial Mantel tests, MMRR provides the independent effect of each variable as beta coefficients, allowing simultaneous comparisons among them. Because all estimated distances are at different scales, we normalized (subtracted the mean and divided by the standard deviation) them to facilitate the interpretation of the beta coefficients.

We performed the MMRR method using the R function provided by Wang (2013) with 10,000 permutations.

In addition, to further explore how our explanatory variables shape the patterns of genetic differentiation observed in the study species, we also implemented a Generalized Dissimilarity Modelling (GDM). This matrix regression technique can fit nonlinear relationships of environmental variables to biological variation using I-spline basis functions (more details in Ferrier et al. 2007). The splines plots are very informative because they provide insights into the total magnitude of biological change as a function of each gradient and where those changes are most pronounced along each gradient.

We fitted the initial model using as response variable our matrices of genetic distance and the previously described matrices of isolation (IBD, IBRt, IBRcs,

IBRlgm and IBE) as predictor variables. Then, we plotted the I-splines to assess how magnitudes and rates of genetic differentiation varied along the gradients represented in our predictor variables. To confirm the significance obtained in the

GDM, we estimated variable importance and significance using matrix permutation and backward elimination (as detailed in Ferrier et al. 2007). After summing the coefficients of the I-splines, we discarded the less contributing predictor, and then using this reduced set of n-1 predictors, we fitted a second GDM model. We ran 500 random permutations and excluded the variable with the least significant contribution to explained deviance in a stepwise procedure. At each step of the procedure, the unique contribution of each variable to total explained deviance was calculated. We repeated the method until all variables retained in the final model made significant unique contributions to explained deviance (P ≤ 0.05). We performed these analyses with the R package ‘gdm’ (Manion et al. 2017).

RESULTS

Intraspecific patterns of genetic differentiation

For the two salamander species in our data set, both of them having distributions in the Pacific, our ABGD analysis recovered two clades. In Bolitoglossa lignicolor, we

found a break between populations in the Central Pacific Region and those extending from the Southern Pacific of Costa Rica to Peninsula de Azuero in Panama. In the case of O. alleni the break seems to be altitudinal, since strictly coastal individuals grouped apart from those individuals in slightly higher elevations.

Among the direct-developing frogs (Terrarana), we found two major clades for Diasporus diastema; one including Caribbean populations from the slopes of

Cordillera de Tilarán to the Chiriquí Province in Panama and other containing samples from Coclé Region in Central Panama. For Pristimantis ridens we found support for a clade represented by Costa Rican populations and a clade containing all Panamanian populations. Among the glass frogs, we found two clades with high support for Sachatamia albomaculata, one on the Caribbean and other on the Pacific versant of Costa Rica. For 16S, for which we have a wider geographic sampling for this species, we found divergence between populations in the Caribbean and Pacific of Costa Rica as well as between the central region of Panama and the eastern portion of the Cordillera de Talamanca, which lays in Panama. In Espadarana prosobleplon, we found four clades with high support: Darien Region, Eastern Canal Zone,

Cordillera Central de Panama and Pacific versant of Costa Rica.

In our hylid species, we found four genetically differentiated groups in

Agalychnis callidryas, two on the Pacific and two on the Caribbean; in Smilisca phaeota, samples grouped in two major clusters, although these are not clearly

differentiated in the geography. In Dendropsophus ebbraccatus we detected the major genetic divergence between samples from southern Pacific of Costa Rica and a group including samples from Central Pacific and all the Caribbean versant

In the ranid Rana (Lithobates) warszewitschii, we found one clade widely distributed in both the Pacific and the Caribbean versants of Costa Rica, which differs from two clades occurring in Panama, one in the Eastern Canal Zone and the other in the Cordillera Central. Finally, in the common cane toad Rhinella horribilis, we detected two clusters, although their geographic distribution is not clearly differentiated.

Relative role of IBD, IBRt, IBRcs and IBE on genetic differentiation

From the MMRR analyses, we found that at least one of the factors here tested, significantly explained the patterns of genetic differentiation in seven out of the 11 species studied (Fig. 2). None of the models tested here explained genetic differentiation in the salamander species (O. alleni and B. lignicolor), the hylid frog

S. phaeota and the toad R. horribilis.In the species where the full model significantly explained genetic differentiations among populations, we found idiosyncratic responses to the predictors tested.

Figure 2. Contribution of different types of isolation in explaining genetic differentiation within our study species. Black bars in the first colum represent the percentange of variance in the genetic data significantly explained by the full model. Second to fifth colums show beta values for the tested predictors, black bars represent predictors significantly associated with patterns of genetic differentiation

Geographic distance played an important role in the genetic differentiation of

P. ridens (βD=0.648, P<0.001), topography (IBRt) explained most genetic divergence for A. callidryas (βT=0.580, P<0.001) and S. albomaculata, for both available genes: 16S (βT=0.580, P<0.001) and COI (βT=1.173, P<0.001). While climatic suitability (IBRcs) represented the main driver of genetic divergence for D. ebracattus (βCS=0.998, P<0.001), D. diastema (βCS=0.816, P<0.001), and L. warszewitschii (βCS=0.947, P<0.001), local environmental dissimilarity (IBE) was the major factor explaining genetic variation within E. prosoblepon (βED=0.486,

P<0.001).

From the GDM we found that, not only the responses of each species are idiosyncratic to the drivers of genetic differentiation but, also the magnitude and rates in which each species are affected along the gradient of variation of each predictor. In figure 3, we show this heterogeneity by presenting the spline plots for the two variables that contribute more within the species in which the full model was significant in explaining patterns of genetic differentiation.

Figure 3. Generalized dissimilarity model-fitted I-splines for variables significantly associated with genetic differentiation in seven of our study species. The maximum height reached by each curve, indicates the total amount of genetic differentiation associated with the respective variable, holding all other variables constant. The shape of each function shows how the rate of genetic differentiation varies along the gradients.

DISCUSSION

Despite its young geological history, ICA hosts an immense number of species, higher than any other region of the globe with similar area (Bagley and Johnson,

2014). Such richness was assembled a combination of dispersal events and in situ diversification, reflected in high levels of endemism (Marshall 1988; Webb 2006;

Kluge & Kessler 2006; Savage & Bolaños 2009; Bogarín et al. 2013). Here, we quantified the role that different components of the landscape has had on driving isolation between populations and restricting gene flow.

In almost all species studied, we found that at least one of the factors tested, significantly explained patterns of spatial genetic differentiation in ICA. However, instead of finding a major force that generally explained intraspecific genetic divergence across the region, our results reveal idiosyncratic responses, so the drivers we tested differentially affect each species. Previous studies focused on describing spatial patterns of amphibian genetic divergence in the ICA, have also showed that phylogeographic histories among species have few patterns in common

(Weigt et al. 2005; Crawford et al. 2007; Wang et al. 2008; Robertson et al. 2009).

In a recent review, Bagley & Johnson (2014), summarized the emerging phylogeographic patterns in the region by collecting over 50 studies dealing with more than 90 nominal taxa distributed in LCA. Such work compiles at least 31 phylogeographic breaks - recovered from mitochondrial DNA markers - in a region

spanning only ~127 000 km2 (Bagley and Johnson, 2014). Our findings support this observation and provide strong evidence of intense processes of in situ lineage diversification promoted not necessarily by a dominant factor but by multiple drivers in the complex landscapes of ICA. By applying the same analyses over a multi- species data set containing the same molecular markers, we were able to make robust comparisons and go a step further, as we aimed to test for causality in the observed patterns of genetic differentiation and disentangle the relative contribution of its drivers. By combining MMRR and GDM approaches, we were able not only to make a partition of the variation explained by each predictor (Wang, 2013; Wang et al.,

2013) but also to quantify the intensity and rate in which genetic divergence is drove along the gradient of each tested variable (Ferrier et al., 2007).

Even geographic distance (IBD), which was not expected to be a good predictor in the roughed landscapes of the ICA, was the major factor in explaining genetic structure in two species, E. prosoblepon and P. ridens. In such cases, we hypothesized that the major geographic gap between our Costa Rican and

Panamanian samples could influence such result. E. prosoblepon is a small species distributed from lowlands to almost 2000 altitude (Kubicki, 2007), despite its wide altitudinal range, reproduction in this species is restricted to streams, a typical feature of glass frogs (Castroviejo-Fisher et al., 2014). It may promote isolation between watersheds instead of along altitudinal gradients, explaining the increase in genetic

differentiation with geographic distance due to restricted dispersal between more distant populations as expected by IBD (Wright, 1943; Slatkin, 1993).

Complementarily, previous studies has documented phylogeographic breaks separating frog lineages from Costa Rica and western Panama from those occurring in Central Panama, probably as result of a the vicariance generated by a Dry Forest

Barrier (Crawford et al., 2007). In P. ridens for example, phylogeographic approaches, revealed a Panamanian lineage showing long-term geographic stasis and another showing rapid geographic expansion occurring from Costa Rica to Honduras

(Wang, Crawford, et al., 2008). These dynamics, although potentially resulting from climatic factors, could explain the higher divergence between our Costa Rican and

Panamanian samples and the signal we recovered of IBD driving such pattern.

The region’s abundant mountainous reliefs reflect a rich geological history mainly dominated by orogeny of volcanic and tectonic origin (Gabb et al., 2007) which results in major physical barriers (IBRt). Within a species, genetic divergence is expected to be greater in areas of higher topographic complexity (Guarnizo and

Cannatella, 2013). In our case, IBRt significantly explained divergence patterns in the red-eye tree frog A. callidryas and the cascade glass frog S. albomaculata. For these species, we analyzed samples from both, the Pacific and the Caribbean versant.

Our results highlight the role of the three main Cordilleras that bisect the region serving as a vicariant barrier between lowland populations in both slopes.

In the case of A. callidryas, the dimension of this effect is evident even at the phenotypic level, previously documented on the variation in flank coloration among individuals from Caribbean and Pacific populations (Robertson and Zamudio, 2009).

Genetic variation assessed in that same work among 20 populations, recovered five well-supported mitochondrial clades, some of which the authors explain by IBD

(Robertson and Zamudio, 2009), the second force in contributing to the patterns of genetic divergence we found. To our knowledge, our study is the first in evaluating genetic divergence among populations of S. albomaculata, our geographic sampling on both sides of the Cordilleras as well as our results are similar to A. callidryas.

Sachatamia albomaculata has been reported to reach the 1500 m elevation (Savage,

2002) but, it is more common below 1000 m (Kubicki, 2007). This fact, in combination with life history traits of this species such as its small size and high levels of phylopatry -restricted to forest covered streams- (Solís et al., 2010), may explain the resistance imposed by topographic barriers not only among versants, but also between peaks within the same mountainous range.

A similar pattern to the one found at the intra-specific level for these two species could have led to speciation in many groups in the region for which sister taxa occur. On each side of major mountain systems, such as several anurans including Dendropsophus microcephalus and D. phlebodes, Oophaga granulifera and O. pumilio or Phyllobates lugubris and P.vittatus (Savage, 2002). Among

reptiles some examples include snakes as Lachesis stenophrys and L. melanocephala

(Solórzano and Cerdas, 2010), the lizards Basiliscus basiliscus and B. vittatus,

Sphaerodactylus graptolaemus and S. homolepis or Leposoma flavimaculatum and

L. reticulatum (Savage, 2002). This pattern occurs even in representatives of groups with higher dispersal abilities such as the birds Amazilia decorata and A. amabilis,

Cotinga ridwayi and C. amabilis as well as Manacus candei and M. aurantiacus

(Prum et al., 2000; Brumfield and Braun, 2001; Stiles et al., 2017). Divergence between Pacific and Caribbean groups is expected to be more evident in eastern

Costa Rica and western Panama, where topographic barriers becomes stronger as mountains reaches their higher elevations (>3000 m) in Cordillera de Talamanca (de

Boer et al., 1995). In western Costa Rica instead, vicariant events may derive from sky island dynamics, because high elevation habitats in those regions are isolated and surrounded by lower intervening valleys. In such cases species tend to have restricted distributions limited to mountain tops, as occur in many microendemic amphibians including Nototriton Guanacaste and N. gamezi, representatives of

Crepidophryne, Incilius periglenes, I. holdridgei and I. fastidiosus (Savage, 2002;

Vaughan and Mendelson, 2007; Abarca et al., 2010).

We also hypothesized that either current or historical barriers imposed by climatic conditions that determine niche suitability (IBRsuit) for each species could promote isolation and influence genetic differentiation in the study region. We found

that such explanation fits in the case of D. ebraccatus, D. diastema and L. warszewitschii. In all cases, the resistance imposed by less suitable regions between sampled localities is the factor that best explains genetic differentiation in these species. Such trend follows the principles of niche conservatism, the tendency of species to retain ancestral ecological characteristics (Wiens and Graham, 2005;

Wiens et al., 2010). At the intraspecific level, it could result either from the fragmentation of continuous distributions due to climatic oscillations that reflect on range contractions (Hewitt, 2003, 2004) or from disjointed distributions that result from different routes of colonization. In both cases populations could become isolated by means of unsuitable conditions in the regions that connect them.

In our fourth hypothesis, we predicted that climatic heterogeneity in the region might promote adaptation to local conditions (IBE) and influence the patterns of genetic differentiation. We found that this scenario explains genetic differentiation in the COI gene of S. albomaculata.

Certainly, one of the most striking characteristics of LCA is the vast variety of climates that converges in such a small area (Coen, 1991). For example, transitions between arid and very humid conditions occur in few hundreds of kilometers in the

Pacific of Costa Rica while temperature may change significantly in short distances along elevation gradients in the mountain systems of the region (Coen, 1991;

Savage, 2002). Variation in species-specific tolerances and local adaptation to such

diverse climates seems to influence distribution of several taxa in the region. It occurs even in groups considered good dispersers, such as birds (Garrigues and

Dean, 2014) and volant mammals like the vespertilionid bats Rhogeessa bickhami and R. io.

Finally, our full model containing all predictors was not able to significantly explain genetic divergence in four species: the salamanders B. lignicolor and O. alleni, the masked tree frog S. phaeota and the toad R. horribilis. These species have several particularities that may account for such lack of relation between our predictors and their genetic structure. Both salamanders have distributions restricted to the Pacific slope of the study region, in elevations below 900 meters (Savage,

2002). This removes the opportunity to test for topographic or climatic barriers along broad altitudinal gradients. Conversely, it also highlights the role that physical barriers and climate play as primary constraints of distributions by limiting species potential for dispersion and establishment, depending on their specific tolerances to abiotic conditions (Grinnell, 1917; Barve et al., 2011).

The two anurans can be considered more generalists in their preferences and thus are expected to have wider distributions. Such distributions can be maintained due to their, broader tolerances and higher dispersal capabilities, probably as result of their intrinsic biological features. For example, S. phaeota a common species, is the largest tree frog among the hylids included in this study (Savage, 2002). This

species has extended breeding through the year, reproduces in small ponds or shallow streamlets even in altered habitats, clutch size can reach the 2000 eggs and tadpoles are resistant enough to survive up to 24 hours out of the water (Valerio,

1971; Savage, 2002). Rhinella horribilis was only recently described as a cryptic species related to the common cane toad R. marina (Acevedo et al., 2016). Rhinella mariana, is widely distributed in the Neotropics, has high dispersal potential and is known for one of the most aggressive invasions in Australia (Phillips et al., 2006,

2010). In their native range these species are common even in altered habitats from sea level to above 2000 meters elevation where they lay up to 25000 eggs in lentic water bodies (Savage, 2002).

Morphological features, high dispersal ability, intrinsic physiological tolerances and behavioral strategies in these large species may explain its distribution and persistence under heterogeneous environmental pressures (Hilje and

Arévalo-Huezo, 2012; Mccann et al., 2014). For these reasons, we consider that barriers affecting other anurans do not necessarily represent major drivers of isolation and gene flow must be higher in this species.

CONCLUSIONS

The convergence of several drivers of isolation and the co-occurrence of such a heterogeneous biota with different life histories, origins and arrival times to the complex landscapes of ICA maximize the chances of genetic differentiation in the region. Different types of isolation such as IBD and IBE (both reviewed in Sexton et al. 2014) or IBR driven by topographic and climatic barriers (Rodríguez et al.,

2015; Thomas et al., 2015; Oliveira et al., 2017) are proved forces involved in the genetic differentiation of different taxa around the globe. Our study highlights how they can act simultaneously and differentially affect co-distributed taxa in a relatively small area. Certainly, the intrinsic characteristics of each species play an important role in how a given species respond to different drivers of isolation; and such interaction between organisms and their environment must be considered when trying to understand patterns of genetic divergence (Paz et al., 2015; Rodríguez et al., 2015). The evolutionary dynamic of ICA is far from the simplistic view that point out vicariance between Caribbean and Pacific clades as the main form of speciation in the region. In situ diversification plays an important role in shaping richness patterns in the region and its biota is undoubtedly underestimated. For that reason, further efforts must be oriented to first document unknown diversity and then add more groups to this kind of analysis.

ACKNOWLEDGEMENTS

Javier Guevara and Sistema Nacional de Areas de Conservación de Costa Rica provided permits to researchers of Museo de Zoología, Universidad de Costa Rica

(MZUCR). We = thank Federico Bolaños (MZUCR) for permission and access to the samples sequenced for this study; Sandra Flechas for his patient collaboration during lab procedures. NAOS Island Laboratories of the Smithsonian Tropical

Research Institute for support during lab work A.G.R acknowledges Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, Brazil (CAPES) for the financial support and Gerardo Chaves (MZUCR) for his teachings during innumerous field trips, as well as for the constant discussion on the herpetology of Central America and his comments on early versions of this manuscript.

AUTHOR CONTRIBUTIONS

A.G.R, C.E.G and G.C.C conceived the study; A.G.R and A.J.C conducted field work and collected DNA sequence data; A.G.R and C.E.G analyzed data; A.G.R,

C.E.G and G.C.C wrote the first drafts of the manuscript; All authors provide vital inputs, discussed, edited, and improved the final version of this study.

REFERENCES

Abarca J, Chaves G, Garcia-Rodriguez A, Vargas R (2010). Reconsidering extinction : rediscovery of Incilius holdridgei (Anura: Bufonidae) in Costa Rica after 25 years. Herpetol Rev 41: 150–152. Acevedo AA, Lampo M, Cipriani R (2016). The cane or marine toad, Rhinella marina (Anura, Bufonidae): two genetically and morphologically distinct species. Zootaxa 4103: 574–586. Anger GR, Dean R (2010). The Birds of Panama: a Field Guide. Zona Tropical Publication: San José. Bacon CD, Silvestro D, Jaramillo C, Smith BT, Chakrabarty P, Antonelli A (2015). Biological evidence supports an early and complex emergence of the Isthmus of Panama. Proc Natl Acad Sci 112: 6110–6115. Bagley JC, Johnson JB (2014). Phylogeography and biogeography of the lower Central American Neotropics: Diversification between two continents and between two seas. Biol Rev 89: 767–790. Barve N, Barve V, Jiménez-Valverde A, Lira-Noriega A, Maher SP, Peterson a. T, et al. (2011). The crucial role of the accessible area in ecological niche modeling and species distribution modeling. Ecol Modell 222: 1810–1819. Blaustein AR, Wake DB, Sousa WP (1994). Amphibian declines: judging stability, persistence, and susceptibility of populations to local and global extinctions. Conserv Biol 8: 60–71. de Boer JZ, Drummond MS, Bordelon MJ, Defant MJ, Bellon H, Maury RC (1995). Cenozoic magmatic phases of the Costa Rican island arc (Cordillera de Talamanca). Geol Soc Am Spec Pap 295: 35–56. Bogarín D, Pupulin F, Arrocha C, Warner J (2013). Orchids without borders: studying the hotspot of Costa Rica and Panama. Lankesteriana 13: 13–26. Brumfield RT, Braun MJ (2001). Phylogenetic relationships in bearded manakins (Pipridae : Manacus) indicate that male plumage color is a misleading taxonomic marker. Condor 103: 248–258. Castroviejo-Fisher S, Guayasamin JM, Gonzalez-Voyer A, Vilà C (2014). Neotropical diversification seen through glassfrogs. J Biogeogr 41: 66–80. Coates AG, Jackson JBC, Collins LS, Cronin TM, Dowtsett HJ, Bybell LM, et al. (1992). Closure of the Isthmus of Panama: The near-shore marine record of Costa Rica and western Panama. Geol Soc Am Bull 104: 814–828. Coen E (1991). Climate. In: Janzen DH (ed) Historia Natural de Costa Rica, Editorial de la Universidad de Costa Rica: San José, Costa Rica, p 822. Collins RA, Boykin LM, Cruickshank RH, Armstrong KF (2012). Barcoding’s next top model: An evaluation of nucleotide substitution models for specimen

identification. Methods Ecol Evol 3: 457–465. Collins RA, Cruickshank RH (2012). The seven deadly sins of DNA barcoding. Mol Ecol Resour 13: 969–975. Crawford AJ, Bermingham E, Carolina PS (2007). The role of tropical dry forest as a long-term barrier to dispersal: a comparative phylogeographical analysis of dry forest tolerant and intolerant frogs. Mol Ecol 16: 4789–807. Doebeli M, Dieckmann U (2003). Speciation along environmental gradients. Nature 421: 259–264. Edgar RC (2004). MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 32: 1792–1797. Ferrier S, Manion G, Elith J, Richardson K (2007). Using generalized dissimilarity modelling to analyse and predict patterns of beta diversity in regional biodiversity assessment. Divers Distrib 13: 252–264. Frost DR (2017). Amphibian species of the world: an online reference. Version 6.0 (August 2017). Electronic Database accessible at http://research.amnh.org/herpetology/amphibia/index.html. American Museum of Natural History, New York, USA. Funk WC, Blouin MS, Corn PS, Maxell B a, Pilliod DS, Amish S, et al. (2005). Population structure of Columbia spotted frogs (Rana luteiventris) is strongly affected by the landscape. Mol Ecol 14: 483–96. Funk WC, Caminer M, Ron SR (2012). High levels of cryptic species diversity uncovered in Amazonian frogs. Proc R Soc B 279: 1806–14. Gabb WM, Lücke OH, Gutiérrez V, Soto G (2007). On the geology of the Republic of Costa Rica. 203: 103–118. Garrigues R, Dean R (2014). The Birds of Costa Rica. Zona Tropical Publication: San José, Costa Rica. Grinnell J (1917). Field tests of theories concerning distributional control. Am Nat 51: 115–128. Guarnizo CE, Cannatella DC (2013). Genetic divergence within frog species is greater in topographically more complex regions. J Zool Syst Evol Res 51: 333– 340. Hewitt G (2003). Ice ages: their impact on species distributions and evolution. In: Rothschild LJ, Lister AM (eds) Evolution on Planet Earth: The Impact of the Physical Environment, Academic Press: New York, pp 339–361. Hewitt GM (2004). Genetic consequences of climatic oscillations in the Quaternary. Philos Trans R Soc Lond B Biol Sci 359: 183–95; discussion 195. Hijmans RJ, van Etten J (2010). raster: Geographic analysis and modeling with raster data. R Packag version 1. Hijmans RJ, Phillips S, Leathwick J, Elith J (2012). Package ‘dismo’. Species distribution modeling. R package version 0.8-11.

Packag. Hilje B, Arévalo-Huezo E (2012). Aestivation in the cane toad Rhinella marina Linnaeus, 1758 (Anura, Bufonidae) during the peak of a dry season in a tropical dry forest, Costa Rica. Herpetol Notes 5: 533–534. Horn S (1990). Timing of deglaciation in the Cordillera de Talamanca, Costa Rica. Clim Res 1: 81–83. Islebe GA, Van Der Borg K, Hooghiemstra H (1995). The Younger Dryas climatic event in the Cordillera de Talamanca, Costa Rica. Geol en Mijnb 74: 281–283. Islebe GA, Hooghiemstra H, van´t Veer R (1996). Holocene vegetation and water level history in two bogs of the Cordillera de Talamanca, Costa Rica. Vegetatio 124: 155–171. Janzen DH (1991). Historia Natural de Costa Rica, Primera (D Janzen, Ed.). Editorial San José, Universidad de Costa Rica: San José, Costa Rica. Keigwin LD (1978). Pliocene closing of the Isthmus of Panama, based on biostratigraphic evidence from nearby Pacific Ocean and Caribbean Sea cores. Geology 6: 630–634. Kimura M (1980). A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. J Mol Evol 16: 111–120. Kluge J, Kessler M (2006). Fern endemism and its correlates: contribution from an elevational transect in Costa Rica. Divers Distrib 12: 535–545. Kubicki B (2007). Ranas de Vidrio de Costa Rica, 1st edn. Instituto Nacional de Biodiversidad: Santo Domingo de Heredia, Costa Rica. Luteyn J (1999). Paramos: A checklist of plant diversity, geographical distribution, and botanical literature. Mem N Y Bot Gard 84: 138–141. Manel S, Holderegger R (2013). Ten years of landscape genetics. Trends Ecol Evol 28: 614–621. Marshall LG (1988). Land mammals and the great American interchange. Am Sci 76: 380–388. Marshall JS (2007). The geomorphology and physiographic provinces of Central America. In: Bundschuh & Alvarado (ed) Central America: Geology, Resources and Hazards,, pp 1–51. Mayr E (1942). Systematics and the origin of species, 1st edn. Columbia University Press: New York. Mccann S, Greenlees MJ, Newell D, Shine R (2014). Rapid acclimation to cold allows the cane toad to invade montane areas within its Australian range. Funct Ecol 28: 1166–1174. McRae BH (2006). Isolation by resistance. Evolution 60: 1551–1561. Montes C, Cardona A, McFadden R, Morón SE, Silva CA, Restrepo-Moreno S, et al. (2012). Evidence for middle Eocene and younger land emergence in central

100

Panama: Implications for Isthmus closure. Bull Geol Soc Am 124: 780–799. Muscarella R, Galante PJ, Soley-Guardia M, Boria RA, Kass JM, Uriarte M, et al. (2014). ENMeval: An R package for conducting spatially independent evaluations and estimating optimal model complexity for Maxent ecological niche models. Methods Ecol Evol 5: 1198–1205. O’Dea A, Lessios HA, Coates AG, Eytan RI, Restrepo-Moreno SA, Cione AL, et al. (2016). Formation of the Isthmus of Panama. Sci Adv 2: 1–12. Oliveira EF, Martinez PA, São-Pedro VA, Gehara M, Burbrink FT, Mesquita DO, et al. (2017). Climatic suitability, isolation by distance and river resistance explain genetic variation in a Brazilian whiptail lizard. Heredity (Edinb) 120: 251–265. Padial JM, De La Riva I (2009). Integrative taxonomy reveals cryptic Amazonian species of Pristimantis (Anura: Strabomantidae). Zool J Linn Soc 155: 97–122. Paz A, Crawford AJ (2012). Molecular-based rapid inventories of sympatric diversity: a comparison of DNA barcode clustering methods applied to geography-based vs clade-based sampling of amphibians. J Biosci 37: 887–896. Paz A, Ibáñez R, Lips KR, Crawford AJ (2015). Testing the role of ecology and life history in structuring genetic variation across a landscape: a trait-based phylogeographic approach. Mol Ecol 24: 3723–3737. Phillips BL, Brown GP, Shine R (2010). Evolutionarily accelerated invasions: the rate of dispersal evolves upwards during the range advance of cane toads. J Evol Biol 23: 2595–2601. Phillips BL, Brown GP, Webb JK, Shine R (2006). Invasion and the evolution of speed in toads. Nature 439: 803. Pinto-Sánchez NR, Ibañez R, Madriñán S, Sanjur OI, Bermingham E, Crawford AJ (2012). The Great American Biotic Interchange in frogs: multiple and early colonization of Central America by the South American genus Pristimantis (Anura: Craugastoridae). Mol Phylogenet Evol 62: 954–972. Posada D (2008). jModelTest: Phylogenetic model averaging. Mol Biol Evol 25: 1253–1256. Prum R, Rice NH, Mobley J, Dimmick W (2000). A preliminary phylogenetic hypothesis for the cotingas (Cotingidae) based on mitochondrial DNA. Auk 117: 236–241. Puillandre N, Lambert A, Brouillet S, Achaz G (2012). ABGD, Automatic Barcode Gap Discovery for primary species delimitation. Mol Ecol 21: 1864–1877. Robertson JM, Duryea MC, Zamudio KR (2009). Discordant patterns of evolutionary differentiation in two Neotropical treefrogs. Mol Ecol 18: 1375– 1395. Robertson JM, Zamudio KR (2009). Genetic diversification, vicariance, and selection in a polytypic frog. J Hered 100: 715–731.

101

Rodriguez-Herrera B, Chinchilla F a., May-collado L (2004). Lista de especies, endemismo y conservación de los de mamíferos de Costa Rica. Rev Mex Mastozool 6: 19–41. Rodríguez A, Börner M, Pabijan M, Gehara M, Haddad CFB, Vences M (2015). Genetic divergence in tropical anurans: deeper phylogeographic structure in forest specialists and in topographically complex regions. Evol Ecol 29: 765– 785. Rousset F (1997). Genetic differentiation and estimation of gene flow from F- statistics under isolation by distance. Genetics 145: 1219–1228. Savage JM (1966). The origins and history of the Central American herpetofauna. Copeia 1966: 719–766. Savage JM (2002). The Amphibians and Reptiles of Costa Rica: A Herpetofauna Between Two Continents, Between Two Seas. University of Chicago press. Savage J, Bolaños F (2009). A checklist of the amphibians and reptiles of Costa Rica: additions and nomenclatural revisions. Zootaxa 2005: 1–23. Selonen V, Hanski IK, Painter JN (2010). Gene flow and natal dispersal in the Siberian flying squirrel based on direct and indirect data. Conserv Genet 11: 1257–1264. Sexton JP, Hangartner SB, Hoffmann AA (2014). Genetic isolation by environment or distance: Which pattern of gene flow is most common? Evolution 68: 1–15. Shah VB, McRae BH (2008). Circuitscape : a tool for landscape ecology. Proc 7th Python Sci Conf: 62–65. Slatkin M (1993). Isolation by distance in equilibrium and non-equilibrium populations. Evolution 47: 264–279. Solís F, Ibáñez R, Chaves G, Savage J, Jaramillo C, Fuenmayor Q, et al. (2010). Sachatamia albomaculata. IUCN Red List Threat Species 2010. Solórzano A, Cerdas L (2010). A New Subspecies of the Bushmaster , Lachesis muta , from southeastern Costa Rica. J Herpetol 20: 463–466. Stiles G, Remsen J V., McGuire JA (2017). The generic classification of the Trochilini (Aves: Trochilidae): reconciling taxonomy with phylogeny. Zootaxa 4353: 401–424. Tamura K, Stecher G, Peterson D, Filipski A, Kumar S (2013). MEGA6: Molecular evolutionary genetics analysis version 6.0. Mol Biol Evol 30: 2725–2729. Thomas L, Kennington WJ, Stat M, Wilkinson SP, Kool JT, Kendrick GA (2015). Isolation by resistance across a complex coral reef seascape. Proc R Soc B Biol Sci 282: 20151217. Valerio C. (1971). Ability of some tropical tadpoles to survive without water. Copeia 2: 364–365. Vaughan A, Mendelson JR (2007). Taxonomy and ecology of the Central American toads of the genus Crepidophryne (Anura: Bufonidae). Copeia 2007: 304–314.

102

Vences M, Thomas M, van der Meijden A, Chiari Y, Vieites DR (2005). Comparative performance of the 16S rRNA gene in DNA barcoding of amphibians. Front Zool 2: 5. Wang IJ (2013). Examining the full effects of landscape heterogeneity on spatial genetic variation: A multiple matrix regression approach for quantifying geographic and ecological isolation. Evolution 67: 3403–3411. Wang IJ, Bradburd GS (2014). Isolation by environment. Mol Ecol 23: 5649–5662. Wang IJ, Crawford AJ, Bermingham E (2008). Phylogeography of the Pygmy Rain Frog (Pristimantis ridens) across the lowland wet forests of isthmian Central America. Mol Phylogenet Evol 47: 992–1004. Wang IJ, Glor RE, Losos JB (2013). Quantifying the roles of ecology and geography in spatial genetic divergence. Ecol Lett 16: 175–82. Wang Y-H, Yang K-C, Bridgman CL, Lin L-K (2008). Habitat suitability modelling to correlate gene flow with landscape connectivity. Landsc Ecol 23: 989–1000. Webb SD (2006). The Great American Biotic Interchange: patterns and processes. Ann Missouri Bot Gard 93: 245–257. Weigt L a, Crawford AJ, Rand a S, Ryan MJ (2005). Biogeography of the túngara frog, Physalaemus pustulosus: a molecular perspective. Mol Ecol 14: 3857–76. Weir JT, Bermingham E, Schluter D (2009). The Great American Biotic Interchange in birds. Proc Natl Acad Sci U S A 106: 21737–42. Weyl R (1980). Geology of Central America. Gebrueder Borntraeger: Berlin- Stuttgart. Wiens JJ, Ackerly DD, Allen AP, Anacker BL, Buckley LB, Cornell H V, et al. (2010). Niche conservatism as an emerging principle in ecology and conservation biology. Ecol Lett 13: 1310–24. Wiens JJ, Graham CH (2005). Niche Conservatism: integrating evolution, ecology, and conservation biology. Annu Rev Ecol Evol Syst 36: 519–539. Wilson JS, Carril OM, Sipes SD (2014). Revisiting the Great American Biotic Interchange through analyses of amphitropical bees. Ecography 37: 791–796. Wollenberg KC, Vieites DR, Glaw F, Vences M (2011). Speciation in little: the role of range and body size in the diversification of Malagasy mantellid frogs. BMC Evol Biol 11: 217. Wright S (1943). Isolation by Distance. Genetics 28: 114–138.

103

-Supplementary Material-

Idiosyncratic responses to drivers of genetic differentiation in the complex landscapes of lower central america

Ecological Niche Modeling details

Model Evaluation and Tuning

We built a series of candidate models, using MaxEnt algorithm with a variety of user-defined settings and provided multiple evaluation metrics to aid in selecting optimal model settings. We built multiple models for each species varying the values of regularization multipliers from 1 to 5 at 0.5 intervals and testing several feature classes (linear, quadratic, hinge, linear) in all their possible combinations. We created the models using a training and test datasets generated using the block partition method, which divides occurrences into four bins based on the lines of latitude and longitude that divide occurrence localities as equally as possible

(Radosavljevic & Anderson 2014)

Model Selection

Between candidate models, we choose the optimal settings assessing the values of 4 metrics in the following order of priority: Omission rate of Minimum training presence threshold, Area under the curve (AUC) for test data (AUCtest), AUC difference between training and test data (AUCdiff) and Akaike Information Criteria

104

(AICc). The area under the curve based on test data (AUCtest) measures model ability to discriminate conditions at withheld occurrence localities from those at background samples (Radosavljevic & Anderson 2014). AUCdiff quantifies model overfitting by comparing training and test AUC, values for this metric should be high in the case of overfit models (Warren & Seifert, 2011). AICc provides information on the relative quality of a model given the data (Burnham & Anderson

2004; Warren & Seifert 2011).

Variables Used

We eliminated variables with VIF values above 10 and kept the following 11 variables: Mean Diurnal Range of Temperature, Isothermality, Temperature seasonality, Temperature Annual Range, Mean Temperature of Wettest Quarter,

Annual Precipitation, Precipitation of Wettest Month, Precipitation of Driest Month,

Precipitation Seasonality, Precipitation of Warmest Quarter and Precipitation of

Coldest Quarter.

105

Table S1. Details on source, accession, locality and coordinates of each sample included in this study.

Species Source Accession Marker Country Locality Latitude Longitude Agalychnis callidryas GenBank FJ489260 16s Costa Rica Cabo Blanco 9.581 -85.125 Agalychnis callidryas GenBank FJ489261 16s Costa Rica Cabo Blanco 9.581 -85.125 Agalychnis callidryas GenBank FJ489262 16s Costa Rica Cabo Blanco 9.581 -85.125 Agalychnis callidryas GenBank FJ489263 16s Costa Rica Cabo Blanco 9.581 -85.125 Agalychnis callidryas GenBank FJ489264 16s Costa Rica Cahuita 9.719 -82.814 Agalychnis callidryas GenBank FJ489265 16s Costa Rica Cahuita 9.719 -82.814 Agalychnis callidryas GenBank FJ489266 16s Costa Rica Cahuita 9.719 -82.814 Agalychnis callidryas GenBank FJ489276 16s Panama El Cope 8.630 -80.592 Agalychnis callidryas GenBank FJ489277 16s Panama El Cope 8.630 -80.592 Agalychnis callidryas GenBank FJ489278 16s Panama El Cope 8.630 -80.592 Agalychnis callidryas GenBank FJ489279 16s Panama El Cope 8.630 -80.592 Agalychnis callidryas GenBank FJ489280 16s Costa Rica Guacimo 10.237 -83.567 Agalychnis callidryas GenBank FJ489281 16s Costa Rica Guacimo 10.237 -83.567 Agalychnis callidryas GenBank FJ489282 16s Costa Rica Guacimo 10.237 -83.567 Agalychnis callidryas GenBank FJ489283 16s Costa Rica Guacimo 10.237 -83.567 Agalychnis callidryas GenBank FJ489284 16s Panama El Valle 8.630 -80.116 Agalychnis callidryas GenBank FJ489285 16s Panama El Valle 8.630 -80.116 Agalychnis callidryas GenBank FJ489286 16s Panama El Valle 8.630 -80.116 Agalychnis callidryas GenBank FJ489288 16s Panama Gamboa 9.123 -79.693 Agalychnis callidryas GenBank FJ489289 16s Panama Gamboa 9.123 -79.693 Agalychnis callidryas GenBank FJ489290 16s Panama Gamboa 9.123 -79.693 Agalychnis callidryas GenBank FJ489291 16s Costa Rica La Selva 10.433 -84.008 Agalychnis callidryas GenBank FJ489292 16s Costa Rica La Selva 10.433 -84.008 Agalychnis callidryas GenBank FJ489293 16s Costa Rica La Selva 10.433 -84.008 Agalychnis callidryas GenBank FJ489294 16s Costa Rica La Selva 10.433 -84.008 Agalychnis callidryas GenBank FJ489295 16s Costa Rica La Selva 10.433 -84.008 Agalychnis callidryas GenBank FJ489296 16s Costa Rica La Selva 10.433 -84.008

106

Agalychnis callidryas GenBank FJ489297 16s Costa Rica Manzanillo 9.633 -82.656 Agalychnis callidryas GenBank FJ489298 16s Costa Rica Manzanillo 9.633 -82.656 Agalychnis callidryas GenBank FJ489299 16s Costa Rica Manzanillo 9.633 -82.656 Agalychnis callidryas GenBank FJ489301 16s Costa Rica Manzanillo 9.633 -82.656 Agalychnis callidryas GenBank FJ489302 16s Costa Rica Manzanillo 9.633 -82.656 Agalychnis callidryas GenBank FJ489307 16s Costa Rica Bandera 9.519 -84.377 Agalychnis callidryas GenBank FJ489308 16s Costa Rica Bandera 9.519 -84.377 Agalychnis callidryas GenBank FJ489309 16s Costa Rica Bandera 9.519 -84.377 Agalychnis callidryas GenBank FJ489310 16s Costa Rica Bandera 9.519 -84.377 Agalychnis callidryas GenBank FJ489311 16s Costa Rica Bandera 9.519 -84.377 Agalychnis callidryas GenBank FJ489312 16s Costa Rica Bandera 9.519 -84.377 Agalychnis callidryas GenBank FJ489313 16s Costa Rica Bandera 9.519 -84.377 Agalychnis callidryas GenBank FJ489314 16s Costa Rica Bandera 9.519 -84.377 Agalychnis callidryas GenBank FJ489315 16s Panama Santa Fe 8.507 -81.114 Agalychnis callidryas GenBank FJ489316 16s Panama Santa Fe 8.507 -81.114 Agalychnis callidryas GenBank FJ489321 16s Costa Rica Siquirres 8.889 -83.477 Agalychnis callidryas GenBank FJ489322 16s Costa Rica San Ramon 10.234 -84.529 Agalychnis callidryas GenBank FJ489323 16s Costa Rica San Ramon 10.234 -84.529 Agalychnis callidryas GenBank FJ489324 16s Costa Rica Tilaran 10.516 -84.960 Agalychnis callidryas GenBank FJ489325 16s Costa Rica Tilaran 10.516 -84.960 Agalychnis callidryas GenBank FJ489327 16s Costa Rica Uvita 9.124 -83.701 Agalychnis callidryas GenBank FJ489328 16s Costa Rica Uvita 9.124 -83.701 Agalychnis callidryas GenBank FJ489331 16s Costa Rica Uvita 9.124 -83.701 Agalychnis callidryas GenBank FJ489333 16s Costa Rica Uvita 9.124 -83.701 Bolitoglossa lignicolor BoLD BSUCR444 16s Costa Rica S. Isidro de 9.677 -84.076 Dota Bolitoglossa lignicolor BoLD BSUCR441 16s Costa Rica Dominical 9.264 -83.872 Bolitoglossa lignicolor GenBank JX434638 16s Panama Santa Clara 8.830 -82.780 Bolitoglossa lignicolor GenBank JX434639 16s Panama Buenos Aires 8.470 -81.510 Bolitoglossa lignicolor BoLD BSUCR442 16s Costa Rica La Gamba 8.675 -83.203

107

Bolitoglossa lignicolor GenBank JX434640 16s Panama Cerro Hoya 7.320 -80.790 Bolitoglossa lignicolor GenBank JX434643 16s Panama Cerro Hoya 7.320 -80.790 Bolitoglossa lignicolor GenBank JX434641 16s Panama Cerro Hoya 7.320 -80.790 Bolitoglossa lignicolor GenBank JX434642 16s Panama Guacá 8.500 -82.430 Bolitoglossa lignicolor GenBank AF218484 16s Costa Rica Osa 9.150 -83.335 Bolitoglossa lignicolor BoLD BSUCR443 16s Costa Rica Potrero Grande 9.098 -83.113 Diasporus diastema GenBank FJ766809 COI Panama Cocle 8.670 -80.590 Diasporus diastema GenBank FJ766810 COI Panama Cocle 8.670 -80.590 Diasporus diastema GenBank KT186558 COI Panama Fortuna 8.719 -82.232 Diasporus diastema GenBank JN991347 COI Costa Rica San Ramon 10.220 -84.540 Diasporus diastema BoLD BSUCR240 COI Costa Rica Sabalito 8.944 -82.753 Diasporus diastema BoLD BSUCR241 COI Costa Rica Sabalito 8.944 -82.753 Diasporus diastema BoLD BSUCR234 COI Costa Rica Balsa 10.186 -84.508 Diasporus diastema BoLD BSUCR097 COI Costa Rica Balsa 10.186 -84.508 Diasporus diastema BoLD BSUCR223 COI Costa Rica Guayacan 10.050 -83.550 Diasporus diastema BoLD BSUCR218 COI Costa Rica Veragua 9.926 -83.188 Diasporus diastema BoLD BSUCR224 COI Costa Rica Tapanti 9.775 -83.797 Dendropsophus ebraccatus GenBank FJ542181 16s Costa Rica Manzanillo 9.633 -82.654 Dendropsophus ebraccatus GenBank FJ542180 16s Costa Rica Manzanillo 9.633 -82.654 Dendropsophus ebraccatus GenBank FJ542179 16s Costa Rica Manzanillo 9.633 -82.654 Dendropsophus ebraccatus GenBank FJ542195 16s Costa Rica Uvita 9.994 -83.032 Dendropsophus ebraccatus GenBank FJ542194 16s Costa Rica Uvita 9.994 -83.032 Dendropsophus ebraccatus GenBank FJ542184 16s Costa Rica Pavones 8.388 -83.140 Dendropsophus ebraccatus GenBank FJ542197 16s Costa Rica Uvita 9.994 -83.032 Dendropsophus ebraccatus BoLD BSUCR267 16s Costa Rica Golfito 8.651 -83.180 Dendropsophus ebraccatus BoLD BSUCR268 16s Costa Rica Piro 8.411 -83.344 Dendropsophus ebraccatus BoLD BSUCR400 16s Costa Rica Altamira 8.784 -83.019 Dendropsophus ebraccatus GenBank FJ542196 16s Costa Rica Uvita 9.994 -83.032 Dendropsophus ebraccatus GenBank FJ542185 16s Costa Rica Pavones 8.387 -83.140 Dendropsophus ebraccatus GenBank FJ542183 16s Costa Rica Pavones 8.387 -83.140

108

Dendropsophus ebraccatus GenBank FJ542182 16s Costa Rica Pavones 8.387 -83.140 Dendropsophus ebraccatus GenBank FJ542193 16s Costa Rica Siquirres 10.092 -83.515 Dendropsophus ebraccatus GenBank FJ542192 16s Costa Rica Siquirres 10.092 -83.515 Dendropsophus ebraccatus GenBank FJ542191 16s Costa Rica Siquirres 10.092 -83.515 Dendropsophus ebraccatus GenBank FJ542189 16s Costa Rica Bandera 9.523 -84.412 Dendropsophus ebraccatus GenBank FJ542188 16s Costa Rica Bandera 9.523 -84.412 Dendropsophus ebraccatus GenBank FJ542186 16s Costa Rica Bandera 9.523 -84.412 Dendropsophus ebraccatus GenBank FJ542190 16s Costa Rica Bandera 9.523 -84.412 Dendropsophus ebraccatus GenBank FJ542187 16s Costa Rica Bandera 9.523 -84.412 Dendropsophus ebraccatus GenBank FJ542178 16s Costa Rica La Selva 10.430 -84.008 Dendropsophus ebraccatus BoLD BSUCR271 16s Costa Rica Veragua 9.926 -83.188 Espadarana prosoblepon GenBank KR863251 16s Panama Cerro Brewster 9.320 -79.289 Espadarana prosoblepon GenBank KR863250 16s Panama Rio Chagres 9.265 -79.508 Espadarana prosoblepon GenBank KR863246 16s Panama Rio Chagres 9.265 -79.508 Espadarana prosoblepon GenBank KR863235 16s Panama Cerro Brewster 9.320 -79.289 Espadarana prosoblepon GenBank KR863241 16s Panama Cerro Brewster 9.320 -79.289 Espadarana prosoblepon GenBank KR863253 16s Panama Cerro Azul 9.231 -79.403 Espadarana prosoblepon GenBank KR863245 16s Panama Cerro Azul 9.231 -79.403 Espadarana prosoblepon BoLD BSUCR056 16s Costa Rica Balsa 10.189 -84.504 Espadarana prosoblepon BoLD BSUCR055 16s Costa Rica Balsa 10.186 -84.508 Espadarana prosoblepon BoLD BSUCR148 16s Costa Rica Londres de 9.462 -84.063 Quepos Espadarana prosoblepon BoLD BSUCR147 16s Costa Rica Potrero Grande 9.098 -83.113 Espadarana prosoblepon GenBank FJ784362 16s Panama El Cope 8.667 -80.592 Espadarana prosoblepon GenBank FJ784363 16s Panama El Cope 8.667 -80.592 Espadarana prosoblepon BoLD BSUCR054 16s Costa Rica Rodeo 9.904 -84.280 Espadarana prosoblepon GenBank KR863252 16s Panama Cana Station 7.756 -77.684 Espadarana prosoblepon GenBank KR863249 16s Panama Cana Station 7.756 -77.684 Espadarana prosoblepon GenBank KR863248 16s Panama Cana Station 7.756 -77.684 Espadarana prosoblepon GenBank KR863243 16s Panama Rio Cana 7.762 -77.724

109

Espadarana prosoblepon GenBank KR863242 16s Panama Rio Cana 7.762 -77.724 Espadarana prosoblepon GenBank KR863238 16s Panama Rio Cana 7.762 -77.724 Espadarana prosoblepon GenBank KR863237 16s Panama Rio Cana 7.762 -77.724 Espadarana prosoblepon GenBank KR863247 16s Panama Cana Station 7.756 -77.684 Espadarana prosoblepon GenBank KR863244 16s Panama Rio Cana 7.762 -77.724 Espadarana prosoblepon GenBank KR863240 16s Panama Rio Cana 7.762 -77.724 Espadarana prosoblepon GenBank KR863239 16s Panama Rio Cana 7.762 -77.724 Espadarana prosoblepon GenBank KR863234 16s Panama Rio Cana 7.762 -77.724 Lithobates warscewitschii GenBank FJ784384 16s Panama El Cope 8.667 -80.592 Lithobates warscewitschii GenBank KR863272 16s Panama Cerro Brewster 9.320 -79.289 Lithobates warscewitschii GenBank KR863282 16s Panama Cerro Brewster 9.320 -79.289 Lithobates warscewitschii GenBank KR863281 16s Panama Rio Chagres 9.265 -79.508 Lithobates warscewitschii GenBank KR863280 16s Panama Cerro Brewster 9.320 -79.289 Lithobates warscewitschii GenBank KR863279 16s Panama Cerro Brewster 9.320 -79.289 Lithobates warscewitschii GenBank KR863276 16s Panama Rio Chagres 9.265 -79.508 Lithobates warscewitschii GenBank KR863275 16s Panama Rio Chagres 9.265 -79.508 Lithobates warscewitschii GenBank KR863274 16s Panama Cerro Azul 9.231 -79.403 Lithobates warscewitschii GenBank KR863271 16s Panama Cerro Brewster 9.320 -79.289 Lithobates warscewitschii GenBank KR863284 16s Panama Cerro Azul 9.231 -79.403 Lithobates warscewitschii GenBank KR863283 16s Panama Cerro Azul 9.231 -79.403 Lithobates warscewitschii GenBank KR863278 16s Panama Cerro Azul 9.231 -79.403 Lithobates warscewitschii GenBank KR863277 16s Panama Cerro Azul 9.231 -79.403 Lithobates warscewitschii GenBank KR863273 16s Panama Cerro Brewster 9.320 -79.289 Lithobates warscewitschii GenBank FJ784454 16s Panama El Cope 8.667 -80.592 Lithobates warscewitschii GenBank KR911917 16s Panama El Cope 8.667 -80.592 Lithobates warscewitschii GenBank KR911916 16s Panama El Cope 8.667 -80.592 Lithobates warscewitschii GenBank KR911918 16s Panama El Cope 8.667 -80.592 Lithobates warscewitschii BoLD BSUCR366 16s Costa Rica Fila Matama 9.618 -83.283 Lithobates warscewitschii BoLD BSUCR364 16s Costa Rica Fila Matama 9.618 -83.268 Lithobates warscewitschii BoLD BSUCR365 16s Costa Rica Fila Matama 9.618 -83.283

110

Lithobates warscewitschii BoLD BSUCR367 16s Costa Rica Fila Matama 9.605 -83.280 Lithobates warscewitschii BoLD BSUCR129 16s Costa Rica Talamanca 9.357 -83.229 Lithobates warscewitschii BoLD BSUCR382 16s Costa Rica Talamanca 9.357 -83.229 Lithobates warscewitschii BoLD BSUCR372 16s Costa Rica Balsa 10.183 -84.510 Lithobates warscewitschii BoLD BSUCR373 16s Costa Rica Veragua 9.926 -83.188 Lithobates warscewitschii BoLD BSUCR370 16s Costa Rica Potrero Grande 9.102 -83.114 Lithobates warscewitschii BoLD BSUCR378 16s Costa Rica La Gamba 8.679 -83.202 Oedipina alleni BoLD BSUCR464 16s Costa Rica S. Isidro de 9.677 -84.076 Dota Oedipina alleni BoLD BSUCR465 16s Costa Rica Londres 9.462 -84.063 Oedipina alleni GenBank AF199209 16s Costa Rica Cerro Zapote 8.750 -82.983 Oedipina alleni GenBank AF199208 16s Costa Rica Damas 9.462 -84.224 Oedipina alleni GenBank AF199207 16s Costa Rica Sirena 8.481 -83.594 Pristimantis ridens GenBank FJ784399 16s Panama El Cope 8.667 -80.592 Pristimantis ridens GenBank FJ784398 16s Panama El Cope 8.667 -80.592 Pristimantis ridens GenBank JN991466 16s Costa Rica Rio Claro 8.740 -82.960 Pristimantis ridens BoLD BSUCR415 16s Costa Rica Veragua 9.926 -83.188 Pristimantis ridens BoLD BSUCR414 16s Costa Rica Balsa 10.186 -84.508 Pristimantis ridens BoLD BSUCR416 16s Costa Rica S. Isidro de 9.677 -84.076 Dota Pristimantis ridens BoLD BSUCR420 16s Costa Rica Potrero Grande 9.117 -83.097 Pristimantis ridens GenBank KR863320 16s Panama Cerro Azul 9.217 -79.422 Pristimantis ridens GenBank KR863318 16s Panama Cerro Azul 9.231 -79.403 Pristimantis ridens GenBank KR863319 16s Panama Cerro Azul 9.231 -79.403 Pristimantis ridens GenBank JN991465 16s Panama Nusgandi 9.317 -78.983 Pristimantis ridens GenBank KR863317 16s Panama Cerro Brewster 9.290 -79.300 Pristimantis ridens GenBank FJ784389 16s Panama El Cope 8.667 -80.592 Pristimantis ridens GenBank FJ784388 16s Panama El Cope 8.667 -80.592 Rhinella horribilis BoLD BSUCR042 16s Costa Rica Veragua 10.877 -84.329 Rhinella horribilis GenBank DQ415563 16s Costa Rica - 10.453 -84.081

111

Rhinella horribilis BoLD BSUCR123 16s Costa Rica Los Chiles 10.947 -84.725 Rhinella horribilis BoLD BSUCR124 16s Costa Rica Medio Queso 11.032 -84.691 Rhinella horribilis BoLD BSUCR041 16s Costa Rica Veragua 10.871 -84.350 Rhinella horribilis BoLD BSUCR127 16s Costa Rica Crucitas 10.877 -84.329 Rhinella horribilis BoLD BSUCR125 16s Costa Rica Golfito 8.604 -83.170 Rhinella horribilis GenBank FJ784357 16s Panama El Cope 8.667 -80.592 Rhinella horribilis BoLD BSUCR126 16s Costa Rica Veragua 9.926 -83.188 Sachatamia albomaculata BoLD BSUCR159 16S Costa Rica Veragua 9.926 -83.188 Sachatamia albomaculata BoLD BSUCR157 16S Costa Rica Talamanca 9.618 -83.268 Sachatamia albomaculata BoLD BSUCR156 16S Costa Rica La Tirimbina 10.402 -84.108 Sachatamia albomaculata BoLD BSUCR158 16S Costa Rica Londres 9.462 -84.063 Sachatamia albomaculata BoLD BSUCR059 16S Costa Rica Rodeo 9.904 -84.280 Sachatamia albomaculata GenBank FJ784392 16S Panama El Cope 8.667 -80.592 Sachatamia albomaculata GenBank FJ784550 16S Panama El Cope 8.667 -80.592 Sachatamia albomaculata GenBank FJ784441 16S Panama El Cope 8.667 -80.592 Sachatamia albomaculata GenBank FJ784474 16S Panama El Cope 8.667 -80.592 Sachatamia albomaculata GenBank FJ784468 16S Panama El Cope 8.667 -80.592 Sachatamia albomaculata GenBank KR863349 16S Panama Cerro Azul 9.231 -79.403 Sachatamia albomaculata GenBank KR863347 16S Panama Cerro Azul 9.231 -79.403 Sachatamia albomaculata GenBank KR863346 16S Panama Cerro Azul 9.231 -79.403 Sachatamia albomaculata GenBank KR863348 16S Panama Cerro Azul 9.231 -79.403 Sachatamia albomaculata BoLD BSUCR158 COI Costa Rica Londres 9.462 -84.063 Sachatamia albomaculata BoLD BSUCR160 COI Costa Rica Potrero Grande 9.098 -83.113 Sachatamia albomaculata BoLD BSUCR156 COI Costa Rica La Tirimbina 10.402 -84.108 Sachatamia albomaculata BoLD BSUCR157 COI Costa Rica Talamanca 9.618 -83.268 Sachatamia albomaculata BoLD BSUCR159 COI Costa Rica Veragua 9.926 -83.188 Sachatamia albomaculata GenBank KR863092 COI Panama Cerro Azul 9.231 -79.403 Sachatamia albomaculata GenBank KR863091 COI Panama Cerro Azul 9.231 -79.403 Sachatamia albomaculata GenBank KR863090 COI Panama Cerro Azul 9.231 -79.403 Sachatamia albomaculata GenBank FJ766595 COI Panama Cocle 8.670 -80.590

112

Sachatamia albomaculata GenBank FJ766594 COI Panama Cocle 8.670 -80.590 Sachatamia albomaculata GenBank FJ766598 COI Panama Cocle 8.670 -80.590 Sachatamia albomaculata GenBank FJ766596 COI Panama Cocle 8.670 -80.590 Sachatamia albomaculata GenBank FJ766599 COI Panama Cocle 8.670 -80.590 Sachatamia albomaculata GenBank KR863089 COI Panama Cerro Azul 9.231 -79.403 Smilisca phaeota GenBank AY326040 16s Costa Rica La Lola 10.100 -83.383 Smilisca phaeota BoLD BSUCR321 16s Costa Rica Veragua 9.926 -83.188 Smilisca phaeota BoLD BSUCR318 16s Costa Rica Balsa 10.189 -84.509 Smilisca phaeota GenBank FJ784433 16s Panama El Cope 8.667 -80.592 Smilisca phaeota GenBank FJ784413 16s Panama El Cope 8.667 -80.592 Smilisca phaeota BoLD BSUCR317 16s Costa Rica Piro 8.411 -83.344 Smilisca phaeota BoLD BSUCR320 16s Costa Rica S. Isidro de 9.677 -84.076 Dota Smilisca phaeota BoLD BSUCR311 16s Costa Rica Crucitas 10.868 -84.345 Smilisca phaeota BoLD BSUCR319 16s Costa Rica Londres 9.462 -84.063

113

CAPÍTULO III***

The role of geography, topography and climate in the acoustic divergence of Neotropical Diasporus frogs

114

The role of geography, topography and climate in the acoustic divergence of Neotropical Diasporus frogs

Adrián García-Rodríguez1,2, Marcelo Araya-Salas3, & Gabriel C. Costa4

1Departamento de Ecologia, Universidade Federal do Rio Grande do Norte,

Natal - RN, Brasil, 59078-900

2Escuela de Biología, Universidad de Costa Rica, San Pedro, 11501-2060 San

José, Costa Rica.

3Laboratory of Ornithology, Cornell University, 159 Sapsucker Woods Road,

Ithaca, New York 14850, USA

4Department of Biology, Auburn University at Montgomery, Montgomery AL

36124.

Correspondence: [email protected]

115

Abstract Background. Acoustic communication is central in the biology of most anuran species. Geographic variation in acoustic signals can led to failure in conspecific recognition, becoming an important mechanism of reproductive isolation and a primer of speciation. Here we documented patterns of acoustic variation among 21 populations of two species of Diasporus frogs. Then, we assessed whether geographic distance, topography, connectivity of suitable habitats and local climate have a role in shaping those patterns. Results. We found deep acoustic divergence in both species. Pacific populations of D. diastema vocalize at lower frequencies than Caribbean populations. Males of D. hylaeformis from Tapantí, showed striking differences in call duration. Topography explained ~30% of the deviance in the acoustic divergence of D.diastema. In D. hylaeformis our model was not able to explain acoustic variation, although, we found a signature of association with isolation by environment and isolation by topography. Conclusions. Most abiotic factors tested here clearly promote isolation among populations due to the complexity of the Costa Rican landscapes. However, only topography –significantly- and climatic dissimilarity –marginally explained patterns of acoustic divergence in these species. Considering the high levels of acoustic variation detected, we conclude that signal evolution in this case is likely determined by a combination of mechanisms operating independently in local scales on isolated populations such as sexual selection, character displacement or genetic drift.

Keywords: Advertisement calls, isolation by distance, isolation by resistance, isolation by environment, generalized dissimilarity modeling

116

Background

In a broad range of taxonomic groups, from insects to mammals, communication is essentially mediated by acoustic signals [1]. These signals have evolved to efficiently transfer information that is encoded by a sender and then is decoded by a receiver [2,3]. However, for this process to be efficient signals must be first recognized and then interpreted on the basis of their spectro-temporal features [4,5]. For organisms depending on acoustic communication for mating purposes, these signals are crucial to discriminate between conspecifics and individuals of other species [6]. This process is possible due to the evolution of highly stereotyped call characteristics that assures receivers the recognition at species level [7,8]. Acoustic signals in this context are even more relevant for organisms that attract mates over relatively long distances [9–11] or interact in complex environments with limited visibility [12,13]. This is the case of many anurans, a group where the vast majority of species are active at night and many reproduces in noisy settings

[14].

Anuran advertisement calls, produced by males in reproductive contexts, must be highly stereotype to minimize energetic costs associated to non-viable crosses [15,16]. Since failure in recognition during mating affects reproductive success, divergence on advertisement calls has been proposed as a pre-mating

117

isolation mechanism [17,18]. Hence, call divergence, has evolutionary implications on population differentiation and may ultimately play a relevant role on speciation process [19–21]. Although several studies have documented intraspecific divergence in anuran advertisement calls, fewer efforts have been oriented to understand what factors promote such patterns of variation [22–25].

Sexual selection, reproductive character displacement, genetic drift and ecological selection, haven been hypothesized as the most important mechanisms driving divergence in the evolution of anuran calls [21,26–28]. In the case of sexual selection, evidence suggest female preferences can select for acoustic parameters that have impact on mating success [29]. Under a scenario of reproductive character displacement, acoustic signals are expected to evolve in order to minimize costs of interspecific competition or maladaptive hybridization when in sympatry with closely related taxa [30]. Under a scenario of genetic drift, is expected that populations breeding further apart have more distinct genomes and therefore more distinct phenotypes that could influence acoustic traits [31]. It has been also proposed that pressures set by ecological factors are responsible of shaping patterns of geographic call divergence [32] as predicted by the Acoustic Adaptation hypothesis (AAH), which expects that structural differences among habitat influences signal evolution through the constraints of signal transmission [33]. A way to understand whether ecologica

118

factors have promoted acoustic divergence, is by evaluating the relationship between geographic and/or environmental variables and geographic patterns of call variation.

Several hypotheses has been proposed in other fields to evaluate causal relation between environmental variables and a given response variable, as is molecular divergence in landscape genetics studies [34,35]. For example, isolation by distance (IBD), states that genetic differentiation is expected to increase with geographic distance due to restricted dispersal among populations

[36,37]. Isolation by resistance (IBR), goes a step further and evaluates how the friction imposed by landscape components such as topography or habitat suitability influence genetic variation [38]. Isolation by Environment (IBE), in contrast, has been recently proposed to describe a pattern in which genetic differences increase with the environmental differences between sampling localities, independently of geographic distance or the resistance imposed by the landscape that connects populations [39].

In this study, we used such reasoning to test whether isolation promoted by different landscape features can explain observed patterns of divergence in acoustic traits within species. To this end, we first documented geographic call variation in two direct-developing frog species of the genus Diasporus, occurring across the complex landscapes of Isthmian Central America. Then,

119

we explored the potential causes of such variation patterns by assessing the role of distance, topography, connectivity of suitable habitats and influence of local climate for each species. To quantify the effect of these factors in shaping acoustic divergence we tested hypotheses and predictions diagramed in figure

1. By assessing acoustic variation within species and its potential drivers we aim to reach a better understanding of the mechanisms promoting divergence and early phases of speciation.

Figure 1. Schematic representation of hypotheses tested in this study and their respective predictions

120

Methods Study Species. Five of the eleven species in the genus Diasporus occur in Costa

Rica [40]. D. diastema is distributed in both slopes of the main cordilleras of the country, from sea level to above 1500 meters elevation, with exception of the dry forest, D. hylaeformis, can be found above 1300 meters in humid lower montane forests in the cordilleras of Talamanca, Volcanica Central, Tilarán and

Guanacaste [41] and D. vocator has been documented on the southwestern portion of the country from 2 to 1600 m (Savage, 2002; A. García-Rodríguez, pers. observ.). The other two species have restricted distributions: D. tigrillo is known only from the valley of the Río Lari, Limón, Province (250-440) [41,42] and D. ventrimaculatus is endemic to a highland valley (2500) in Cordillera de

Talamanca known as Valle del Silencio [43]. All species are abundant and highly vocal, call from perches hidden into dense vegetation from where produce calls characterized by short duration and high frequencies, especially during the rainy season [41,42,44]. We focused our analyses on D. diastema and D. hylaeformis, the two species with wider distributions in Costa Rica. The wider geographic spread of these species allow us to document the acoustic characteristics of several populations in order to assess geographic variation and test for potential factors (Fig.1) promoting those patterns across complex landscapes.

121

Field Work. We documented advertisement calls of our study species all over

Costa Rica. We recorded calling males at 23 different localities across all the regions covered by the distribution of the genus in the country, at elevations from the sea level to near 2800 m altitude (Fig. 2). It includes the Caribbean and the Pacific lowlands as well as the four main NW-trending volcanic cordilleras that bisects the country. Calls were recorded with a Sennheiser ME66 shotgun microphone and a Marantz PMD-660 digital recorder (.wav file format; 44.1

Khz sampling rate and 16 bit accuracy), a SONY TCM-500EV analogic recorder (La Selva and La Guaria sites) or a SONY TCM-150 analogic recorder

(Volcán Tenorio and Fortuna sites). We digitalized these analogic recordings using the software Adobe Audition at a 44.1 kHz sample rate and sample size of 16 bits.

We recorded all males by positioning the microphone at less than 2 meters of distance. In order to avoid replicated recordings of the same male, at each locality we walked linearly while searching for calling individuals. When collected, voucher specimens were deposited at the collection of Herpetology of the Museo de Zoología, Universidad de Costa Rica (MZUCR).

122

Figure 2. Localities sampled for the 2 species of Diasporus analyzed in this study. Acoustic Analyses. For each species, we identified individual calls in recordings through an automatic detection procedure, based on amplitude and duration thresholds applied within the frequency range of the species. From the calls detected in each recording we selected those with the highest signal-to- noise ratio, for a maximum of 8 calls per recording. We visually inspected the spectrograms of the detected calls, removed undesired sounds or calls that overlapped in time with other sounds and manually adjusted time and frequency of selections when necessary using.

123

We analyzed calls from audio recordings belonging to 170 males (D. diastema = 98, D. hylaeformis = 72). For each male we analyzed a mean of

6.7±1.3 calls, for a total of 1241 calls evaluated. We measured 26 acoustic parameters (detailed in table S1). Acoustic parameters were BoxCox- transformed to improved normality. We also removed collinear variables, identified as those with a mean-absolute Pearson correlation coefficient higher than 0.95. We used unsupervised random forest (RF) [45] on the acoustic parameters to identify underlying acoustic structure within species and to evaluate the relative contribution of the parameters to overall acoustic differences. Briefly, this method randomizes the data to create a new synthetic data set [46]. These two data sets (“original” and “synthetic”) are then used in a two-class classification model using supervised RF, which creates multiple decision trees using different parameter subsets to discriminate between the classes [45]. This predictive model generates a proximity matrix, which contains counts of times in which each pair of items are found in the same tree node. Hence, this matrix represents the similarity between calls as more similar calls are expected to be together more often across trees. Thus, we converted it to a distance matrix by subtracting it from 1 and used it in subsequent analysis as a measure of pairwise acoustic dissimilarity. Individual models were built for each species. Distance matrices were calculated between all calls, between

124

individuals and between sites (based on average pairwise similarities between individuals and sites, respectively). This procedure was run separately for each species. All acoustic analyses were done using the packages tuneR [47],

Seewave [48], and warbler [49].

Potential Abiotic Drivers of Acoustic Variation. We tested whether patterns of IBD, IBR or IBE, are able to explain variation in acoustic structure at the intra-specific level in our study species. To this end we used coordinates recorded in almost all the study sites using a GPS Garmin 63st. In few cases, sites were georeferenced a posteriori using the software QGIS 2.16.3. We used such geographic information to create pairwise distance matrices among localities. To estimate IBD, we calculated euclidean distances in kilometers among localities of the same species using the R package raster [50].

To estimate IBR, we considered two factors acting as potential barriers between localities: topography (IBRt) and poor climatic suitability (IBRs). We estimated resistance distances using such factors as independent friction layers based on a circuit-theory approach conducted in the program Circuitscape V.4.0

[51]. To calculate topographic cost distances among localities, we first generated a friction layer that accounts for topographic complexity. To this, we used a layer of elevation at 30-second resolution (~1km at the equator, http://www.worldclim.org/) and calculated the standard deviation of differences

125

between 5x5 adjacent elevations using the R package raster [50]. This procedure has been demonstrated to more accurately represent topographic roughness than elevation range, which only indicates the strength of a gradient within a cell

[52].

To estimate distances mediated by climatic suitability (IBRs) among localities we first generated distribution models for each species. To create them, we gathered occurrence points, for each species, from the Global

Biodiversity Information Facility and the Herpetology Collection held at Museo de Zoología at the University of Costa Rica. Then we projected and clean those occurences for georreferencing errors by visualizing geographic outliers and excluding points outside the known altitudinal range for each species. We used as predictors the 19 bioclimatic variables available for current conditions at www.worldclim.org at a resolution of 30 arc seconds (~1km at the Equator). For each species we ran 52 candidate models using a Maximum Entropy algorithm

[53,54] with different parametrizations, varying regularization multipliers and features classes [55]. All these procedures were conducted in R using the packages ENMeval, dismo and raster [50,56,57]. We determined the best models as those having a combination of the highest mean AUC values and the lowest mean omission rate of the 10th percentile. Finally, we used as friction

126

surface the inverse of suitability derived from the selected ENM’s for each species.

To test for IBE, we estimated environmental dissimilarity among locality pairs, independently of the climatic conditions in the areas that separate them.

For each sampled locality, we extracted the values for the 19 bioclimatic variables available in the Worldclim data set (worldclim.org) at 30 arcseconds resolution (~1km at the equator). Then, we estimated Euclidean pairwise distances in the multidimensional space using the function dist in R, which computes specific distances between the rows of a multivariate matrix. All data used in this calculation is derived from temperature and precipitation means

[58], which are variables known to have a direct or indirect influence (i.e humidity) on sound transmission [59]. Then, we consider this metric a proxy to test the hypothesis of acoustic adaptation.

Generalized Dissimilarity Modelling. Using the acoustic distances between localities as response matrix and the IBD, IBRt, IBRs and IBE matrices as predictors we conducted a Generalized Dissimilarity Modelling (GDM). GDM, is a distance-based statistical approach that uses regression techniques to relate geographic or environmental distances and dissimilarities in a biological trait between sites [60]. Contrary to other distance-based approaches, such as

Mantel, Partial Mantel and Multiple Matrix of Regression with Randomization

127

(MMRR), GDM is able to fit the nonlinear responses, commonly encountered in ecological datasets [61]. This technique deals with nonlinear relationships of environmental variables to biological variation by using I-spline basis functions

(more details in [60]). Such splines plots provide insights into the total magnitude of biological change as a function of each gradient and where those changes are most pronounced along each gradient [62].

Results

Acoustic variation. From the supervised Random Forest analysis on the data set containing both species we obtained a test data error rate of 0.1650

(e.g.83.5% of the calls correctly classified, ntree=10000; mtry=5, accuracy p- value<0.0001). The most important variables contributing to this classification were: spectral flatness, mean frequency, duration and standard deviation of frequency.

At the intra-specific level for D. diastema we obtained a classification accuracy of 88.04% on the test data set (ntree=10000; mtry=5; accuracy p- value<0.0001). In this case, the most important variables were mean frequency, spectral flatness, standard deviation of frequency and the interquartile frequency range (Fig 3). In D. hylaeformis, the classification accuracy of the RF analyses was of 94.65% on the test data and the most important variables for

128

this classification were mean frequency, duration, spectral flatness and minimum dominant frequency (Fig 4).

Isolation levels among populations. After estimating IBD matrices, we found geographic linear distances, among localities, ranging between ~1 (La

Selva- La Guaria) and 314 km (Piro-Volcan Tenorio) for D. diastema and between seven (Queverí-Tapantí) and 113 km (Tapantí-Monteverde) for the highland frog D. hylaeformis. In the case of IBR based on topographic features, we confirmed from the resulting connectivity voltage maps, that among studied populations of D. diastema the most evident geographic barriers are the NW- trending volcanic cordilleras separating the Caribbean and Pacific lowlands.

129

Figure 3. Variation in the four acoustic variables that most contributed in the discrimination among of calls of Diasporus diastema among sites: mean frequency; spectral flatness, standard deviation of frequency, interquartile of frequency range .

130

Figure 4. Variation in the four acoustic variables that most contributed in the discrimination among of calls of Diasporus hylaeformis among sites: mean frequency, call duration, spectral flatness and minimum dominant frequency.

131

For D. hylaeformis instead, the intermontane valleys, below 1300 m represent a major barrier that limits connectivity, especially among the sites sampled for this species. The most evident barrier separates the locality we sampled at

Cordillera de Tilaran and the rest of localities sampled within the Cordillera de

Talamanca-Cordillera Central complex. In terms of bioclimatic suitability

(IBRs), this separation is not evident, and all areas between localities of D. hylaeformis seems to be accessible when accounting for this variable as a resistance factor (Fig 5).

Figure 5. IBR among sites based on topographic barriers and bioclimatically unsuitable regions. A and B correspond to isolation by habitat suitability and topography, for D. diastema, respectively. C and D represent the same variables for D. hylaeformis.

132

Climatic dissimilarity among studied sites (IBE) for D. diastema was higher between Agua Buena in the southwestern (South Pacific) and Cahuita in the northwestern (South Caribbean) portion of the country. As expected due to their geographic proximity, the most similar pair of sites for this species was La

Selva-Guaria. In D. hylaeformis, local climatic conditions are more different between Dantas and Tapantí; and more similar between Montserrat and Volcán

Poás, although this two localities are not the closer ones.

Drivers of acoustic divergence. We fitted a generalized dissimilarity model

(GDM) of the acoustic divergence as a function of the different types of isolation. The GDM explained 30.68% of the deviance the in acoustic divergence of D.diastema (full model p-value< 0.001). In terms of variable importance, for D. diastema isolation by topography and isolation by distance were the two variables that contributed in explaining acoustic divergence (Fig.

6a). For D. hylaeformis the full model was not able to significantly explain acoustic variation (full model p-value=0.162), although, we obtained a signature of association with isolation by environment and isolation by topography (Fig. 6d). The shape of the I-splines of all these variables were all similar, depicting increases in divergence with the increase in the degree of isolation (Fig. 6). However, in D. diastema the rate of change in acoustic traits varies exponentially as topographic isolation increases, while the rate of change

133

associated to IBD in this species and to IBE in D. hylaeformis showed a logarithmic growth.

134

Figure 6. Variable importance and generalized dissimilarity model-fitted I-splines for variables associated with acoustic divergence in D. diastema (A-C) and D. hylaeformis (D- F). The maximum height reached by each curve, indicates the total amount of divergence associated with the respective variable, holding all other variables constant. The shape of each function shows how the rate of acoustic divergence varies along the tested gradients.

135

Discussion

Our study provide the first quantitative evidence of intra-specific acoustic variation in the direct-developing frogs Diasporus diastema and D. hylaeformis. High classificatory power of our RF algorithm (>86% in D. diastema and ~95% in D. hylaeformis) reveals deep acoustic intra-specific divergences among populations of each species. Lower discrimination power in D. diastema could be affected by the potential inclusion of meta-populations from near localities that may obscured optimal discrimination. Higher discrimination power in D. hylaeformis results from deeper mean divergences among localities.

For D. diastema we found that calls tend to be more similar among localities within the Pacific and within the Caribbean, than between versants. The most variable feature in the calls of this species was mean frequency, with a trend of

Pacific populations to vocalize at lower frequencies. In most anurans, frequency is constrained by body size, larger males, are expected to have vocal cords of larger mass, making them able to produce lower frequencies. Although out of the scope of our study, we have evidence that males from Pacific tend to be bigger than those from the Caribbean (García-Rodríguez, unpublished data). In the case of D. hylaeformis, we did not found a clear geographic pattern of variation, nevertheless, calls from Tapantí were notably different among all sampled sites. Males from this locality produces remarkably long calls, with more than twice the duration of all

136

other populations. In several species, longer call duration has been proposed as an indicator of genetic quality [63,64], such evidence support the idea that females are enabled to choose males of higher quality thorough their mating preferences (i.e good genes hypothesis [65]) and highlights the importance of biotic pressures on the evolution of calls. Certainly, is not easy to explain why only females from that locality may specifically select for that trait. Then, habitat particularities affecting signal transmission such as vegetation structure, noise sources or even the composition of the local soundscape may have influenced this striking variation.

Intra-specific geographic variation in acoustic signals as documented here, has been long reported in anurans (reviewed by [66]). Through time it also has become a topic of interest in evolutionary research due to its potential as a mechanism of reproductive isolation and incipient speciation [67]. Ever since such patterns of variation in communication systems can provide insights on drivers of evolutionary processes, we evaluated the potential role of several abiotic forces (e.g. distance, topography, habitat suitability and climatic dissimilarity) in shaping them. Our results showed that irregular topographies likely influenced divergence patterns in the advertisement calls of D. diastema. Contrary, none of the factors tested significantly explained acoustic variation within D. hylaeformis, although climatic dissimilarity among sites showed an association with acoustic divergence.

137

In D. diastema, detected differences between versants highlights the role of topographic barriers. This pattern is expectable, considering that this species rarely occur above 1500 m altitude [41], the cordilleras of the country may represent a major vicariant barrier. For D. diastema and other lowland species distributed in both Pacific and Caribbean versants, IBRt should be more evident in the eastern portion of the country, where Cordillera de Talamanca reaches elevations above

3500 meters [68]. Talamanca has a dynamic geological history with an uplift rate of approximately 1km/1Ma [69] that has turned into the most evident physical barrier for lowland organisms in Costa Rica, affecting for example, phylogeographic patterns in several taxa [reviewed in 72]. In the eastern portion of the country

(Cordillera de Tilaran and Guanacaste), mountain passes are lower, however influenced by dry conditions from the Northern Pacific [68]. It may be the factor limiting dispersal between versants in this region, maintaining between versant isolation and indirectly increasing the effect of the mountainous barriers.

In D. hylaeformis, patterns of acoustic variation could be better explained by local biotic pressures. For example, local female choices based on call duration may act as a directional pressure selecting and fixing longer calls in D.hyalaeformis from

Tapantí. Interestingly, in our dataset, Tapantí is the only locality where we have found sympatric populations of D. diastema and D.hylaeformis, with males calling at less than 1km from the other species (García-Rodríguez, unpublished data).

138

Hence, another possibility is that substantial differences in the call of D. hylaeformis there, could be driven by the syntopic occurrence of both species as predicted by pre-reproductive character displacement: mate attracting signals differ more in sympatric than in allopatric areas of closely related species [71]. Such shifts are hypothesized to arise in order to minimize agonistic hetero-specific interactions or the production of costly hybrids, however evidence of this phenomena associated with acoustic traits are still scarce [31]. While the focus of our study was to assess the potential role of abiotic factors drivers of isolation in explaining acoustic divergence, the above cited factors could be understand as biotic mechanisms driven by interactions among individuals and should be addressed more deeply in future studies with these species. Finally although not statistically significant, we recovered a signal of correlation between climatic dissimilarity and acoustic divergence. It also could give insights of local pressures such as climate or vegetation structure shaping call structure according to the acoustic adaptation hypothesis [33].

Previous studies testing the influence of IBD on acoustic divergence, including some conducted in the same landscapes of Costa Rica, have found correlations between linear distance and acoustic dissimilarity in birds [72–74], frogs [22,75,76] and rodents [59]. In some cases as in the Andean frog Colostethus palmatus [77] or the singing mice Scotinomys teguina [59], these patterns are also explained by genetic distances, suggesting an important contribution of genetic drift

139

in shaping patterns of acoustic variation. Contrary, in the strawberry poison frog

Dendrobates pumilio and the tree frog Hyla leucophyllata, among population call variation follows a geographic cline without a strict acoustic separation between genetic groups [22,78]. It reflects that acoustic divergence is not necessarily predicted by genetic divergence or at least, that genes used in those studies evolved slower genes underlying sexual signals [22]. We found a small contribution of IBD in the GDM of D. diastema, in our opinion, IBR based on topographic complexity which better explained acoustic variation, is a more robust metric to describe patterns of isolation in the irregular landscapes of Costa Rica. In this case it also likely reflects genetic drift as the mechanism leading to major acustic differentiation between Pacific and Caribbean populations.

Conclusions

Most abiotic factors tested here, failed in significantly explain patterns of acoustic divergence in our study species, even though they clearly promoted patterns of isolation among populations due to the topographic and climatic heterogeneity of the Costa Rican landscapes. Then, given the high levels of acoustic variation detected, we conclude that signal evolution in this case is likely determined by a combination of mechanisms like sexual selection [e.g. 22,81,82], character displacement [e.g 73,83–85] or genetic drift [e.g 23,29,76,77] operating independently in local scales on isolated populations. Data presented here comes

140

from a genus with a complex taxonomy due to the potential existence of cryptic diversity [41,44,84]. A recently published molecular analysis for this group supported the hypothesis of masked cryptic diversity within the genus and highlighted the potential use of bioacoustics as a powerful approach to solve this issue [44]. Information on acoustic divergence have been successfully incorporated as a complementary source of evidence in the field of integrative taxonomy

[23,42,85,86]. In many anuran groups, the definition of species boundaries based merely on morphological approaches becomes a hard task due to their highly conserved morphology [87,88]. In these cases, bioacoustic approaches has proved to be useful and has helped disentangling the taxonomy of cryptic complexes where many species were masked under one name [88–90]. From that perspective, our study adds valuable information for future studies concerned with addressing such questions.

Acknowledgements

We are profoundly grateful to all the people involved in the fieldwork conducted to obtain the recordings analyzed in this study: Víctor Acosta, Kathia Alfaro,

Esmeralda Arévalo, Erick Arias, Gilbert Barrantes, Eliana Faria de Oliveira,

Francisco Fonseca, Sofía Granados, Castiele Holanda Bezerra, Brian Kubicki,

141

Daniela Masís, Francesca Protti, Sofía Rodríguez, Luis Sandoval, Vinicius São

Pedro, Rodolfo Vargas, Beatriz Willink and Héctor Zumbado. Branko Hilje and

Mark Wainwright provided recordings from La Selva and La Guaria, and

Monteverde, respectively. AGR thanks Federico Bolaños, Gilbert Barrantes and

Gerardo Chaves for initial discussions on this topic.

Funding

This research was partially funded by the National Geographic Society [grant number W-346-14]. AGR was supported by Coordenação de Aperfeiçoamento de

Pessoal de Nível Superior, Brazil.

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

AGR and GCC designed the study; AGR collected data; AGR and MAS analyzed data; AGR draft the manuscript, subsequently improved by all co-authors.

References

1. Whitlow WL. Animal bioacoustics. J. Acoust. Soc. Am. 1999;106:1204.

142

2. Marler P. Animal communication signals: we are beginning to understand how the structure of animal signals relates to the function they serve. Science (80-. ).

[Internet]. 1967;157:769–74. Available from: http://www.sciencemag.org/content/157/3790/769.abstract

3. Seyfarth RM, Cheney DL, Bergman T, Fischer J, Zuberbühler K,

Hammerschmidt K. The central importance of information in studies of animal communication. Anim. Behav. 2010;80:3–8.

4. Ryan MJ. Species recognition and sexual selection as a unitary problem in animal communication. Evolution (N. Y). 1993;47:647–57.

5. Ord TJ, King L, Young AR. Contrasting theory with the empirical data of species recognition. Evolution (N. Y). 2011;65:2572–91.

6. Okamoto KW, Grether GF. The evolution of species recognition in competitive and mating contexts: the relative efficacy of alternative mechanisms of character displacement. Ecol. Lett. 2013;16:670–8.

7. Schwartz JJ. The importance of spectral and temporal properties in species and call recognition in a neotropical treefrog with a complex vocal repertoire. Anim.

Behav. 1987;35:340–7.

8. Ryan MJ, Phelps SM, Rand AS. How evolutionary history shapes recognition

143

mechanisms. Trends Cogn. Sci. 2001;5:143–8.

9. Ryan M, Kime N. Selection on long-distance acoustic signals. In: Simmons AM,

Popper AN FR, editor. Acoust. Commun. New York: Springer-Verlag; 2003. p.

225–74.

10. Michelsen A, Nocke H. Biophysical aspects of sound communication in insects. Adv. In Insect Phys. 1974;10:247–96.

11. Oliveira DAG, Ades C. Long-distance calls in Neotropical primates. An. Acad.

Bras. Cienc. 2004. p. 393–8.

12. Haddad CFB, Giaretta AA. Visual and acoustic communication in the brazilian torrent frog, Hylodes asper (Anura: Leptodactylidae). Herpetologica. 1999;55:324–

33.

13. Kime NM. The transmission of advertisement calls in Central American frogs.

Behav. Ecol. 2000;11:71–83.

14. Wells KD. The ecology & behavior of Amphibians. Chicago and London: The

University of Chicago Press; 2008.

15. Kelley DB. Vocal communication in frogs. Curr. Opin. Neurobiol. 2004. p.

751–7.

16. Duellman WE, Trueb L. Biology of Amphibians. Second. Baltimore,

144

Maryland: Johns Hopkins University Press; 1994.

17. Littlejohn MJ. Premating isolation in the Hyla ewingi complex Anura:

Hylidae). Evolution (N. Y). 1965;19:234–43.

18. Blair WF. Mating call and stage of speciation in the Microhyla olivacea–M. carolinensis complex. Evolution (N. Y). 1955;9:469–480.

19. Hoskin CJ, Higgie M. Speciation via species interactions: the divergence of mating traits within species. Ecol. Lett. 2010;13:409–20.

20. Funk WC, Murphy M., Hoke KL, Muths SM, Amburgey, Lemmon EM, et al.

Elevational speciation in action? Restricted gene flow associated with adaptive divergence across an altitudinal gradient. J. Evol. Biol. 2016;29:241–52.

21. Boul KE, Funk WC, Darst CR, Cannatella DC, Ryan MJ. Sexual selection drives speciation in an Amazonian frog. Proc. R. Soc. B [Internet]. 2007 [cited

2011 Jul 23];274:399–406. Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1702375&tool=pmcent rez&rendertype=abstract

22. Pröhl H, Hagemann S, Karsch J, Höbel G. Geographic variation in male sexual signals in strawberry poison frogs (Dendrobates pumilio). Ethology [Internet].

2007 [cited 2011 Sep 19];113:825–37. Available from:

145

http://doi.wiley.com/10.1111/j.1439-0310.2007.01396.x

23. Heyer WR, García-Lopez JM, Cardoso AJ. Advertisement call variation in the

Leptodactylus mystaceus species complex (Amphibia: Leptodactylidae) with a description of a new sibling species. Amphibia-Reptilia. 1996;17:7–31.

24. Asquith A, Altig R, Zimba P. Geographic variation in the mating call of the green treefrog Hyla cinerea. Am. Midl. Nat. [Internet]. 1988;119:101–10.

Available from: http://links.jstor.org/sici?sici=0003-

0031(198801)119:1%3C101:GVITMC%3E2.0.CO;2-M

25. Hasegawa Y, Ueda H, Sumida M. Clinal geographic variation in the advertisement call of the wrinkled frog, Rana rugosa. Herpetologica.

1999;55:318–24.

26. Hoskin CJ, Higgie M, Mcdonald KR, Moritz C. Reinforcement drives rapid allopatric speciation. Nature. 2005;437:10–3.

27. Röhr DL, Paterno GB, Camurugi F, Juncá FA, Garda AA. Background noise as a selective pressure: stream-breeding anurans call at higher frequencies. Org.

Divers. Evol. 2016;16:269–73.

28. Lee KH, Shaner PJL, Lin YP, Lin SM. Geographic variation in advertisement calls of a microhylid frog - testing the role of drift and ecology. Ecol. Evol.

146

2016;6:3289–98.

29. Ritchie MG. The shape of female mating preferences. Proc. Natl. Acad. Sci. U.

S. A. 1996;93:14628–31.

30. Lemmon EM. Diversification of conspecific signals in sympatry: geographic overlap drives multidimensional reproductive character displacement in frogs.

Evolution. 2009;63:1155–70.

31. Wilkins MR, Scordato ESC, Semenov GA, Karaardiç H, Shizuka D, Rubtsov

A, et al. Global song divergence in barn swallows (Hirundo rustica ): exploring the roles of genetic , geographical and climatic distance in sympatry and allopatry.

Biol. J. Linn. Soc. 2018;1–25.

32. Wilkins MR, Seddon N, Safran RJ. Evolutionary divergence in acoustic signals: causes and consequences. Trends Ecol. Evol. 2013;28:156–66.

33. Morton ES. Ecological sources of selection on avian sounds. Am. Nat.

1975;109:17–34.

34. Spear SF, Balkenhol N, Fortin M-J, McRae BH, Scribner K. Use of resistance surfaces for landscape genetic studies: considerations for parameterization and analysis. Mol. Ecol. 2010;19:3576–91.

35. Manel S, Holderegger R. Ten years of landscape genetics. Trends Ecol. Evol.

147

2013. p. 614–21.

36. Wright S. Isolation by Distance. Genetics. 1943;28:114–38.

37. Slatkin M. Isolation by distance in equilibrium and non-equilibrium populations. Evolution (N. Y). 1993;47:264–79.

38. McRae BH. Isolation by resistance. Evolution (N. Y). 2006;60:1551–61.

39. Wang IJ, Bradburd GS. Isolation by environment. Mol. Ecol. 2014. p. 5649–

62.

40. Frost DR. Amphibian species of the world: an online reference. Version 6.0

(August 2017). Electronic Database accessible at http://research.amnh.org/herpetology/amphibia/index.html. American Museum of

Natural History, New York, USA. 2017.

41. Savage JM. The Amphibians and Reptiles of Costa Rica: a herpetofauna between two continents, between two seas. University of Chicago press; 2002.

42. García-Rodríguez A, Arias E, Chaves G. Multiple lines of evidence support the species status of the poorly known Diasporus tigrillo and the recently described

Diasporus citrinobapheus (Anura: Eleutherodactylidae). Neotrop. Biodivers.

2016;2:59–68.

43. Chaves G, García-Rodríguez A, Mora A, Leal A. A new species of dink frog

148

(Anura: Eleutherodactylidae: Diasporus) from Cordillera de Talamanca, Costa

Rica. Zootaxa. 2009;2088:1–14.

44. Batista A, Köhler G, Mebert K, Hertz A, Veselý M. An integrative approach to reveal speciation and species richness in the genus Diasporus (Amphibia: Anura:

Eleutherodactylidae) in eastern Panama. Zool. J. Linn. Soc. 2016;178:267–311.

45. Breiman L. Random Forests. Mach. Learn. 2001;45:5–32.

46. Afanador NL, Smolinska A, Tran TN, Blanchet L. Unsupervised random forest: a tutorial with case studies. J. Chemom. 2016;30:232–41.

47. Ligges U, Krey S, Mersmann O, Schnackenberg S. tuneR: Analysis of music.

[Internet]. 2013. Available from: //r-forge.r-project.org/projects/tuner/.

48. Sueur J, Aubin T, Simonis C. Seewave: a free modular tool for sound analysis and synthesis. Bioacoustics. 2008;18:213–26.

49. Araya-Salas M, Smith-Vidaurre G. warbleR: an r package to streamline analysis of animal acoustic signals. Methods Ecol. Evol. 2017;8:184–91.

50. Hijmans RJ, van Etten J. raster: Geographic analysis and modeling with raster data. R Packag. version. 2010.

51. Shah VB, McRae BH. Circuitscape : a tool for landscape ecology. Proc. 7th

Python Sci. Conf. 2008;62–5.

149

52. McCarroll D, Nesje A. Rock surface roughness as an indicator of degree of rock surface weathering. Earth Surf. Process. Landforms. 1996;21:963–77.

53. Elith J, Phillips S, Hastie T. A statistical explanation of MaxEnt for ecologists.

Divers. Distrib. 2011;17:43–57.

54. Phillips SJ, Anderson RP, Schapire RE. Maximum entropy modeling of species geographic distributions. 2006;190:231–59.

55. Radosavljevic A, Anderson RP. Making better Maxent models of species distributions: complexity, overfitting and evaluation. J. Biogeogr. 2014;41:629–43.

56. Muscarella R, Galante PJ, Soley-Guardia M, Boria RA, Kass JM, Uriarte M, et al. ENMeval: An R package for conducting spatially independent evaluations and estimating optimal model complexity for MaxEnt ecological niche models.

Methods Ecol. Evol. 2014;5:1198–205.

57. Hijmans RJ, Phillips S, Leathwick J, Elith J. Package “dismo”. Species distribution modeling. R package version 0.8-11.

2012;

58. Hijmans RJ, Cameron S, Parra J. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 2005;25:1965–78.

59. Campbell P, Pasch B, Pino JL, Crino OL, Phillips M, Phelps SM. Geographic

150

variation in the songs of neotropical singing mice: testing the relative importance of drift and local adaptation. Evolution (N. Y). 2010;64:1955–72.

60. Ferrier S, Manion G, Elith J, Richardson K. Using generalized dissimilarity modelling to analyse and predict patterns of beta diversity in regional biodiversity assessment. Divers. Distrib. 2007;13:252–64.

61. Ferrier S. Mapping spatial pattern in biodiversity for regional conservation planning: Where to from here? Syst. Biol. 2002;51:331–63.

62. Oliveira EF, Martinez PA, São-Pedro VA, Gehara M, Burbrink FT, Mesquita

DO, et al. Climatic suitability, isolation by distance and river resistance explain genetic variation in a Brazilian whiptail lizard. Heredity (Edinb). 2017;120:251–

65.

63. Welch AM, Semlitsch RD, Gerhardt HC. Call duration as an indicator of genetic quality in male gray tree frogs. Science (80-. ). 1998;280:1995–7.

64. Doty G V, Welch AM. Advertisement call duration indicates good genes for offspring feeding rate in gray tree frogs (Hyla versicolor). Behav. Ecol. Sociobiol.

2001;150–6.

65. Hamilton W, Zuk M. Heritable true fitness and bright birds: a role for parasites? Science (80-. ). 1982;218:384–7.

151

66. Gerhardt HC. The evolution of vocalization in frogs and toads. Annu. Rev.

Ecol. Syst. 1994;25:293–324.

67. Endler JA. Signals, signal conditions, and the direction of evolution. Am. Nat.

1992;139:125–53.

68. Janzen DH. Historia Natural de Costa Rica. Primera. Janzen D, editor. San

José, Costa Rica: Editorial San José, Universidad de Costa Rica; 1991.

69. Gräfe K, Frisch W, Villa IM, Meschede M. Geodynamic evolution of southern

Costa Rica related to low-angle subduction of the Cocos Ridge: Constraints from thermochronology. Tectonophysics. 2002;348:187–204.

70. Bagley JC, Johnson JB. Phylogeography and biogeography of the lower

Central American Neotropics: diversification between two continents and between two seas. Biol. Rev. 2014;89:767–790.

71. Höbel G, Gerhardt HC. Reproductive character displacement in the acoustic communication system of green tree frogs (Hyla cinerea). Evolution (N. Y).

2003;57:894–904.

72. Wei C, Jia C, Dong L, Wang D, Xia C, Zhang Y, et al. Geographic variation in the calls of the Common Cuckoo (Cuculus canorus): isolation by distance and divergence among subspecies. J. Ornithol. 2014;156:533–42.

152

73. Irwin DE, Thimgan MP, Irwin JH. Call divergence is correlated with geographic and genetic distance in greenish warblers (Phylloscopus trochiloides): a strong role for stochasticity in signal evolution? J. Evol. Biol. 2008;21:435–48.

74. Sosa-López JR, Mennill DJ, Navarro-Sigüenza AG. Geographic variation and the evolution of song in Mesoamerican rufous-naped wrens Campylorhynchus rufinucha. J. Avian Biol. 2013;44:27–38.

75. Pröhl H, Koshy R, Mueller U, Rand S, Ryan M. Geographic variation of genetic and behavioral traits in northern and southern Túngara frogs. Evolution (N.

Y). 2006;60:1669–79.

76. Amézquita A, Lima AP, Jehle R, Castellanos L, Ramos Ó, Crawford AJ, et al.

Calls, colours, shape, and genes: a multi-trait approach to the study of geographic variation in the Amazonian frog Allobates femoralis. Biol. J. Linn. Soc.

2009;98:826–38.

77. Bernal XE, Guarnizo C, Lüddecke H. Geographic variation in advertisement call and genetic structure of Colostethus palmatus (Anura, Dendrobatidae) from the Colombian Andes. 2009;61:395–408.

78. Lougheed SC, Austin JD, Bogart JP, Boag PT, Chek A a. Multi-character perspectives on the evolution of intraspecific differentiation in a neotropical hylid frog. BMC Evol. Biol. 2006;6:23.

153

79. Searcy WA, Andersson M. Sexual selection and the evolution of song. Annu.

Rev. Ecol. Syst. 1986;17:507–33.

80. Ryan MJ, Rand AS. The sensory basis of sexual selection for complex calls in the tungara frog , Physalaemus pustulosus. Evolution (N. Y). 1990;44:305–14.

81. Kirschel ANG, Blumstein DT, Smith TB. Character displacement of song and morphology in African tinkerbirds. Proc. Natl. Acad. Sci. 2009;106:8256–61.

82. Lemmon EM. Diversification of conspecific signals in sympatry: Geographic overlap drives multidimensional reproductive character displacement in frogs.

Evolution (N. Y). 2009;63:1155–70.

83. Pfennig KS, Ryan MJ. Character displacement and the evolution of mate choice: an artificial neural network approach. Philos. Trans. R. Soc. Lond. B. Biol.

Sci. 2007;362:411–9.

84. Ibáñez R, Rand S, Jaramillo C. The Amphibians of Barro Colorado Nature

Monument, Soberania Natural Park and Adjacent Areas. 1999.

85. Wycherley J, Doran S, Beebee TJC. Male advertisement call characters as phylogeographical indicators in European water frogs. Biol. J. Linn. Soc.

2002;77:355–65.

86. Angulo AA, Cocroft RB. Species identity in the genus Adenomera (Anura :

154

Leptodactylidae) in southeastern Peru. Herpetologica. 2003;59:490–504.

87. Bickford D, Lohman DJ, Sodhi NS, Ng PKL, Meier R, Winker K, et al. Cryptic species as a window on diversity and conservation. Evolution (N. Y). 2006;22.

88. Padial JM, De La Riva I. Integrative taxonomy reveals cryptic Amazonian species of Pristimantis (Anura: Strabomantidae). Zool. J. Linn. Soc. 2009;155:97–

122.

89. Vences M, Wake DB. Speciation, species boundaries and phylogeography of amphibians. In: Heatwole HH, Tyler M, editors. Amphib. Biol. (Vol. 6). Chipping

Norton.: Surrey Beatty & Sons; 2007. p. 2613–2669.

90. Goicoechea N, De La Riva I, Padial JM. Recovering phylogenetic signal from frog mating calls. Zool. Scr. 2010;39:141–54.

155

-Supplementary Material-

The role of geography, topography and climate in the acoustic divergence of Neotropical Diasporus frogs

Table S1. Acoustic variables measured for each recording analyzed in this study

Acoustic Variable Definition Mean frequency Weighted average of frequency by amplitude (in kHz) Minimum dominant frequency Lower frequency bound of the call Start dominant frequency Dominant frequency measurement at the start of the signal Frequency median The frequency at which the signal is divided in two frequency intervals of equal energy (in kHz) Mean dominant frequency Average of dominant frequency measured across the acoustic signal Mean peak frequency Frequency with highest energy from the mean spectrum First quartile of frequency The frequency at which the signal is divided in two frequency intervals of 25% and 75% energy respectively (in kHz) End dominant frequency Dominant frequency measurement at the end of the signal Maximum dominant frequency Higher frequency bound of the call Dominant frequency slope Slope of the change in dominant frequency through time ((enddom-startdom)/duration).Units are kHz/s. Dominant frequency Range Range of dominant frequency measured across the acoustic signal Cumulative absolute difference between adjacent measurements of dominant frequencies divided by the dominant Modulation index frequency range. 1 means the signals is not modulated. Standard deviation frequency SD of frequency weighted by amplitude Time entropy Energy distribution on the time envelope. Pure tone ~ 0; noisy ~ 1. Duration Length of signal (in s) First quartile time The time at which the signal is divided in two time intervals of 25% and 75% energy respectively (in s). Third quartile time The time at which the signal is divided in two time intervals of 75% and 25% energy respectively (in s). Time median The time at which the signal is divided in two time intervals of equal energy (in s)

156

Interquartile time range Time range between 'time.Q25' and 'time.Q75' (in s). Interquartile frequency range Frequency range between 'freq.Q25' and 'freq.Q75' (in kHz) Third quartile frequency The frequency at which the signal is divided in two frequency intervals of 75% and 25% energy respectively (in kHz) Spectral flatness Similar to sp.ent (Pure tone ~ 0; noisy ~ 1). Entropy Product of time and spectral entropy sp.ent * time.ent. Spectral entropy Energy distribution; pure tone ~ 0; noisy ~ 1 Skewness Asymmetry of the spectrum Kurtosis Peakedness of the spectrum

157

CONSIDERAÇÕES FINAIS Nesta dissertação pretendi avaliar a influência da heterogeneidade ambiental contida em regiões complexas, na geração de padrões de variação em diversas dimensões biológicas. Meu interesse foi abordar essa questão de maneira complementar, explorando tanto processos macroevolutivos, como por exemplo a diversificação de grupos taxonômicos superiores quanto processos microevolutivos, como a diferenciação de linhagens e divergência de características comportamentais na escala intraespecífica.

O estudo traz valiosas informações que revelam a importância das regiões montanhosas como motores evolutivos que sustentam ricas biotas numa escala global. Mostra também como múltiplas particularidades das paisagens complexas como a heterogeneidade climática e a irregularidade topográfica são capazes de promover fases incipientes da especiação a nível regional, funcionando como “combustível desses motores”. Demostrei que as taxas de especiação de anfíbios são mais rápidas em áreas montanhosas numa escala global. O padrão encontrado é robusto e se mantem, na maioria dos casos, quando é desconstruído e testado a nível de regiões zoogeográficas.

Além disso, apresento evidência comparativa e quantitativa de que a evolução numa região complexa específica pode ser promovida por múltiplas forças ao invés de ser explicada por uma pressão exclusiva. Por exemplo,

158

encontramos que a variação genética em 11 espécies que co-ocorrem no Istmo da América Central é induzida por diversos fatores abióticos da região, como são a heterogeneidade climática e a topografia montanhosa. O efeito desses fatores varia em intensidade dependendo da espécie, provavelmente devido as histórias de vida particulares de cada organismo. Também encontrei que em fases de especiação ainda mais incipientes, como a divergência em sinais acústicos envolvidas na reprodução, os padrões de variação não necessariamente são explicados pelas forças que geram isolamento entre populações. Neste caso, processos estocásticos que atuam a nível local sob essas populações isoladas são os prováveis responsáveis da evolução acústica. Porém, o efeito indireto das paisagens complexas gerando altos níveis de isolamento certamente acelera esse processo de evolução independente.

Como todo estudo, o meu trabalhalho possui algumas limitações. A espacialização de padrões evolutivos é certamente uma tarefa complexa que só agora está começando a ser abordada. Porém, achamos válida e robusta a forma em que procedemos com este objetivo no primeiro capítulo, mais ainda considerando a escala geográfica e a quantidade de espécies envolvidas no estudo. Não duvidamos que no futuro serão desenvolvidos métodos para lidar melhor com a projeção de taxas evolutivas no espaço. No segundo capítulo, o uso exclusivo de sequências mitocondriais tem suas fraquezas, mas a natureza

159

de um estudo comparativo incluindo múltiplas espécies, limita a incorporação de mais marcadores quando eles não estão disponíveis para todas as espécies.

A incorporação de mais marcadores e mais grupos taxonômicos dever ser o objetivo em futuro estudos. No terceiro capítulo, consideramos que os objetivos foram atingidos, porém a incorporação de mais populações especificamente na espécie de altas altitudes poderia ajudar a ter evidências mais conclusivas com relação ao peso das barreiras testadas como drivers de evolução dos cantos.

Em definitiva as montanhas do mundo têm um papel muito importante na evolução, montagem e manutenção da biodiversidade. Proporcionam um fascinante modelo de estudo, com réplicas distribuídas pelo globo, com diferentes particularidades como idades, configurações, regimes climáticos e rugosidade de terreno que permitem testar sua influência nos processos evolutivos. Pessoalmente tenho interesse de continuar explorando essas questões de maneira integrativa desde as perspectivas macro e microevolutivas.

Por exemplo pretendo continuar essa linha de macroevolução para mapear na escala global quais regiões do mundo funcionam como bombas e quais como museus de espécies. Também para testar se as altas taxas de especiação achadas nessa dissertação podem explicar padrões biogeográficos gerais como os gradientes latitudinais de diversidade. A nível micro meus esforços serão orientados a incorporar informação genômica que permita estudar associações

160

entre genótipos e variantes climáticas especificas ao longo de gradientes climáticos comuns em regiões montanhosas, para entender melhor o processo de seleção adaptativa e incrementar nosso ainda incompleto conhecimento do mecanismo de evolução parapátrica.

Finalmente, gostaria de complementar meu programa de pesquisa com estudos que incorporem particularidades da configuração geomorfológica das montanhas na avaliação de dinâmicas de contração e expansão de amplitudes de distribuição de espécies em cenário futuros de mudança climática. Uma caracterização, não só de quais são os grupos biológicos mais ameaçados, mas também de quais são os sistemas montanhosos que por suas características físicas são mais vulneráveis nesse contexto. Essa informação é estritamente necessária para definir prioridades e informar as ações de conservação dessas regiões, que temos demostrado são muito importantes para o origem e manutenção da biodiversidade.

161