Phylogeography of Andean Frogs

(Thesis Format: Integrated Article)

by

Daria Koscinski

Graduate Program in Biology

A thesis submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Faculty of Graduate Studies The University of Western Ontario London, Ontario, Canada

© Daria Koscinski 2008 Library and Bibliotheque et 1*1 Archives Canada Archives Canada Published Heritage Direction du Branch Patrimoine de I'edition

395 Wellington Street 395, rue Wellington Ottawa ON K1A0N4 Ottawa ON K1A0N4 Canada Canada

Your file Votre reference ISBN: 978-0-494-39291-1 Our file Notre reference ISBN: 978-0-494-39291-1

NOTICE: AVIS: The author has granted a non­ L'auteur a accorde une licence non exclusive exclusive license allowing Library permettant a la Bibliotheque et Archives and Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par telecommunication ou par Plntemet, prefer, telecommunication or on the Internet, distribuer et vendre des theses partout dans loan, distribute and sell theses le monde, a des fins commerciales ou autres, worldwide, for commercial or non­ sur support microforme, papier, electronique commercial purposes, in microform, et/ou autres formats. paper, electronic and/or any other formats.

The author retains copyright L'auteur conserve la propriete du droit d'auteur ownership and moral rights in et des droits moraux qui protege cette these. this thesis. Neither the thesis Ni la these ni des extraits substantiels de nor substantial extracts from it celle-ci ne doivent etre imprimes ou autrement may be printed or otherwise reproduits sans son autorisation. reproduced without the author's permission.

In compliance with the Canadian Conformement a la loi canadienne Privacy Act some supporting sur la protection de la vie privee, forms may have been removed quelques formulaires secondaires from this thesis. ont ete enleves de cette these.

While these forms may be included Bien que ces formulaires in the document page count, aient inclus dans la pagination, their removal does not represent il n'y aura aucun contenu manquant. any loss of content from the thesis. Canada THE UNIVERSITY OF WESTERN ONTARIO FACULTY OF GRADUATE STUDIES

CERTIFICATE OF EXAMINATION

Supervisor Examiners

Dr. Paul Handford Dr. Ben Evans

Supervisory Committee Dr. Elizabeth MacDougall-Shackleton

Dr. Marc-Andre Lachance Dr. Bryan Neff

Dr. Brian Luckman Dr. Cheryl Pearce

Dr. Bryan Neff

The thesis by

Daria Koscinski

entitled:

Phylogeography of Andean Frogs

is accepted in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Date: April 11,2008 Chair of the Thesis Examination Board Dr. Jeff Chen

ii ABSTRACT

The mtDNA genealogical patterns of two co-distributed frogs, Hypsiboas andinus (Hylidae; 262 individuals, 26 sites) and Pleurodema borellii (Leptodactylidae; 131 individuals, 20 sites), were investigated in a little studied region of the Andes Mountains (northwestern Argentina and Bolivia) to further our understanding of diversification in the Neotropics. Phylogeographic analyses of//, andinus indicated significant genetic structure. The two main lineages diverged between 2-6 million years ago, possibly isolated during Andean uplift; current lineage sympatry may result from secondary contact due to range expansion or may reflect cryptic species. Within the most extensive clade, geographic structuring was consistent with probable habitat fragmentation and subsequent range expansion in the Pleistocene (<2 mya), a time of extensive climatic cycling and vegetational shifts. Average divergence among clades is lower than found in other Neotropical taxa, highlighting the importance of Pleistocene climatic history in diversification in the Andes. Three models of movement among populations ofH. andinus were compared at two spatial scales. Effective distances, which incorporate habitat suitability for H. andinus, explained more variation in genetic differentiation than straight-line distance at both spatial scales. In heterogeneous landscapes where barriers to dispersal may be common, effective distances better explain patterns of population differentiation. In contrast to H. andinus, Pleurodema borellii exhibited signs of deeper temporal divergence although the geographic distribution of clades was similar. The three main lineages, separated about 4-10 mya, are nearly allopatric, suggesting a long history of isolation that may reflect cryptic species. All divergence events in P. borellii are older than those in H. andinus and we did not detect genetic divergence in P. borellii shallow enough to be influenced by the recent Pleistocene climatic cycling (<2 mya). Differences in the temporal patterns of divergence may be due to different biogeographic history of the two species, as H. andinus is a more recent (likely <10 mya) immigrant to the Andes. At similar spatial scales, genetic differentiation appeared lower in H. andinus than in P. borellii even though the latter is considered a "weedy species".

in Keywords: phylogeography, phylogeny, diversification, landscape genetics, population genetics, Andes, refugia, effective distance, partial Mantel's test, mitochondrial DNA, effective population size, Hypsiboas andinus, Pleurodema borellii

IV CO-AUTHORSHIP

Research towards this thesis was performed with several collaborators. All research was done under the supervision and financial support of Drs. P. Handford and S.C. Lougheed. Additional contribution to data collection and analyses for chapter 2 were provided by Dr. P.L. Tubaro (Museo Argentino de Ciencias Naturales, Argentina) and S. Sharp (undergraduate thesis student, Queen's University). The concept for chapter 3 was inspired by A.G. Yates, who performed the GIS data analyses.

A modified version of Chapter 2 is published in Molecular Ecology: Daria Koscinski, Paul Handford, Pablo L. Tubaro, Sarah Sharp, and Stephen C. Lougheed (2008) Pleistocene climatic cycling and diversification of the Andean treefrog, Hypsiboas andinus. Molecular Ecology 17: 2012-2025.

A modified version of Chapter 3 is under review at the Journal of Biogeography: Daria Koscinski, Adam G. Yates, Paul Handford, and Stephen C. Lougheed. Effects of landscape and history on diversification of a montane, stream-breeding amphibian.

Chapter 4 has not yet been submitted for publication: Daria Koscinski, Paul Handford, and Stephen C. Lougheed. Comparative phylogeography of two Andean frogs, Hypsiboas andinus and Pleurodema borellii.

v ACKNOWLEDGEMENTS

Provincial governments and National Parks of Argentina kindly permitted us to work in Argentina. I wish to thank all the people who assisted with field collections over the years: Nina Craig, Emily Croteau, Marcelo Romano, Gabriela Ibarguchi, Mike Rae, Pablo Tubaro, Marcos Vaira, Isabelle Bisson, Maria Ines Bonansea, Jose Segovia, Carlos and Sylvia Cufiado. Thank you to Jose Padial and Daiana Ferraro for sharing tissue samples. I am also grateful to many laboratory assistants over the years especially Nadine Sharpe, Sarah Sharp and Tanya Berkers. Thank you also to Zhaohua Chen for assistance with GIS analyses. Thank you to my committee for support and input throughout the years: Bryan Neff, Brian Luckman and M. Andre Lachance. A special thank you to Andre Lachance for providing me with laboratory space at UWO, and above all for being a willing and amazing mentor. Funding for these projects was provided by NSERC grants to Stephen Lougheed and Paul Handford; NSERC scholarship, OGSST scholarship, Sigma Xi Grants in Aid of Research, Gaige Award from the American Society of Ichthyologists and Herpetologists, and various University of Western Ontario awards. Many thanks to the Environment and Sustainability program for providing funding. Finally, but most importantly, I owe a great debt of gratitude to Paul Handford and Stephen C. Lougheed for providing motivation, inspiration, thought-provoking conversation, interesting distractions, late nights of frog catching, and unwavering support on my j ourney.

VI TABLE OF CONTENTS

CERTIFICATE OF EXAMINATION / ii ABSTRACT / iii CO-AUTHORSHIP /v ACKNOWLEDGEMENTS / vi TABLE OF CONTENTS / vii LIST OF TABLES/x LIST OF FIGURES / xi LIST OF APPENDICES / xiii LIST OF ABBREVIATIONS AND SYMBOLS / xiv

CHAPTER 1. GENERAL INTRODUCTION /1 1.1 Introduction /1 1.2 Study Area: Northwestern Argentina / 5 1.3 Frogs as study organisms / 9 1.4 Outline of thesis /11 1.5 References/12

CHAPTER 2. PLEISTOCENE CLIMATIC CYCLING AND DIVERSIFICATION OF THE ANDEAN TREEFROG, HYPSIBOAS ANDINUS118 2.1 Introduction/18

2.2 Methods / 22

2.2.1 Field Methods/22

2.2.2 DNA extraction, PCR and sequencing / 22

2.2.3 Analyses / 25

2.2.4 Timing of divergence / 26

2.3 Results / 27

2.3.1 Molecular diversity / 2 7

vn 2.3.2 Pbylogenetic analyses on the combined dataset / 28

2.3.3 Phylogeographic patterns / 28

2.3.4 Timing of divergence / 34

2.4 Discussion / 36

2.4.1 Divergent lineages within the H. andinuslH. riojanus complex / 36

2.4.2 Phylogeography of H. andinus 138

2.4.3 Diversification in the Andes / 40

2.5 References / 42

CHAPTER 3. EFFECTS OF LANDSCAPE AND HISTORY ON DIVERSIFICATION

OF A MONTANE, STREAM-BREEDING AMPHIBIAN / 49

3.1 Introduction / 49

3.1.1 Study Area/50

3.2 Methods/51

3.3 Results / 56

3.3.1 Genetic Differentiation / 56

3.3.2 Entire data set / 57

3.3.3 Subset of data set/59

3.4 Discussion / 60

3.4.1 Effects of scale/ 65

3.4.2 Effects of habitat classification / 65

3.5 References / 67 CHAPTER 4. COMPARATIVE PHYLOGEOGRAPHY OF TWO ANDEAN FROGS,

HYPSIBOAS ANDINUS AND PLEURODEMA BORELLII I 72

4.1 Introduction / 72

4.2 Methods / 73

4.2.1 Study species / 73

4.2.2 Sampling / 74

4.2.3 PCR and sequencing / 77

4.2.4 Phylogenetic analyses / 77

4.2.5 Phylogeographic analyses / 77

4.2.6 Timing of divergence / 78

4.2.7 Population-level statistical analyses / 78

4.3 Results/80

4.3.1 Molecular diversity / 80

4.3.2 Phylogenetic analyses / 80

4.3.3 Phylogeographic patterns / 82

4.3.4 Timing of divergence / 84

4.3.5 Population-level statistical analyses / 87

4.4 Discussion/ 91

4.4.1 Divergent lineages of P. borellii 191

4.4.2 Phylogeography of P. borellii 193

4.4.3 Comparative phylogeography / 94

4.4.4 Ecology and life history / 97

4.5 References/100

IX CHAPTER 5. SUMMARY AND CONCLUSIONS /107

5.1 Summary and conclusions/107

5.2 References/114

APPENDICES/118

VITA/128

x LIST OF TABLES

Table Description Page

1.1 Summary of climate and vegetation patterns of Northwestern 6 Argentina.

2.1 Sources of tissues, sample sizes and distribution of haplotypes for 23-24 mtDNA control region analyses for each population ofHypsiboas andinus.

2.2 Nested clade analysis results. 31

2.3 Inferences made for each clade in network #1 found to have significant 33 geographical associations.

2.4 Estimates of divergence time based on analyses using BEAST. 34

3.1 Sample sizes for mtDNA control region analyses for each population of 53 Hypsiboas andinus.

3.2 Resistance values used for each habitat in least-cost path analysis for 57 each type of effective distance measured.

3.3 Correlations between genetic distance and different geographic 60 distances using partial Mantel's tests for the entire data set.

3.4 Correlations between genetic distance and different geographic 62 distances using partial Mantel's tests at the smaller scale (<100 km).

4.1 Sources of tissues and sample sizes for each population of Pleurodema 75 borellii.

4.2 Nested clade analysis results. 84

4.3 Inferences made for each clade in network #1 found to have significant 85 geographical associations.

4.4 Diversity indices and mismatch distributions for each clade. 85

4.5 Estimates of divergence time based on analyses using BEAST. 89

4.6 AMOVA results for both species. 89

XI LIST OF FIGURES

Figure Description Page

1.1 Generalized patterns of genetic relationships and geographic 4 distributions for each phylogeographic category.

1.2 Distribution and climate diagrams of different habitat types of 8 Northwestern Argentina.

2.1 Sampling localities of Hypsiboas andinus in Northwestern Argentina 20 and Bolivia, and H. riojanus.

2.2 Phylogenetic relationships of Argentine and Bolivian Hypsiboas 29 andinus and H. riojanus based on Bayesian analyses of cytochrome b and control region sequences.

2.3 Maximum parsimony networks using mtDNA control region from 259 30 individuals.

2.4 Geographical distribution of 4-step clades from network #1. 35

3.1 Relationships among populations of Hypsiboas andinus based on 55 analyses in Chapter 2.

3.2 Locations of sharp genetic discontinuities in H. andinus likely 58 corresponding to barriers to gene flow.

3.3 Relationship between pair-wise simple and detailed habitat effective 59 distances and Euclidean distance.

3.4 Routes between populations 1 l-12a-20 using Euclidean distance, 61 simple and detailed habitat effective distances.

4.1 Sampling localities of Pleurodema borellii in Northwestern Argentina 76 and Bolivia.

4.2 Phylogenetic relationships of Argentine and Bolivian Pleurodema 81 borellii based on Bayesian analyses of cytochrome b and 16S.

4.3 Maximum parsimony networks using mtDNA from 130 individuals. 83

4.4 Geographical distribution of 3-step clades from network #1. 86

4.5 Locations of sharp genetic discontinuities in Pleurodema borellii likely 88 corresponding to barriers to gene flow.

xn Figure Description Page

4.6 Bayesian Skyline plots for both species. 90

Xlll LIST OF APPENDICES

Appendix Description Page

1 Voucher specimens and GenBank numbers for Hypsiboas andinus. 118

2 Population pair-wise FST values for entire data set for Hypsiboas 121 andinus.

3 Population pair-wise FST values for analyses of populations 122 < 100km apart for Hypsiboas andinus.

4 Voucher specimens for Pleurodema borellii. 123

5 Copy of animal care protocol. 125

6 Copy of copyright agreement with Molecular Ecology for 126 publication of chapter 2.

xiv LIST OF ABBREVIATIONS AND SYMBOLS

°c degree Celsius \xL microlitres bp base pair masl metres above sea level mtDNA mitochondrial DNA mya million years ago Ne effective population size Nem effective number of migrants ng nanogram PCR Polymerase Chain Reaction

XV 1

CHAPTER 1. GENERAL INTRODUCTION

1.1 Introduction Central to our understanding of diversification is an understanding of the factors that facilitate or impede the genetic differentiation of populations, perhaps eventually leading to speciation. Although there has been much interest in the processes of diversification, many questions remain about the mechanisms promoting differentiation of populations, especially in understudied regions of high biodiversity such as the Neotropics. South America encompasses a disproportionately large fraction of global biodiversity (World Conservation Monitoring Centre 1992), yet our understanding of the timing of, and mechanisms responsible for, its origins lags far behind that of the northern temperate biota of North America and Europe (e.g. Hewitt 1996, 2000, Avise et al. 1998, Johnson & Cicero 2004, Soltis et al. 2006, Waltari et al. 2007). Neotropical montane regions in particular have been proposed as important centres of species diversification (Simpson 1979, Fjeldsa 1994, Moritz et al. 2000). Thus, research focusing on factors that may have facilitated or impeded the genetic differentiation of taxa inhabiting Andean mountain habitats is of great relevance to our understanding of diversification and speciation in the Neotropics. Gene flow, through the movement of breeding individuals among populations (migration), is a powerful homogenizing force, spreading genetic variants (alleles) among populations. Gene flow is often measured as the number of migrants (Nem) among populations, based on the effective population size (Ne) and the migration rate (m)

(Wright 1951, Slatkin 1985). When gene flow is high (Nem > 1), the genetic make-up of populations remains similar, but when gene flow among populations is reduced (Nem < 1) populations may begin to differentiate through mutation, genetic drift, and/or natural selection acting independently in each (partially) isolated population (Wright 1951, Slatkin 1985). The extent of gene flow, or connectivity, among populations depends on several ecological and life history (demographic) characteristics of the given species, as well as on historical events that may influence the movement of individuals (biogeographic). 2

Ecology and life history characteristics can have profound impacts on connectivity. Highly vagile species (adults or larvae) typically show little genetic structuring, as expected (e.g. eels Avise et al. 1986, sea cucumbers Uthicke & Benzie 2003, pearl oysters Yu & Chu 2006, coral reef fish Haney et al. 2007), while more sedentary species show genetic differentiation even at small spatial scales (e.g. frogs Lougheed et al. 1999, butterflies Keyghobadi et al. 2005, snails Holland & Cowie 2007). Although dispersal ability is an important prerequisite for gene flow, other life history characteristics may constrain its impact. For example, many species undertake long distance migrations, and hence have great dispersal capability, but yet may be highly philopatric, returning to breed at or near the site where they were hatched or born. As a result, the distance traveled by an individual during its annual cycle is much greater than the distance traveled by its gametes (e.g. Wenink et al. 1996, Alvarado Bremer et al. 2005, Vaha et al. 2007). Habitat-restricted species also show stronger genetic differentiation among geographic areas than generalist species due to a low propensity for movement and hence low gene flow (Schauble & Moritz 2001, Stuart-Fox et al. 2001). Similarly, fecundity, an important life history characteristic, may relate positively to gene flow, and is highly correlated with gene flow in darters, a stream inhabiting fish (Turner et al. 1996, Turner & Trexler 1998). Species laying a few large eggs show lower gene flow than species producing many small eggs (Turner & Trexler 1998). Low fecundity may lead to smaller effective population sizes, with obvious implications for the effects of genetic drift and migration within these populations (Turner et al. 1996, Turner & Trexler 1998). Genetic drift can produce rapid differentiation in small,

genetically isolated populations. To maintain levels of high gene flow (i.e. Nem > 1), populations of a smaller effective size would also require higher migration rates. Through their effects on gene flow and population size, characteristics affecting population demographics such as movement of individuals and recruitment of young may strongly influence contemporary patterns of connectivity among populations. The signature of contemporary processes on intraspecific genetic diversity, however, may overlie pre-existing genetic structure imposed by historical processes (Templeton 1998). Current estimates of gene flow may be confounded by historical events, so examination of contemporary processes must control for these possible effects. 3

Phylogeography, the study of patterns of spatial distribution of genealogical lineages across landscapes (Avise 1998), provides a crucial link between population level processes and broad landscape patterns. A phylogeographic-analytical approach uses both a spatial and a temporal scale to infer factors that have shaped contemporary distributions of genetic diversity. The strength of this approach is that it can differentiate between current gene flow patterns and historical events, enabling discrimination between restricted gene flow through isolation-by-distance, and historical events such as fragmentation events involving no genetic exchange, or range expansions (be they contiguous range expansions or colonization events). The connections between demography and phylogenetic history provide a range of possible spatial distributions of genetic lineages across landscapes, each with a unique signature (figure 1.1) (Avise 2000). Patterns of populations that differ strongly genetically (large genetic gaps) suggest historical divergence, likely in allopatry (category I and II). The contemporary geographic distributions of lineages, however, suggest that the history of these two categories differ. In category I, the divergent lineages are currently allopatric and likely continue to be separated by extrinsic barriers. In category II, lineages that diverged in allopatry have come into secondary contact, perhaps due to the removal of the extrinsic barriers that separated them historically, and currently occupy the same geographic range. Patterns of shallow genetic population differentiation (III, IV, V) can also be distinguished by examining the geographic distribution of the groups. Taxa that show shallow differentiation but have geographically localized distributions (III) are often examples of populations that have restricted gene flow, but were in historical contact recently. Processes such as genetic drift or natural selection may promote the differentiation of these recently connected groups. High-gene-flow species often show patterns typical of category IV - shallow genetic differentiation and sympatric distribution of lineages. The high dispersal capacity of these organisms results in the wide geographic distribution of haplotypes. Category IV patterns are also often associated with rapid population expansions, where populations increase in numbers quickly and expand to new geographic areas, resulting in the characteristic "star phylogeny". Category V is intermediate between III and IV, where populations have been in contact in recent history and currently have low or modest gene 4 flow. The most common signature of this category is a pattern of widespread common haplotypes with a few geographically localized unique haplotypes. By examining the spatial and temporal patterns of population differentiation using a phylogeographic approach, we can better understand the factors that influence the diversification process.

Geographic Ph^fo- Gene Representation geographic Tree Category Pattern Region 1 Region 2 Region 3

large / f,n.wwnlfrw»"H5lj in iimi i M*w*Hffl*-'~*'nj ^iwi*MH-.-^X'*<^^yu^**3£ J genetic gaps

»-4>-HB)

large n geneH e aaps (i—m—n l—m—n l—m-^T) -» r N t \X'—y~*~B x—-y—* "N

no large ni genetic gaps CZD—CZD—CZD

(a a a )

no large (b b b ) IV genetic gaps

c * c « )

no large ¥ (© — a a a—®) genetic gaps Figure 1.1. Generalized patterns of genetic relationships and geographic distributions for each phylogeographic category. Small letters (e.g. a, x) represent haplotypes and their distribution among regions is indicated with grey ovals. Lines between haplotypes represent genetic relationships where increased number of slashes indicates more genetic differences and hence deeper differentiation. Adapted from Avise (2000). 5

Avise (1998, 2000) stressed the importance of assessing patterns of genetic variation in multiple taxa to evaluate the impact of common historical processes in shaping phylogeographic patterns across landscapes. Concordant phylogeographic patterns for multiple taxa suggest that common historical barriers have affected gene flow and driven differentiation, and the specific patterns often suggest the identity and timing of those factors. Discordance in patterns among co-distributed taxa may, on the other hand, highlight the relative importance of differences in ecology, demography and distributional history among species (Avise 1998). Examining phylogeographic patterns of multiple species representing a range of life history and ecological attributes can thus provide for explicit tests of the relative importance of demographic processes vs. historical events in shaping patterns of differentiation among populations. Fewer molecular phylogeographic studies of Neotropical species exist, compared to Holarctic species, with a decided skew in the former towards lowland taxa of lower latitudes (e.g. Patton et al. 1994, Lougheed et al. 1999, Costa 2003, Lougheed et al. 2006). Our understanding of diversification in the habitats associated with the Andes Mountains in South America is limited by the paucity of data currently available. My study investigates the factors that affect population differentiation in this highly understudied region to improve our understanding of the processes of diversification that generate biodiversity.

1.2 Study Area: Northwestern Argentina Contemporary Northwestern Argentina possesses great topographic diversity associated with the Andes Mountains, which generate great spatial variation in precipitation and temperature, reflected in a great complexity of habitats (Czajka & Vervoorst 1956, Cabrera 1976, Vervoorst 1982; see Handford 1988 for English-language summary) (table 1.1, figure 1.2). East of the Andean system, rainfall is low resulting in the semi-arid chaco on the flat terrain, which is slightly tilted towards the east (Czajka & Vervoorst 1956, Cabrera 1976, Vervoorst 1982, Handford 1988). The open drought-deciduous woodlands and grasslands of the western chaco are dominated by small leguminous trees and shrubby cacti. 6

Table 1.1. Summary of climate and vegetation patterns of Northwestern Argentina. For geographic distributions of each habitat and detailed climate diagrams refer to figure 1,2. Data compiled from Czajka & Vervoorst (1956), Cabrera (1976), Vervoorst (1982), Cagnolo et al. (2006). Masl = meters above sea level. Annual Elevation Precipitation Habitat (masl) (mm) Typical Vegetation Chaco <350 500-700 Semi-arid, open drought-deciduous woodlands and grasslands, small leguminous trees, shrub cacti. Transition 350-500 700-900 Open monsoon forest, dominated by forest large trees, some epiphytes. Chaco serrano *400- 750 Semi-arid open low woodland with (montane 1300 numerous cacti. Forms complex mosaic Chaco) with other assemblages. Yungas 500-1300 850-3000 Dense rain forests rich in epiphytes and (montane bromeliads, mixture of large evergreen rainforest) and deciduous trees, according to local rainfall. Alisos (alder) 1300- 500 Mesic woodland dominated by the woodland 2700 alder Alnus jorullensis Grassland 1500- 400-500 Alpine meadows dominated by grasses 3000 and forbs. Monte desert **1500- 80-200 Arid scrublands, mostly in inter- scrub 3000 montane valleys: large columnar cacti, scattered leguminous trees, Larrea, creosote bush; some gallery woodland Puna scrub >3400 <300 Sparse grass- and scrub-lands, grading into desert on dry ridges in the west and moister grasslands in the east. Extensive salt flats. * Usually found on the lowest ridges of the easternmost mountain chain or on drier mountain slopes. **Only found in valleys west of the eastern ranges.

Humid air masses travelling from the northeast in the summer bring heavy precipitation, which falls primarily on the east-facing slopes of the easternmost ranges. As a result, longitudinal strips of forest stretch along the east-facing slopes south from Bolivia to the province of Catamarca, Argentina (-28° S). The character and composition 7

of the forests and woodlands found along the eastern slopes of the mountains change dramatically with increasing altitude (table 1.1). On the flats immediately adjacent to the base of the slopes is the so-called transition forest, an open drought-deciduous monsoon forest of tall trees with a dense understory and some epiphytes. This Transition forest grades into the more humid and dense montane forests of the mid-elevations dominated by tall trees, epiphytes, and bromeliads. In the wettest regions, this forest remains green year-round as is typical of lowland tropical rainforests, but drier portions also include drought-deciduous species. Above this forest system, the precipitation and temperature decrease and deciduous alder (Alnus jorullensis) woodlands dominate. The montane forests are essentially continuous except for a narrow invasion of arid thorn scrub habitat in the province of Salta (figure 1.2), which isolates the southernmost tip of this forest peninsula (Brown et al. 2001). Above the alder woodlands (~1800m) is a series of alpine meadows dominated by grasses and forbs. The lowest ridges of the easternmost mountain chain and drier mountain slopes contain the chaco serrano (montane chaco), a semi-arid open low woodland with numerous arborescent cacti (Cagnolo et al. 2006). This woodland forms complex mosaics with other assemblages such as the transition forest, grasslands and Monte desert scrub. In the study area, the Monte is found exclusively in inter-montane valleys to the west. These arid scrublands are dominated by large columnar cacti, scattered leguminous trees, and creosote bush. Some gallery woodlands exist along streams. The Monte grades above into the puna, a high altitude desert with sparse grass and scrub vegetation. Extensive salt flats are found at these high altitudes (>3400 masl). The Andean cordillera runs north/south along the Western margin of South America, spanning approximately 8000 km from about 10°N to 57°S, and is about 200- 400 km wide (Iriondo 1999). Major uplift of the central Andean plateaus in northwestern Argentina and Bolivia began about 10 million years ago (mya; Miocene) and continues today (Zeil 1979, Gregory-Wodzicki 2000). This orogeny substantially altered atmospheric circulation and patterns of precipitation (Hooghiemstra & van der Hammen 1998, Gregory-Wodzicki 2000), as well as the flow of rivers (Hooghiemstra & van der Hammen 1998). Furthermore glacial-interglacial cycling throughout the Pleistocene had pronounced effects on precipitation and temperature (Iriondo 1999). Cool and dry glacial 8

La Quiaca (3458 masl) 9.4° 321 mm | | Chaco HHH Alisos Oran (357 masl) 21.4° 886 mm [soj Transition Forest J| Grassland Ml Chaco serrano | | Monte Yungas \-'\ Puna

ilSBftlfltiltll t F • I * J J Figure 1.2. Distribution and climate diagrams of different habitat types of Northwestern Argentina. Climate diagrams display monthly temperature (left axis, °C) and precipitation (right axis, mm). Mean annual temperature and precipitation is listed below location name. 9 periods resulted in compression of vegetation zones in the mountains, contraction of forests in the lowlands, and general northward migration of vegetation systems (Gentry 1982, Hooghiemstra & van der Hammen 1998, Seltzer 2002). The vegetation systems then expanded and moved southward with the return of warm and moist interglacials. Given the complex history of the topography and of the climate and vegetation regimes in the Andean region during the last 10 million years, organisms are likely to have experienced a wide array of environmental conditions, changing even over short geographic distances, and their ranges may plausibly have been fragmented over their evolutionary history. The rate and timing of Andean uplift, together with associated changes in climate and vegetation, are believed to have been centrally important in influencing the diversification and distribution of neotropical species (birds: Fjeldsa 1994, Garcia-Moreno & Fjeldsa 2000, Burns & Naoki 2004; rodents: Reig 1986; herpetofauna: Cei 1986, Lynch 1986, Zamudio & Greene 1997). Dispersal and colonization of newly created habitats of the Andes has been proposed for some taxa (e.g. hummingbirds Bleiweiss 1998), and many authors have suggested that the Andes fragmented species ranges resulting in small populations that diverged in isolation (Lynch 1986, Reig 1986, Zamudio & Green 1997, Garcia-Moreno & Fjeldsa 2000). An understanding of the historical and contemporary distribution of habitats contributes to our understanding of diversification and contemporary distributions of organisms.

1.3 Frogs as study organisms Amphibian diversity is greater in South America than on any other continent (Duellman 1999). The highest anuran diversity in South America occurs along the Andes Mountains, with 720 species of frogs, 95% of them endemic to the Andes (Duellman 1999). Amphibians display a spectacular diversity of reproductive modes, the greatest among vertebrates (Duellman & Trueb 1986). Twenty-nine modes have been described, ranging from aquatic eggs and larvae (most primitive) through to direct development of froglets from terrestrial eggs (Duellman & Trueb 1986). The variety of modes is best represented in tropical areas: the Neotropical region contains species that use 21 of the 29 modes, 8 of which are unique to the region (Duellman & Trueb 1986). The most 10 generalized and primitive mode is the laying of eggs in ponds (Mode 1) or streams (Mode 2) with tadpoles feeding in the same site. Adaptations to drier environments and a lowered dependence on unpredictable water bodies have evolved independently in many lineages and include for example foam nests (Mode 8 and 9), eggs in terrestrial cavities (Modes 12-16), direct development of froglets from terrestrial eggs (Mode 17), or internal fertilization (Mode 28-29) (Duellman & Trueb 1986). The move towards terrestriality has involved changes in fecundity for females - laying fewer but larger eggs rather than many small eggs. These changes affect demographic factors that may influence population genetic structure, as described above. Frogs generally show restricted movement (less than 10 km) due to their small size, biphasic life history, dependence on water, philopatry, and particular habitat requirements (reviewed in Marsh & Trenham 2001). Limited dispersal in montane regions has been reported for several amphibians, and mountain ridges have been implicated as strong barriers to movement, leading to marked differentiation among populations (Lougheed et al. 1999, Funk et al. 2005, Lowe et al. 2006). Hypsiboas andinus (formerly Hyla andina; new designation suggested by Faivovich et al. 2005, supported by Wiens et al. 2005) is a moderate-sized tree frog (50- 60 mm) that inhabits deeply-incised river valleys that penetrate into the Andes, from the province of Catamarca, Argentina (28°S) to northern Bolivia (16°S). It is mostly found near water along streams, ditches or flooded areas (Cei 1980, Duellman et al. 1997, personal observation) in various habitats, from lowland montane forests (500 masl) to montane grasslands (>1500 masl), but is not found in the semi-arid chaco to the east nor in high altitude Monte and puna habitats (Duellman et al. 1997). Hypsiboas andinus eggs are deposited in masses of approximately 600 eggs (M. Vaira unpublished) secured to submerged vegetation in streams (Cei 1980, personal observation). Larvae are aquatic and develop at the site of oviposition. Generation time for the species is not known but is likely one year, as for the closely related//, pulchellus (Basso & Kehr 1992). Pleurodema borellii is a moderate-sized (40-55 mm) leptodactylid frog that inhabits a wide range of Andean habitats from the province of La Rioja, Argentina (30°S) to the province of Jujuy, Argentina (23°S). P. borellii uses open, often disturbed, habitats throughout the moist montane forests, Monte, puna, and the western margin of the chaco 11

(Cei 1980). This species reproduces in small, temporary ponds or pools, as well as in drier habitats enabled by the protective foam nests it builds to house developing eggs (Duellman & Trueb 1986). Typically between 1000 and 2500 eggs (M. Vaira unpublished) are deposited in foam nests floating on the surface of shallow pools (Cei 1980, personal observation). Larvae are aquatic and develop at the site of oviposition. Direct measurements of generation time are not available for the species. Breeding is highly concentrated in the wettest part of the year (Halloy & Fiarlo 2000, Vaira 2002) and likely results in a generation time of one year since juveniles that emerge would necessarily have to wait until the next summer to breed. Given their limited vagility, diverse fecundities, breeding biology and breeding habitats, frogs of South America offer excellent possibilities for an examination of the factors involved in evolutionary divergence, diversification and speciation.

1.4 Outline of thesis My study explores the factors that facilitate or impede the genetic differentiation of populations of taxa inhabiting Andean mountain habitats to further our understanding of diversification and speciation in the Neotropics. In Chapter 2,1 investigate the mtDNA genealogical patterns in 262 individuals of the frog Hypsiboas andinus from 23 sites across the Eastern ranges of the Andes Mountains in Argentina and 3 sites in Bolivia. Chapter 3 focuses on the roles of landscape features in shaping patterns of contemporary and historical genetic diversification among populations of H. andinus across two spatial scales (< 100km, and entire study area of 700km). I examined three models of movement to test which of these explained the most variation in pair-wise population genetic differentiation: (i) traditional straight-line distance, and two effective distances calculated using a least-cost path analysis using a (ii) simple habitat classification, and (iii) a detailed habitat classification. In Chapter 4,1 present the mtDNA geneological patterns for 131 individuals of Pleurodema borellii from 21 sites. I then compare the phylogeographic patterns of P. borellii and H. andinus to evaluate the influence of common geographic history and divergent life history on diversification in the Andes. 12

1.5 References Alvarado Bremer JR, Vifias J, Mejuto J, Ely B, Pla C (2005) Comparative phylogeography of Atlantic bluefin tuna and swordfish: the combined effects of vicariance, secondary contact, introgression, and population expansion on the regional phytogenies of two highly migratory pelagic fishes. Molecular Phylogenetics and Evolution 36, 169-187.

Avise JC, Helfman GS, Saunders NC, Hales LS (1986) Mitochondrial DNA Differentiation in North Atlantic Eels: Population Genetic Consequences of an Unusual Life History Pattern. Proceedings of the National Academy of Sciences 83, 4350-4354.

Avise JC, Walker D, Johns GC (1998) Speciation durations and Pleistocene effects on vertebrate phylogeography. Proceedings of the Royal Society of London Series B- Biological Sciences 265, 1707-1712.

Avise JC (2000) Phylogeography: The history and formation of species. Harvard University Press, Cambridge.

Avise JC (1998) The history and purview of phylogeography: a personal reflection. Molecular Ecology 7, 371-379.

Basso NG, Kehr AI (1992) Estructura poblacional y crecimiento de Hyla pulchella pulchella (Anura: Hylidae). Acta Zoologica Lilloana 41, 143-148.

Bleiweiss R (1998) Origin of hummingbird faunas. BiologicalJournal oftheLinnean Society 65, 77-97.

Brown AD, Grau HR, Malizia LR, Grau A (2001) Argentina. In: Bosques nublados del neotropico (eds. Kappelle M, Brown AD). Instituto Nacional de Biodiversidad, Costa Rica.

Burns KJ, Naoki K (2004) Molecular phylogenetics and biogeography of Neotropical tanagers in the genus Tangara. Molecular Phylogenetics and Evolution 32, 838- 854.

Cabrera AL (1976) Regiones fitogeograficas Argentinas. In: Enciclopedia Argentina de Agricultura y Jardineria, 2nd edition, (ed. Parodi LR), pp. 1-85. Editorial Acme, Buenos Aires.

Cagnolo L, Cabido M, Valladares G (2006) Plant species richness in the Chaco Serrano Woodland from central Argentina: Ecological traits and habitat fragmentation effects. Biological Conservation 132, 510-519.

Cei JM (1980) Amphibians of Argentina Monitore Zoologico Italiano N. S Monografia 2, Florence. 13

Cei JM (1986) Speciation and adaptive radiation in Andean Telmatobius frogs. In: High altitude tropical biogeography (ed. Vuilleumier F, and M. Monasterio). Oxford University Press, New York.

Costa LP (2003) The historical bridge between the Amazon and Atlantic Forest of Brazil: a study of molecular phylogeography with small mammals. Journal of Biogeography 30, 71-86.

Czajka W, Vervoorst F (1956) Die naturraumliche Gliederung Nordwest-Argentiniens. Petermanns Geographische Mitteilungen 100, 89-102 & 196-208.

Duellman WE (1999) Patterns and distributions of amphibians: A global perspective. The John Hopkins University Press, Baltimore.

Duellman WE, Trueb L (1986) Biology of Amphibians. The Johns Hopkins University Press, Baltimore.

Duellman WE, de la Riva I, Wild ER (1997) Frogs of the Hyla armata and Hyla pulchella groups in the Andes of South America, with definitions and analyses of phylogenetic relationships of Andean groups of Hyla. Scientific Papers of the Natural History Museum of the University of Kansas 3, 1-41.

Faivovich J, Haddad CFB, Garcia PCA, et al. (2005) Systematic review of the frog family Hylidae, with special reference to Hylinae: Phylogenetic analysis and taxonomic revision. Bulletin of the American Museum of Natural History 294, 1- 240.

Fjeldsa J (1994) Geographical patterns for relict and young species of birds in Africa and South America and implications for conservation priorities. Biodiversity and Conservation 3, 207-226.

Funk WC, Blouin MS, Corn PS, et al. (2005) Population structure of Columbia spotted frogs (Rana luteiventris) is strongly affected by the landscape. Molecular Ecology 14, 483-496.

Garcia-Moreno J, Arctander P, Fjeldsa J (1999) A Case of Rapid Diversification in the Neotropics: Phylogenetic Relationships among Cranioleuca Spinetails (Aves, Furnariidae). Molecular Phylogenetics and Evolution 12, 273-281.

Gentry AH (1982) Neotropical floristic diversity: phytogeographical connections between Central and South America, Pleistocene climatic fluctuations, or an accident of the Andean orogeny? Annals of the Missouri Botanical Garden 69, 557-593.

Gregory-Wodzicki KM (2000) Uplift history of the Central and Northern Andes: A review. Geological Society of America Bulletin 112, 1091-1105. 14

Halloy M, and J. M. Fiano (2000) Oviposition site selection in Pleurodema borellii (Anura: Leptodactylidae) may be influenced by tadpole presence. Copeia 2000, 606-609.

Handford P (1988) Trill rate dialects in the Rufous-collared sparrow, Zonotrichia capensis, in northwestern Argentina. Canadian Journal of Zoology 66, 2658-2670.

Haney RA, Silliman BR, Rand DM (2007) A multi-locus assessment of connectivity and historical demography in the bluehead wrasse (Thalassoma bifasciatum). Heredity 98, 294-302.

Hewitt G (2000) The genetic legacy of the Quaternary ice ages. Nature 405, 907-913.

Hewitt GM (1996) Some genetic consequences of ice ages, and their role in divergence and speciation. BiologicalJournal of the Linnean Society 58, 247-276.

Holland BS, Cowie RH (2007) A geographic mosaic of passive dispersal: population structure in the endemic Hawaiian amber snail Succinea caduca (Mighels, 1845). Molecular Ecology 16, 2422-2435.

Hooghiemstra H, van der Hammen T (1998) Neogene and Quaternary development of the neotropical rain forest: the forest refugia hypothesis, and a literature overview. Earth-Science Reviews 44, 147-183.

Iriondo M (1999) Climatic changes in the South American plains: records of a continent- scale oscillation. Quaternary International SI I S%, 93-112.

Johnson NK, Cicero C (2004) New mitochondrial DNA data affirm the importance of Pleistocene speciation in North American birds. Evolution 58, 1122-1130.

Keyghobadi N, Roland J, Strobeck C (2005) Genetic differentiation and gene flow among populations of the alpine butterfly, Parnassius smintheus, vary with landscape connectivity. Molecular Ecology 14, 1897-1909.

Lougheed SC, Austin JD, Bogart JP, Boag PT, Chek AA (2006) Multi-character perspectives on the evolution of intraspecific differentiation in a neotropical hylid frog. BMC Evolutionary Biology 6, 23.

Lougheed SC, Gascon C, Jones DA, Bogart JP, Boag PT (1999) Ridges and rivers: a test of competing hypotheses of Amazonian diversification using a dart-poison frog (Epipedobates femoralis). Proceedings of the Royal Society of London B 266, 1829-1835.

Lowe WH, Likens GE, McPeek MA, Buso DC (2006) Linking direct and indirect data on dispersal: Isolation by slope in a headwater stream salamander. Ecology 87, 334- 339. 15

Lynch JD (1986) Origins of the High Andean Herpetological fauna. In: High altitude tropical biogeography (ed. Vuilleumier F, and M. Monasterio). Oxford University Press, New York.

Marsh DM, Trenham PC (2001) Metapopulation dynamics and amphibian conservation. Conservation Biology 15, 40-49.

Moritz C, Patton JL, Schneider CJ, Smith TB (2000) Diversification of rainforest faunas: An integrated molecular approach. Annual Review of Ecology and Systematics 31, 533-563.

Patton JL, Dasilva MNF, Malcolm JR (1994) Gene genealogy and differentiation among arboreal spiny rats (Rodentia, Echimyidae) of the Amazon basin - A test of the riverine barrier hypothesis. Evolution 48, 1314-1323.

Reig OA (1986) Diversity patterns and differentiation of High Andean rodents. In: High altitude tropical biogeography (ed. Vuilleumier F, and M. Monasterio). Oxford University Press, New York.

Schauble CS, Moritz C (2001) Comparative phylogeography of two open forest frogs from eastern Australia. Biological Journal oftheLinnean Society 74, 157-170.

Seltzer GO, Rodbell DT, Baker PA, et al. (2002) Early warming of Tropical South America at the last glacial-interglacial transition. Science 296, 1685-1686.

Simpson BB (1979) Quaternary Biogeography of the High Montane Regions of South America. In: The South American Herpetofauna: Its Origin, Evolution, and Dispersal, (ed. Duellman WE), pp. 157-188. The University of Kansas, Lawrence.

Slatkin M (1985) Gene flow in natural populations. Annual Review of Ecology and Systematics 16, 393-430.

Soltis DE, Morris AB, McLachlan JS, Manos PS, Soltis PS (2006) Comparative phylogeography of unglaciated eastern North America. Molecular Ecology 15, 4261-4293.

Stuart-Fox DM, Schneider CJ, Moritz C, Couper PJ (2001) Comparative phylogeography of three rainforest-restricted lizards from mid-east Queensland. Australian Journal of Zoology 49, 119-127.

Templeton AR (1998) Nested Clade Analysis of phylogeographic data: testing hypotheses about gene flow and population history. Molecular Ecology 7, 381- 397.

Turner TF, Trexler JC (1998) Ecological and historical associations of gene flow in Darters (Teleostei: Percidae). Evolution 52, 1781-1801. 16

Turner TF, Trexler JC, Kuhn DN, Robinson HW (1996) Life-history variation and comparative phylogeography of Darters (Pisces: Percidae) from the North American Central Highlands. Evolution 50, 2023-2036.

Uthicke S, Benzie JAH (2003) Gene flow and population history in high dispersal marine invertebrates: mitochondrial DNA analysis of Hblothuria nobilis (Echinodermata: Holothuroidea) populations from the Indo-Pacific. Molecular Ecology 12, 2635- 2648.

Vaha JP, Erkinaro J, Niemela E, Primmer CR (2007) Life-history and habitat features influence the within-river genetic structure of Atlantic salmon. Molecular Ecology 16, 2638-2654.

Vaira M (2002) Anurans of a subtropical montane forest in northwestern Argentina: ecological survey and a proposed list of species of conservation concern. Biodiversity and Conservation 11, 1047-1062.

Vervoorst F (1982) Noroeste. In: Conservacion de la vegetacion natural de la Republica Argentina. Serie Conservacion de la Naturaleza. XVHIJornadas Argentinas de Botdnica, pp. 9-24. Fundacion Miguel Lillo, Tucuman.

Waltari E, Hoberg EP, Lessa EP, Cook JA (2007) Eastward Ho: phylogeographical perspectives on colonization of hosts and parasites across the Beringian nexus. Journal of Biogeography 34, 561-574.

Wenink PW, Baker AJ, Rosner H-U, Tilanus MGJ (1996) Global mitochondrial DNA phylogeography of holarctic breeding dunlins (Calidris alpina). Evolution 50, 318- 330.

Wiens JJ, Fetzner JW, Parkinson CL, Reeder TW (2005) Hylid frog phylogeny and sampling strategies for speciose clades. Systematic Biology 54, 719-748.

World Conservation Monitoring Centre (1992) Global biodiversity. Status of the Earth's Living Resources (ed. Groombridge B). Chapman & Hall, London.

Wright S (1951) The genetical structure of populations. Annals of Eugenics 15, 323-354.

Yu DH, Chu KH (2006) Low genetic differentiation among widely separated populations of the pearl oyster Pinctada fucata as revealed by AFLP. Journal of Experimental Marine Biology and Ecology 333, 140-146.

Zamudio KR, and H. W. Greene (1997) Phylogeography of the bushmaster (Lachesis muta: Viperidae): implications for neotropical biogeography, systematics, and conservation. Biological Journal of the Linnean Society 62, 421-442. 17

Zeil W (1979) The Andes as a geodynamic body. In: The Andes: a geological review., p. 126-184. Gebruder Borntraeger, Berlin. 18

CHAPTER 2. PLEISTOCENE CLIMATIC CYCLING AND DIVERSIFICATION OF THE ANDEAN TREEFROG, HYPSIBOAS ANDINUS.

2.1 Introduction South America encompasses a disproportionately large fraction of global biodiversity (World Conservation Monitoring Centre 1992), yet our understanding of the mechanisms responsible for its origins lags far behind that of the northern temperate biota of North America and Europe (e.g. Hewitt 1996,2000, A vise et al. 1998, Johnson & Cicero 2004, Soltis et al. 2006, Waltari et al. 2007). Comparatively fewer molecular phylogeographic studies of Neotropical compared to Holarctic species exist, with a decided skew in the former towards lowland taxa of lower latitudes (e.g. Patton et al. 1994, Lougheed et al. 1999, Costa 2003, Lougheed et al. 2006). On average, divergence for Neotropical taxa has been suggested to be deeper among lineages within "species" and among species as well (Hackett & Rosenberg 1990, Chek et al. 2003). However, such conclusions typically derive from broad-scale comparisons between biogeographic realms, rather than between comparable biomes, latitudes or regions within realms (but see Martin and McKay 2004). Montane regions have been proposed as important centres of species diversification (Simpson 1979, Fjeldsa 1994, Moritz et al. 2000). Thus, research focusing on factors that plausibly may have facilitated or impeded the genetic differentiation of taxa inhabiting Andean mountain habitats is of great relevance to our understanding of diversification and speciation in the Neotropics. Given the extensive latitudinal span of the Andes, the distinct orogenic and climatic histories of different regions (Simpson 1979), and particularly the evidence that the southern Andes were greatly affected by the last glacial maximum (Clapperton 1993, McCulloch et al. 2000), current evidence suggests that patterns and causes of diversification may vary along the entire Andean axis, particularly for the south-central region that may more closely mirror patterns from the temperate regions rather than lowland tropical ones. In the present study, we investigate timing and causes of diversification in a montane treefrog from southern Bolivia and northern Argentina. Diversification of South American taxa has been influenced by uplift of the Andes and Pleistocene climatic cycling (e.g. Fjeldsa 1994, Lougheed et al. 2000, Chesser 2000, 19

Hubert & Renno 2006). The Andes began to rise in the late Cretaceous more than 60 million years ago (my a), although it was not until the mid-Miocene (11-14 my a) that central regions began to exceed 1000 meters above sea level (masl) (Potts & Behrensmeyer 1992, Gregory-Wodzicki 2000). Major orogenesis of the ranges occurred in the last 10 my, resulting in the uplift to current elevations of greater than 4000 masl (Gregory-Wodzicki 2000). Global cooling began 2.4 mya, most notably with the advance of glaciers at high latitudes, with intense climatic cycling in the last 0.9 my (Potts & Behrensmeyer 1992). Tropical regions also experienced variations in temperature and aridity associated with glacial and interglacial cycles, altering the distribution and extent of various habitats (Vuilleumier 1971, Prance 1982, Potts & Behrensmeyer 1992, Hooghiemstra & van der Hammen 1998, Burnham & Graham 1999, Weng et al. 2007). The Andes thus have a rich geological history of uplift and climate change, with necessarily pronounced effects on the distribution of flora and fauna. Contemporary Northwestern Argentina possesses great diversity and complexity of habitats associated with the Eastern ranges of the Andes Mountains reflecting great spatial variation in precipitation and temperature (Handford 1988). The area is generally arid to semi-arid; however, heavy precipitation from humid summer air masses traveling from the northeast falls primarily on the eastern slopes of the Andean front ranges, resulting in longitudinal strips of forest stretching along the mountains south from Bolivia (an extension of the lowland Amazonian forests) to the province of Catamarca (~28° S). These montane forests are essentially continuous except for a narrow invasion of arid thorn scrub habitat in the province of Salta (~25° S, figure 2.1), which separates the southernmost tip of this forest peninsula (Brown et al. 2001), which is the southern limit of this humid forest ecotype. Away from these east-facing slopes rainfall is much lower resulting in the semi-arid chaco on the flat and slightly tilted terrain to the east, and the arid interior valleys and high plateau to the west (Handford 1988), both of which are possible dispersal barriers for taxa dependent on moist forests. Historical changes in elevation and climate likely resulted in changes in connectivity of habitats, which may have promoted diversification of taxa in this region. 20

64° 62° 60" i_L m: <1000ma.s.I. | | 18H Im N 1000-3000 m a.s.i. ||| >3000 m a.s.i. [ | 20* A

; Bolivia

»*»

:fsl

2

24 A 0 100 200 400 600 ZD Kilometers Figure 2.1. Sampling localities of Hypsiboas andinus in Northwestern Argentina (circles) and Bolivia (squares), and H. riojanus (triangle). Site names and sample sizes are listed in table 2.1 (p. 23). 21

Frogs are excellent taxa for studies of diversification and speciation as limited vagility tends to promote differentiation, and species vary markedly in fecundity and breeding habitats. Our chosen study species, Hypsiboas andinus (formerly Hyla andina; new name suggested by Faivovich et al. 2005, supported by Wiens et al. 2005), is a moderate-sized treefrog (50-60 mm) in the large family Hylidae (Faivovich et al. 2005). Hypsiboas andinus is a member of the Hypsiboas pulchellus Group (formerly Hyla pulchella Group), comprising 25 species distributed across southern South America, from the Brazilian lowlands in the state of Sao Paulo to the highlands of Peru, Bolivia and Argentina (Faivovich et al. 2004). Recent molecular phylogenies show that H. andinus is closely related to other Andean members of the group (H. balzani, H. marianitae, H. riojanus) (Faivovich et al. 2004,2005, Wiens et al. 2005). The relation between H. andinus and its putative sister species H. riojanus is unclear (Faivovich et al. 2004); therefore we obtained samples of H. riojanus and H. balzani for phylogenetic analyses to delineate an appropriate group for phylogeographic analyses (more detailed analyses of the four Andean taxa are underway, see discussion). Hypsiboas andinus occurs largely in deeply incised river valleys that penetrate into the Andes Mountains, from the province of Catamarca, Argentina (approximately 28°S) to north Bolivia (16°S), spanning various habitats with rising altitude, from humid montane forests (500 masl) to montane grasslands (>1500 masl) (Duellman etal. 1997). The complex distribution of habitats across Northwestern Argentina and the complex tectonic and glacial history of the region suggest a range of possible patterns of historical and present-day connectivity scenarios that could have impacted patterns of genetic differentiation of populations. Our study seeks (1) to better understand the roles of aforementioned contemporary and historical factors in shaping the distribution of genetic diversity of H. andinus across its range in Northwestern Argentina, the southern edge of the "tropical forest" ecosystem, and (2) to compare the levels of divergence and underlying causes of diversification in this southern montane frog species to other, lower latitude, Neotropical taxa, and to Nearctic anurans. 22

2.2 Methods Samples included in this study were either (i) field collected by the authors (258) or (ii) tissues loaned from the Museo Nacional de Ciencias Naturales (MNCN), Madrid, Spain (6) (table 2.1).

2.2.1 Field methods A total of 257 individuals of H. andinus from 23 sites across the Argentine portion of the species' range, approximately 700 km north to south, were collected under provincial and federal permits during several field expeditions (figure 2.1, table 2.1). We also collected H. riojanus at one site for outgroup analyses. For each locale, streams and surrounding areas were searched between 21:00- 03:00 depending on calling density and ease of capture. Males were located by advertisement calls and captured by hand (females and tadpoles were sampled if encountered). One or two frogs per site were killed by submersion in an anaesthetic (MS222). Specimens were used to verify identification and build a permanent reference collection. From these, liver tissue was removed and stored in 70% ethanol for subsequent molecular work. The frogs were then fixed in 10% formalin, stored in 70% ethanol, and deposited at Argentine Institutions (appendix 1). For the remainder, one or two toes were clipped to provide tissue for genetic studies and the frogs released at the site of capture. Toes were stored in 70% ethanol until DNA extraction.

2.2.2 DNA extraction, PCR and sequencing Samples collected in 1987 and 2001 were extracted using standard phenol/chloroform protocols (Sambrook & Russel 2001) and DNA was stored in IX Tris EDTA buffer. Samples collected in 2004-2006 and those on loan from MNCN were extracted using the DNEasy Tissue Kit® (Qiagen). A 663 bp fragment of mitochondrial control region was amplified using primers ControlP(H) and Wrev(L) from Goebel et al. (1999). PCR cocktails contained 10 ng genomic DNA, 2.5 mM MgCl2,4 |xg BSA, 0.2 \xM dNTP, 2 units Taq polymerase in a total volume of 20 ^L. For samples collected in 1987 and 2001, PCR products were electrophoresed in 1% agarose and purified by a modified "freeze-squeeze" method 23

Table 2.1. Sources of tissues and sample sizes for mtDNA control region analyses for each population. Site numbers refer to figure 2.1 (p. 20). Samples were collected by DK and/or SCL and/or PH in 1987, 2001, 2004, 2005 or 2006 unless otherwise noted. Haplotype diversity (h) and nucleotide diversity expressed as percent (it) are shown (standard deviations in parentheses) for all individuals in the population (values excluding lineage #2 individuals, where applicable, are italicized). Results from mismatch distributions for each population are as follows: ns- not significant (mismatch distribution is not significantly different from sudden expansion model), unim - unimodal distribution, bimod - bimodal distribution. Mismatch Site h K distribution 1 Tarija, Boliviaa 1

2 Santa Victoria, Salta 2 0 0

3 San Andres, Salta 10 0.93(0.06) 1.36(0.82) unim - ns

4 Tilcara, Jujuy 11 0.65(0.11) 0.26 (0.22) unim - ns

5 P. N. Calilegua, Jujuy 10 0.53(0.18) 0.24 (0.21) unim - ns

6 Lozano, Jujuy 19 0.57(0.13) 0.37 (0.28) unim - ns 7 Villa Monte, Jujuy 6 0 0

8 Maiz Gordo, Jujuy 10 0.36(0.16) 0.10 (0.12) unim - ns

9 Ruta de Cornisa, Salta 7 0.52 (0.24) 0.60 (0.43)

10 P. N. El Rey, Salta 5 0.70 (0.22) 0.24 (0.23)

11 Molinos, Salta 3 1.00(0.27) 2.00(1.62)

12 S. F. de Escoipe, Salta 31 0.64(0.09) 0.36 (0.26) unim - ns 13 Rio Piedras, Salta 5 0.40 (0.24) 0.12(0.15)

14 Rosario de la Frontera, Salta 7 0.29 (0.20) 0.08(0.11)

15 San Pedro, Tucuman 11 0.85(0.09) 2.44(1.38) bimod - ns 0.82 (0.10) 1.17(0.73) 16 Rio El Nio, Tucuman 20 0.81 (0.07) 0.92 (0.56) unim - ns Mismatch

Site h K distribution 17 Villa Nogues, Tucuman 14 0.85(0.07) 2.80(1.54) bimod - ns 0.80(0.10) 0.86(0.55) 18 La Angostura, Tucuman unim - ns 19 0.52(0.12) 0.44(0.31) 19 Rio Los Sosa, Tucuman 20 0.85(0.06) 0.78(0.49) unim - ns 20 Haulfin, Catamarca 4 0 0 21 Las Estancias, Catamarca bimod - ns 16 0.81(0.09) 2.40(1.31) 22 Rio San Ignacio, Tucuman 0.75(0.11) 0.48(0.34) bimod - P=0.022 21 0.89(0.04) 4.65(2.41) 23 Poman, Catamarca 0.69(0.15) 0.35(0.28) 6 0.80(0.17) 1.06(0.72) 24 Sanogasta, La Rioja 1 (H. riojanus) 25 Cochabamba, Bolivia3 2 1.0(0.5) 1.49(1.62)

26 Santa Cruz, Bolivia3 2 0 0

Tissue loaned from the Museo Nacional de Ciencias Naturales, Madrid, Spain.

(Thuring et al. 1975), squeezing the frozen gel pieces through filtered tips (200 \xL, Ultident Scientific, St. Laurent, QC) by centrifugation for two minutes at 7600 x g. Fragments were sequenced with the ControlP(H) primer using the Thermo Sequenase Radio-labeled terminator cycle sequencing kit according to manufacturer's protocols (Amersham BioSciences, Baie d'Urfe, QC), electrophoresed on vertical 6% polyacrylamide gels (PAA) and exposed to autoradiography film for 24-36 hours. All autoradiographs were scored by hand. For samples collected in 2004-2006 and those on loan from MNCN, PCR products were purified (Multiscreen; Millipore, Bedford, MA), and sequencing was performed using BigDye Terminator chemistry (version 3.1) and analyzed on ABI 3730x1 sequencers (Applied BioSystems, Foster City, CA). Bidirectional reads for a subset of 10 individuals resulted in identical sequences providing greater confidence that variation among haplotypes is not an artifact of PCR or sequencing error. 25

A 358bp fragment of cytochrome b was amplified for a subset of samples representing the diversity of control region haplotypes (see below), all Bolivian H. andinus samples, one H. riojanus, and one H. balzani using primers MVZ15-L from Goebel et al. (1999) and a modified version of H15149 (Lougheed et al. 1999). PCR and sequencing protocols were the same as for the control region fragment for samples collected in 2004-2006.

2.2.3 Analyses Sequence alignments were made in Clustal X, ver. 1.83 (Thompson et al. 1997) with subsequent visual verification. We used Arlequin ver. 2.000 (Schneider et al. 2000) to calculate haplotype and nucleotide diversity indices for control region sequences for each population using Kimura-2-Parameter (K2P) molecular distance. We categorized the results according to Grant and Bowen (1998) as reflecting (1) prolonged bottleneck

(/K0.5 and JI<0.5%), (2) rapid population growth from ancestral population with low effective population size (h>0.5 and Jt<0.5%), (3) a brief bottleneck (h<0.5 and JT>0.5%), or (4) stable population with large historical effective population size, or secondary contact among differentiated lineages (A>0.5 and JO0.5%). Although the boundaries appear sharp, Grant and Bowen (1998) have included nucleotide diversities up to 0.71% in category 2. We performed mismatch distributions for all populations with sample size greater than 10 and tested against a model of sudden expansion using the generalized non-linear least-squares approach (Schneider & Excoffier 1999). Model validity was evaluated using 1000 parametric bootstraps as implemented in Arlequin. To generate our phylogenetic trees, we used a combined data set of control region and cytochrome b for a subset of H. andinus samples collected by us in Northwestern Argentina (n=33, representing 18 of the 23 populations); one H. riojanus collected by us in Northwestern Argentina; five H. andinus collected in Bolivia plus one H. balzani as outgroup. We used ModelTest (v. 3.7, Posada & Crandall 1998), executed with PAUP*, to select the best model of evolution as estimated by the Akaike Information Criterion (AIC) for subsequent maximum likelihood (ML) analysis. ML analyses were run using 10 random-addition replicates and we evaluated support of the resulting topologies using 26

100 nonparametric bootstraps with PAUP*. We also performed Bayesian analyses (MR.BAYES 3.1.2, Huelsenbeck & Ronquist 2001, Ronquist & Huelsenbeck 2003) to compare topologies and support. Bayesian analyses were performed using two independent runs of 106 generations implementing Metropolis-coupled MCMC using four incrementally heated Markov chains, sampling every 100 generations. Convergence of the two runs was assumed if the average standard deviation of the split frequencies was less than 0.01 and the Potential Scale Reduction Factor (PSRF) was at 1.00. We used only control region data for phylogeographic analyses. Based on the results of our phylogenetic analyses, we included all H. andinus from Northwestern Argentina and one individual from southern Bolivia, as well as one individual of H. riojanus. TCS version 1.21 (Clement et al. 2000) was used to generate a Maximum Parsimony Network (MPN) representing the genealogical relationships among control region haplotypes. A Nested Clade Analysis (NCA; Templeton et al. 1992, Templeton et al. 1995, Templeton 1998) was applied to integrate genealogical information with geographic distributions. Ambiguities were resolved following the rules of Templeton and Sing (1993). All distance measure calculations and permutations (n=1000) were assessed using GeoDis version 2.5 (Posada et al. 2000) and the significant results were then interpreted using a published inference key (11 November 2005, available at http://darwin.uvigo.es/). Cognizant of criticisms of NCA due to the possibility of false- positives (Petit 2007), we performed additional analyses to aid us in interpreting patterns. These included the supplementary tests for secondary contact where appropriate (Templeton 2001), mismatch distribution analyses and comparisons of haplotype (h) and nucleotide diversity (JT) for each of the main clades at the total cladogram level in Arlequin.

2.2.4 Timing of divergence To estimate the time of divergence among clades we used only cytochrome b sequences because molecular evolution of this gene is better understood than that of control region. We calculated corrected average pairwise differences among the clades using the K2P corrected molecular distance using Arlequin. We applied a slower poikilothermic molecular clock typically used for anuran cytochrome b (see Austin et al. 27

2004) with a rate of change between 0.5 and 1% per million years. We also used a coalescent approach as implemented in the program BEAST v. 1.4.6 (Drummond & Rambaut 2007). We performed two independent runs of 20 million generations each with burn-ins of 2000000, which were then combined in TRACER v. 1.4 (Rambaut & Drummond 2007). We employed a GTR + I + G model of evolution with 6 rate categories and assumed a relaxed lognormal clock (using rates of divergence of 0.5 and 1% per million years). Parameters were sampled every 1000 generations. All other initial parameters settings were the default provided by BEAST v. 1.4.6.

23 Results 2.3.1 Molecular diversity A total of 340 bp of the mitochondrial control region were obtained for each of 257 H. andinus individuals sampled from across the entire Argentine range of the species, five individuals from Bolivia, plus one individual ofH. riojanus (GenBank Accession EU403157-EU403420). Sixty-eight haplotypes were defined by 91 polymorphic sites (60 parsimony informative), differing by 1-41 base-pair substitutions. One shared deletion was observed in two individuals (representing two different haplotypes), and one insertion was detected in the outgroup. Accordingly, gaps were treated in subsequent analyses as a 5th state (where possible). We found an overall AT nucleotide bias (T = 32.6%, A = 34.2%, G = 12.0%, C = 21.2%), as is characteristic of vertebrate mtDNA (Saccone et al. 1987). Haplotype and nucleotide diversity indices for each population are listed in table 2.1. A total of 358 bp of cytochrome b was obtained for the subset of samples used for phylogenetic analyses (GenBank Accession EU403117-EU403156). A total of 22 haplotypes were defined by 59 polymorphic sites (33 parsimony informative), differing by 1-37 base-pair substitutions. As for control region, an overall AT nucleotide bias was evident (T = 34.6%, A = 24.2%, G = 15.9%, C = 25.3%). No stop codons were found in the fragment. 28

2.3.2 Phylogenetic analyses on the combined dataset The combined data set of control region and cytochrome b comprised 699 bp for a subset of 33 individuals of H. andinus from Northwestern Argentina and five from Bolivia, one individual ofH. riojanus, and one individual of H. balzani as an outgroup.The best-fit model of evolution was the general time reversible model with a set proportion of invariable sites (1=0.3863), with gamma distributed rates (G=0.3505) (GTR+I+G). Trees derived from both ML and Bayesian methods were similar, with generally stronger support for the clades obtained by Bayesian analyses (figure 2.2). Two main clades were evident. Lineage #1 contained H. andinus from Northwestern Argentina and one of five individuals of//, andinus from Bolivia (southern Bolivia, site #1). Lineage #2 contained the single H. riojanus plus several H. andinus individuals from the southern portion of the Argentine range. The remaining four H. andinus samples from Bolivia (all in northern Bolivia) did not form a single clade but are basal to both lineage #1 and #2.

2.3.3 Phylogeographic patterns We used all 257 individuals from Argentina as well as one sample from southern Bolivia and one sample of//, riojanus for phylogeographic analyses. Northern Bolivian samples were excluded due to small sample size and sparse coverage of that portion of the species range. The resulting two maximum parsimony networks could not be connected at the 95% parsimony level and were separated by a minimum of 21 mutational steps exceeding the parsimony limit of seven steps. We therefore analysed these independently (figure 2.3). Network #1 contained only individuals identified as H. andinus, and network #2 contained the single H. riojanus sample plus additional samples identified as H. andinus as found in the phylogenetic analyses (lineage #1 and #2, respectively). Some ambiguous loops in network #1 could not be resolved following the rules of Templeton and Sing (1993) (figure 2.3); therefore we used the combined data set tree as a guide for the nesting procedure. Hierarchical nesting resulted in 4 nested levels in network #1. Seven nested clades showed significant geographical associations of haplotypes based on the results of our permutation tests (table 2.2). Two clades (2-1, 2-6) -23 -29 Lineage 1: North 28 1.00 h26 0. !£j— 3

0.99kiJ— 7 n-6 Lineage 1: Central 1.00 -5 15 10 1 -1 — 16 14 0.97 39 I-37•5 0 1.00 u37 ^45 43 Lineage 1: South -47 0.73[—3 1 31 °H h-35 31 t -32 Lineage 1: single 0.72 •30^" haplotype -CE 0.85 CF -CB 1.00 Lineage 2 •— CA 1.00 CI 0.971-(C J CM B3 1-001i BB1 Northern Bolivia B2 H. balzani 0.01

Figure 2.2. Phylogenetic relationships of Argentine and Bolivian Hypsiboas andinus and H. riojanus based on Bayesian analyses of cytochrome b and control region sequences, using H. balzani as an outgroup. Posterior probability values are indicated for each branch, filled circles indicate values less than 0.70. Haplotype codes correspond to those shown in figure 2.3 (p. 30) and listed in table 2.1 (p. 23). Boldface haplotypes: #28 = southern Bolivia, population #1; #CJ = sample ofH. riojanus, population #24. 30

network #1 "W if rx % ? /VS

network #2

;.CB_

lr\

1-3 CK—CI • ®

Figure 2.3. Maximum parsimony networks using mtDNA control region from 259 individuals. Filled circles indicate missing or unsampled haplotypes and font size approximates relative sample size for each haplotype. Nested clades are indicated with rectangles: thin solid lines for 1-step clades, dotted for 2-step clades, thick solid for 3-step clades, and dashed for 4-step clades. The two networks could not be connected at the 95% parsimony level and were analysed separately. Underlined haplotypes are those included in the combined dataset for phylogenetic analyses (haplotype CJ = H. riojanus). 31

Table 2.2. Nested clade analysis results for network #1 (a) and network #2 (b) clades tested using 1000 permutations. Significant geographical associations from permutation tests indicated in boldface. Refer to figure 2.3 (p. 30) for clade numbers, a. Network #1 Clade Permutational Chi-square statistic Probability 1-1 161.01 0.016 1-2 4.95 0.888 1-3 4.00 0.171 1-5 0.54 1.000 1-8 6.00 0.189 1-9 3.00 0.318 1-12 32.34 0.139 1-13 25.42 0.052 1-16 12.00 0.280 1-17 3.00 0.329 1-18 0.12 1.000 2-1 137.90 <0.001 2-2 0.47 1.000 2-3 0.22 1.000 2-4 0.83 1.000 2-6 60.81 0.007 3-2 19.00 <0.001 3-3 0.24 1.000 3-5 12.00 0.012 4-2 9.95 0.111 4-3 32.46 0.001 Total Cladogram 408.30 <0.001 b. Network #2 Clade Permutational Chi-square statistic Probability 1-1 38.16 0.019 2-1 1.48 1.000 2-2 5.00 0.208 2-3 3.00 1.000 3-1 20.00 0.002 Total Cladogram 19.29 0.014

showed evidence of restricted gene flow, two (1-1,4-3) showed a signal of contiguous range expansion, and three clades (3-2, 3-5, total cladogram) showed patterns suggestive of past fragmentation, possibly coupled with range expansion or long distance colonization (table 2.3). At the total cladogram level, three distinct clades showed a 32

north to south distribution with little geographical overlap, namely clade 4-2 (north), the large clade 4-1 (central) and clade 4-3 (south) (figure 2.4). We assigned clade 4-1 as interior and 4-2 and 4-3 as tips because the former contained haplotype #1, which had the highest outgroup probability and therefore was likely to be the oldest haplotype in the network (Castelloe & Templeton 1994). A single distinct haplotype (#30) could not be grouped with any of the three clades; the haplotype was from two individuals from two different populations (sites #15 and 16). The most common haplotype (#1) was found in 65 individuals from 9 populations. Clade 1-1, which contained haplotype #1, exhibited a star-burst haplotype topology typical of the inferred range expansion. Six populations (#6,9,11,15, 16, and 17) showed patterns consistent with secondary contact among divergent lineages based on the supplemental tests from Templeton (2001) (data not shown). Populations #6 and #9 contained haplotypes from clades 4-1 and 4-2, whereas the remaining populations shared haplotypes from clades 4-1 and 4-3. In all cases the majority of individuals in a population contained haplotypes from their "home" clade and only one or two individuals contained haplotypes from a geographically adjacent clade. Populations #21 and #22 also contained haplotypes from network #1 (clade 4-3) and network #2. We looked for evidence of range expansion at the population level using mismatch distribution, and evaluating haplotype (h) and nucleotide (K) diversity using the approach of Grant and Bowen (1998). Mismatch distributions for all populations (with n>10) except Rio San Ignacio (#22) were consistent with population growth, as the distributions of pair-wise differences did not differ significantly from the sudden expansion model (p>0.05). Many populations showed high h and low it, also indicative of population growth according to Grant and Bowen (1998) (table 2.1). Populations #15, 17, and 21 showed bimodal distributions, although the deviation was not significant. Population #22 differed significantly from the sudden expansion model and showed a strongly bimodal distribution. Populations #15, 17, 21, and 22 also showed both high h and it implying large stable populations or secondary contact between differentiated lineages (Grant & Bowen 1998). For populations containing haplotypes from networks #1 and #2 we calculated h and it for all individuals and excluding individuals with network #2 haplotypes (table 2.1). 33

For network #1, clade 4-1 (central) and clade 4-3 (south) exhibited patterns characteristic of rapid population expansion using both mismatch distributions (4-1: x=l .163, p=0.67,4-3: T=2.441, p=0.39) and haplotype and nucleotide diversity patterns

(4-1: h=0.69 and JI=0.355%, 4-3: h=0.S2 and JT=0.588%). Clade 4-2 (north) had a multimodal mismatch distribution (p=0.07), and both high h (0.87) and jt (1.256%) suggesting either a stable population or secondary contact. Hierarchical nesting resulted in four levels for network #2. Three clades showed significant geographical associations (table 2.2) but evaluation of inferences was not possible due to low sample size (tip/interior status could not be determined). All individuals in populations #20 and #23 contained network #2 haplotypes, although some individuals from 4 other populations also contained network #2 haplotypes (#15,17,21, 22). For network #2, clade 3-1 showed high haplotype diversity (0.91) and low to moderate nucleotide diversity (0.627%) likely due to rapid population expansion. Given

Table 2.3. Inferences made for each clade in network #1 found to have significant geographical associations. Refer to figure 2.3 (p. 30) for clade numbers. Clade Chain of Inference Inference 1-1 1-2-11-12-NO Contiguous range expansion 2-1 1-2-3-5-6-7-YES Restricted gene flow/dispersal but with some long distance dispersal 2-6 1-2-3-5-6-7-YES Restricted gene flow/dispersal but with some long distance dispersal 3-2 1-2-3-4-9-NO Allopatric fragmentation 3-5 1-2-11-12-13-YES Long distance colonization possibly coupled with subsequent fragmentation OR past fragmentation followed by range expansion 4-3 1-2-11-12-NO Contiguous range expansion Total 1-2-11-12-13-YES Long distance colonization possibly coupled with Cladogram subsequent fragmentation OR past fragmentation followed by range expansion 34 the three unique and divergent haplotypes found in clade 3-2 of network #2, the haplotype and nucleotype diversity were both high (h=1.0, jt=1.319%).

2.3.4 Timing of divergence We estimated divergence times using cytochrome b at two distinct levels. At the deepest level, between network #1 (H. andinus clade) and network #2 (H. riojanus clade), the corrected average pairwise K2P molecular distance was 3.1%. Using rates of 0.5 and 1% per million years suggests a divergence time between the H. andinus clade and H. riojanus clade of 3-6 mya. Within network #1 only, the K2P distances of less than 1% (north-central=0.84%, central-south=0.56%, north-south=0.27%) imply divergence times between 1 and 2 mya. Divergence times calculated using a coalescent approach (table 2.4) were similar. Again the deepest divergences for network #1 were all less than 2 million years ago, placing them within the Pleistocene (table 2.4).

Table 2.4. Estimates of divergence time based on analyses using BEAST (see text for details). The analyses were run using a molecular clock of 1% and 0.5% per million years. Clades are identified as for the phylogeographic analyses. Refer to figure 2.3 (p. 30) for clade numbers. TMRCA = time to most recent common ancestor, mya = million years ago.

TMRCA (mya)

Clade 1% 0.5%

Networks #1 & #2 2.167 4.331

Network #1 0.920 1.846

Clade 4-1 0.144 0.286

Clade 4-2 0.528 1.048

Clade 4-3 0.840 1.684

Network #2 0.792 1.600 35

68° 66° 64° 62°

Figure 2.4. Geographical distribution of 4-step clades from network #1. Populations within the 3 main groups of network #1 that also contain haplotypes from network #2 are indicated with unfilled circles. Populations #20,23 and 24 contain only network #2 haplotypes. Populations #25 and 26 were not included in the analyses. Shading represents elevation as in figure 2.1 (p. 20). 36

2.4 Discussion Both historical and contemporary connectivity among populations play important roles in shaping evolutionary trajectories of lineages. In highly heterogeneous regions like the Andes, there is high potential for isolation and for spatial habitat variation to affect the movement of individuals. Unsurprisingly, Hypsiboas andinus showed significant phylogeographic structuring across its Argentine range. The patterns likely represent an interaction between contemporary restriction of gene flow and historical barriers that caused range fragmentation with later secondary contact. Our study supports the growing body of research illustrating how historical events have impacted genetic diversity in the Andes, and especially the role more recent Pleistocene climate cycling may play in population differentiation.

2.4.1 Divergent lineages within the H. andinus/H. riojanus complex While the relationships among H. andinus and other members of the H. pulchellus Group are uncertain (Barrio 1965, Duellman et al. 1997, Faivovich et al. 2004), recent work unambiguously shows Andean members of the group (H. balzani, H. marianitae, H. andinus, H. riojanus) to be monophyletic within the broadly distributed (eastern Brazilian and Argentinian lowlands westward to the highlands of Peru, Bolivia and Argentina) H. pulchellus Group (Faivovich et al. 2004, 2005, Wiens et al. 2005). Although H. andinus and H. riojanus together form a monophyletic group (Faivovich et al. 2004), our findings indicate that the relationship between these two taxa may be complicated. A single individual of H. andinus from southern Bolivia (population #1, haplotype #28) clustered with populations from northern Argentina (northern lineage, clade 4-2), but the four samples from northern Bolivia (populations #25 and #26) are genealogically separate. The taxonomic status of Bolivian and Argentine populations of H. andinus is not well established (the studies by Faivovich et al. 2004 and Faivovich et al. 2005 did not contain samples from Bolivia). Our preliminary analyses imply that H. riojanus is nested within H. andinus, making H. andinus a paraphyletic taxon. We did find two distinct, divergent lineages, but 27 individuals from 6 populations, 5 of which were asserted to contain only H. andinus (Cei 1980, Duellman et al. 1997), cluster with H. riojanus (figures 2.2, 2.3). This may be due to a sympatric distribution of two 37 morphologically similar taxa (i.e. cryptic species) in the provinces of Tucuman and Catamarca, or to secondary contact between divergent lineages. Nuclear loci are currently being developed to distinguish between these two scenarios. All samples collected at our sites #20 and #23 cluster with H. riojanus, and 4 other sites contain both mitochondrial lineages. The only morphological feature known to distinguish the two species is the presence of a dashed white or cream dorsolateral stripe that begins behind the eyes in H. andinus (Cei 1980, Faivovich et al. 2004), but the extensive colour and pattern variability seen in H. andinus shows that this is not a reliable diagnostic feature. Given that individuals representing both mitochondrial lineages can be found breeding in the same locale, only strong reproductive isolating mechanisms would prevent hybridization, if indeed the two are different species. Call differences, presumably the main cue for mate selection, are minimal between the taxa (Barrio 1965) although playback and female choice experiments have not been reported. Limited sonographic analyses of H. andinus from five of the populations sampled suggest that calls are highly variable within and among populations (Liadsky 2003). The 3.1% divergence at cytochrome b between the lineage representing H. andinus alone (network #1) and that containing both H. riojanus + H. andinus (network #2) is lower than between-species divergences reported for many Neotropical frogs (Chek et al. 2001, Symula et al. 2003, Camargo et al. 2006). The two dating methods we used suggest a divergence time between 2 and 6 million years ago between these networks, a time spanning both major uplift of the Andes Mountains, and the formation of arid high altitude habitats (Gregory-Wodzicki 2000) that would have been significant barriers to movement. During uplift, populations would have become isolated, especially if drainage patterns, the presumed main paths of connectivity, changed. Following such isolation, subsequent range expansion, perhaps during warmer and moister interglacial periods, would have resulted in secondary contact of the H. riojanus lineage and southern populations of H. andinus. The current evidence — limited if any morphological and call differences and shallow genetic divergence -- suggests that the two taxa are probably not reproductively isolated species. 38

2.4.2 Phylogeography ofH. andinus Our NCA and mismatch analyses implied that both demographic and historical processes affected the patterns of differentiation in H. andinus. Frogs generally have restricted movements (less than 10 km) due to their small size, biphasic life history, water dependence, and philopatry (reviewed in Marsh & Trenham 2001). Hypsiboas andinus is a mountain stream breeder, and is never found far from water (Cei 1980, personal observation). Andean streams flow through various grassland, forest, and scrub habitats as they descend from the mountains and unite in the eastern semi-arid chaco, which is outside of the range ofH. andinus. Connectivity of populations therefore would be constrained by the drainage patterns connecting streams, and/or by the probability of their crossing through moist habitats that might lie between water courses. Given that the streams flow through steep valleys and that connectivity of drainages is limited within the range of the species, and given its physiological limitations to movement, we expect that dispersal for this species is usually low except within the immediate local stream system. Limited dispersal in montane regions has been reported for several amphibians and topographical ridges have been implicated as strong barriers that promote marked differentiation among populations (Lougheed et al. 1999, Funk et al. 2005, Lowe et al. 2006). Similar to other montane frogs, H. andinus shows evidence of restricted gene flow resulting in the genetic structuring of populations. Five clades (1-1,3-2, 3-5,4-3, total cladogram) were inferred to show patterns consistent with past fragmentation, long distance colonization and/or range expansion. Given the presumptive low vagility ofH. andinus, especially among drainage basins, we believe that long distance movement is unlikely in this species. The distribution of H. andinus in highly heterogeneous montane habitats would suggest that fragmentation may have a much more pronounced effect. At the total cladogram level for network #1, both NCA on the control region fragment and phylogenetic analysis of the combined dataset support four lineages: clades 4-1,4-2, and 4-3, plus one divergent haplotype (#30). The three main clades of network #1 form a north-south series with some geographic overlap between them (figure 2.4), and probably diverged less than 2 mya. This span falls within the Pleistocene and corresponds to major climatic cycling and probable vegetational changes in the northern 39

Argentine montane vegetation. Major global cooling and cycling between glacial vs. interglacial, reflecting cool and arid vs. warm and moist periods, respectively, was most pronounced in the last 0.9 my (Potts & Behrensmeyer 1992, Hooghiemstra & van der Hammen 1998, Burnham & Graham 1999). Tropical regions experienced climate instability resulting in contraction and expansion of different habitat types, especially tropical forests (Vuilleumier 1971, Prance 1982, Potts & Behrensmeyer 1992, Hooghmiestra & van der Hammen 1998, Burnham & Graham 1999, Weng et al. 2007). Effects of climatic cycling on montane regions are particularly difficult to reconstruct given the rugged topography, the complexity of related changes in temperature and precipitation, and the interplay between the two factors, but there is little doubt that Pleistocene climatic cycling resulted in major shifts in Andean habitat. At glacial maxima snow lines may have descended by as much as 1000m, and global temperatures may have decreased by 6°C (Vuilleumier 1971, Potts & Behrensmeyer 1992). High altitude habitats expanded to lower altitudes (Vuilleumier 1971, Prance 1982, Weng et al. 2007) while in the lowlands drier habitats also expanded in many areas (Vuilleumier 1971, Prance 1982, Burnham & Graham 1999), fragmenting moist forests. Vegetational distribution undoubtedly strongly influenced the movements and ranges of many animal taxa, likely isolating populations. Hypsiboas andinus currently does not inhabit the high altitude puna nor the low altitude seasonally dry chaco: it is restricted to the moister woodlands and grasslands typically found on mountain slopes between these two drier habitats. Thus we propose that Hypsiboas andinus populations became isolated during glacial maxima in the pockets of mesic forest that remained. Paleoclimatic data from northwestern Argentina are too limited to identify unambiguously potential mesic refugia that correspond to each of the three clades identified here. However, two contemporary well-defined areas, at ~22°S and ~27°S, of especially moist habitat could have retained their mesic character during glacial dry periods and thus supported montane forests and dependent populations of taxa such as Hypsiboas andinus. These two areas also contain high numbers of endemic taxa and have been suggested as potential refugia during glacial intervals (Brown et al. 2001, Quiroga & Premoli 2007). Wet sites in the northern area (~22°S) around the Argentine/Bolivian border in the Upper Bermejo River Basin currently receive more than 40

2300 mm of rain annually (Grau & Brown 2000). The southern mesic region, west of Concepcion, Tucuman (~27°S), receives more than 1400 mm precipitation annually (Grau & Veblen 2000). The high elevation Cordilleras (> 5000 masl) to the west of these regions trap much of the precipitation from the humid summer air masses moving south- westward. Additionally, the concave shape of the Cordillera Aconquija to the west of Concepcion contributes to the trapping of rainfall. On a larger scale, the similar concave topography in the northern Andes (between 5° and 15° south) has been implicated in maintaining moist Amazon rainforest during the same glacial periods (Hooghiemstra & van der Hammen 1998). The area between these two sites (~24° to 26° south) currently contains narrow bands of rainforest (Sierras de Metan, Lumbrera, Santa Barbara) that are separated by chacoan arid thorn scrub, effectively isolating the southernmost extension of moist Andean forest. During dry glacial periods the arid thorn scrub almost certainly extended its range, probably completely severing the connection between the southern (26° to 27° S) and the northern (<24° S) moist forests. The three network #1 clades described here correspond closely with these three areas: northern moist, central drier, southern moist. Clade 4-3, found in the southern moist region, and clade 4-1, found in the central dry region, show evidence of secondary contact of divergent lineages, as well as contiguous range expansion based on the NCA, mismatch distributions and haplotype and nucleotide diversity patterns, lending further support to range contraction during dry glacials and subsequent range expansion during moister interglacial periods. Further work in NW Argentina will clarify the role of historical climatic fluctuation, and the resultant habitat changes, in isolation and diversification of Hypsiboas andinus and other taxa.

2.4.3 Diversification in the Andes Depths of divergence among populations have been shown to differ across latitudes (as does species richness), both being greater at lower latitudes (Martin & McKay 2004). The differentiation events inferred for the whole H. andinus and H. riojanus complex (2- 6 my a), as well as for clades within the northern network #1 (less than 2 my a), are more recent than those found for many other Neotropical taxa (birds: Voelker 1999, Burns & Naoki 2004, Cheviron et al. 2005; mammals: Patton et al. 2000, Costa 2003; frogs: Chek 41 et al. 2001, Symula et al. 2003, Camargo et al. 2006, Lougheed et al. 2006). These studies suggest that much of the diversification in the Amazonian region occurred during the Pliocene, whereas few taxa have been shown to have levels of divergence shallow enough to coincide with Pleistocene climatic fluctuation (Garcia-Moreno etal. 1999, Chesser 2000, Noonan & Gaucher 2005, Rull 2006). These latter levels of divergence are more typical of divergence times of temperate anurans (Green et al. 1996, Masta et al. 2002, Martinez-Solano 2004, Nielsen et al. 2006). Although relatively few studies exist for the Neotropics, especially prominent are the increasing number of datasets, including ours, supporting the much shallower divergences of vertebrate taxa in the Andes as compared to the Amazon Basin (reviewed in Moritz et al. 2000) suggesting that the Andes may not mirror patterns typically associated with lowland low-latitude South America. Factors that affect diversification rates in high latitude temperate regions such as seasonality, climatic cycling, range shifting and recolonization (Martin & McKay 2004), may also contribute to shallower divergences for higher altitude regions as well as high latitude regions in southern South America. Faivovich et al. (2004) suggested that the Hypsiboas pulchellus Group originated in the Atlantic Forest of Southeastern Brazil. The four taxa discussed above are the only Andean representatives in this clade - likely a result of a recent dispersal event of the ancestor into the Andes. Since the late Tertiary, the Atlantic Forest was probably intermittently connected by two or three pathways with the lowland Amazon forests (Costa 2003), and this may have been the route taken by the ancestor of the Andean members of the H. pulchellus Group to reach the Andes. The low level of differentiation and distinct geographic distribution from other members of the H. pulchellus Group suggests a radiation of the Andean taxa perhaps less than ten million years ago. However, future work with more extensive sampling and additional markers will clarify evolutionary affinities and refine estimates of the timing of divergence, particularly of the Bolivian populations of H. andinus. 42

2.5 References

Austin JD, Lougheed SC, Boag PT (2004) Discordant temporal and geographic patterns in maternal lineages of eastern north American frogs, Rana catesbeiana (Ranidae) and Pseudacris crucifer (Hylidae). Molecular Phylogenetics and Evolution 32,799- 816.

Avise JC, Walker D, Johns GC (1998) Speciation durations and Pleistocene effects on vertebrate phylogeography. Proceedings of the Royal Society of London Series B- Biological Sciences 265,1707-1712.

Barrio A (1965) Las subespecias de Hyla pulchella Dumeril y Bibron (Anura, Hylidae). Physis 69,115-128.

Brown AD, Grau HR, Malizia LR, Grau A (2001) Argentina. In: Bosques nublados del neotropico (eds. Kappelle M, Brown AD). Instituto Nacional de Biodiversidad, Costa Rica.

Burnham RJ, Graham A (1999) The History of Neotropical Vegetation: New Developments and Status. Annals of the Missouri Botanical Garden 86,546-589.

Burns KJ, Naoki K (2004) Molecular phylogenetics and biogeography of Neotropical tanagers in the genus Tangara. Molecular Phylogenetics and Evolution 32, 838-854.

Camargo A, De Sa RO, Heyer WR (2006) Phylogenetic analyses of mtDNA sequences reveal three cryptic lineages in the widespread neotropical frog Leptodactylus fuscus (Schneider, 1799) (Anura, Leptodactylidae). Biological Journal of the Linnean Society 87, 325-341.

Castelloe J, Templeton AR (1994) Root probabilities for intraspecific gene trees under neutral coalescent theory. Molecular Phylogenetics and Evolution 3,102-113.

Cei JM (1980) Amphibians of Argentina Monitore Zoologico Italiano N. S Monografia 2, Florence.

Chek AA, Austin JD, Lougheed SC (2003) Why is there a tropical-temperate disparity in the genetic diversity and taxonomy of species? Evolutionary Ecology Research 5,69- 77.

Chek AA, Lougheed SC, Bogart JP, Boag PT (2001) Perception and history: Molecular phylogeny of a diverse group of Neotropical frogs, the 30-chromosome Hyla (Anura: Hylidae). Molecular Phylogenetics and Evolution 18, 370-385.

Chesser RT (2000) Evolution in the High Andes: The phylogenetics of Muscisaxicola Ground-Tyrants. Molecular Phylogenetics and Evolution 15,369-380. 43

Cheviron ZA, Hackett SJ, Capparella AP (2005) Complex evolutionary history of a Neotropical lowland forest bird (Lepidothrix coronata) and its implications for historical hypotheses of the origin of Neotropical avian diversity. Molecular phylogenetics and Evolution 36, 338-357.

Clapperton CC (1993) Quaternary Geology and Geomorphology of South America. Elsevier, Amsterdam.

Clement M, Posada D, Crandall KA (2000) TCS: a computer program to estimate gene genealogies. Molecular Ecology 9, 1657-1659.

Costa LP (2003) The historical bridge between the Amazon and Atlantic Forest of Brazil: a study of molecular phylogeography with small mammals. Journal of Bio geography 30,71-86.

Drummond AJ, Rambaut A (2007) BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evolutionary Biology 7, 214.

Duellman WE, de la Riva I, Wild ER (1997) Frogs of the Hyla armata and Hyla pulchella groups in the Andes of South America, with definitions and analyses of phylogenetic relationships of Andean groups of Hyla. Scientific Papers of the Natural History Museum of the University of Kansas 3,1-41.

Faivovich J, Garcia PCA, Ananias F, et al. (2004) A molecular perspective on the phylogeny of the Hyla pulchella species group (Anura, Hylidae). Molecular Phylogenetics and Evolution 32,938-950.

Faivovich J, Haddad CFB, Garcia PCA, et al. (2005) Systematic review of the frog family Hylidae, with special reference to Hylinae: Phylogenetic analysis and taxonomic revision. Bulletin of the American Museum of Natural History 294, 1-240.

Fjeldsa J (1994) Geographical patterns for relict and young species of birds in Africa and South America and implications for conservation priorities. Biodiversity and Conservation 3, 207-226.

Funk WC, Blouin MS, Corn PS, et al. (2005) Population structure of Columbia spotted frogs (Rana luteiventris) is strongly affected by the landscape. Molecular Ecology 14, 483-496.

Garcia-Moreno J, Arctander P, Fjeldsa J (1999) A Case of Rapid Diversification in the Neotropics: Phylogenetic Relationships among Cranioleuca Spinetails (Aves, Furnariidae). Molecular Phylogenetics and Evolution 12, 273-281.

Goebel AM, Donnelly JM, Atz ME (1999) PCR primers and amplification methods for 12S ribosomal DNA, the control region, cytochrome oxidase I, and cytochrome b in Bufonids and other frogs, and an overview of PCR primers which have amplified DNA in amphibians successfully. Molecular Phylogenetics and Evolution 11,163- 199. 44

Grant WS, Bowen BW (1998) Shallow population histories in deep evolutionary lineages of marine fishes: insights from sardines and anchovies and lessons for conservation. Journal of Heredity 89,415-426.

Grau A, Brown AD (2000) Development threats to biodiversity and opportunities for conservation in the mountain ranges of the Upper Bermejo River Basin, NW Argentina and SW Bolivia. Ambio 29,445-450.

Grau HR, Veblen TT (2000) Rainfall variability, fire and vegetation dynamics in neotropical montane ecosystems in north-western Argentina. Journal of Biogeography 27, 1107-1121.

Green DM, Sharbel TF, Kearsley J, Kaiser H (1996) Postglacial range fluctuations, genetic subdivision and speciation in the Western North American Spotted Frog Complex, Ranapretiosa. Evolution 50, 374-390.

Gregory-Wodzicki KM (2000) Uplift history of the Central and Northern Andes: A review. Geological Society of America Bulletin 112,1091-1105.

Hackett SJ, Rosenberg KV (1990) Comparison of phenotypic and genetic differentiation in South American antwrens Formocariidae. Auk 107,473-489.

Handford P (1988) Trill rate dialects in the Rufous-collared sparrow, Zonotrichia capensis, in northwestern Argentina. Canadian Journal of Zoology 66, 2658-2670.

Hewitt G (2000) The genetic legacy of the Quaternary ice ages. Nature 405,907-913.

Hewitt GM (1996) Some genetic consequences of ice ages, and their role in divergence and speciation. Biological Journal of the Linnean Society 58, 247-276.

Hooghiemstra H, van der Hammen T (1998) Neogene and Quaternary development of the neotropical rain forest: the forest refugia hypothesis, and a literature overview. Earth-Science Reviews 44, 147-183.

Hubert N, Renno J-F (2006) Historical biogeography of South American freshwater fishes. Journal of Biogeography 33, 1414-1436.

Huelsenbeck JP, Ronquist F (2001) MRBAYES: Bayesian inference of phylogeny. Bioinformatics 17, 754-755.

Johnson NK, Cicero C (2004) New mitochondrial DNA data affirm the importance of Pleistocene speciation in North American birds. Evolution 58,1122-1130.

Liadsky D (2003) Call variation in the Andean Treefrog, Hyla andina Bachelor of Science (Honours) thesis, Queen's University. 45

Lougheed SC, Austin JD, Bogart JP, Boag PT, Chek AA (2006) Multi-character perspectives on the evolution of intraspecific differentiation in a neotropical hylid frog. Bmc Evolutionary Biology 6,16.

Lougheed SC, Freeland JR, Handford P, Boag PT (2000) A Molecular Phylogeny of Warbling-Finches (Poospiza): Paraphyly in a Neotropical Emberizid Genus. Molecular Phylogenetics and Evolution 17, 367-378.

Lougheed SC, Gascon C, Jones DA, Bogart JP, Boag PT (1999) Ridges and rivers: a test of competing hypotheses of Amazonian diversification using a dart-poison frog (Epipedobates femoralis). Proceedings of the Royal Society of London B 266, 1829- 1835.

Lowe WH, Likens GE, McPeek MA, Buso DC (2006) Linking direct and indirect data on dispersal: Isolation by slope in a headwater stream salamander. Ecology 87, 334-339.

Marsh DM, Trenham PC (2001) Metapopulation dynamics and amphibian conservation. Conservation Biology 15, 40-49.

Martin PR, McKay JK (2004) Latitudinal variation in genetic divergence of populations and the potential for future speciation. Evolution 58, 938-945.

Martinez-Solano I (2004) Phylogeography of Iberian Discoglossus (Anura : Discoglossidae). Journal of Zoological Systematics and Evolutionary Research 42, 298-305.

Masta SE, Sullivan BK, Lamb T, Routman EJ (2002) Molecular systematics, hybridization, and phylogeography of the Bufo americanus complex in Eastern North America. Molecular Phylogenetics and Evolution 24, 302-314.

McCulloch RD, Bentley MJ, Purves RS, et al. (2000) Climatic inferences from glacial and palaeoecological evidence at the last glacial termination, southern South America. Journal of Quaternary Science 15, 409-417.

Moritz C, Patton JL, Schneider CJ, Smith TB (2000) Diversification of rainforest faunas: An integrated molecular approach. Annual Review of Ecology and Systematics 31, 533-563.

Nielson M, Lohman K, Daugherty CH, et al. (2006) Allozyme and mitochondrial DNA variation in the tailed frog (Anura: Ascaphus): The influence of geography and gene flow. Herpetologica 62, 235-258.

Noonan BP, Gaucher P (2005) Phylogeography and demography of Guianan harlequin toads (Atelopus): diversification within a refuge. Molecular Ecology 14,3017-3031.

Patton JL, Da Silva MNF, Malcolm JR (2000) Mammals of the Rio Jurua and the evolutionary and ecological diversification of Amazonia. Bulletin of the American Museum of Natural History 244, 1-306. 46

Patton JL, Dasilva MNF, Malcolm JR (1994) Gene genealogy and differentiation among arboreal spiny rats (Rodentia, Echimyidae) of the Amazon basin - A test of the riverine barrier hypothesis. Evolution 48,1314-1323.

Posada D, Crandall KA (1998) Modeltest: testing the model of DNA substitution. Bioinformatics 14, 817-818.

Posada D, Crandall KA, Templeton AR (2000) GeoDis: A program for the Cladistic Nested Analysis of the Geographical Distribution of Genetic Haplotypes. Molecular Ecology 9,487-488.

Potts R, Behrensmeyer AK (1992) Late Cenozoic terrestrial ecosystems. In: Terrestrial ecosystems through time - evolutionary paleoecology of terrestrial plants and animals (eds. Behrensmeyer AK, Damuth JD, DiMichele WA, et al), pp. 419-541. Chicago University Press, Chicago.

Prance GT (1982) A Review of the Phytogeographic Evidences for Pleistocene Climate Changes in the Neotropics. Annals of the Missouri Botanical Garden 69, 594-624.

Quiroga MP, Premoli AC (2007) Genetic patterns in Podocarpus parlatorei reveal the long-term persistence of cold-tolerant elements in the southern Yungas. Journal of Biogeography 34,447-455.

Rambaut A, Drummond A J (2007) Tracer vl.4, Available from http: //beast .bio .ed .ac .uk/Tracer.

Ronquist F, Huelsenbeck JP (2003) MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19, 1572-1574.

Rull V (2006) Quaternary speciation in the Neotropics. Molecular Ecology 15,4257- 4259.

Saccone C, Attimonelli M, Sbisa E (1987) Structural elements highly preserved during the evolution of the D-loop-containing region in vertebrate mitochondrial DNA. Journal of Molecular Evolution 26, 205-211.

Sambrook, J., and D. W. Russel. 2001. Molecular cloning: a laboratory manual, 3rd ed. Cold Spring Harbor Laboratory Press, New York.

Schneider S, Excoffier L (1999) Estimation of past demographic parameters from the distribution of pairwise differences when the mutation rates vary among sites: Application to human mitochondrial DNA. Genetics 152, 1079-1089.

Schneider S, Roessli D, Excoffier L (2000) Arlequin ver. 2.000: A software for population genetics data analysis. Genetics and Biometry Laboratory, University of Geneva, Switzerland. 47

Simpson BB (1979) Quaternary Biogeography of the High Montane Regions of South America. In: The South American Herpetofauna: Its Origin, Evolution, and Dispersal, (ed. Duellman WE), pp. 157-188. The University of Kansas, Lawrence.

Soltis DE, Morris AB, McLachlan JS, Manos PS, Soltis PS (2006) Comparative phylogeography of unglaciated eastern North America. Molecular Ecology 15,4261- 4293.

Symula R, Schulte R, Summers K (2003) Molecular systematics and phylogeography of Amazonian poison frogs of the genus Dendrobates. Molecular Phylogenetics and Evolution 26,452-475.

Templeton AR (1998) Nested Clade Analysis of phylogeographic data: testing hypotheses about gene flow and population history. Molecular Ecology 7, 381-397.

Templeton AR (2001) Using phylogeographic analyses of gene trees to test species status and processes. Molecular Ecology 10,779-791.

Templeton AR, Crandall KA, Sing CF (1992) A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping and DNA sequence data. III. Cladogram estimation. Genetics 132,619-633.

Templeton AR, Routman E, Phillips CA (1995) Separating population structure from population history: a cladistic analysis of the geographical distribution of mitochondrial DNA haplotypes in the Tiger Salamander, Ambystoma tigrinum. Genetics 140, 767-782.

Templeton AR, Sing CF (1993) A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping. IV. Nested analyses with cladogram uncertainty and recombination. Genetics 134, 659-669.

Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG (1997) The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research 2A, 4876-4882.

Thuring RWJ, Sanders JPM, Borst P (1975) A freeze-squeeze method for recovering long DNA from agarose gels. Analytical Biochemistry 66,213-220.

Voelker G (1999) Dispersal, Vicariance, and Clocks: Historical Biogeography and Speciation in a Cosmopolitan Passerine Genus (Anthus: Motacillidae). Evolution 53, 1536-1552.

Vuilleumier BS (1971) Pleistocene changes in the fauna and flora of South America. Science 173,771-780.

Waltari E, Hoberg EP, Lessa EP, Cook JA (2007) Eastward Ho: phylogeographical perspectives on colonization of hosts and parasites across the Beringian nexus. Journal of Biogeography 34,561-574. 48

Weng C, Hooghiemstra H, Duivenvoorden J (2007) Response of pollen diversity to the climate-driven altitudinal shift of vegetation in the Colombian Andes. Philosophical Transactions of the Royal Society B: Biological Sciences 362, 253-262.

Wiens JJ, Fetzner JW, Parkinson CL, Reeder TW (2005) Hylid frog phylogeny and sampling strategies for speciose clades. Systematic Biology 54, 719-748.

World Conservation Monitoring Centre (1992) Global biodiversity. Status of the Earth's Living Resources (ed. Groombridge B). Chapman & Hall, London. 49

CHAPTER 3. EFFECTS OF LANDSCAPE AND HISTORY ON DIVERSIFICATION OF A MONTANE, STREAM-BREEDING AMPHIBIAN

3.1 Introduction Investigations of the historical impact of landscape features on genetic diversity and differentiation have a long history in evolutionary ecology (e.g. Gentry 1982, Capparella 1991, Gascon et al. 2000, Costa 2003). The discipline of landscape genetics, however, is quite new and as of yet its perspectives are under-utilized in studies of natural populations (Manel et al. 2003). For example, recently developed assays of variation in high-resolution molecular markers combined with novel spatial statistical approaches can provide significant insight into the causes that underlie the distribution of genetic and adaptive diversity across spatial and temporal scales spanning a few meters or generations to entire species' ranges and millennia (reviewed in Manel et al. 2003, Storfer et al. 2006). Such analyses can identify how contemporary and historical landscape features have interacted to influence individual movements, and thus, ultimately gene flow. Studies of spatial patterns in genetic diversity, however, have often attributed a primary (sometimes exclusive) role to historical processes like vicariance, and complementary assessments of contemporary landscape-level processes are urgently needed to fully understand the generation and maintenance of genetic diversity (Vucetich & Waite 2003). Response to landscape features depends on the life history characteristics of a species. Highly vagile species (adults or larvae) typically have little genetic structuring due to their ability to disperse long distances (e.g. eels Avise et al. 1986, sea cucumbers Uthicke & Benzie 2003, pearl oysters Yu & Chu 2006, coral reef fish Haney et al. 2007), while more sedentary species show genetic differentiation even at small spatial scales (e.g. frogs Lougheed et al. 1999, butterflies Keyghobadi et al. 2005, snails Holland & Cowie 2007). Although potential dispersal ability is an important prerequisite for gene flow, other life history characteristics may constrain its impact. For example, many species undergo long distance migrations, hence they have great dispersal capability, but some are highly philopatric returning to breed at or near the site where they were hatched or born - i.e. the distance traveled by an individual during its annual cycle is much 50 greater than the distance traveled by its gametes (e.g. Wenink et al. 1996, Alvarado Bremer et al. 2005, Vaha et al. 2007). Modeling the movement of animals is an important contribution to interpreting patterns of genetic variation (contemporary and historical) across landscapes. Isolation by distance (Wright, 1943), one of the best studied spatial genetic relationships, is often calculated based on matrices of straight-line distances between sampling locations. Genetic differentiation among populations may not be well explained by such measures if the species' required habitat does not occur in a direct line between two sites. Recent explicit considerations of landscape features in a topographically heterogeneous region have demonstrated the importance of measuring biologically meaningful dispersal pathways (Keyghobadi et al. 1999, Funk et al. 2005, Spear et al. 2005). Past studies, however, have tended to look at the influence of landscape at one of two levels without explicitly linking the two: (i) coarsely at a large scale (>100 km) in a biogeographic or phylogeographic context, or (ii) in detail at a fine scale (hundreds of meters to several kilometers) where contemporary processes are more prominent than historical processes. Here we examine the effect of incorporating different levels of detail regarding habitat distribution at different scales to explain patterns of genetic diversity. Our study addresses questions such as: Do analyses using distance metrics based on intricate habitat distribution matter only at smaller scales? Is such detail required at larger scales where differentiation may be more likely due to historical fragmentation rather than isolation by distance? Does a simultaneous consideration of phylogeographic and landscape genetic perspectives provide insights unavailable to either in isolation?

3.1.1 Study area Contemporary Northwestern Argentina possesses great topographic diversity and habitat complexity associated with the Andes Mountains reflecting great spatial variation in precipitation and temperature (Handford 1988). The area is generally arid to semi-arid; however, heavy precipitation from humid air masses travelling from the northeast in the summer falls primarily on the eastern slopes of the Andean ranges, resulting in longitudinal strips of forest stretching along those eastern slopes south from Bolivia to the province of Catamarca (-28° S). These montane forests are essentially continuous 51 except for a narrow invasion of arid thorn scrub habitat in the province of Salta (-25° S, figure 1), which separates the southernmost tip of this forest peninsula (Brown et al. 2001). Away from these mesic east-facing slopes rainfall is much lower resulting in the semi-arid chaco on the flat and slightly tilted terrain to the east, and the arid interior valleys and high plateau to the west (Czajka & Vervoorst 1956, Cabrera 1976, Vervoorst 1982, Handford 1988). These arid habitats are possible dispersal barriers for taxa dependent on mesic environments. This complex distribution of habitats may affect present-day connectivity among populations. Furthermore, historical changes in elevation and climate likely changed the connectivity among habitats, and hence populations, which may have promoted diversification of taxa in this region. In particular, the area between -24° to 26° south currently contains narrow bands of rainforest (Sierras de Metan, Lumbrera, Santa Barbara) that are separated by ehacoan arid thorn scrub vegetation, effectively isolating the southern-most extension of moist Andean forest. During dry glacial periods the arid thorn scrub almost certainly would have extended its range, probably completely severing the connection between the southern (26° to 27° S) and the northern (<24° S) moist forests. Using a phylogeographic approach, we found significant genetic differentiation among populations of tree frogs (Hypsiboas andinus) in this region, due to both contemporary and historical factors (Chapter 2). The aim of this study is to better understand the roles of landscape features in explaining patterns of contemporary and historical genetic diversification among populations at various scales.

3.2 Methods Given their limited vagility, diverse fecundities, breeding biology and breeding habitats frogs offer excellent possibilities for examination of the factors involved in evolutionary divergence, diversification and speciation. Hypsiboas andinus (formerly Hyla andina; new designation suggested by Faivovich et al. 2005; supported by Wiens et al. 2005) is a moderate-sized tree frog (50-60 mm) that inhabits deeply-incised river valleys that penetrate into the Andes Mountains, from the province of Catamarca, Argentina (28°S) to northern Bolivia (16°S). It is mostly found near water along streams, ditches or flooded areas (Cei 1980, Duellman et al. 1997, personal observation) in 52 various habitats, from lowland humid montane forests (500 meters above sea level [masl]) to montane grasslands (>1500 masl), but is not found in the semi-arid chaco to the east nor in high altitude Monte and puna habitats (Duellman et al. 1997). Hypsiboas andinus eggs are deposited in masses containing approximately 600 eggs (M. Vaira, unpublished) secured to submerged vegetation in streams (Cei 1980, personal observation). Larvae are aquatic and develop at the site of oviposition. Generation time for the species is not known but is likely one year, as for the closely related H. pulchellus (Basso &Kehr 1992). Detailed methods for sample collection and molecular procedures are described in chapter 2. A total of 247 individuals from 23 populations are included in these analyses (table 3.1). We used a 340 bp fragment of the mitochondrial control region to assess genetic differentiation among populations (GenBank accession numbers EU403157- EU403413). To calculate population pair-wise FST, we used the Kimura-2-Parameter (K2P) genetic distance, which corrects for multiple substitutions and different rates of substitution between transitions and transversions as implemented in Arlequin ver. 2.000 (Schneider et al. 2000). Well-supported mitochondrial lineages of H. andinus across the study area show strong differentiation that we attributed to historical range fragmentation based on phylogenetic and nested clade analyses (figure 3.1, chapter 2). At the deepest level, two divergent lineages appear to be in secondary contact in the southern portion of the species range. Within the widespread lineage #1, recent (<2 mya) fragmentation has resulted in three subclades distributed in a north to south series (figure 3.1, chapter 2). To further investigate potential causes of these divergence events and the locations of possible barriers to gene flow, we used Monmonier's maximum difference algorithm as implemented in Barrier v. 2.2 (Manni et al. 2004, Manni & Guerard 2004). We calculated one barrier for the entire study area (all individuals in all populations). The analyses were then repeated (two barriers were calculated) for individuals in populations containing lineage #1 only (excluded: all individuals at sites #20 and 23, some individuals at sites # 15, 17, 21 and 22). Given that vegetation types of Northwestern Argentina are strongly related to temperature and precipitation, and hence to elevation, latitude and aspect, we used a 53

Table 3.1. Sample sizes for mtDNA control region analyses for each population and the type of habitat found at the site. Site numbers refer to figure 3.1. Samples were collected by DK and/or collaborators* in 1987, 2001, 2004,2005 or 2006. Site n Habitat type 2 Santa Victoria, Salta 2 Grassland 3 San Andres, Salta 10 Yungas 4 Tilcara, Jujuy 11 Grassland 5 P. N. Calilegua, Jujuy 10 Yungas 6 Lozano, Jujuy 19 Yungas 7 Villa Monte, Jujuy 6 Yungas 8 Maiz Gordo, Jujuy 10 Transition Forest 9 Ruta de Cornisa, Salta 7 Yungas 10 P. N. El Rey, Salta 5 Yungas 11 Molinos, Salta 3 Monte 12a S. F. de Escoipe, Salta 19 Yungas 12b S. F. de Escoipe, Salta 12 Yungas 13 Rio Piedras, Salta 5 Chaco serrano 14 Rosario de la Frontera, Salta 7 Yungas 15 San Pedro, Tucuman 11 Yungas 16 Rio El Nio, Tucuman 20 Yungas 17 Villa Nogues, Tucuman 14 Yungas 18 La Angostura, Tucuman 19 Grassland 19 Rio Los Sosa, Tucuman 20 Yungas 20 Haulfin, Catamarca 4 Monte 21 Las Estancias, Catamarca 16 Grassland 22 Rio San Ignacio, Tucuman 21 Transition Forest 23 Poman, Catamarca 6 Monte * Stephen C. Lougheed, Paul Handford. 54

90 m resolution digital elevation map (DEM) and the "reclass" function of the ArcGIS 9.1 Spatial Analyst extension (ESRI 2005) to create a digital vegetation layer based on published detailed distributions of habitat types (Czajka & Vervoorst 1956, Cabrera 1976, Vervoorst 1982, Handford 1988) and personal observations of authors (55+ cumulative years of experience in the region). Because there were no digital layers depicting the surface hydrology of the region available we used the same 90 m DEM to estimate the locations of all streams and rivers in the region using the ArcHydro 9.1 extension in ArcGIS 9.1 (ESRI 2005). Based on the created vegetation and hydrology layers, as well as the ecological requirements (Cei 1980, Duellman et al. 1997) and known locations of H. andinus, we defined suitable habitat as shown in table 3.2. In the chaco serrano and Monte, the frogs would be restricted to movement only within waterways ("within streams only") as the habitat outside of the riparian zone would be unsuitable. We examined three models of dispersal to test which explained the most variation in pair-wise population genetic differentiation: i) straight-line distance, ii) simple effective distance, and iii) detailed effective distance. Isolation by distance has traditionally been examined using straight-line (or Euclidean) distance between sites and is used here as the null model. Straight-line distance between sampled populations of H. andinus was calculated using the ArcGIS 9.1 Spatial Analyst extension. Neither landscape structure nor elevation was included in this measure. We used a least-cost approach (Michels et al. 2001, Adriaensen et al. 2003) to calculate two effective distances - simple and detailed - that differ in the amount of detail used to define classes of habitats and therefore their resistance-to-movement or "cost" values. The effective distances were calculated by first developing least cost grids for each of the sampling sites. Cost values for surrounding pixels were assigned based on habitat suitability such that as suitability decreases the greater the cost of movement through the pixel. Completely unsuitable habitats (i.e. habitats in which H. andinus cannot survive) were considered absolute barriers to movement and were assigned impassable cost values. Cost values were assigned in relative terms to each other- i.e. one habitat is twice as costly to travel through than another - based primarily on our experience of abundance of the species in each type of habitat. This least cost surface was 55

67° 65° 63° 61°

0 50100 200 300 •czmzzzzza^MB Kilometers

Figure 3.1. Distribution of distinct mitochondrial control region lineages of Hypsiboas andinus across populations of Northwestern Argentina (chapter 2). Unfilled circles = lineage #2 only, black circles = lineage #1 only, mixed black and white circles = both lineages. Distribution of the three subclades within lineage #1 across populations is indicated with solid and dotted lines. The north and south clade do not overlap with each other, but both overlap with the central clade. Shaded regions represent suitable habitat fox H. andinus. 56 then used to generate the paths of "least-cost" between all sampling sites. The lengths of these paths were then calculated as the effective distance. For the simple distance, we coded the cost of movement through all suitable habitat equally (two categories: suitable vs. not suitable) whereas for the detailed distance, we partitioned the suitable habitat into more categories to define cost values for each vegetation type separately (table 3.2). Partial Mantel's tests (Mantel 1967) were performed to look for correlations between each of the three geographic distance matrices and the K2P genetic distance matrix, as well as percent of the variation explained by each type of geographic distance. Significance of correlations was tested using 2000 random permutations. All partial Mantel's tests were performed in FSTAT ver. 2.9.3 (Goudet 2001). We performed the analyses at two scales to look for differences in explanatory power at different scales: (i) for the entire data set (23 populations, about 760 km North to South), and (ii) a subset of populations containing only lineage #1 with pair-wise Euclidean distances of 100 km or less (total of 36 pairs, appendix 2). Because removal of values greater than 100 km resulted in incomplete matrices, we unfolded the matrices and assessed significance of correlations using vector permutation of the genetic distance column as implemented in Permute! ver. 3.4a9 (Legendre etal. 1994, Casgrain 2001) using 999 permutations. To examine the effect of landscape features in explaining historical differentiation we used a matrix of clade distance between populations based on which genetic lineages were present in each compared population (0 = same lineages present in both populations, 1 = different lineages, 0.5 = share one lineage). Analyses were therefore run with the clade distance matrix as the first explanatory variable before the addition of geographic distances. The analyses were run for the entire data set (identifying the two deeply divergent lineages #1 and #2) as well as for the smaller dataset based on the three subclades within mitochondrial lineage #1.

3.3 Results 3.3.1 Genetic differentiation Population pair-wise FST values for the subset data set ranged from -0.19 to 0.92 (appendix 2) and for the entire data set from -0.19 to 1.00 (appendix 3). For the entire range, Monmonier's maximum difference algorithm, implemented in Barrier, identified a 57

Table 3.2. Resistance values used for each habitat in least-cost path analysis for each type of effective distance measured. Simple Detailed Habitat Stream Resistance Resistance Resistance Habitat Suitability Values Suitability Values Values Chaco No No - - Transition forest Yes ][ Yes 30 20 Chaco serrano River only River only - 80 Yungas Yes 1I Yes 10 10 Alisos Forest Yes ][ Yes 40 30 Grassland Yes ][ Yes 45 35 Monte River only River only - 80 Puna No No - -

sharp genetic discontinuity in the southern portion of the range suggesting a possible barrier to gene flow (figure 3.2a). Within network 1 only, two potential barriers were identified across the range (figure 3.2b), which correspond well with the subclades identified using a phylogeographic approach (chapter 2).

3.3.2 Entire data set Although all three geographic distances were highly correlated (both effective distances to Euclidean: r = 0.88, p<0.001), effective distances for some pair-wise comparisons were much greater in magnitude than the Euclidean distances (figure 3.3). In particular, pair-wise comparisons involving populations #11, 20 and 23 showed the greatest difference (arrow in figure 3.3, paths for 1 l-20-12a shown in figure 3.4). Shorter pair-wise Euclidean distances appear to be more similar for all measures. Effective geographic distances explained more variation in the genetic data set than did Euclidean distance (table 3.3). Effective distances had significant partial 58

a) 67* 65° b) 67° 65°

22°

4 . frier 1 24°_ 6 * 9, '10 12a &b .13 '11 . 11*: 26°

. "17 ^Barrier2 20 19 Barrier 1 21* 28a

23

Figure 3.2. Locations of sharp genetic discontinuities (solid black lines) in Hypsiboas andinus: a) across the entire study area, and b) across a portion of the study area corresponding to populations containing individuals with lineage #1 haplotypes as identified using mitochondrial control region sequences. Population pair-wise FST values were analysed using Monmonier's maximum difference algorithm as implemented in Barrier (Manni et al. 2004). The genetic discontinuities likely correspond to barriers to gene flow among the populations. Thin pale grey lines = Delaunay triangulation, thicker grey lines =. Vorono'i tessellation, black points = sampling sites, grey points = virtual points used for the Voronoi' tessellation along map edges. 59 correlations after controlling for the effect of straight-line distance. Using the best model, the simple classification for the effective geographic distance, 36% of the variation in pair-wise population FST was explained. The clade distance matrix was significantly correlated to pair-wise FST (r = 0.38, r = 0.14, p<0.0001). All distances maintained significant partial correlations with pair- wise FST when the clade distance matrix was added to the analyses (table 3.3). The variation explained increased only in the model with Euclidean distance alone (20% to 24%).

3.3.3 Subset of data set Euclidean geographic distance was not significantly correlated with the matrix of genetic distances at the less than 100 km scale and explained only five percent of the total variation; therefore we did not include it when testing effective distances (table 3.4). Only the simple effective distance was significantly correlated with genetic distance, explaining 20% of the variation. The clade distance matrix for lineage #1 was

Figure 3.3. Relationship between pair-wise simple and detailed habitat effective distances and Euclidean distance. Grey line is line of equality. Arrow indicates pair-wise effective distances that are much larger than Euclidean distances, see text for more details. Table 3.3. Correlations between genetic distance and different geographic distances using partial Mantel's tests for the entire data set. Each model was tested with and without the clade distance matrix. Significance of correlations is indicated as follows: *<0.05, **<0.01, ***<0.001. Clade distance matrix Partial Variance Partial Variance Correlation explained Correlation explained Route Variable (r) (r2) (r) (r2) Null straight Euclidean 0.45 0.20 0.31 0.24 line distance*** Simple Euclidean -0.45 0.36 -0.31 0.36 classification distance*** Simple habitat 0.40 0.35 distance*** Detailed Euclidean -0.45 0.34 -0.31 0.34 classification distance*** Detailed habitat 0.37 0.32 distance***

significantly correlated with genetic distance (r = 0.51, r2 = 0.26, p = 0.001) and improved percent variation explained for all three distances. Only the simple effective distance had a significant partial correlation after addition of the clade distance matrix.

3.4 Discussion At both scales, effective distances explained more variation in pair-wise genetic differentiation than straight-line geographic distance. The least-cost distances, based on the simple classification, performed better than the more detailed habitat classification. We were also able to control for the effects of historical fragmentation, and therefore evaluate the effect of different distances on the remaining genetic variation. Our results suggest that landscape features play a prominent role in historical and contemporary 61

Euclidean Simple Habitat HI HI in Detailed Habitat

Alisos Forest

Streams in Chaco Serrano

Yungas

Transition Forest

Grasslands

Figure 3.4. Routes between populations #1 l-12a-20 using Euclidean distance, simple and detailed habitat effective distances. Shades of grey indicate type of habitat according to the detailed habitat classification (see table 3.2). Simple classification weights all suitable habitats equally. Table 3.4. Correlations between genetic distance and different geographic distances using partial Mantel's tests at the smaller scale (<100 km). Each model was tested with and without the clade distance matrix. Because Euclidean distance was not significantly correlated with genetic distance, it was not included in the effective distance models. Significance of correlations is indicated as follows: *<0.05, **<0.01, ***<0.001. Clade distance matrix Partial Variance Partial Variance Correlation explained Correlation explained Route Variable (r) (r2) (r) (r2) Null straight Euclidean 0.21 0.05 0.13 0.28 line distance Simple Simple habitat 0.45 (**) 0.20 0.29 (*) 0.33 classification distance Detailed Detailed habitat 0.32 0.10 0.18 0.29 classification distance

genetic structuring of populations ofH. andinus at both large and small scales. At the larger scale, the main historical genetic discontinuity among the two lineages is found in the southern portion of the range (figure 3.2a). Populations #20 and 23 contain lineage #2 only, while population #22 has both lineages present in equal proportion. Although two other populations contain both lineages the majority of individuals have lineage #1 haplotypes. The location of the putative barrier suggests that populations #15, 17, 21 and 22 likely are sites of secondary contact between these two divergent lineages. Barriers such as prominent topographic features and rivers have long been correlated to genetic divergence of populations on either side (e.g. Bermingham & Avise 1986, Lougheedefa/. 1999, Keyghobadi etal 1999, Funketal. 2005). Funked/. (2005) found that mountain ridges, which do not have suitable habitat, strongly affected population differentiation of Columbia spotted frogs [Rana luteiventris). Populations that 63

had a short straight-line distance but were on opposite sides of ridges showed high genetic divergence, but populations not separated by ridges were genetically similar over longer geographic distances (Funk et al. 2005). Since ridges are clearly acting as barriers, the effective or ecological distance between populations is much longer than the straight- line as frogs would need to travel a much larger distance to "go around" the ridge. The presence of unsuitable habitat in our study similarly resulted in much longer pair-wise distances since movement was modelled to avoid these areas. If these habitats served as barriers to movement and hence the populations showed higher genetic differentiation, then models using effective distance should better explain the variation in pair-wise population FST- Indeed, we found that populations surrounded largely by unsuitable habitat (e.g. #11, 20, 23) had effective distances to other populations that were much larger than the Euclidean pathways (figures 3.3 and 3.4). These populations were usually found where movement would necessarily be restricted only to riparian habitat along streams. Such a pattern is found in many aquatic species that move exclusively through specific corridors such as waterways (e.g. Turner & Trexler 1998, Fetzner & Crandall 2003, Spear et al. 2005, Vaha et al. 2007). Our findings suggest that peripheral populations typically found in the most environmentally extreme parts of the species range are most likely to be genetically differentiated from other populations but are also most subject to underestimation of population pair-wise geographic distances when using Euclidean distance. This bias may have consequences for detecting patterns of isolation by distance and distinguishing historical from contemporary process. Vucetich and Waite (2003)

found peripheral populations of grassland birds to have lower Ne than core populations, suggesting that genetic drift might have a much more pronounced effect in populations along edges of the species range. Differentiation in peripheral populations ofH. andinus could be due to historical isolation; however, contemporary processes such as genetic drift may also play a key, but often undetected, role through the loss of genetic diversity. To examine the effect of historical differentiation among the two major mitochondrial lineages in H. andinus, we used a design matrix to identify relationships among populations based on which lineages were present in each population. The clade distance matrix was significantly correlated with pair-wise FST, as would be expected 64 given that it describes differences among populations due to historical processes, and it alone explained 14% of the total variation in the genetic dataset. Although all geographic distances were still significantly correlated with pair-wise FST after controlling for history, only the model with solely Euclidean distance showed an improvement in percent variation explained. Effective distances appear to account for the variation due to historical factors even without addition of the clade distance matrix. Because the effective distance routes avoid currently unsuitable habitat, they likely better encompass not only contemporary but also historical connectivity among populations. We also found a significant correlation between the clade distance matrix for lineage #1 and the genetic differentiation at the smaller scale (<100 km). The clade distance matrix alone explained 26% of the variation, and increased the percent variation explained for all three geographic distances. Given what we know of the history of the region, unsuitable habitat was likely more extensive during glacial maxima, isolating populations in remaining moist refugia. Two contemporary well-defined areas, at ~22°S and ~27°S, of moist habitat may have retained their mesic character during glacial dry periods and thus may have supported montane forests and dependent populations of taxa. These mesic forests are separated by arid chacoan scrub that likely expanded during dry glacials severing the connection between these mesic forests. Assuming contraction of moist habitats at mid-altitudes during the Pleistocene, the current distribution of the yungas may closely represent suitable habitat for H. andinus during this time. Such a model predicts severing of mesic habitats between sites #12a/b and 15, which may represent the barrier between the north and south refugia for the subclades in our study. Under this model, the Sierra de Santa Barbara mountain chain (sites #7, 8, and 10) would also be isolated and may have served as the refugium for the central subclade. Putative barriers identified in this study supported our previous findings of differentiated subclades across the study area. At the smaller scale, only the simple effective distance was significantly correlated with genetic distance, which suggests that habitat features may be driving differentiation rather than distance alone. Previous studies of salamanders (Spear et al. 2005, Giordano et al. 2007) and frogs (Rowe et al. 2000, Funk et al. 2005) have suggested that features such as rivers, sea water, forests, meadows, and elevation rather 65 than geographic distance alone may affect movement of individuals and hence population differentiation. Even at the smaller scale in our study, populations separated by unsuitable habitat may be driving the correlation between genetic differentiation and distance. Some pair-wise effective distances were greater than 200km (twice the Euclidean distance) due to avoidance of unsuitable habitat. Patterns of connectivity and isolation at this scale may result in differentiation due to contemporary processes such as limited gene flow among populations separated by unsuitable habitat, as well as the historical isolation described above.

3.4.1 Effects of scale All distances were significantly correlated at the larger scale, but only the simple effective distance showed a statistically significant relation at the small scale (with or without the clade distance matrices). The difference in patterns suggests that different processes may drive genetic differentiation at various scales. Several studies have found that a pattern of isolation by distance is not as evident at very small scales (review in Newman & Squire 2001, Stepien et al. 2007, Wilmer & Wilcox 2007). Structure at smaller scales may be strongly affected by metapopulation and extinction/recolonization dynamics (Rowe et al. 2000, Newman & Squire 2001, Wilmer & Wilcox 2007), differences in effective population size (Vucetich & Waite 2003), and behavioural traits such as site fidelity (Stepien et al. 2007, Vaha et al. 2007). At larger scales, historical processes such as fragmentation and range expansion may leave a stronger signature in genetic differentiation among populations or groups.

3.4.2 Effects of habitat classification Although both effective distances performed better than Euclidean, the simple habitat effective distance explained the most variation at both scales. The critical determinant of population differentiation appears to be the availability of any suitable habitat rather than small differences in suitability. Although many studies use indices (e.g. genetic differentiation) to infer movement of organisms, the interaction between the organism and the landscape is rarely quantified at suitable scales due to logistical limitations of sampling at sufficient density (but see Keyghobadi et al. 2005). Effective 66 distances among populations and levels of connectivity will vary for different groups depending on life history attributes (D'Eon et al. 2002), but little is known about many species in the Neotropics. Our findings suggest that dispersal in H. andinus may depend on factors other than the fine-scale differences in habitat we included in the detailed classification, or that historical changes in phytogeography, which do not match patterns of contemporary habitats, have affected genetic differentiation. Further investigation of movement patterns for this (and other) species will shed light on the processes responsible for genetic diversification in this region. Given our understanding of the limitations in life history of H. andinus, it is not surprising that effective distances better explain patterns of genetic differentiation. Our study demonstrates that using effective distance can better explain patterns of differentiation in populations, especially in heterogeneous landscapes where barriers to dispersal may be common, and that combining phylogeographic and landscape genetic perspectives provides more comprehensive insights into the causes of the present day distribution of genetic diversity. Using a least cost path analysis may identify corridors of movement between populations that are biologically more realistic and more relevant to understanding landscape level processes that are critical for conservation planning. 3.5 References

Adriaensen F, Chardon JP, De Blust G, et al. (2003) The application of 'least-cost' modelling as a functional landscape model. Landscape and Urban Planning 64, 233- 247.

Alvarado Bremer JR, Vifias J, Mejuto J, Ely B, Pla C (2005) Comparative phylogeography of Atlantic bluefin tuna and swordfish: the combined effects of vicariance, secondary contact, introgression, and population expansion on the regional phylogenies of two highly migratory pelagic fishes. Molecular Phylogenetics and Evolution 36, 169-187.

Avise JC, Helfman GS, Saunders NC, Hales LS (1986) Mitochondrial DNA Differentiation in North Atlantic Eels: Population Genetic Consequences of an Unusual Life History Pattern. Proceedings of the National Academy of Sciences 83, 4350-4354.

Basso NG, Kehr AI (1992) Estructura poblacional y crecimiento de Hyla pulchella pulchella (Anura: Hylidae). Acta Zoologica Lilloana 41, 143-148.

Bermingham E, Avise JC (1986) Molecular zoogeography of freshwater fishes in the Southeastern United States. Genetics 113, 939-965.

Brown AD, Grau HR, Malizia LR, Grau A (2001) Argentina. In: Bosques nublados del neotropico (eds. Kappelle M, Brown AD). Institute Nacional de Biodiversidad, Costa Rica.

Cabrera AL (1976) Regiones fitogeograficas Argentinas. In: Enciclopedia Argentina de Agricultura y Jardineria, 2nd edition, (ed. Parodi LR), pp. 1-85. Editorial Acme, Buenos Aires.

Capparella AP (1991) Neotropical avian diversity and riverine barriers. Acta Congressus Internationalis Ornithologici 20, 307-316.

Casgrain P (2001) Permute! Multiple regression over distance, ultrametric and additive matrices with permutation test (ver. 3.4 alpha 9). Available from http://www.bio.umontreal.ca/Casgrain/en/labo/permute/index.html

Cei JM (1980) Amphibians of Argentina Monitore Zoologico Italiano N. S Monografia 2., Florence.

Costa LP (2003) The historical bridge between the Amazon and Atlantic Forest of Brazil: a study of molecular phylogeography with small mammals. Journal of Biogeography 30, 71-86. 68

Czajka W, Vervoorst F (1956) Die naturraumliche Gliederung Nordwest-Argentiniens. Petermanns Geographische Mitteilungen 100, 89-102 & 196-208.

D'Eon R, Glenn SM, Parfitt I, Fortin MJ (2002) Landscape connectivity as a function of scale and organism vagility in a real forested landscape. Conservation Ecology 6, 10. [online]

Duellman WE, de la Riva I, Wild ER (1997) Frogs of the Hyla armata and Hyla pulchella groups in the Andes of South America, with definitions and analyses of phylogenetic relationships of Andean groups of Hyla. Scientific Papers of the Natural History Museum of the University of Kansas 3, 1-41.

Faivovich J, Haddad CFB, Garcia PCA, et al. (2005) Systematic review of the frog family Hylidae, with special reference to Hylinae: Phylogenetic analysis and taxonomic revision. Bulletin of the American Museum of Natural History 294, 1-240.

Fetzner JW, Jr., Crandall KA (2003) Linear habitats and the nested clade analysis: an empirical evaluation of geographic versus river distances using an Ozark Crayfish (Decapoda: Cambaridae). Evolution 57, 2101-2118.

Funk WC, Blouin MS, Corn PS, et al. (2005) Population structure of Columbia spotted frogs (Rana luteiventris) is strongly affected by the landscape. Molecular Ecology 14, 483-496.

Gascon C, Malcolm JR, Patton JL, et al. (2000) Riverine barriers and the geographic distribution of Amazonian species. Proceedings of the National Academy of Sciences of the United States of America 97, 13672-13677.

Gentry AH (1982) Neotropical floristic diversity: phytogeographical connections between Central and South America, Pleistocene climatic fluctuations, or an accident of the Andean orogeny? Annals of the Missouri Botanical Garden 69, 557-593.

Giordano AR, Ridenhour B J, Storfer A (2007) The influence of altitude and topography on genetic structure in the long-toed salamander (Ambystoma macrodactulym). Molecular Ecology 16, 1625-1637.

Goudet J (2001) FSTAT, a program to estimate and test gene diversities and fixation indices (version 2.9.3). Available from http://www.unil.ch/izea/softwares/fstat.html.

Handford P (1988) Trill rate dialects in the Rufous-collared sparrow, Zonotrichia capensis, in northwestern Argentina. Canadian Journal of Zoology 66, 2658-2670.

Haney RA, Silliman BR, Rand DM (2007) A multi-locus assessment of connectivity and historical demography in the bluehead wrasse (Thalassoma bifasciatum). Heredity 98, 294-302. 69

Holland BS, Cowie RH (2007) A geographic mosaic of passive dispersal: population structure in the endemic Hawaiian amber snail Succinea caduca (Mighels, 1845). Molecular Ecology 16,2422-2435.

Keyghobadi N, Roland J, Strobeck C (1999) Influence of landscape on the population genetic structure of the alpine butterfly Parnassius smintheus (Papilionidae). Molecular Ecology 8, 1481-1495.

Keyghobadi N, Roland J, Strobeck C (2005) Genetic differentiation and gene flow among populations of the alpine butterfly, Parnassius smintheus, vary with landscape connectivity. Molecular Ecology 14, 1897-1909.

Legendre P, Lapointe F-J, Casgrain P (1994) Modeling brain evolution from behavior: a permutational regression approach. Evolution, 48, 1487-1499.

Lougheed SC, Gascon C, Jones DA, Bogart JP, Boag PT (1999) Ridges and rivers: a test of competing hypotheses of Amazonian diversification using a dart-poison frog (Epipedobates femoralis). Proceedings of the Royal Society of London B 266, 1829- 1835.

Manel S, Schwartz MK, Luikart G, Taberlet P (2003) Landscape genetics: combining landscape ecology and population genetics. Trends in Ecology & Evolution 18, 189- 197.

Manni F, Guerard E, Heyer E. (2004) Geographic patterns of (genetic, morphologic, linguistic) variation: how barriers can be detected by "Monmonier's algorithm". Human Biology, 76, 173-190.

Manni F, Guerard E (2004) Barrier vs. 2.2. Manual of the user. Population genetics team, Museum of Mankind (Musee de l'Homme), Paris [Publication distributed by the authors].

Mantel N (1967) The detection of disease clustering and a generalized regression approach. Cancer Research 27, 209-220.

Michels E, Cottenie K, Neys L, et al. (2001) Geographical and genetic distances among zooplankton populations in a set of interconnected ponds: a plea for using GIS modelling of the effective geographical distance. Molecular Ecology 10, 1929-1938.

Newman RA, Squire T (2001) Microsatellite variation and fine-scale population structure in the wood frog (Rana sylvatica). Molecular Ecology 10, 1087-1100.

Rowe G, Beebee Tjc, Burke T (1998) Phylogeography of the natterjack toad Bufo calamita in Britain: genetic differentiation of native and translocated populations. Molecular Ecology 7, 751-760. 70

Schneider S, Roessli D, Excoffier L (2000) Arlequin ver. 2.000: A software for population genetics data analysis. Genetics and Biometry Laboratory, University of Geneva, Switzerland.

Spear SF, Peterson CR, Matocq MD, Storfer A (2005) Landscape genetics of the blotched tiger salamander (Ambystoma tigrinum melanostictum). Molecular Ecology 14, 2553-2564.

Stepien CA, Murphy DJ, Strange RM (2007) Broad- to fine-scale population genetic patterning in the smallmouth bass Micropterus dolomieu across the Laurentian Great Lakes and beyond: an interplay of behaviour and geography. Molecular Ecology 16, 1605-1624.

Storfer A, Murphy MA, Evans JS, et al. (2006) Putting the landscape' in landscape genetics. Heredity 98, 128-142.

Turner TF, Trexler JC (1998) Ecological and historical associations of gene flow in Darters (Teleostei: Percidae). Evolution 52, 1781-1801.

Uthicke S, Benzie JAH (2003) Gene flow and population history in high dispersal marine invertebrates: mitochondrial DNA analysis of Holothuria nobilis (Echinodermata: Holothuroidea) populations from the Indo-Pacific. Molecular Ecology 12, 2635-2648.

Vaha JP, Erkinaro J, Niemela E, Primmer CR (2007) Life-history and habitat features influence the within-river genetic structure of Atlantic salmon. Molecular Ecology 16, 2638-2654.

Vervoorst F (1982) Noroeste. In: Conservation de la vegetation natural de la Republica Argentina. Serie Conservation de la Naturaleza. XVIIIJornadas Argentinas de Botdnica, pp. 9-24. Fundacion Miguel Lillo, Tucuman.

Vucetich JA, Waite TA (2003) Spatial patterns of demography and genetic processes across the species' range: Null hypotheses for landscape conservation genetics. Conservation Genetics 4, 639-645.

Wenink PW, Baker AJ, Rosner H-U, Tilanus MGJ (1996) Global mitochondrial DNA phylogeography of holarctic breeding dunlins (Calidris alpina). Evolution 50, 318- 330.

Wiens JJ, Fetzner JW, Parkinson CL, Reeder TW (2005) Hylid frog phylogeny and sampling strategies for speciose clades. Systematic Biology 54, 719-748.

Wilmer JW, Wilcox C (2007) Fine scale patterns of migration and gene flow in the endangered mound spring snail, Fonscochlea accepta (Mollusca: Hydrobiidae) in arid Australia. Conservation Genetics 8, 617-628. 71

Wright S (1943) Isolation by distance. Genetics, 28, 114-13 8.

Yu DH, Chu KH (2006) Low genetic differentiation among widely separated populations of the pearl oyster Pinetada fucata as revealed by AFLP. Journal of Experimental Marine Biology and Ecology 333, 140-146. 72

CHAPTER 4: COMPARATIVE PHYLOGEOGRAPHY OF TWO ANDEAN FROGS, HYPSIBOAS ANDINUS AND PLEURODEMA BORELLII

4.1 Introduction Life history characteristics can profoundly impact population genetic structure through their effects on dispersal and population size, which in turn influence gene flow and genetic drift. Habitat-restricted species tend to show stronger genetic differentiation among geographic areas than generalist species due to a lower propensity for movement and hence low gene flow (Schauble & Moritz 2001, Stuart-Fox et al. 2001, Crawford et al. 2007). Highly vagile species or species with broadcast spawning, for example, typically show little genetic structuring (e.g. eels Avise et al. 1986, sea cucumbers Uthicke & Benzie 2003, pearl oysters Yu & Chu 2006, coral reef fish Haney et al. 2007), while more sedentary species show genetic differentiation even at small spatial scales (e.g. frogs Lougheed et al. 1999, butterflies Keyghobadi et al. 2005, snails Holland & Cowie 2007). Dispersal ability is an important prerequisite for gene flow, but other life history attributes may constrain its effect. For example, species that undertake long distance migrations have great potential dispersal capability, but return to breed at or near the site where they were hatched or born (highly philopatric) - thus the distance traveled by an individual during its annual cycle would be much greater than the distance traveled by its gametes (e.g. Wenink et al. 1996, Alvarado Bremer et al. 2005, Vaha et al. 2007). Fecundity also may relate positively to gene flow. Fecundity was highly correlated with gene flow in Eiheostoma darters, stream inhabiting fish (Turner et al. 1996, Turner & Trexler 1998), where species laying a few large eggs showed lower gene flow than species producing many small eggs (Turner & Trexler 1998). Low fecundity may lead to smaller effective population sizes, with obvious implications for the effects of genetic drift and emigration for such populations (Turner et al. 1996, Turner & Trexler 1998). Genetic drift facilitates rapid differentiation among small, genetically isolated populations. To maintain levels of high gene flow (i.e. Nem > 1), populations of a smaller effective size would also require proportionately higher migration rates. Thus, through their effects on gene flow and population size, species' characteristics affecting 73 population demographics, such as movement of individuals and recruitment of young, may strongly influence contemporary patterns of connectivity among populations. Comparing phylogeographic patterns of multiple species with distinct life history and ecological attributes allows for explicit assessment of the relative importance of demographic processes vs. historical events in shaping differentiation among populations. Concordant phylogeographic patterns for co-distributed taxa imply that common historical barriers once affected gene flow and produced differentiation to similar degrees even in species with different ecological and life history characteristics; specific genealogical patterns often suggest the identity and timing of those factors (Avise 1998, 2000). Alternatively, discordant patterns among co-distributed taxa may highlight the relative importance of differences in ecology and demography among species in shaping contemporary patterns of genetic diversity or suggest different evolutionary histories (Avise 1998). Given their diverse fecundities, breeding biology and breeding habitats, frogs offer excellent possibilities for an examination of the factors involved in evolutionary divergence, diversification and speciation. The focal species of this study, Hypsiboas andinus and Pleurodema borellii, are broadly sympatric in northwestern Argentina and Bolivia but differ markedly in the types of habitats they occupy, the bodies of water used for breeding, and their fecundity. This study seeks to better understand the roles of contemporary and historical factors in shaping the distribution of genetic diversity of P. borellii across its range in Northwestern Argentina, and to assess the influence of ecology and life history on genetic structure through a comparison of the patterns found for P. borellii (reported here) and for H. andinus (reported in chapter 2).

4.2 Methods 4.2.1 Study species Hypsiboas andinus is a moderate-sized tree frog (50-60 mm) that inhabits deeply- incised river valleys that penetrate into the Andes Mountains. It is mostly found near water along streams, ditches or flooded areas (Cei 1980, Duellman et al. 1997, personal observation) in various habitats, from lowland montane forests (500 masl) to montane grasslands (>1500 masl), but is not found in the low elevation (<400 masl) semi-arid 74 chaco to the east nor in high altitude puna (>3400 masl) habitats (Duellman et al. 1997). Pleurodema borellii is a moderate-sized leptodactylid frog (40-55 mm) that inhabits a wide range of Andean habitats. Pleurodema borellii uses open, often disturbed, habitats throughout the moist montane forests and grasslands, but also drier habitats such as the Monte desert scrub, the high altitude puna grass- and scrub-lands, and the western margin of the chaco thornscrub (Cei 1980). Hypsiboas andinus eggs are deposited in masses of approximately 600 eggs (M. Vaira unpublished) secured to submerged vegetation in streams (Cei 1980, personal observation). For P. borellii, typically between 1000 and 2500 eggs (M. Vaira unpublished; Halloy & Fiano 2000: mean for 11 clutches = 1277+/- 352) are deposited in foam nests floating on the surface of small, temporary ponds or pools (Cei 1980, personal observation). Foam nests are believed to protect the developing eggs from desiccation (Duellman and Trueb 1986) and likely allow P. borellii to reproduce in arid habitats (e.g. chaco). Larvae of both species are aquatic and develop at the site of oviposition. Generation time for H. andinus is not known but is likely one year, as for the closely related//, pulchellus (Basso & Kehr 1992). Direct measurements of generation time are not available for P. borellii. Breeding in P. borellii is highly concentrated in the wettest part of the year (Halloy & Fiano 2000, Vaira 2002) and likely results in a generation time of one year, since juveniles that emerge would necessarily have to wait until the next summer to breed. Nothing is known of the activities of either adults or young of the year following the drying-up of the breeding pools.

4.2.2 Sampling Samples included in this study were either (i) field collected (//. andinus = 258, P. borellii = 127), or (ii) tissues loaned from the Museo Nacional de Ciencias Naturales (MNCN), Madrid, Spain (//. andinus = 6, P. borellii = 2) or the Museo de La Plata, La Plata, Argentina (P. borellii = 2) (tables 2.1 and 4.1). Individuals of//, andinus from 23 sites, and of P. borellii from 20 sites, across the Argentine portion of the species' range were collected under provincial and federal permits during several field expeditions (table 4.1, figure 4.1). Hypsiboas riojanus and P. tucumana at one site each were also collected for outgroup analyses. Detailed field 75

Table 4.1. Sources of tissues and sample sizes for Pleurodema borellii. Site numbers refer to Figure 4.1. Samples were collected by DK and collaborators* in 2004, 2005 or 2006 unless otherwise noted. Haplotype diversity Qi) and nucleotide diversity expressed as percent (jt) are shown (standard deviations in parentheses) for all individuals in the population. For sites #7 and 19 italicized values are those for individuals belonging to network #1 only. Site n h n 1 Tarija, Bolivia3 2 1.00(0.50) 0.11(0.16) 2 Abra Pampa, Jujuy 5 0.70 (0.22) 0.27 (0.20) 3 Tres Cruces, Jujuy 1 - - 4 San Andres, Salta 8 0.75(0.10) 0.11(0.09) 5 Tropico de Capricornio, Jujuy 5 0.60(0.18) 0.07 (0.07) 6 P. N. Calilegua, Jujuy 9 0.72(0.16) 0.12(0.10) 7 Tiraxi, Jujuy 15 0.36(0.14) 0.78 (0.43) 13 0.15(0.13) 0.02(0.03) 8 Villa Monte, Jujuy 16 0.83 (0.07) 0.26(0.17) 9 Maiz Gordo, Jujuy 9 0.86 (0.09) 0.34 (0.22) 10 Ruta de Cornisa, Salta 9 0.83 (0.10) 0.11(0.09) 11 P. N. El Rey, Salta 1 - - 12 Pulares, Salta 10 0.73(0.10) 0.11(0.09) 13 Ojo de Agua, Salta 8 0.93 (0.08) 0.26(0.18) 14 San Pedro, Tucuman 7 0.29(0.20) 0.10(0.09) 15 Yerba Buena, Tucuman 3 0.67(0.31) 0.15(0.15) 16 La Angostura, Tucuman 9 0.42(0.19) 0.02 (0.04) 17 Infernillo, Tucuman 3 0.67(0.31) 0.07 (0.09)

18 Las Estancias, Catamarca 1 - - 19 Embalse Escaba, Tucuman 7 0.90(0.10) 1.92(1.11) 3 0.67(0.31) 0.07(0.10) 20 San Ignacio 2 1.00(0.5) 0.33 (0.39) * Stephen C. Lougheed, Paul Handford. aTissue loaned from the Museo Nacional de Ciencias Naturales, Madrid, Spain. Two of the three tissues loaned from the Museo de La Plata, La Plata, Argentina. sampling methods are described in chapter 2. At site #14 we collected eggs from foam nests of P. borellii to augment sample size. Only one egg per nest was used for analyses. Voucher specimens are listed in Appendix 4.

86fl S4C

\_\ <1000 masl MM n Bolivia 1000-3000 masl ^v '_!__* 22H ff i II >3Q00 masl *m / dentin a 2«

N

24

12i

26°-H 14.

17i 15 m

200 300 Z3 Kilometers 28^ Figure 4.1. Sampling localities of Pleurodema borellii in Northwestern Argentina (circles) and Bolivia (square), and the outgroup taxon Pleurodema tucumana (triangle). Site names and sample sizes for P. borellii are listed in table 4.1 (p. 74). 77

4.2.3 PCR and sequencing DNA extraction, PCR and sequencing methods are described in chapter 2. For if. andinus only, a 663 bp fragment of control region was amplified using primers ControlP(H) and Wrev(L) from Goebel et al. (1999). For P. borellii only, a 538bp fragment of the mitochondrial 16S rRNA gene was amplified for all individuals using primers 16SaL and 16SbrH (Goebel et al. 1999). A 358bp fragment of cytochrome b was amplified for a subset of H. andinus samples representing the diversity of control region haplotypes, and all P. borellii using primers MVZ15-L from Goebel et al. (1999) and a modified version of H15149 (Lougheed et al. 1999). These genes were chosen to achieve the best resolution of relationships within each species. Sequence alignments were made in Clustal X, ver. 1.83 (Thompson et al. 1997) with subsequent visual verification.

4.2.4 Phylogenetic analyses To allow comparison of genealogical histories, we generated phylogenetic trees using a combined data set of control region and cytochrome b for a subset of//, andinus samples plus H. balzani as outgroup (chapter 2). For P. borellii, we used a combined data set of cytochrome b and 16S for all samples plus P. tucumana as an outgroup. All analyses were performed as described in chapter 2.

4.2.5 Phylogeographic analyses We used only control region data for phylogeographic analyses of//, andinus (described in chapter 2). P. borellii analyses were conducted on the combined genetic data set for all samples. TCS version 1.21 (Clement et al. 2000) was used to generate a Maximum Parsimony Network (MPN) representing the genealogical relationships among haplotypes. A Nested Clade Analysis (NCA; Templeton et al. 1992, Templeton et al. 1995, Templeton 1998) was applied to integrate genealogical information with geographic distributions. Ambiguities were resolved following the rules of Templeton and Sing (1993). All distance measure calculations and permutations (n=1000) were assessed using GeoDis version 2.5 (Posada et al. 2000) and the significant results were then interpreted using the most recently published inference key (11 November 2005, 78 available at http://darwin.uvigo.es/). Due to the possibility of false-positives in NCA (Petit 2008), we performed additional analyses to aid us in interpreting patterns. These included the mismatch distribution analyses and comparisons of haplotype (h) and nucleotide diversity (JI) using Kimura-2-Parameter (K2P) molecular distance for each of the main clades at the total cladogram level in Arlequin ver. 2.000 (Schneider et al. 2000). To investigate locations of potential barriers to gene flow for P. borellii, we used Monmonier's maximum difference algorithm as implemented in Barrier v. 2.2 (Manni et al. 2004, Manni & Guerard 2004). We compared these results to those found for H. andinus (chapter 3).

4.2.6 Timing of divergence To estimate the time of divergence among clades we used only cytochrome b sequences because the molecular evolution of this gene is better understood than that of control region or 16S. We calculated average pairwise differences among the clades using the K2P molecular distance corrected for within-clade variation using Arlequin. We applied the slower poikilothermic molecular clock typically used for anuran cytochrome b (see Austin et al. 2004) with a rate of change between 0.5 and 1% per million years. We also used a coalescent approach as implemented in the program BEAST v. 1.4.6 (Drummond & Rambaut 2007). We performed two independent runs of 20 million generations each with burn-ins of 2 million, which were then combined in TRACER v. 1.4 (Rambaut & Drummond 2007). We employed a GTR +1 + G model of evolution with 6 rate categories and assumed a relaxed lognormal clock (using rates of divergence of 0.5 and 1% per million years). Parameters were sampled every 1000 generations. All other initial parameter settings were the default provided by BEAST v. 1.4.6.

4.2.7 Population-level statistical analyses We used Arlequin to calculate haplotype (h) and nucleotide (jt) diversity indices for each population using the K2P molecular distance. We categorized the results according to Grant and Bowen (1998) as reflecting (1) prolonged bottleneck (h<0.5 and 79

3t<0.5%), (2) rapid population growth from ancestral population with low effective population size (h>0.5 and Jt<0.5%), (3) a brief bottleneck (/z<0.5 and Jt>0.5%), or (4) stable population with large historical effective population size, or secondary contact among differentiated lineages (7z>0.5 and Jt>0.5%). Although the boundaries between categories are presented as sharply delineated, Grant and Bowen (1998) have included nucleotide diversities up to 0.71% in category 2. We also performed mismatch distributions for all populations with sample size greater than 10 and tested against a model of sudden expansion using the generalized non-linear least-squares approach (Schneider & Excoffier 1999). The validity of the model was evaluated using 1000 parametric bootstraps as implemented in Arlequin. We used a coalescent approach to calculate effective population size from DNA sequences using Bayesian Skyline plots (Drummond et al. 2005) for the two species. We performed runs of 80 million generations with burn-ins of 8 million in BEAST v. 1.4.6. We employed a GTR +1 + G model of evolution with 6 rate categories and assumed a relaxed lognormal clock (using a molecular clock of 1% per million years). We used a model of constant population size between sampling points with 10 intervals (m = 10). Parameters were sampled every 1000 generations. All other initial parameter settings were the default provided by BEAST v. 1.4.6. Skyline plots were produced using TRACER v. 1.4. The proportion of variation among clades and among populations (within lineage #1 only, see results) for each species was assessed using Analysis of Molecular Variance (AMOVA) as implemented in Arlequin. We used two methods to assess if there was a difference in the levels of differentiation for the two species (within lineage #1 only, see results). Population pair- wise number of base pair differences (corrected for within population variation) were computed using the K2P molecular distance for each species. This estimate of genetic distance was then divided by the population pair-wise geographic distance to produce a genetic distance per kilometer for each species (Martin & McKay 2004). The difference between the means of each species was compared to 1000 replicates sampled from the genetic distances shuffled between species. The difference was considered significant if the value appeared in fewer than 5% of the replicates. We also evaluated the relationship 80

between genetic and geographic distances for each species using Mantel's tests as implemented in Le Progiciel R (Casgrain & Legendre 2001). The slopes of this relationship were then compared between the two species, again using 1000 replicates. All resampling statistics were performed using the Resampling Stats Add-in for Microsoft Excel ver. 3.2 (Resampling Stats, Inc.).

4.3 Results Primary results pertaining to H. andinus are described in chapters 2 and 3 (pp. 18- 70).

4.3.1 Molecular diversity A total of 535 bp of 16S and 364 bp of cytochrome b were obtained for 130 individuals of P. borellii. We found an overall AT nucleotide bias that is characteristic of vertebrate mtDNA (Saccone et al. 1987). No stop codons were found in the fragments of cytochrome b.

4.3.2 Phylogenetic analyses The best-fit model of evolution selected using the Akaike Information Criterion for the combined data set of cytochrome b and 16S was the general time-reversible model with a set proportion of invariable sites, and with rates gamma-distributed (GTR+I+G). Trees derived from both ML and Bayesian methods recovered similar topologies, with generally stronger support for clades from Bayesian analyses (figure 4.2). Three well-supported lineages with little geographic overlap are evident. The most divergent lineage (#3) contained solely individuals collected in northern Jujuy and Bolivia (sites #1, 2, 3 and 5, figure 4.1), and it may represent P. borellii's putative sister taxon, P. cinerea. The remaining two lineages showed an unusual geographic distribution. Lineage #2 was found only in the central portion of the range, whereas lineage #1 was found northeast and southwest, but not in the central portion. All three clades within lineage #1 were mutually entirely allopatric. 81

0.98r 8

0.98 ~r 6 0.95- 1 Lineage 1: North - 4 .00 - 2 - 3 L5 7 j— 21 20 hJ -1 39 • 14 1.00 18 Lineage 1: Sierra de Santa Barbara • 19 • 17 • 16 1.00 • 15 1.00 12 L 11 0.97-- r 24 -TJ.99J- 25 26 Lineage 1: South 100~b 29 y 23 0.87TP L 22 28 0.98HT27 1.00 mr 36 H.37 1.00" if 38 u 39 0.85- jf- 41 T_ 40 Lineage 2: Central 0-74 £ 1 00-" 35 1.00 -31 -43 -33 L 32 I- 42 47 fC45 44 1.00 I- 46 Lineage 3: Northern Argentina & Southern Bolivia 50 I ,-48 -TOOL 49 -P. tucumana 0.1

Figure 4.2. Phylogenetic relationships of Argentine and Bolivian Pleurodema borellii based on Bayesian analyses of 16S and cytochrome b, using P. tucumana as an outgroup. Posterior probability values are indicated for each branch, filled circles indicate values less than 0.70. Haplotype codes correspond to those shown in figure 4.3 (p. 82). 82

4.3.3 Phylogeographic patterns The three lineages found in our phylogenetic analyses corresponded to three maximum parsimony networks that could not be connected at the 95% parsimony level. Network #1 (equivalent to lineage #1) and network #2 (lineage #2) were separated by at least 23 steps, and network #1 and #3 (lineage #3) by 27 steps. Relationships among the networks were assessed using phylogenetic approaches (above) and then phylogeographic patterns for each network were analyzed individually. Sample sizes for networks #2 and #3 were low and results of permutation tests were either non-significant or inconclusive (data not shown). Hierarchical nesting of network #1 resulted in three nested levels (figure 4.3). At the total cladogram level, we assigned clade 3-1 as interior and all others as tips since clade 3-1 contained haplotype #1, which had the highest outgroup probability and therefore is most likely to be the oldest haplotype in the network (Castelloe & Templeton 1994). Five nested clades showed significant geographical associations of haplotypes based On results of our permutation tests (table 4.2). One clade (1-1) had an inconclusive outcome, one clade (1-5) showed evidence of restricted gene flow, and three clades (3-1, 3-3, total cladogram) had patterns suggestive of past fragmentation, possibly coupled with range expansion or long distance colonization (table 4.3). At the total cladogram level, the three distinct clades had no geographical overlap (figure 4.4). Networks #2, #3, and all three clades at the total cladogram level for network #1 exhibited patterns characteristic of rapid population expansion using both mismatch distributions and haplotype and nucleotide diversity patterns (table 4.4). We calculated three barriers using Monmonier's maximum difference algorithm, implemented in Barrier, to examine correspondence with the phylogenetic and phylogeographic analyses. All three barriers correspond well with the three main lineages identified (figure 4.5). The analyses identified a sharp genetic discontinuity (Barrier 1) for P. borellii in the southern portion of the range (figure 4.5, ~27°S, between sites #15 and #16) as was found fori/, andinus (figure 3.2a, ~27°S, between sites #21 and #22), suggesting a possible barrier to gene flow for both species. 83

network #1 10 14*

•Rat 1-B -1 1-1 3-2 2-3* 12 8 1=2. 3-1 "ffe^: • ••inttiiiiiii* -«- 3-3 11-*- :2-5 ^ r* 22 #**KB**Sftft* t'2-4 /Jo 14 H-9| : 1-11. \ 23 27 28 13 -15 i- 4 2-7:

-3| 19' 21 If 16 25 -29 si-5 24^ -lal3_ 17 26 1-10 V 2-6 •*

network #3 2-1 42, *. network #2 « |lllllfllllilllllllil«.

2-2 : 4 7 46 2-1 40, j ( /-< (—45-50—44^ -•—41 < Mbf- * •* > < > 2?^* • ( • 11 » < M 34 * | 1- 3 • 8—49H ! .J 35

Figure 4.3. Maximum parsimony networks using mtDNA 16S and cytochrome b genes of Pleuroderna borellii. Filled circles indicate missing or unsampled haplotypes and font size approximates relative sample size for each haplotype. Nested clades are indicated with rectangles: thin solid lines for 1-step clades, dotted for 2-step clades, and thick solid for 3-step clades. The three networks could not be connected at the 95% parsimony level and were analysed separately. 84

4.3.4 Timing of divergence So that we might attempt direct comparisons of the two frog taxa, we estimated divergence times for splits between major clades using cytochrome b for both species. For H. andinus, at the deepest level, between network #1 (H. andinus clade) and network #2 (H. riojanus clade), the approximate time of divergence was 2-6 mya (table 2.4). Within network #1 only, divergence times were all less than 2 million years ago, placing them within the Pleistocene (table 2.4).

Table 4.2. Nested clade analysis results for Pleurodema borellii network #1 tested using 1000 permutations. Significant geographical associations from permutation tests indicated in boldface. Refer to figure 4.3 (p. 82) for clade numbers. Clade Permutational Chi-square statistic Probability 1-1 21.82 0.037 1-2 1.88 0.398 1-5 34.15 0.002 1-9 0.44 1.000 1-10 3.15 0.486 2-1 6.98 0.042 2-3 2.22 0.405 2-4 0.53 1.000 2-6 0.36 1.000 2-7 2.00 1.000 3-1 14.86 0.001 3-2 1.86 0.435 3-3 29.25 0.001 Total Cladogram 166.00 <0.001 85

Table 4.3. Inferences made for each clade in Pleurodema borellii network #1 found to have significant geographical associations. Refer to figure 4.3 (p. 82) for clade numbers. Clade Chain of Inference Inference 1-1 1-2-Tip/Interior status Inconclusive outcome cannot be determined 1-5 1-2-3-4-NO Restricted gene flow with isolation by distance 3-1 1-2-3-5-6-13-YES Long distance colonization possibly coupled with subsequent fragmentation OR past fragmentation followed by range expansion 3-3 1-2-3-5-15-NO Past fragmentation and/or long distance colonization Total 1-19-20-2-11-12-13 Long distance colonization possibly coupled Cladogram YES with subsequent fragmentation OR past fragmentation followed by range expansion

Table 4.4. Haplotype diversity (h), nucleotide diversity expressed as percent (JV), and results from mismatch distributions are shown for each network of Pleurodema borellii. Timing of population expansion estimated by x = 2/j,t, where [i is the mutation rate per generation per gene and t is time; mismatch distribution is not significantly different from the sudden expansion model when p>0.05. Refer to figure 4.3 (p. 82) for clade numbers. Clade n h JC Mismatch distribution Network #1 83 0.90 0.989 x=14.371,p=0.16 Clade 3-1 39 0.64 0.091 x-0.967,p=0.12 Clade 3-2 26 0.90 0.300 x=2.635,p=0.51 Clade 3-3 18 0.74 0.190 x=3.262, p=0.65 Network #2 35 0.79 0.210 x=2.502, p=0.66 Network #3 13 0.88 0.228 x=7.931,p=0.71 86

668 64*

LJ Boliv1 vii a 22M /^gentina

Uneage 1, Clacfe 3*1: North

24M Lineage 1, Clade 3*2: Sierra de Saute Barbara

26°-H

615 16 Lineage 1, Clade 3-3: Soutti

o a 10Q 200 3Q0i mmm mmmm 28°—I m ^ Figure 4.4. Geographical distribution of genetic lineages within Pleurodema borellii. Populations indicated with the same colour belong to the same lineage/network: black = network #1 with 3-step clades grouped using solid black lines, white = network #2, grey = network #3. Populations #7 and #19 are indicated with black and white circles as each population contains individuals belonging to networks #1 and #2. Shading represents elevation as in figure 4.1. 87

For P. borellii, at the deepest levels, the corrected average pairwise differences among lineage #3 and lineages #1 and #2 combined using the K2P molecular distance was 3.82%. Using the rates of 1 or 0.5% per million years suggested divergence among the clades of approximately 4-8 million years ago (mya). Lineages #1 and #2 differed by 4.93% suggesting a divergence time of 5-10 mya. Divergences among the three clades within lineage #1 suggested divergence times between 1-7 mya (3-1 & 3-2 = 3.43%; 3-1 & 3-3 = 2.34%; 3-2 & 3-3 = 1.33%). Estimates from coalescent analyses corresponded closely with the estimates from mean percent sequence divergence (table 4.5). The deepest divergences were among the main lineages (#1, #2 and #3) ranging from 4.2-10 mya. Within network #1, the three clades appeared to have diverged between 2-4 mya. The majority of divergence events for P. borelli likely occurred in the late Pliocene (>2 mya). Divergence events within P. borelli are thus substantially older than those within H. andinus.

4.3.5 Population-level statistical analyses We looked for evidence of range expansion at the population level by evaluating haplotype (h) and nucleotide (jt) diversity using the approach of Grant and Bowen (1998). Many populations of P. borellii showed high h and low jt indicative of population growth (table 4.1). Mismatch distributions for populations #7, 8 and 10 were not significantly different from the sudden expansion model, but population #7 did show a bimodal distribution. Population #7 showed low h and moderate jt indicative of a brief or prolonged bottleneck. Population #19 showed both high h and vindicative of secondary contact between differentiated lineages. Only populations #7 and 19 contained haplotypes from both network #1 and #2. Within network #1 of each species, the proportion of the total variation contributed by among-population comparisons was higher in P. borellii than in H. andinus (table 4.6). Skyline plots of the effective population size were very similar for both species and the 95% highest posterior density (HPD) intervals overlapped substantially (figure 4.6). Recent (<1 mya) population growth is evident in both species and supports findings using mismatch distributions and haplotype and nucleotide diversity. 88

66°

22°-

Barrier 2 24®-

Barrier 3

8 26 Barrier 1

28°

Figure 4.5. Locations of sharp genetic discontinuities (Barriers 1, 2 and 3), likely corresponding to barriers to gene flow among populations of Pleurodema borellii. Population pair-wise FST were analysed using Monmonier's maximum difference algorithm as implemented in Barrier (Manni et al. 2004). Thin grey lines = Delaunay triangulation; thicker grey lines = Voronoi' tessellation; black points = numbered sampling sites (see table 4.1, p. 75); grey points = virtual points for the Voronoi tessellation along map edges. Compare to patterns found for if. andinus (figure 3.2, p. 58). 89

Table 4.5. Estimates of divergence time for lineages of Pleurodema borellii based on analyses using BEAST (see text for details). The analyses were run using a molecular clock of 1% or 0.5% per million years. Refer to figure 4.3 (p. 82) for clade numbers. TMRCA = time to most recent common ancestor, mya = million years ago.

TMRCA (:mya )

Clade 1% 0.5%

Networks #1, #2 & #3 4.831 10.059

Networks #1 & #2 4.181 8.684

Within Network #1 1.955 3.975

Clade 3-1 0.552 1.133

Clade 3-2 0.613 1.242

Clade 3-3 1.263 2.541

Within Network #2 0.879 1.786

Within Network #3 0.976 2.023

Table 4.6. Percentage of variation (degrees of freedom) partitioned among and within populations for each species based on an Analysis of Molecular Variance (AMOVA). Analyses were run on populations belonging to network #1 only within each species. Pleurodema borellii Hypsiboas andinus Among populations 87% (11) 60% (20) Within populations 13% (71) 40% (207) 90

1000

CD o

CD N too

C o JXS ^z> 1 o u LLI

0.1 2 3 Time (mya)

1000n

CD o

CD 100 N CO C g z> Q. O Q_ (D >

LU

2 3 Time (mya) Figure 4.6. Bayesian skyline plots for H. andinus (top) and P. borellii (bottom). The thick solid line is the median estimate, and the pale grey lines show the 95% HPD limits. Both species show population growth less than 1 mya. 91

Both species showed a positive relationship between genetic distance and geographic distance (Mantel's test, P. borellii r = 0.48, p = 0.011; H. andinus r = 0.65, p = 0.001) and their slopes were not significantly different (P. borellii slope = 0.0161, H. andinus slope = 0.0109, 1000 replicates, p=0.24). Population pair-wise genetic distances per kilometer, however, were significantly higher in P. borellii (0.0657) than in H. andinus (0.0134) (1000 replicates, pO.OOl).

4.4 Discussion Both historical and contemporary connectivity among populations play important roles in shaping differentiation and eventual diversification among lineages. Hypsiboas andinus and Pleurodema borellii show significant phylogeographic structuring across their Argentine ranges. Given the great potential for isolation and for spatial habitat variation to affect the movement of individuals in the Andes, the patterns of genetic structuring likely reflect interactions between contemporary restriction of gene flow and historical barriers that caused range fragmentation with possible later secondary contact. The spatial distribution of haplotypes is somewhat similar in the two species, though the timing of divergence differs significantly, suggesting different evolutionary histories. The differences in life history and ecology of the two species may have played an important, yet surprising, role in differentiation of the two species.

4.4.1 Divergent lineages of P. borellii We found three divergent lineages of P. borellii in samples from northwestern Argentina and southern Bolivia. Lineage #3 occurs in the northernmost part of the range sampled in this study. This area corresponds well with the distribution of the sister taxon to P. borellii, P. cinerea, believed to be ecologically divergent as it inhabits the arid high altitude Andean plateau (the puna) (Cei 1980). McLister et al. (1991) studied five populations, three within the range of P. cinerea and two within the range of P. borellii, and found little genetic differentiation at 15 allozyme loci between the two putative species. In fact, one population of P. cinerea clustered genetically with those of P. borellii. The described differences in ecology may not translate to genetically differentiated taxa. Our study included individuals from high altitudes (sites #2, 3, 5, 17) 92 as well as low altitudes (all remaining sites) throughout the range of both species, including two individuals of P. borellii from southern Bolivia. We found divergent mtDNA lineages arranged in a north-south series rather than according to low vs. high altitude. Two low altitude individuals from southern Bolivia (#1) clustered within lineage #3, while high altitude samples further south in Argentina (#17) clustered within lineage #1. Unambiguous samples of P. borellii and P. cinerea from high altitude sites in Bolivia are required to further clarify the relationship between these divergent lineages. Calls produced by males are the main cue for mate selection in frogs (e.g. Ryan & Rand 1990, Ryan et al. 1992, Lampert et al. 2003). The limited call data available for these two species indicate a difference in the call pulse rate, with P. borellWs rate twice that of P. cinerea (McLister et al. 1991). Analyses of calls recorded at our site #12 confirm a pulse rate of approximately 120 pulses per second (data not shown) as reported for P. borellii by McLister et al. (1991). Although these differences may be enough to indicate species status, McLister et al. (1991) point out that a strong positive correlation exists between temperature and pulse rate in anurans. Since P. cinerea occurs at high altitudes the calls were recorded at lower temperatures than those for P. borellii suggesting that the rate difference could be due to temperature alone. Call data under controlled conditions and female choice experiments are needed to establish the taxonomic status of these species. The three divergent lineages are nearly allopatric suggesting a long history of isolation. These lineages are estimated to have become separated about 4-10 mya, a time of rapid uplift of the Andes and the formation of novel arid high altitude habitats (Gregory-Wodzicki 2000). This orogeny substantially altered atmospheric circulation and patterns of precipitation (Hooghiemstra & van der Hammen 1998, Gregory-Wodzicki 2000), as well as the flow of rivers (Hooghiemstra & van der Hammen 1998). These changes in the landscape may have significantly affected connectivity of populations perhaps leading to the divergence of the lineages we have described. Lineages #2 and #3 show mismatch distributions and haplotype and nucleotide diversity patterns consistent with population growth. If range expansion accompanies population growth then secondary contact of these divergent lineages may occur. Lineage #2 is currently found in 93 sympatry with lineage #1 at the northern and southern edges of its range across our study area (sites #7 and #19).

4.4.2 Phylogeography of P. borellii At the total cladogram level, network #1 contains three entirely allopatric clades. The mechanism inferred for the divergence was either long distance colonization, possibly coupled with subsequent fragmentation, or past fragmentation followed by range expansion. Given that clades 3-2 and 3-3, which are more closely related to each other, are separated by more than 200 km (where only individuals belonging to network #2 are found), we would suggest that long distance colonization is unlikely. The mismatch distribution and haplotype and nucleotide analyses also suggest that all three clades have undergone population growth and perhaps range expansion, lending support to the alternative of fragmentation followed by range expansion. The divergence of these three clades likely occurred approximately 1-7 mya suggesting that the events may predate Pleistocene climatic cycling. Once again, the uplift of the Andes may have played a critical role in the fragmentation of these clades. For example, a close correspondence of the distribution of clade 3-2 to the isolated mountain range of Sierra de Santa Barbara suggests that the populations became isolated within this mountain range (figure 4.4). As mentioned above, individuals of both lineage #1 and #2 are found at two sites (#7 and 19). Population expansion was detected in lineage #2 as well as the north and south clades of lineage #1, which may account for the apparent secondary contact at these sites. The geographic distribution of these lineages is puzzling since genetic exchange appears to have continued between the three clades of lineage #1 but not with the central lineage #2, although it is found directly in between. This "leap-frog" pattern may indicate that other populations not sampled may link the clades of lineage #1 or that lineage #1 was more extensive in the past but is now extinct across the central part of our study area, possibly replaced by lineage #2. The genetic divergence between these two lineages is nearly 5% at cytochrome Z>, which suggests that long-term isolation has led to the differentiation between these lineages. 94

4.4.3 Comparative phylogeography Although both species show strong genetic structure and similar phylogeographic patterns across the study area, the timing of divergence differs dramatically (tables 2.4 and 4.5). Three possibilities exist to explain the patterns found for the two species. First, the patterns may be similar across the study area but the timing of differentiation events may differ due to different biogeographic history. Second, while the patterns and the timing of events may be similar, differences in the rate of molecular evolution in each species gives the apparent timing difference. Last, the depth of divergence may differ due to historical demographic factors for each species even though the barriers leading to the differentiation were the same. The deepest divergences are estimated to have originated within H. andinus IH. riojanus between 2-6 mya, and in P. borellii I P. cinerea between 4-10 mya. The timing does suggest a strong influence of the Andean uplift in population differentiation in both species but the effects were clearly different for each species. The dates for if. andinus of 2-6 my may be related to the recent dispersal of the entire Andean species group (H. balzani, H. marianitae, H. andinus, H. riojanus) into the Andes from the Atlantic Forest of eastern Brazil (Faivovich et al. 2004). Connections between the Atlantic Forest of southeastern Brazil and the Amazon lowland forests at the foot of the Andes have been suggested for terrestrial species (small mammals, Costa 2003; plants, Por 1992) as well as obligate freshwater species (Characiformes fish, Hubert & Renno 2006) in the last 10 my. The major route of dispersal was likely through the Parana river basin, which drains major rivers of the Andes of northwestern Argentina, Paraguay and southern Brazil flowing into the Atlantic. There are no molecular phylogenies for the group of Pleurodema species. Duellman and Veloso (1977) suggested that the origin of the group was likely in the Andes of southern South America. The basis of this biogeographic scenario, however, is a non-molecular phylogeny that does not correspond well with the limited molecular data available for the group (mtDNA 16S sequences for 6 of 12 species, data not shown). Drawing cautious support from Duellman and Veloso (1977) and given the deep genetic divergence among lineages we found, suggests that the taxa likely have existed in northwestern Argentina for much longer than 10 my. The differences in timing of these deepest divergence events in the two taxa, therefore, are 95 more probably due to different biogeographic histories of the two species than to the influence of life history attributes on gene flow. Factors such as body size, generation time, population size, metabolic rate, and clutch size (independently or correlated) may affect the rate of molecular evolution due to their effects on DNA mutation and repair, leading to different rates among taxa (Martin & Palumbi 1993, Bromham 2002, Crawford 2003, Pulquerio & Nichols 2007). The clades within lineage #1 of each species do broadly overlap but the timing of divergence differs. Using the coalescent analyses, the time of the most recent common ancestor (TMRCA) for H. andinus lineage #1 occurred 1-2 mya (table 2.4), and for P. borellii lineage #1 occurred 2-4 mya (table 4.5). These analyses used two different rates of evolution of mtDNA (1% and 0.5%). The TMRCA for the 0.5% rate for #. andinus and the TMRCA for the 1% rate for P. borellii both correspond to 2 mya, implying that a two-fold molecular rate difference between the species would be required to attain the same date. Although the two species differ in fecundity, they are similar in body size, generation time, and likely metabolic rates hence a two-fold difference in rates is highly unlikely (Martin & Palumbi 1993, Crawford 2003, Pulquerio & Nichols 2007). Analyses of genetic discontinuity for both species suggest some common areas of genetic divergence, and possible barriers, across this part of the Andes. Both species show the most supported barrier to gene flow across the southern portion of the study area (~27°S, figures 3.1 and 3.2a for if. andinus, figures 4.4 and 4.5 for P. borellii), although there is currently no obvious geological, hydrological or habitat-based disjunction. The divergence among populations in this area dates to approximately 2-6 mya in H. andinus (lineages #1 and #2) and 5-10 mya in P. borellii (clade 3-3 of lineage #1 and lineage #2). The general concordance of the distributions of clades within lineage #1 of each species further suggests possible common historical vicariant events. A historical fragmentation event may have driven the divergence in both species but the demographic history of each species may have affected the genetic imprint left by the event. Large or highly structured populations tend to show deeper lineage separation in response to a barrier because (i) more ancestral polymorphisms were available which might be fixed in each of the sundered lineages, or (ii) already differentiated and spatially structured lineages may characterize the sundered lineages (Avise 2000). 96

Similar geographic patterns with different temporal signatures may also be due to local extinction and reoclonization events across the barrier, especially if the barrier is intermitten (e.g. climatic cycling) (Avise 2000). If the barrier formed over a period of time or the species response varied then the barrier may have been more effective (and acted longer) for one species than another. Such events would erase the phylogenetic signal of previous differentiation events leading to differences in temporal patterns. The different temporal patterns found in our study may be due to differences in demographic history of P. borellii and//, andinus. Only one detailed molecular study for northwestern Argentina is available for comparison. Quiroga and Premoli (2007) studied a cold-tolerant tree (Podocarpus parlatorei) inhabiting the uppermost levels of the Yungas montane forests across northwestern Argentina and Bolivia. This species shows genetically distinct northern, central and southern clades generally matching those found in our study, particularly those of//, andinus. Quiroga and Premoli (2007) suggested that, as this species is adapted to cool climates, populations of P. parlatorei would have expanded to a more continuous range at lower altitudes (and lower latitude) during glacial maxima, to become more fragmented during warmer periods. Although this pattern of movement is opposite to that expected for the frogs in this study - expansion during warmer and more mesic episodes - any cycling between cool and warm periods may result in genetic divergence of populations isolated in suitable refugia. In northwestern Argentina, three potential refugia may be identified based on species distributions (Brown et al. 2001), and the genetic divergence patterns described for these three co-distributed species (//. andinus and P. borellii, this study; P. parlatorei, Quiroga & Premoli 2007). The northern (centred ~22°S) and southern (~27°S) refugia have been previously postulated based on high species richness and endemism (Brown et al. 2001) and currently have a highly mesic climate, possibly indicative of a mesic climate in the past. Our results suggest a third refugium, situated in the Sierra de Santa Barbara (~24.5°S) given the genetically distinct lineages associated with this area and the current discontinuous distribution of yungas forests and possible presence of mesic habitats in the past (see chapter 3). Our study supports an important role of mountain habitats in maintaining biodiversity through the preservation of many different lineages that are able to shift altitudinally to avoid 97 unfavourable habitats (and possibly extinction). In addition, fragmentation and reconnection among populations provide opportunities for differentiation and possibly speciation.

4.4.4 Ecology and life history The described patterns of genetic differentiation highlight the importance of differences in life history and ecology of the two species. Given the expected correlates of life history and gene flow, we expected//, andinus to show stronger differentiation due to its lower fecundity and its limited ability to move outside of the riparian zones of its breeding streams. Pleurodema borellii breeds opportunistically in temporary pools of water in a wide range of habitats, including unusual choices such as swimming pools (Halloy & Fiano 2000), giving the appearance of a high-dispersal habitat generalise This opportunistic breeder is found in high densities when rains begin to fill shallow depressions. Regardless of these characteristics, population connectivity nevertheless appears to be higher in H. andinus, even across habitats that are more likely to be barriers for this aquatic tree frog, such as the semi-arid scrubland (chaco and chaco serrano) separating the Sierra de Santa Barbara (P. borellii sites #7, 8, 11, figure 4.4) from all other populations. Both species showed a significant positive relationship between genetic and geographic distances, and the regression slopes did not differ. Genetic distances per kilometer, however, are significantly higher in P. borellii and the proportion of among- population variance is also higher in P. borellii (within lineage #1 of each species). Thus, although P. borellii is found in a wider range of habitats than H. andinus, and appears to breed opportunistically, the genetic data show higher differentiation, and hence lower levels of gene flow, in P. borellii than H. andinus at similar spatial scales. This result is counter-intuitive given that several generalist species have been found to display higher gene flow than habitat-restricted species (Schauble & Moritz 2001, Stuart-Fox et ah 2001, Crawford et al. 2007). Factors other than fecundity and habitat specialization may therefore play a critical role in driving patterns of differentiation. Genetic measures of dispersal, however, may underestimate movement if immigrants into new populations have low reproductive success (reviewed in Slatkin 98

1985, Bohonak 1999). Individuals may be capable of moving long distances but may not be well adapted to the habitat, climate, competitors, or other features of the new site and fail to reproduce there, hence leaving no genetic trace of their dispersal. In amphibians dispersal distance is predicted to be greater for juveniles than adults suggesting that the main agents of gene flow among populations are juveniles (reviewed in Smith & Green 2006). InH. andinus, passive or active dispersal may also take place when the eggs or tadpoles are moved downstream with currents. Downstream gene flow has been reported for some amphibians (Kraaijeveld-Smit et al. 2005, Measey et al. 2007). Pleurodema borellii, on the other hand, has no potential for movement at these stages since eggs and tadpoles are isolated in small closed-basin pools. The ephemeral nature and usual lack of connectivity among breeding pools of P. borellii may have a strong influence on population demographics. If dispersal is indeed highest in juveniles, then any factors that might affect the success of a breeding pool may limit gene flow. These pools are often small and vulnerable to desiccation, leading to high competition among the tadpoles for food and strong pressure to reach metamorphosis rapidly (Halloy & Fiano 2000). Larger (and usually older) tadpoles will cannibalize smaller tadpoles and females will avoid ovipositing at sites that already contain high densities of developed tadpoles (Halloy & Fiano 2000). It is also possible to observe complete failure if these pools dry up before metamorphosis is complete (personal observation). The stochasticity associated with breeding in these pools suggests that there may be high temporal variability in reproductive success and large temporal fluctuations in population size, which is known to reduce the effective population size substantially (Wright 1938, Vucetich et al. 1997). The skyline plots analyses, did not, however, detect smaller effective population size in P. borellii as compared to H. andinus. Philopatry may also limit the hypothesized large dispersal capabilities of P. borellii. Since our study employed mtDNA, a maternally inherited marker, the dispersal patterns are specific to females. Patterns of dispersal depend on many factors, including mating system, which may influence sex-biased dispersal. Three broad categories of theory summarize the advantages to dispersal from natal site for each sex: (i) competition for resources, (ii) competition for mates, and (iii) inbreeding avoidance (Greenwood 99

1980, Perrin & Mazalov 2000). General patterns are available for well-studied groups such as mammals (typically polygynous; males disperse), and birds (typically monogamous; females disperse) but little is known about this matter in amphibians (Goudet et al. 2002). Amphibians typically show a polygynous mating system with strong male-male competition for territories and strong female preference for mates (Duellman & Trueb 1986). Females also tend to breed once per season whereas males may breed with many females (Duellman & Trueb 1986), therefore the cost of inbreeding would be higher for females than males. These aspects of the mating system suggest that females should disperse from the natal site to avoid inbreeding, and that males should remain (even though polygyny predicts the opposite) as familiarity with the area would lead to better success in obtaining high quality territories and therefore attracting females (Austin et al. 2003). Recent molecular studies of amphibians have found female biased dispersal (Austin et al. 2003, Palo et al. 2004), weak male biased dispersal (Lampert et al. 2003), as well as no sex bias in dispersal (Smith & Green 2006, Cabe et al. 2007). Of particular interest is the study of Physalaemus pustulosus (Lampert et al. 2003), a species that shares many life history attributes with P. borellii as it is abundant, inhabits a wide range of habitats (often disturbed), has a lek-mating system, breeds in temporary ponds depositing eggs in foam nests, and has hypothesized high dispersal rates. Across our study area P. borellii was often found breeding in pools with another member of the genus, Physalaemus biligonigerus. If a male bias in dispersal exists in P. borellii as found for P. pustulosus, then the strong genetic structure found in our study may be linked to female philopatry in P. borellii. Phylogeographic patterns are typically categorized as concordant or discordant, but recent studies have suggested that intermediate patterns may exist (Sullivan et al. 2000, Steele & Storfer 2007). The two species in our study show such an intermediate pattern - concordance in parts of the distribution and discordance in others. In addition, the effect of temporal differences, if they exist, may significantly affect the extent to which the phylogeographic patterns are concordant. Life history data is scarce for most non-model organisms, but our study suggests that the influence of such characteristics on differentiation is likely significant and that patterns are not easily generalized across taxa. 4.5 References

Alvarado Bremer JR, Vinas J, Mejuto J, Ely B, Pla C (2005) Comparative phylogeography of Atlantic bluefin tuna and swordfish: the combined effects of vicariance, secondary contact, introgression, and population expansion on the regional phylogenies of two highly migratory pelagic fishes. Molecular Phylogenetics andEvolution 36, 169-187.

Austin JD, Davila JA, Lougheed SC, Boag PT (2003) Genetic evidence for female-biased dispersal in the bullfrog, Rana catesbeiana (Ranidae). Molecular Ecology 12, 3165- 3172.

Austin JD, Lougheed SC, Boag PT (2004) Discordant temporal and geographic patterns in maternal lineages of eastern north American frogs, Rana catesbeiana (Ranidae) and Pseudacris crucifer (Hylidae). Molecular Phylogenetics and Evolution 32, 799- 816.

Avise JC (1998) The history and purview of phylogeography: a personal reflection. Molecular Ecology 7, 371-379.

Avise JC (2000) Phylogeography: The history and formation of species Harvard University Press, Cambridge.

Avise JC, Helfman GS, Saunders NC, Hales LS (1986) Mitochondrial DNA Differentiation in North Atlantic Eels: Population Genetic Consequences of an Unusual Life History Pattern. Proceedings of the National Academy of Sciences 83, 4350-4354.

Avise JC, Walker D, Johns GC (1998) Speciation durations and Pleistocene effects on vertebrate phylogeography. Proceedings of the Royal Society of London Series B- Biological Sciences 265, 1707-1712.

Basso NG, Kehr AI (1992) Estructura poblacional y crecimiento de Hyla pulchella pulchella (Anura: Hylidae). Acta Zoologica Lilloana 41, 143-148.

Bohonak AJ (1999) Dispersal, Gene Flow, and Population Structure. The Quarterly Review of Biology 74, 21-45.

Bromham L (2002) Molecular Clocks in Reptiles: Life History Influences Rate of Molecular Evolution. Molecular Biology and Evolution 19, 302-309.

Cabe PR, Page RB, Hanlon TJ, et al. (2007) Fine-scale population differentiation and gene flow in a terrestrial salamander (Plethodon cinereus) living in continuous habitat. Heredity 98, 53-60.

Casgrain P, Legendre P (2001) The R Package for multidimensional and spatial analysis (version 4, release 6). Universite de Montreal, Montreal. 101

Castelloe J, Templeton AR (1994) Root probabilities for intraspecific gene trees under neutral coalescent theory. Molecular Phylogenetics and Evolution 3, 102-113.

Cei JM (1980) Amphibians of Argentina Monitore Zoologico Italiano N. S Monografia 2.

Clement M, Posada D, Crandall KA (2000) TCS: a computer program to estimate gene genealogies. Molecular Ecology 9, 1657-1659.

Costa LP (2003) The historical bridge between the Amazon and Atlantic Forest of Brazil: a study of molecular phylogeography with small mammals. Journal of Biogeography 30,71-86.

Crawford AJ (2003) Relative rates of nucleotide substitution in frogs. Journal of Molecular Evolution 57, 636-641.

Crawford AJ, Bermingham E, Polania C (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. Molecular Ecology 16, 4789-4807.

Drummond AJ, Rambaut A (2007) BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evolutionary Biology 7, 214.

Drummond AJ, Rambaut A, Shapiro B, Pybus OG (2005) Bayesian Coalescent Inference of Past Population Dynamics from Molecular Sequences. Molecular Biology and Evolution 22, 1185-1192.

Duellman WE, Trueb L (1986) Biology of Amphibians The Johns Hopkins University Press, Baltimore.

Duellman WE, Veloso MA (1977) Phylogeny of Pleurodema (Anura: Leptodactylidae): a biogeographic model. Occasional Papers of the Museum of Natural History, The University of Kansas 64, 1-46.

Duellman WE, de la Riva I, Wild ER (1997) Frogs of the Hyla armata and Hyla pulchella groups in the Andes of South America, with definitions and analyses of phylogenetic relationships of Andean groups of Hyla. Scientific Papers of the Natural History Museum of the University of Kansas 3, 1-41.

Faivovich J, Garcia PCA, Ananias F, et al. (2004) A molecular perspective on the phylogeny of the Hyla pulchella species group (Anura, Hylidae). Molecular Phylogenetics and Evolution 32, 938-950.

Goebel AM, Donnelly JM, Atz ME (1999) PCR primers and amplification methods for 12S ribosomal DNA, the control region, cytochrome oxidase I, and cytochrome b in Bufonids and other frogs, and an overview of PCR primers which have amplified DNA in amphibians successfully. Molecular Phylogenetics and Evolution 11, 163- 199. Goudet J, Perrin N, Waser P (2002) Tests for sex-biased dispersal using bi-parentally inherited genetic markers. Molecular Ecology 11,1103-1114. Grant WS, Bowen BW (1998) Shallow population histories in deep evolutionary lineages of marine fishes: insights from sardines and anchovies and lessons for conservation. Journal of Heredity 89, 415-426.

Greenwood P (1980) Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour 28, 1140-1162.

Gregory-Wodzicki KM (2000) Uplift history of the Central and Northern Andes: A review. Geological Society of America Bulletin 112, 1091-1105.

Halloy M, Fiafio JM (2000) Oviposition site selection in Pleurodema borellii (Anura: Leptodactylidae) may be influenced by tadpole presence. Copeia 2000, 606-609.

Haney RA, Silliman BR, Rand DM (2007) A multi-locus assessment of connectivity and historical demography in the bluehead wrasse (Thalassoma bifasciatum). Heredity 98, 294-302.

Holland BS, Cowie RH (2007) A geographic mosaic of passive dispersal: population structure in the endemic Hawaiian amber snail Succinea caduca (Mighels, 1845). Molecular Ecology 16, 2422-2435.

Hooghiemstra H, van der Hammen T (1998) Neogene and Quaternary development of the neotropical rain forest: the forest refugia hypothesis, and a literature overview. Earth-Science Reviews 44, 147-183.

Hubert N, Renno J-F (2006) Historical biogeography of South American freshwater fishes. Journal of Biogeography 33, 1414-1436.

Keyghobadi N, Roland J, Strobeck C (2005) Genetic differentiation and gene flow among populations of the alpine butterfly, Parnassius smintheus, vary with landscape connectivity. Molecular Ecology 14, 1897-1909.

Kraaijeveld-Smit FJL, Beebee TJC, Griffiths RA, Moore RD, Schley L (2005) Low gene flow but high genetic diversity in the threatened Mallorcan midwife toad Alytes muletensis. Molecular Ecology 14, 3307-3315.

Lampert KP, Rand AS, Mueller UG, Ryan MJ (2003) Fine-scale genetic pattern and evidence for sex-biased dispersal in the tungara frog, Physalaemus pustulosus. Molecular Ecology 12, 3325-3334.

Lougheed SC, Gascon C, Jones DA, Bogart JP, Boag PT (1999) Ridges and rivers: a test of competing hypotheses of Amazonian diversification using a dart-poison frog (Epipedobates femoralis). Proceedings of the Royal Society of London B 266, 1829- 1835. 103

Manni F, Guerard E (2004) Barrier vs. 2.2. Manual of the user. Population genetics team, Museum of Mankind (Musee de l'Homme), Paris [Publication distributed by the authors].

Manni F, Guerard E, Heyer E (2004) Geographic patterns of (genetic, morphologic, linguistic) variation: how barriers can be detected by "Monmonier's algorithm". Human Biology 76, 173-190.

Martin PR, McKay JK (2004) Latitudinal variation in genetic divergence of populations and the potential for future speciation. Evolution 58, 938-945.

Martin AP, Palumbi SR (1993) Body Size, Metabolic Rate, Generation Time, and the Molecular Clock. Proceedings of the National Academy of Sciences of the United States of America 90, 4087-4091.

McLister JD, Lougheed SC, Bogart JP (1991) Elecrrophoretic and Vocalization Comparisons among 3 Leptodactylid Frogs (Pleurodema Spp) from Northwestern Argentina. Canadian Journal of Zoology 69, 2397-2403,

Measey GJ, Galbusera P, Breyne P, Matthysen E (2007) Gene flow in a direct- developing, leaf litter frog between isolated mountains in the Taita Hills, Kenya. Conservation Genetics 8, 1177-1188.

Palo JU, D. Lesbarreres, D. S. Schmeller, C. R. Primmer, and J. Media (2004) Microsatellite marker data suggest sex-biased dispersal in the common frog Rana temporaria. Molecular Ecology 13, 2865-2869.

Perrin N, Mazalov V (2000) Local Competition, Inbreeding, and the Evolution of Sex- Biased Dispersal. The American Naturalist 155, 116-127.

Petit RJ (2008) The coup de grace for the nested clade phylogeographic analysis? Molecular Ecology 17, 516-518.Posada D, Crandall KA (1998) Modeltest: testing the model of DNA substitution. Bioinformatics 14, 817-818.

Por FD (1992) Sooretama: the Atlantic rainforest of Brazil. SPB Academic Publishing, The Hague.

Posada D, Crandall KA, Templeton AR (2000) GeoDis: A program for the Cladistic Nested Analysis of the Geographical Distribution of Genetic Haplotypes. Molecular Ecology 9, 487-488.

Pulquerio MJF, Nichols RA (2007) Dates from the molecular clock: how wrong can we be? Trends in Ecology & Evolution 22, 180-184.

Quiroga MP, Premoli AC (2007) Genetic patterns in Podocarpus parlatorei reveal the long-term persistence of cold-tolerant elements in the southern Yungas. Journal of Biogeography 34, 447-455. Rambaut A, Drummond AJ (2007) Tracer vl.4, Available from http://beast.bio.ed.ac.uk/Tracer. Ryan MJ, Rand AS (1990) The sensory bias of sexual selection for complex calls in the Tungara frog, Physalaemus pustulosus (sexual selection for sensory exploitation). Evolution 44, 305-314.

Ryan MJ, Perrill SA, Wilczynski W (1992) Auditory tuning and call frequency predict population-based mating preferences in the Cricket Frog, Acris crepitans. The American Naturalist 139, 1370-1383.

Saccone C, Attimonelli M, Sbisa E (1987) Structural elements highly preserved during the evolution of the D-loop-containing region in vertebrate mitochondrial DNA. Journal of Molecular Evolution 26, 205-211.

Schauble CS, Moritz C (2001) Comparative phylogeography of two open forest frogs from eastern Australia. BiologicalJournal of the Linnean Society 74, 157-170.

Schneider S, Excoffier L (1999) Estimation of past demographic parameters from the distribution of pairwise differences when the mutation rates vary among sites: Application to human mitochondrial DNA. Genetics 152, 1079-1089.

Schneider S, Roessli D, Excoffier L (2000) Arlequin ver. 2.000: A software for population genetics data analysis. Genetics and Biometry Laboratory, University of Geneva, Switzerland.

Slatkin M (1985) Gene flow in natural populations. Annual Review of Ecology and Systematics 16, 393-430.

Smith MA, Green DM (2006) Sex, isolation and fidelity: unbiased long-distance dispersal in a terrestrial amphibian. Ecography 29, 649-658.

Steele CA, Storfer A (2007) Phylogeographic incongruence of codistributed amphibian species based on small differences in geographic distribution. Molecular Phylogenetics and Evolution 43, 468-479.

Stuart-Fox DM, Schneider CJ, Moritz C, Couper PJ (2001) Comparative phylogeography of three rainforest-restricted lizards from mid-east Queensland. Australian Journal of Zoology 49, 119-127.

Sullivan J, Arellano E, Rogers DS (2000) Comparative Phylogeography of Mesoamerican Highland Rodents: Concerted versus Independent Response to past Climatic Fluctuations. The American Naturalist 155, 755-768.

Templeton AR (1998) Nested Clade Analysis of phylogeographic data: testing hypotheses about gene flow and population history. Molecular Ecology 7, 381-397. 105

Templeton AR (2001) Using phylogeographic analyses of gene trees to test species status and processes. Molecular Ecology 10, 779-791.

Templeton AR, Crandall KA, Sing CF (1992) A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping and DNA sequence data. III. Cladogram estimation. Genetics 132, 619-633.

Templeton AR, Routman E, Phillips CA (1995) Separating population structure from population history: a cladistic analysis of the geographical distribution of mitochondrial DNA haplotypes in the Tiger Salamander, Ambystoma tigrinum. Genetics 140, 767-782.

Templeton AR, Sing CF (1993) A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping. IV. Nested analyses with cladogram uncertainty and recombination. Genetics 134, 659-669.

Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG (1997) The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research 24, 4876-4882.

Turner TF, Trexler JC (1998) Ecological and historical associations of gene flow in Darters (Teleostei: Percidae). Evolution 52,1781-1801.

Turner TF, Trexler JC, Kuhn DN, Robinson HW (1996) Life-history variation and comparative phylogeography of Darters (Pisces: Percidae) from the North American Central Highlands. Evolution 50, 2023-2036.

Uthicke S, Benzie JAH (2003) Gene flow and population history in high dispersal marine invertebrates: mitochondrial DNA analysis of Holothuria nobilis (Echinodermata: Holothuroidea) populations from the Indo-Pacific. Molecular Ecology 12, 2635-2648.

Vaha JP, Erkinaro J, Niemela E, Primmer CR (2007) Life-history and habitat features influence the within-river genetic structure of Atlantic salmon. Molecular Ecology 16, 2638-2654.

Vaira M (2002) Anurans of a subtropical montane forest in northwestern Argentina: ecological survey and a proposed list of species of conservation concern. Biodiversity and Conservation 11, 1047-1062.

Vucetich JA, Waite TA, Nunney L (1997) Fluctuating Population Size and the Ratio of Effective to Census Population Size. Evolution 51, 2017-2021.

Wenink PW, Baker AJ, Rosner H-U, Tilanus MGJ (1996) Global mitochondrial DNA phylogeography of holarctic breeding dunlins (Calidris alpina). Evolution 50, 318- 330.

Wright S (1938) Size of a population and breeding structure in relation to evolution. Science 87, 430-431. 106

Yu DH, Chu KH (2006) Low genetic differentiation among widely separated populations of the pearl oyster Pinctada fucata as revealed by AFLP. Journal of Experimental Marine Biology and Ecology 333, 140-146. CHAPTER 5. SUMMARY AND CONCLUSIONS

5.1 Summary and Conclusions This thesis describes the genetic differentiation of populations of two co- distributed species of Andean frogs to contribute to our understanding of the timing and dynamics of diversification and speciation. Few molecular phylogeographic studies of Neotropical species exist, certainly when compared to Holarctic species, and, within the Neotropics, our understanding of diversification in biomes other than the lowland forests of the Amazon Basin is limited by the paucity of data for any major taxon. Studies of taxa inhabiting Andean mountain habitats, a region acknowledged as central to much evolutionary diversification of South America during the Neogene, and yet largely unstudied from this perspective, contribute important first steps in our developing understanding of the processes of diversification that generate biodiversity. This study highlights important issues surrounding species designation in amphibians, particularly in lower-latitude regions. As a group, amphibians are structurally conservative yet notoriously variable in superficial characters (e.g. colouration, patterning) (Cherry et al. 1978); nevertheless, they often display deep genetic differentiation (e.g. Wynn & Heyer 2001, Lougheed et al. 2006, Fouquet et al. 2007). This all leads to difficulty in assigning species categories. A major impediment to understanding is a lack of experimental and observational data on mating system and reproductive compatibility among populations differentiated to varying degrees. In addition, we lack basic data for many species of amphibians (especially molecular) as is evidenced by the continued increase in new species descriptions (Duellman 1999). The great diversity of amphibians, particularly anurans, in the tropics (Duellman 1999) may explain in part why this group is poorly understood. The disparity in research intensity in temperate vs. tropical regions leads to considerable uncertainty when attempting to infer general patterns for tropical and subtropical amphibians. In particular, great caution seems warranted in studying taxa that are imperfectly circumscribed, as evidenced by the discoveries described here regarding the relationships of various deeply differentiated lineages found within each of the two study taxa. 108

The taxonomic status of Bolivian and Argentine populations ofH. andinus is not well established (studies by Faivovich et al. 2004 and Faivovich et al. 2005 did not contain samples from Bolivia). The four samples from northern Bolivia (~17°S) used in this study were genealogically separate from all other H. andinus samples from northwestern Argentina and southern Bolivia, and may represent a distinct species. Within northwestern Argentine samples, we found two distinct lineages (3.1% divergence at cytochrome b), but 27 individuals from 6 populations, 5 of which were asserted to contain only H. andinus (Cei 1980, Duellman et al. 1997), clustered with the putative sister taxon, H. riojanus. This may be due to a sympatric distribution of two morphologically similar taxa (i.e. cryptic species) in the provinces of Tucuman and Catamarca, or to secondary contact between divergent lineages. Given that morphologically indistinguishable individuals representing both mitochondrial lineages can be found breeding in the same locale, only strong reproductive isolating mechanisms would prevent hybridization, if indeed the two are different species. Call differences are minimal between the taxa (Barrio 1965) although playback and female choice experiments have not been reported. Limited sound spectrographic analysis of//. andinus from five of the populations sampled suggested that calls are highly variable within and among populations (Liadsky 2003). The current evidence—i.e. limited morphological and call differences, if any; shallow genetic divergence—suggests that these two taxa are probably not reproductively isolated species. Our discovery of distinct northern Bolivian lineages implies that H. riojanus is nested within H. andinus, making H. andinus a paraphyletic taxon. We found three divergent lineages (>3.8% at cytochrome b) of P. borellii in samples from northwestern Argentina and southern Bolivia. Lineage #3 corresponds well with the distribution of the sister taxon to P. borellii, P. cinerea, which is believed to be ecologically divergent as it inhabits the arid high altitude Andean plateau (the puna) (Cei 1980). The described differences in ecology may not translate to genetically differentiated taxa however, since the divergent mtDNA lineages were arranged in a north-south series rather than according to low vs. high altitude. The limited call data available for these two species (McLister et al. 1991) indicate a difference in the call pulse rate, with P. borellii''s rate twice that of P. cinerea. Although these differences may 109 be consistent with species status, McLister et al. (1991) point out that a strong positive correlation exists between temperature and pulse rate in anurans. Since P. cinerea occurs at high altitudes the calls were recorded at lower temperatures than those for P. borellii suggesting that the rate difference could be due to temperature alone. Unambiguous samples of P. borellii and P. cinerea from high altitude sites in Bolivia are required to further clarify the relationship between these divergent lineages. The geographic distribution of P. borellii lineages #1 and #2 is puzzling since genetic exchange appears to have continued within lineage #1 but not with the central lineage #2, although it is found directly in between. This "leap-frog" pattern may indicate that other unsampled populations may link the clades of lineage #1, or that lineage #1 was more extensive in the past but is now extinct across the central part of our study area, possibly replaced by lineage #2. The genetic divergence between these two lineages is nearly 5% at cytochrome b, which suggests that long-term isolation has led to the differentiation between these lineages. Secondary contact of individuals of lineages #1 and #2 is found only at two sites. This novel finding may provide support for distinct genetic species within a morphologically uniform group of P. borellii across northwestern Argentina. The levels of divergence found among genealogical lineages within Pleurodema borellii and Hypsiboas andinus may represent distinct species, a finding that could be corroborated by mate choice experiments to evaluate reproductive isolation. Most studies of evolutionary processes in Neotropical taxa have focused on lowland forest taxa, especially those associated with the Amazon Basin (e.g. mammals: Patton et al. 2000; frogs: Gascon et al. 1998, Symula et al. 2003; birds: Haffer 1969, Nores 1999) or the scrubland and mountains of southern South America (e.g. mammals: Kim et al. 1998; fish: Ruzzante et al. 2006; plants: Pastorino & Gallo 2002, Muellner et al. 2005). Hitherto, such studies have provided the basis for our understanding of the processes of diversification anywhere in the Neotropics. The entire Neotropical realm, however, spans a great range in latitude (~10°N to ~50°S) and altitude (sea level to greater than 6000 masl), thus encompassing many different biomes. In particular, subtropical habitats (areas -25° to 35° latitude) and those associated with the Andes are understudied. 110

Traditional views of diversification in the Neotropics, based on lowland rainforests, have focused, among other theories, on the very long-term environmental stability of the region and an accumulation of taxa that diverged prior to the Pleistocene (reviewed in Moritz et al. 2000, Patton et al. 2000, Chek et al. 2001, Burns & Naoki 2004, Lougheed et al. 2006). Recent work has suggested that patterns outside the Amazon Basin may differ significantly from this traditional view (Garcia-Moreno et al. 1999, Chesser 2000, Noonan & Gaucher 2005, Rull 2006, Weir 2006). Patterns of diversification found in unambiguous//, andinus (i.e. disregarding lineage #2, comprising individuals associated with H. riojanus) suggest a strong role for Pleistocene glacial cycling, but the divergence events for P. borellii are deeper— apparently pre-Pleistocene. These co-distributed taxa thus clearly show different temporal signatures even though they inhabit the same area. A strong possibility in explaining the temporal differences in divergence may stem from the apparently recent dispersal of the ancestor of the H. andinus species group into the Andes from the Atlantic Forests of Eastern Brazil (Faivovich et al. 2004) during cyclical connections of these forest habitats in the last 10 my (Por 1992, Costa 2003, Hubert-Renno 2006). Such a pattern of diversification in the Andes would therefore necessarily be more recent than those of the lowland forests (i.e. Amazon Basin) because the (presumably not fully genetically representative) ancestors of the H. andinus species in the Andes would have reached these newly formed habitats relatively recently. Pleurodema borellii, on the other hand, was likely present in this region of the Andes prior to the arrival ofH. andinus, hence the genealogical relationships could still reflect older divergence events. There is no genetic signal of divergence during Pleistocene climatic cycling in P. borellii, suggesting that lineages that were sundered during older uplift events remained isolated throughout the late Pliocene and Pleistocene. Both species do, however, show similar temporal patterns of population growth less than 1 mya. The detected range expansions (possibly due to the population growth) in H. andinus have led to more extensive secondary contact between divergent lineages than is evident in P. borellii. Even though there was a lack in temporal concordance, the two species showed similarities in the geographic distributions of genetic lineages across northwestern Argentina. The north to south series of lineages identified in both frog taxa in this study Ill was also found for a cold-tolerant tree, Podocarpus parlatorei, inhabiting the uppermost levels of the Yungas montane forests (Quiroga & Premoli 2007). Our results support the northern (centred ~22°S) and southern (~27°S) refugia that have been previously postulated based on high species richness and endemism (Brown et al. 2001) and current highly mesic climate, possibly indicative of a mesic climate in the past. The genetically distinct lineages associated with the Sierra de Santa Barbara (~24.5°S) in all three species (H. andinus and P. borellii, this study; P. parlatorei, Quiroga & Premoli 2007) and the current discontinuous distribution of yungas forests there, suggest a third, previously unidentified, refugium for this region. Through the preservation of many different lineages that are able to shift altitudinally to avoid unfavourable habitats (and possibly extinction), mountain habitats play an important role in maintaining biodiversity. In addition, fragmentation and reconnection among populations provide opportunities for differentiation and possibly speciation. Life history characteristics can profoundly impact population genetic structure through their effects on dispersal and population size, which in turn influence gene flow and genetic drift (e.g. Turner & Trexler 1998, Crawford et al. 2007, Haney et al. 2007). The two species in this study differ in life history characters (fecundity, habitat specialization), but the patterns of genetic differentiation expected based on these characters were not borne out. Although both species showed evidence of similar population sizes and a pattern consistent with isolation by distance, differentiation was significantly higher in P. borellii than H. andinus at the same spatial scale. We had expected H. andinus to show stronger differentiation due to its lower fecundity and its limited ability to move outside of the riparian zones of its breeding streams. Pleurodema borellii breeds opportunistically in temporary pools of water in a wide range of habitats, including unusual choices such as swimming pools (Halloy & Fiano 2000), suggesting that it is an opportunistic high-dispersal habitat generalist. Our results therefore suggest that factors other than fecundity and habitat specialization may play a critical role in driving patterns of differentiation. Indeed the connectivity of streams may facilitate passive dispersal in H. andinus at early reproductive stages (eggs and tadpoles) that is not possible for P. borellii due to the isolation likely to be typical of their breeding pools. Given that direct life history data is scarce for most non-model organisms and that 112 patterns are not easily generalized across taxa, the influence of such characteristics on differentiation may be underestimated. We were able to investigate the importance of landscape features in affecting movement of individuals ofH. andinus because its ecology is better understood than that of P. borellii. Recent explicit considerations of landscape features in a topographically heterogeneous region have demonstrated the importance of measuring biologically meaningful dispersal pathways (Keyghobadi et al. 1999, Funk et al. 2005, Spear et al. 2005). Geographic distances based on pathways that incorporate habitat suitability explained patterns of genetic differentiation in H. andinus better than conventional Euclidean (straight-line) distance. This pattern was largely driven by populations surrounded almost entirely by unsuitable habitat, which had effective distances to other populations that were much larger than Euclidean pathways. These populations were found where movement would necessarily be restricted to riparian habitat tightly associated with the streams. Such a pattern is found in many aquatic species that move exclusively through specific corridors such as waterways (e.g. Turner & Trexler 1998, Fetzner & Crandall 2003, Spear et al. 2005, Vaha et al. 2007). Differentiation in peripheral populations of H. andinus could be due to historical isolation; however, contemporary processes such as genetic drift may also play a key role through the loss of genetic diversity. Especially in heterogeneous landscapes where barriers to dispersal may be common, combining phylogeographic and landscape genetic perspectives provides more comprehensive insights into the causes of the present day distribution of genetic diversity. Further investigation of movement patterns for H. andinus and other species will shed light on the processes responsible for genetic diversification in northwestern Argentina and similar regions. The patterns described above lead to several conclusions. First, all divergences in P. borellii are older than all those in H. andinus. Indeed, the genetic structure we found within the main lineage (#1) of H. andinus is more recent (<2mya) than that found in taxa inhabiting lowland Amazon forests. Such shallow intraspecific structure has,also been found in other Andean taxa (e.g. Weir 2006, Ribas et al. 2007). Pleurodema borellii divergences, however, are more typical of the depths found for Amazonian taxa. Patterns in these two species suggest that the Andes may have more diverse temporal 113 differentiation patterns - old and recent - supporting the recent body of work cautioning the extrapolation of lowland forest patterns to other biomes of the Neotropics. Second, our understanding and delimitation of species is restricted in many cases to descriptions of external morphology, which clearly were not reliable indicators for the two species studied. Even phylogenetic studies, which typically tend to use sparse geographic sampling, may not detect the extent of genetic differentiation found within morphologically indistinguishable taxa. Molecular phylogeographic studies are urgently needed to delineate appropriate "ingroups" for population level studies. Generally, in such little-studied regions, received species taxonomies should be treated as hypotheses to be tested rather than as definitive statements about the actual reality of species-level differentiation. Finally, the differences in temporal patterns of differentiation of the two species were likely due to different biogeographic histories, not to differences in life history. Our predictions of levels of gene flow based on inferences about life history were not borne out. Given the dearth of empirical studies, we strongly caution against extrapolation from ecology and inferred life history patterns to levels of gene flow. Many of these characteristics have been loosely correlated to gene flow (e.g. fecundity), but only a few have been measured directly (e.g. mark-recapture studies of dispersal). 114

5.2 References

Barrio A (1965) Las subespecias de Hyla pulchella Dumeril y Bibron (Anura, Hylidae). Physis69, 115-128.

Brown AD, Grau HR, Malizia LR, Grau A (2001) Argentina. In: Bosques nublados del neotropico (eds. Kappelle M, Brown AD), pp. 623-659. Instituto Nacional de Biodiversidad, Costa Rica.

Burns KJ, Naoki K (2004) Molecular phylogenetics and biogeography of Neotropical tanagers in the genus Tangara. Molecular Phylogenetics and Evolution 32, 838-854.

Chek AA, Lougheed SC, Bogart JP, Boag PT (2001) Perception and history: Molecular phylogeny of a diverse group of Neotropical frogs, the 30-chromosome Hyla (Anura: Hylidae). Molecular Phylogenetics and Evolution 18, 370-385.

Cherry LM, Case SM, Wilson AC (1978) Frog perspective on the morphological difference between humans and chimpanzees. Science 200, 209-211.

Chesser RT (2000) Evolution in the High Andes: The phylogenetics of Muscisaxicola Ground-Tyrants. Molecular Phylogenetics and Evolution 15, 369-380.

Cei JM (1980) Amphibians of Argentina Monitore Zoologico Italiano N. S Monografia 2, Florence.

Costa LP (2003) The historical bridge between the Amazon and Atlantic Forest of Brazil: a study of molecular phylogeography with small mammals. Journal of Biogeography 30, 71-86.

Crawford AJ, Bermingham E, Polania C (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. Molecular Ecology 16, 4789-4807.

Duellman WE (1999) Patterns and distributions of amphibians: A global perspective. The John Hopkins University Press, Baltimore.

Duellman WE, de la Riva I, Wild ER (1997) Frogs of the Hyla armata and Hyla pulchella groups in the Andes of South America, with definitions and analyses of phylogenetic relationships of Andean groups of Hyla. Scientific Papers of the Natural History Museum of the University of Kansas 3, 1-41.

Faivovich J, Garcia PCA, Ananias F, et al. (2004) A molecular perspective on the phylogeny of the Hyla pulchella species group (Anura, Hylidae). Molecular Phylogenetics and Evolution 32, 938-950. 115

Faivovich J, Haddad CFB, Garcia PCA, et al. (2005) Systematic review of the frog family Hylidae, with special reference to Hylinae: Phylogenetic analysis and taxonomic revision. Bulletin of the American Museum of Natural History 294, 1-240.

Fetzner JW, Jr., Crandall KA (2003) Linear habitats and the nested clade analysis: an empirical evaluation of geographic versus river distances using an Ozark Crayfish (Decapoda: Cambaridae). Evolution 57, 2101-2118.

Funk WC, Blouin MS, Corn PS, et al. (2005) Population structure of Columbia spotted frogs (Rana luteiventris) is strongly affected by the landscape. Molecular Ecology 14, 483-496.

Fouquet A, Vences M, Salducci M-D, et al. (2007) Revealing cryptic diversity using molecular phylogenetics and phylogeography in frogs of the Scinax ruber and Rhinella margaritifera species groups. Molecular Phylogenetics and Evolution 43, 567-582.

Garcia-Moreno J, Arctander P, Fjeldsa J (1999) A Case of Rapid Diversification in the Neotropics: Phylogenetic Relationships among Cranioleuca Spinetails (Aves, Furnariidae). Molecular Phylogenetics and Evolution 12, 273-281.

Gascon C, Lougheed SC, Bogart JP (1998) Patterns of genetic population differentiation in four species of Amazonian frogs: A test of the Riverine Barrier Hypothesis. Biotropica 30, 104-119.

Haffer J (1969) Speciation in Amazonian forest birds. Science 165, 131-137.

Halloy M, Fiano JM (2000) Oviposition site selection in Pleurodema borellii (Anura: Leptodactylidae) may be influenced by tadpole presence. Copeia 2000, 606-609.

Haney RA, Silliman BR, Rand DM (2007) A multi-locus assessment of connectivity and historical demography in the bluehead wrasse {Thalassoma bifasciatum). Heredity 98, 294-302.

Hubert N, Renno J-F (2006) Historical biogeography of South American freshwater fishes. Journal of Biogeography 33, 1414-1436.

Keyghobadi N, Roland J, Strobeck C (1999) Influence of landscape on the population genetic structure of the alpine butterfly Parnassius smintheus (Papilionidae). Molecular Ecology 8, 1481 -1495.

Kim I, Phillips CJ, Monjeau JA , etal. (1998) Habitat islands, genetic diversity, and gene flow in a Patagonian rodent. Molecular Ecology 7, 667-678.

Liadsky D (2003) Call variation in the Andean Treefrog, Hyla andina Bachelor of Science (Honours) thesis, Queen's University. 116

Lougheed SC, Austin JD, Bogart JP, Boag PT, Chek AA (2006) Multi-character perspectives on the evolution of intraspecific differentiation in a neotropical hylid frog. BMC Evolutionary Biology 6, 23.

McLister JD, Lougheed SC, Bogart JP (1991) Electrophoretic and Vocalization Comparisons among 3 Leptodactylid Frogs {Pleurodema Spp) from Northwestern Argentina. Canadian Journal of Zoology 69, 2397-2403.

Moritz C, Patton JL, Schneider CJ, Smith TB (2000) Diversification of rainforest faunas: An integrated molecular approach. Annual Review of Ecology and Systematics 31, 533-563.

Muellner AN, Tremetsberger K, Stuessy T, Baeza CM (2005) Pleistocene refugia and recolonization routes in the southern Andes: insights from Hypochaeris palustris (Asteraceae, Lactuceae). Molecular Ecology 14, 203-212.

Noonan BP, Gaucher P (2005) Phylogeography and demography of Guianan harlequin toads (Atelopus): diversification within a refuge. Molecular Ecology 14, 3017-3031.

Nores M (1999) An alternative hypothesis for the origin of Amazonian bird diversity. Journal of Biogeography 26, 475-485.

Pastorino MJ, Gallo LA (2002) Quaternary evolutionary history of Austrocedrus chilensis, a cypress native to the Andean-Patagonian forest. Journal of Biogeography 29,1167-1178.

Patton JL, Da Silva MNF, Malcolm JR. (2000) Mammals of the Rio Jurua and the evolutionary and ecological diversification of Amazonia. Bulletin of the American Museum of Natural History 244, 1-306.

Por FD (1992) Sooretama: the Atlantic rainforest of Brazil. SPB Academic Publishing, The Hague.

Quiroga MP, Premoli AC (2007) Genetic patterns in Podocarpus parlatorei reveal the long-term persistence of cold-tolerant elements in the southern Yungas. Journal of Biogeography 34, 447-455.

Ribas CC, Moyle RG, Miyaki CY, Cracraft J (2007) The assembly of montane biotas: linking Andean tectonics and climatic oscillations to independent regimes of diversification in Pionus parrots. Proceedings of the Royal Society B: Biological Sciences 274, 2399-2408.

Rull V (2006) Quaternary speciation in the Neotropics. Molecular Ecology 15, 4257- 4259. 117

Ruzzante DE, Walde SJ, Cussac VE, et al. (2006) Phylogeography of the Percichthyidae (Pisces) in Patagonia: roles of orogeny, glaciation, and volcanism. Molecular Ecology 15,2949-2968.

Spear SF, Peterson CR, Matocq MD, Storfer A (2005) Landscape genetics of the blotched tiger salamander (Ambystoma tigrinum melanostictum). Molecular Ecology 14,2553-2564. Symula R, Schulte R, Summers K (2003) Molecular systematics and phylogeography of Amazonian poison frogs of the genus Dendrobates. Molecular Phylogenetics and Evolution 26, 452-475.

Turner TF, Trexler JC (1998) Ecological and historical associations of gene flow in Darters (Teleostei: Percidae). Evolution 52, 1781-1801.

Vaha JP, Erkinaro J, Niemela E, Primmer CR (2007) Life-history and habitat features influence the within-river genetic structure of Atlantic salmon. Molecular Ecology 16, 2638-2654.

Weir JT (2006) Divergent timing and patterns of species accumulation in lowland and highland neotropical birds. Evolution 60, 428-855.

Wynn A, Heyer WR (2001) Do geographically widespread species of tropical amphibians exist? An estimate of genetic relatedness within the neotropical frog Leptodactylus fuscus (Schneider 1799) (Anura Leptodactylidae). Tropical Zoology 14, 255-285. 118

APPENDICES

Appendix 1. Voucher specimens of Hypsiboas andinus collected during the study. Some specimens have not been assigned voucher numbers yet and are listed as pending. GenBank Accession numbers for mtDNA control region are listed for each specimen. Cytochrome b accession numbers listed in parentheses where applicable. Collector number Locality Museum3 Voucher GenBank Accession Number number Number JMP856 1 MNCN 5947 EU403414 (cyt. bEU403150) DK-04-784 2 MCN pending EU403412 DK-04-786 2 MCN pending EU403413 DK-04-343 3 MCN 1203 EU403276 DK-04-344 3 MCN 1203 EU403277 (cyt. bEU403129) DK-01-166" 4 MACN 39037 DK-04-382 5 MCN 1204 EU403294 DK-04-383 5 MCN 1204 EU403295 DK-01-167b 6 MACN 39038 DK-04-151 7 MCN 1200 EU403254 (cyt.bEU403126) DK-04-152 7 MCN 1200 EU403255 DK-04-321 8 MCN 1202 EU403274 (cyt. bEU403128) DK-04-322 8 MCN 1202 EU403275 DK-04-232 9 MCN 1201 EU403264 DK-04-233 9 MCN 1201 EU403265 DK-04-758 11 MCN pending EU403402 DK-04-759 11 MCN pending EU403403 DK-04-760 11 MCN 1209 EU403404 DK-01-125 12 MACN 39035 EU403173 DK-01-126 12 MACN 39036 EU403174 DK-04-763 14 MCN pending EU403405 DK-04-764 14 MCN pending EU403406 119

Collector number Locality Museum3 Voucher GenBank Accession Number number Number DK-04-765 14 MCN pending EU403407 DK-04-766 14 MCN pending EU403408 DK-04-767 14 MCN pending EU403409 DK-04-768 14 MCN pending EU403410 DK-04-780 14 MCN pending EU403411 DK-04-547 16 MCN 1208 EU403384 (cyt. bEU403144) DK-04-548 16 MCN 1208 EU403385 DK-04-478 17 MCN 1207 EU403341 DK-04-479 17 MCN 1207 EU403342 JPB14516 18 CMNAR 33140 EU403189 JPB14519 18 CMNAR 33140 EU403190 JPB14523 18 CMNAR 33140 EU403191 JPB14524 18 CMNAR 33140 EU403192 DK-04-438 18 MCN 1205 EU403308 DK-04-439 18 MCN 1205 EU403309 DK-01-188 19 FML 16112 EU403226 DK-04-465 20 MCN pending EU403333 (cyt. bEU403139) DK-01-097 21 MACN 39031 EU403157 DK-04-503 22 MCN 1206 EU403364 DK-04-504 22 MCN 1206 EU403365 DK-04-459 23 MCN pending EU403327 DK-04-460 23 MCN pending EU403328 DK-04-461 23 MCN pending EU403329 (cyt. bEU403137) DK-01-091 24 MACN 39028 EU403415 (cyt. bEU403155) JMP297 25 MNCN 4086 EU403416 (cyt. bEU403151) JMP298 25 MNCN 4087 EU403417 (cyt. bEU403152) JMP395 26 MNCN 6083 EU403419 (cyt. bEU403154) 120

Collector number Locality Museum3 Voucher GenBank Accession Number number Number JMP396 26 MNCN 6069 EU403418 (cyt bEU403153) a Museum codes: CMNAR: Canadian Museum of Nature, Ottawa, Canada. MACN: Museo Argentino de Ciencias Naturales, Buenos Aires, Argentina. MCN: Museo de Ciencias Naturales, Universidad de Salta, Salta, Argentina. MNCN: Museo Nacional de Ciencias Naturales, Madrid, Spain. FML: Fundacion Miguel Lillo, Tucuman, Argentina. b Samples in process, not included in this study. Appendix 2. Population pair-wise FST values for Hypsiboas andinus for populations that have pair-wise Euclidean distances of less than 100km used in the subset analyses (total of 36 pairs). Only individuals with Network #1 haplotypes were included in the analyses. (Chapter 3) 3 4 5 6 7 8 9 10 11 12a 12b 13 14 .15 16 17 18 19 21 22 3 4 0.60 5 0.51 0.89 6 0.37 0.83 7 0.92 8 0.04 9 -0.04 10 0.04 -0.07 -0.02 11 12a 0.08 0.69 12b -0.07 0.51 0.07 13 0.00 0.06 -0.03 14 -0.19 15 0.47 16 0.07 0.34 17 0.02 0.52 18 0.23 0.05 19 0.09 0.04 0.09 21 0.09 0.32 0.20 22 0.45 0.29 0.11 122

^^ C"> co 1 (-1 CM u £i i CM C8 CM d 43 o o m t— i co r^ CM d d on >, f- m O i r~ CM i•n* "3 CM -g d d d •4-* CM CO T— o co« en i en •* en 2 d d d d 3 en CO in en CO 00 i o en T- *r en T(3U T d d d d d _»-( ~ •-P en m CO CM in CM 4a> r- i o o f~ o CM t^ 4> *~ d d d 9 d d • rj +•> io r^ CO CO o> en £3 <£> i CO co m O) CO oo • 1—1 d d d d d d* d T3 ^~ »5 m in r^ r^ oo CO CO CO 3 ir> , CM o o r-o CM r^ co T~ d d d d d d d d C ^„ o r- r- CO CO en in co *• •* • CM o co 00 CO en CO • o co & co i T— CM o CM 00 CO o> CO •<*e n ** o d d d d d d d d d ca J3. CO o o •* m m CM m i^ co CM CM o o CO CO i~- co en CO •* en d d d d d d d d d d d s fo r

a CO t-- CO i^ en 00 •* CM en 00 CD CO m T— CM 1 o o o CO >«• CO co en •* m en ^R d d d d d d d d d d d d R a> ^_ CO CM en co CM ^_ t o ^ CM m t

1 •«s- - T— Y— CM in CO h- o co "^ o in T m V en • O •» o o o o CM *-CO f»- CD en CO •* en fo r 9 d d d d d d d d d d d d d d rf\ iv. o (O 00 CM •* co co co m f~ CM 00 en en CO m T~ T— lue ! 00 o o o (O o o o co co co co en CO >*e n CO ci d o d d d d O O d d O d d d d d >t- T— •* CO •u- •»r CO •* CM CO en CM in o •* CM 1 o o o C^—O o o o o CM o co 00 CO q CO •* e•n* £ r-- c> d T— o • ci d d i d d d d d d d d d d •

• CO CD CM en r^ CD CO •>* '5 CO • 9s o o od

& CO CM T_ r~ 00 o> en ^ o ,_ CO CM •* CO o CO co o •* in i 00 en I-. 00 r~-oo i^ en a> in r»-m 00 00 en m m en oa d °O> d Ci ci d d d d d d d d d d d d d d

T— J3 en i^ o CO CM r^ d d d d d d d d d d d d d d o o T_ CO en in 1^ in en 00 •* CO V m in 00 CO m CO o en in OH CO i 03 m m •* in CO •* CO m "i- io CM •fl- CM co m oo CO CO 00 , O © o ci o ci o' d d d d>* d d d d d d d d d d X T— CO O i^ CM o m CO o en en •* in CO CO m r- CM o en en CM i T- O) oo 00 o CD f~ en •t oo r~ en en CO CO CO o•o* h- p co CO 03 d O d o ci ci ci d d d d d d d d d d ^— d d d pend i CO .o T— t— a o CM CM CO •* m CO N- 00 en o CM CO < CM CO •«* m CO r^ oo en CM CM CM CM 123

Appendix 4.Voucher specimens of Pleurodema borellii collected during the study. Some specimens have not been assigned voucher numbers yet and are listed as pending. Collector number Locality Museum3 Voucher Number number JMP481 1 MNCN 5931 JMP482 1 MNCN 5945 16520 2 CMNAR 33165 17707 2 GMNAR 33154 17736 2 CMNAR 33154 17740 2 CMNAR 33154 17741 2 CMNAR 33154 17747 2 CMNAR 33154 15106 3 CMNAR 33671 DK-04-325 4 MCN 1216 DK-04-326 4 MCN 1216 DK-04-118 5 MCN 1220 DK-04-126 5 MCN 1220 DK-04-166 6 MCN 1211 DK-04-167 6 MCN 1211 DK-04-100 7 MCN 1219 DK-04-142 8 MCN 1210 DK-04-143 8 MCN 1210 DK-04-308 9 MCN 1215 DK-04-309 9 MCN 1215 DK-04-234 10 MCN 1213 DK-04-235 10 MCN 1213 DK-04-282 11 MCN 1214 DK-04-191 12 MCN 1212 DK-04-769 13 MCN pending DK-04-770 13 MCN pending DK-04-774 13 MCN pending Collector number Locality Museum3 Voucher Number number DK-04-775 13 MCN pending DK-04-776 13 MCN pending DK-04-777 13 MCN pending DK-04-778 13 MCN pending DK-04-779 13 MCN pending DK-04-476 15 MCN 1218 MLPA3909 15 MLPA 3909 MLPA3910 15 MLPA 3910 DK-04-437 16 MCN 1217 DK-04-446 18 MCN pending DK-04-488 20 MCN 1221 DK-04-489 20 MCN 1221 14509 outgroup CMNAR 33147 a Museum codes: CMNAR: Canadian Museum of Nature, Ottawa, Canada. MACN: Museo Argentino de Ciencias Naturales, Buenos Aires, Argentina. MCN: Museo de Ciencias Naturales, Universidad de Salta, Salta, Argentina. MLPA: Museo de La Plata, La Plata, Argentina. October 14, 2004

This Is the Original Approval of this protocol* *A Full Protocol Submission will be required in 2008*

Dear Or. Handford:

Your "Application to Use Animals for Research or Teaching" entitled:

"Comparative Phyfogeography of Andean Frogs - Field Study" Funding Agency - NSERC - Grant «S195A3

has been approved by the University Council on Animal Care. This approval is valid from October 14,2004 to October 31,2005. The number for this project is #2004-086-10.

1. This number must be indicated when ordering animals for this project 2. Animals for other projects may not be ordered under this number. 3. If no number appears please contact this office when grant approval is received. If the application for funding is not successful and you wish to proceed with the project request that an internal scientific peer review be performed by the Animal Use Subcommittee office. 4. Purchases of animals other than through this system must be cleared through the ACVS office. Health certificates will be required.

ANIMALS APPROVED FOR 1 YR. PAIN LEVEL • B

Frogs • Hylaandina adult M/F 150 Eleutherodactylus discoidalis adult M/F 150 Pleurodema boreJIii adult M/F - 150

STANDARD OPERATING PROCEDURES Procedures in this protocol should be carried out according to the following SOPs. Please contact the Animal Use Subcommittee office (661 -2111 ext 86770) in case of difficulties or if you require copies. SOP'S are also available at http://www.uwo.ca/animal/acvs

REQUIREMENTS/COMMENTS Please ensure that individuai(s) performing procedures on live animals, as described in this protocol, are familiar with the contents of this document

c.c. Approved Protoco»>ll • PJtandforaP^Handfora.. D. KoscinskiKoscir , J. Wasylenko-Weber Approval Latter . yOK-yfjTKoscinskic , J. Wasylenko-Weber

University Council on Animal Care • The University of Western Ontario Animal Use Subcommittee • Health Sciences Centre * London. Ontario • N6A 5C1 • Canada 126

Blackwell .NfeA- SB*, CAKX&K. T»ararici.(iyroJ£J^^ '"• Manuscript no, (if known):. : CB.J3S&. f Names of all authors in the aider m which Siey appear in the Artie %fc.l..tt.&hdfo^ C- lagged In order for your Article to be distributed a* widely as possible in Molecular Ecology (the Journal) you gram Blackwell Publishing Ud (Blackwel Publiihing) an excluslvs licence to pubJIsh tho above Article including the at>«tr»cl In printed and electronic form, in at languages, and to administer subsidiary lights agreements with third parties for the full period of copyright and aH renewals, extensions, revisions and revivals. The Article is deemed to include all material submitted for publication with the exception of Letters, and includes the text, figures, tables, author contact delate and »«supplementary material accompanying the Article. Please read this form carefully, sign at the bottom (if your employer owns copyright In your work, arrange for your employer to sign where marked), and return the ORIGINAL to the address below as quickly as possible. As author, younwnaii me copyright owner of the Aroeto. unless copyright is owned by your employer (US federal Government authors please note: your Arfde is In the public domain.)

Your Article will not be published unless art Exclusive Licence Form has been signed and received by Blackwell Publishing.

Please note: You retain the following rights to recuse the Article, as long as you do not sell or reproduce the Article or any part of it for commensal purposes (i.e. for monetary gain on your own account or on that of a third party, or for indirect financial gain by a commercial entity). These rights apply without needing to seek permission from Blackwell Publishing.

• Prior to acceptance: We ask that as part of the publishing process you acknowledge that the Article has been submitted to the Journal, you will not prejudice acceptance If you use the unpublished Article, In torn and content as submitted for publication in the Journal, in the fallowing ways: o sharing print or electronic copies of the Article wHh colleagues; o posting an electronic version of the Article on your own personal website, on your employer's website/repository and on free public servers in your subject area. • After acceptance: Provided that you give appropriate ac*rw»*adgement to the Journal, and Blackwell Publishing, and full bibliographic reference for the Article when it is published, you may use the accepted version of the Article as originally submitted for publication in the Journal, and updated to include any amendments made after peer review, in the following ways: o you may share print or electronic copies of the Article with colleagues; o you may use at or part of the Article and abstract, without revision or modification, in personal comptlationt or other publications of your own work: o you may use the Article within your employer's institution or company for educational or research purposes, including use in course packs: o 12 months after publication you may post an electronic version of the Article on your own personal .website, on your employer's website/repository and on froo public servers In your subject area. Electronic versions of the accepted Article must include a link to the published version of the Article together with the Mowing text 'The definitive version is avateble at www.aiackwgll-svnerav.com'.

Please note that you are not permitted to post the Blackwell Pubishing PDF version of the Article online.

All requests by third parties to re-use the Article in whole or in part win be handled by Blackwell Publishing. Any permission fees will be retained by the Journal. All requests to adapt substantial parts of the Article in another publication (including publication by Blackwell Publishing) will be subject to your approval (which is deemed to be given if we have not heard from you within 4 weeks of your approval being sought by us writing to you at your last notified address). Please address any queries to icwrraisriohtstaoxon,b'acKwelirjuMish»ia.coni.

In signing this Agreement 1. You hereby warrant that this Article is an original work, has not been rxfctshed before and is not being considered for publication elsewhere In its final form either in printed or electronic form; 2. You herebywarrant that you have obtained permission from the oorjyrigM hokfer to reproduce in the Article (m all media including print and electronic form) material not owned by you, and that you have acknowledged the source: 3. You hereby warrant that this Article contains no violation of any existing copyright or other third party right or any material of an obscene. Indecent, libellous or otherwise unlawful rature and that to the best of your knowledge this Article does not infringe the rights of others; 4. You hereby warrant that in the case of a multi-authored Article you have obtained, in writing, authorization to enter into this Agreement on their behalf and that all coauthors have wad and agreed the terms of this Agreement; 5. You warrant that any formula or dosage given is accurate and will not if property followed injure any person; 6. You will Indemnify and keep indemnified the Editors and Blackwell Publishing against all claims and expenses (including legal costs and expenses) arising from any breach of this warranty and the other warranties on your behalf in this Agreement. By signing this Agreement you agree that Blackwell Publishing may arrange for me Article to be: • Published in the above Journal, and sold or distributed, on ttsown , or with other related material: • Published in multi-contributor book form or other edited wmpilations by Slackwell Publishing; • Reproduced and/or distributed (including the abstract) throughout the world in printed, electronic or any other medium whether now known or Hereafter devised, in all languages, and to authorize third parties (including Reproduction Rights Organizations) to do the same; • You agree to Blackwell Publishing using any Images from the Article on the cover of the Journal, and in any marketing material. You authorize Blackwell Publishing to act on your behalf to defend the copyright in the Article if anyone should infringe it, although there is no obligation on Bladcweil Publishing to act in this way. As the Author, copyright in the Article remains in your name (or your employer's name if your employer owns copyright in your work). Blackwell Publishing undertakes that every copy of the Article published by Blackwell Publishing will include the full bibliographic reference for your Article, together with the copyright statement

M BOX A: tob e completed If copyright belongs to you

D BOX B: tob e completed IT copyright belongs to your employer (e.g. HMSO, CSIRO) The copyright holder grants Blackwell Publishing an exclusive licence to publish the Article including the abstract in printed and electronic form, in all languages, and to administer subsidiary rights agreements with third partes forth e full period of copyright and all renewals, extensions, revisions and revivals. Print Name of Copyright holder: This will be printed on the copyright line on each page of the Article. It is your responsibility to provide the correct Information of the copyright holder.

D BOX C: tob e completed If the Article Is In the public domain (e.g. US Federal Government employees) You certify that the Article is in the public domain. No licence to publish is therefore necessary. ^^ r \/ < I ' Signature (on behalf of all co-authorlborsHfanvls (If «nv»l Printt namename:: .>R~i!r.'..jSft«.....&QXSQr...f^Qs&ilJ§X>.L,..?.W!*> 1 .,...„ D^ ZcmAOlQB.. If your employer claims copyright in your work, this form must also be signed below by a person authorized to sign for and on behalf of your employer, as confirmation that your employer accepts the terms of this licence. Signature (on behalf of the employer of the author (s)) Print name: Print name of employer: Date:.... _ The rights conveyed in this licence will only apply upon acceptance of your Article for publication. Data Protection: The Publisher may store your name and contact details in electronic format in order to correspond with you about the publication of your Article in the Journal. We would 8ke to contact you from time to time with information about new Blackwell publications and services in your subject area. (For European contributors, this may involve transfer of your personal data outside the European Economic Area.) Please check the following boxes if you are happy to be contacted in this way:

O(conventionalmailing) Q(vi8e-mail)

Please return the signed form to: Professor Harry Smith FRS, Managing BMor: Molecular Ecology, Division of Plant Sciences, University of, Nottingham, Sutton Bonington Campus, Loughborough, LB12 5RD. UK. 128

DARIA KOSCINSKI

Department of Biology University of Western Ontario 1151 Richmond St. N London ON N6A 5B7

EDUCATION

Ph.D. (Biology with Environment and Sustainability) Department of Biology, University of Western Ontario, London ON Expected completion April 2008

M.Sc. (Biology) Department of Biology, Queen's University, Kingston ON May 2003

B.Sc. (Honours in Biology) Queen's University, Kingston ON May 2000

ACADEMIC AWARDS

2007 Western Graduate Research Scholarship 2007 Ontario Graduate Scholarship in Science and Technology 2007 Western Graduate Thesis Research Award ($ 1300) 2006 UWO Environmental Science In-stream Scholarship 2006 Western Graduate Research Scholarship 2006 Ontario Graduate Scholarship in Science and Technology 2006 Western Graduate Thesis Research Award ($750) 2005 UWO Environmental Science Entrance Scholarship 2005 Western Graduate Research Scholarship 2005 American Society of Ichthyologists and Herpetologists Gaige Award ($500 US) 2004 Sigma Xi Grants-In-Aid of Research ($900 US) 2004 Western Graduate Research Scholarship 2004 • NSERC Research Award (PGS-B) 2001 NSERC Research Award (PGS-A) 2001 Ontario Graduate Scholarship (declined) 2000 Queen's Graduate Award (M. Sc.) 2000 NSERC Undergraduate Student Research Award 1997 Queen's Annie Bentley Lillie Prize in First Year Calculus 1996 Queen's Trillium Scholarship for Undergraduate studies 1996 Canadian Merit Scholarship Foundation- Honour Citation 1996 University of Toronto National Book Award RELEVANT RESEARCH EXPERIENCE

Doctoral Research: Department of Biology, University of Western Ontario January 2004 - present Comparative phylogeography of Andean frogs. Advisors: Drs. Paul Handford and Stephen Lougheed

Research Associate: Molecular Diagnostic Laboratory, London Health Sciences Centre, London ON December 2002 - December 2003 Developing new diagnostic tests for genetic disorders. Advisor: Dr. Peter Ainsworth

Master's Research: Department of Biology, Queen's University September 2000 - May 2003 Assessing the Importance of Geographic Isolation and Ecological Selection in Diversification of an Andean frog. Advisor: Dr. Stephen Lougheed

Undergraduate Honours Research: Department of Biology, Queen's University September 1999 - April 2000 Historical and ecological correlates of diversification in an Amazonian frog. Advisors: Drs. Stephen Lougheed and Peter Boag NSERC Undergraduate Student Research: Department of Biology, Queen's University May 1999-August 1999 Molecular ecology research assistant investigating diversification in tropical frogs. Advisors: Drs. Stephen Lougheed and Peter Boag Research Assistant: Department of Biology, Queen's University September 1997 -April 1999 Research assistant for various projects in a molecular ecology laboratory. Advisors: Drs. Stephen Lougheed and Peter Boag

Aquatic Habitat Technician: Ministry of Natural Resources, Aylmer ON Summer 1997 Assisted in surveying local streams for quality of habitat and species diversity. Organized and prepared extensive database. Advisor: Delbert Miller

PUBLICATIONS AND PRESENTATIONS

Articles published in refereed journals: Ainsworth PJ, Koscinski D, Fraser BP, Stuart JA. 2004. Family cancer histories predictive of a high risk of hereditary non-polyposis colorectal cancer associate significantly with a genomic rearrangement in hMSH2 or hMLHl. Clinical Genetics 66: 183-188. 130

Koscinski D, Handford P, Tubaro PL, Sharp S, Lougheed SC. 2008. Pleistocene climatic cycling and diversification of the Andean treefrog, Hypsiboas andinus. Molecular Ecology 17: 2012-2025.

Articles submitted to refereed journals:

Koscinski D, Yates AG, Handford P, Lougheed SC. Effects of landscape and history on diversification of a montane, stream-breeding amphibian. Journal of Biogeography.

Other contributions:

Koscinski D, Handford P, Lougheed SC. September 2007 (oral presentation). Comparative phylogeography of Andean frogs. Canadian Amphibian and Reptile Conservation Network. Queen's University.

Koscinski D, Handford P, Lougheed SC. May 2007 (oral presentation). Comparative phylogeography of Andean frogs. Canadian Society for Ecology and Evolution Annual Meeting. University of Toronto.

Koscinski D, Yates AG, Handford P, Lougheed SC. April 2007 (oral presentation). The perfect pair: GIS and population genetics. Environmental Research Western Earth Day Colloquium. University of Western Ontario.

Koscinski D, Sharp S, Tubaro PL, Handford P, Lougheed SC. June 2006 (oral presentation). Diversification of the Andean treefrog, Hyla andina. Society for the Study of Evolution Annual Meeting. Stony Brook University.

Koscinski D. April 2003 (oral presentation). Assessing the Importance of Geographic Isolation and Ecological Selection in Diversification of an Andean frog. Ecology, Evolution and Behaviour Research Seminars. Biology Department, Queen's University.

Koscinski D, Tubaro PL, Lougheed SC. June 2002 (poster). Determinates of differentiation in an Andean frog: Isolation vs. Selection. Society for the Study of Evolution Annual Meeting. University of Illinois.

Elmer KR, Koscinski D, Gascon C, Davila JA, Bogart JP, Boag PT, Lougheed SC. June 2002 (poster). Historical and ecological factors in diversification of Amazonian frogs. Society for the Study of Evolution Annual Meeting. University of Illinois.

Koscinski D, Tubaro PL, Lougheed SC. April 2002 (poster). Determinates of differentiation in an Andean frog: Isolation vs. Selection - Preliminary data. Ontario Ecology and Ethology Colloquium. Queen's University.

RELEVANT TEACHING EXPERIENCE

Sessional Lecturer for Biology 441F: Special Topics in Evolution Department of Biology, University of Western Ontario, London ON September - December 2006 and 2007 Designing and delivering course for 40 upper year students. 131

Teaching Assistant Department of Biology, University of Western Ontario, London ON January 2004 - April 2007

Department of Biology, Queen's University, Kingston ON September 2000 - April 2001

Responsibilities included organizing lab experiments and tutorials, grading papers and seminars.

COMMUNITY AND ACADEMIC SERVICE

Chair of Naturalization Project, EnviroWestern January 2005 - August 2007 • Initiated, organized and implemented long-term ecological restoration program on campus with a large network of community and Western partners. • Participated in organization and planning for native wildflower and community gardens on campus. • Solicited and attained financial and material support from local and national sponsors. Graduate Education Committee student representative, Society of Biology Graduate Students, University of Western Ontario September 2004 - August 2006 • Represented graduate students at Graduate Education Committee meetings, provided input into program development, evaluated student applications for program entrance and various scholarships. • Organized graduate student retreat to discuss graduate education issues in Biology.

Member of EnviroWestern July 2004 - January 2005 • Assisted in co-ordinating a multitude of projects centering around a campus-wide waste reduction awareness campaign • Designed posters and leaflets, participated in skits and networked with relevant members of the Western community to further the goals of club. Chair of Schoolyard Naturalization Project, Society for Conservation Biology and Poison Park Public School September 1999 - October 2001 • Designed, organized and implemented long term tree planting program at a local elementary school to provide a stimulating environment for school children. • Solicited and attained financial support from local and national sponsors. • Project recognized with the prestigious Ecology Award for initiating local programs with global impact.

Co-chair of Queen's Biology Departmental Student Council September 1999 - April 2000 • Represented undergraduate students at departmental staff meetings, provided input into course development, participated in faculty evaluations, chaired council meetings, and organized social activities. PROFESSIONAL ASSOCIATIONS AND CERTIFICATION Membership: Canadian Society for Ecology and Evolution Society for the Study of Evolution American Society of Ichthyologists and Herpetologists Society for the Study of Amphibians and Reptiles Thames Talbot Land Trust

Certification:

Certified Seed Collector in Ontario, Ministry of Natural Resources