Received: 3 February 2017 | Revised: 3 July 2017 | Accepted: 12 July 2017 DOI: 10.1111/mec.14260
ORIGINAL ARTICLE
Idiosyncratic responses to climate-driven forest fragmentation and marine incursions in reed frogs from Central Africa and the Gulf of Guinea Islands
Rayna C. Bell1,2,3 | Juan L. Parra4 | Gabriel Badjedjea5 | Michael F. Barej6 | David C. Blackburn7,8 | Marius Burger9,10 | Alan Channing11 | Jonas Maximilian Dehling12 | Eli Greenbaum13 | Vaclav Gvo zd ık14,15 | Jos Kielgast16,17 | Chifundera Kusamba18 | Stefan Lotters€ 19 | Patrick J. McLaughlin20 | Zoltan T. Nagy6,21 | Mark-Oliver Rodel€ 6 | Daniel M. Portik2,22 | Bryan L. Stuart23 | Jeremy VanDerWal24,25 | Ange Ghislain Zassi-Boulou26 | Kelly R. Zamudio3
1Department of Vertebrate Zoology, National Museum of Natural History, Smithsonian Institution, Washington, DC, USA 2Museum of Vertebrate Zoology, University of California, Berkeley, CA, USA 3Department of Ecology and Evolutionary Biology, Cornell University, Ithaca, NY, USA 4Grupo de Ecolog ıa y Evolucion de Vertebrados, Instituto de Biolog ıa, Universidad de Antioquia, Medell ın, Colombia 5Departement d’Ecologie et Biodiversite des ressources Aquatiques, Centre de Surveillance de la Biodiversite, Kisangani, Democratic Republic of the Congo 6Museum fur€ Naturkunde - Leibniz Institute for Evolution and Biodiversity Science, Berlin, Germany 7Florida Museum of Natural History, University of Florida, Gainesville, FL, USA 8Department of Herpetology, California Academy of Sciences, San Francisco, CA, USA 9African Amphibian Conservation Research Group, Unit for Environmental Sciences and Management, North-West University, Potchefstroom, South Africa 10Flora Fauna & Man, Ecological Services Ltd., Tortola, British Virgin Islands 11Biodiversity and Conservation Biology Department, University of the Western Cape, Bellville, South Africa 12Abteilung Biologie, Institut fur€ Integrierte Naturwissenschaften, Universit€at Koblenz-Landau, Koblenz, Germany 13Department of Biological Sciences, University of Texas at El Paso, El Paso, TX, USA 14Institute of Vertebrate Biology, Czech Academy of Sciences, Brno, Czech Republic 15Department of Zoology, National Museum, Prague, Czech Republic 16Section of Freshwater Biology, Department of Biology, University of Copenhagen, Copenhagen, Denmark 17Center for Macroecology, Evolution and Climate, Natural History Museum of Denmark, Copenhagen, Denmark 18Laboratoire d’Herpetologie, Departement de Biologie, Centre de Recherche en Sciences Naturelles, Lwiro, Democratic Republic of the Congo 19Biogeography Department, Trier University, Trier, Germany 20Department of Biology, Drexel University, Philadelphia, PA, USA 21Royal Belgian Institute of Natural Sciences, Brussels, Belgium 22Department of Biology, University of Texas, Arlington, TX, USA 23North Carolina Museum of Natural Sciences, Raleigh, NC, USA 24Centre for Tropical Biodiveristy & Climate Change, College of Science and Engineering, James Cook University, Townsville, Qld, Australia 25Division of Research and Innovation, eResearch Centre, James Cook University, Townsville, Qld, Australia 26Institut National de Recherche en Sciences Exactes et Naturelles, Brazzaville, Republique du Congo
| Molecular Ecology. 2017;26:5223–5244. wileyonlinelibrary.com/journal/mec © 2017 John Wiley & Sons Ltd 5223 5224 | BELL ET AL.
Correspondence Rayna C. Bell, Department of Vertebrate Abstract Zoology, National Museum of Natural Organismal traits interact with environmental variation to mediate how species History, Smithsonian Institution, Washington, DC, USA. respond to shared landscapes. Thus, differences in traits related to dispersal ability Email: [email protected] or physiological tolerance may result in phylogeographic discordance among co-
Funding information distributed taxa, even when they are responding to common barriers. We quantified Explorer’s Club; American Philosophical climatic suitability and stability, and phylogeographic divergence within three reed Society; Sigma Xi; Society of Systematic Biologists; Mario Einaudi Center for frog species complexes across the Guineo-Congolian forests and Gulf of Guinea International Studies; Cornell Graduate archipelago of Central Africa to investigate how they responded to a shared climatic School; Andrew W. Mellon Foundation; EEB Paul P. Feeny Fund; EEB Paul Graduate and geological history. Our species-specific estimates of climatic suitability through Fellowship; University of California time are consistent with temporal and spatial heterogeneity in diversification among President’s Postdoctoral Fellowship; Museum of Comparative Zoology the species complexes, indicating that differences in ecological breadth may partly Herpetology Division at Harvard University; explain these idiosyncratic patterns. Likewise, we demonstrated that fluctuating sea BIOTA projects of the Federal Ministry of Education and Research; BIOTA Germany levels periodically exposed a land bridge connecting Bioko Island with the mainland Grant numbers 01 LC 0017 and 01 LC Guineo-Congolian forest and that habitats across the exposed land bridge likely 0025; US National Science Foundation, Grant/Award Number: DEB-1309171, DEB- enabled dispersal in some species, but not in others. We did not find evidence that 1202609, DEB-1145459; National rivers are biogeographic barriers across any of the species complexes. Despite Geographic Society, Grant/Award Number: 8556-08, 8868-10; California Academy of marked differences in the geographic extent of stable climates and temporal esti- Sciences; Percy Sladen Memorial Fund; mates of divergence among the species complexes, we recovered a shared pattern IUCN/SSC Amphibian Specialist Group; Department of Biology at Villanova of intermittent climatic suitability with recent population connectivity and demo- University; University of Texas at El Paso; graphic expansion across the Congo Basin. This pattern supports the hypothesis Czech Science Foundation, Grant/Award Number: 15-13415Y; Ministry of Culture of that genetic exchange across the Congo Basin during humid periods, followed by the Czech Republic, Grant/Award Number: vicariance during arid periods, has shaped regional diversity. Finally, we identified DKRVO 2017/15; National Museum, Grant/ Award Number: 00023272 many distinct lineages among our focal taxa, some of which may reflect incipient or unrecognized species.
KEYWORDS climatic refugia, ecological niche modelling, Hyperolius, land-bridge island, lineage divergence, riverine barriers
1 | INTRODUCTION species recognition and mate choice may indirectly affect the distri- bution of genetic diversity via sexual selection, assortative mating Trait variation among and within species impacts how populations and inbreeding avoidance (reviewed in Zamudio, Bell, & Mason, respond to environmental variation, which can result in both tempo- 2016). Consequently, interactions between species-specific traits and ral and spatial discrepancy in the distribution of phylogeographic environmental variation may result in phylogeographic discordance structure among co-distributed species (Massatti & Knowles, 2014; among co-distributed taxa, even when they are responding to com- Papadopoulou & Knowles, 2015; Paz, Iba nez,~ Lips, & Crawford, mon landscape barriers (Papadopoulou & Knowles, 2016). Here we 2015). For example, body size and physiological breadth are directly investigate how co-distributed, colour polymorphic reed frog species related to dispersal ability and the capacity to colonize new environ- (Hyperoliidae: Hyperolius) that vary in ecological breadth have ments; thus, these traits likely influence the frequency of migration responded to a shared climatic history across Central Africa. and gene flow among subdivided populations (Papadopoulou & The Lower Guineo-Congolian forests extend across the Congo Knowles, 2015; Rodr ıguez et al., 2015). Likewise, traits related to Basin from the Albertine Rift in East Africa westward to the Atlantic
FIGURE 1 (a) Major biogeographic features of the Lower Guinean and Congolian forests of Central Africa. Country abbreviations: AO (Angola), BI (Burundi), CA (Central African Republic), CD (Democratic Republic of the Congo), CM (Cameroon), EG (Equatorial Guinea), GA (Gabon), KE (Kenya), RC (Republic of Congo), RW (Rwanda), UG (Uganda). (b) Full-parameter modelled distribution of current suitable climate and dynamic stability estimated over the last 120 ky for the Hyperolius tuberculatus complex, H. ocellatus complex, H. c. olivaceus and H. c. cinnamomeoventris BELL ET AL. | 5225
(a)
Sanaga River CA
Bioko Island CM Congo River Príncipe Island EG UG KE São Tomé Island RC GA CD Ogooué RW River BI
Lower Guinean Forests Congolian Forests AO 0 km 400 km (b) H. tuberculatus H. ocellatus
Suitability 0.89 Longitude
−10 00.74 10 0.41
−20 −100 1020304050 Latitude
Stability 0.96 0.48 0 H. c. olivaceus H. c. cinnamomeoventris 5226 | BELL ET AL.
Ocean (Figure 1; Linder et al., 2012) and host much of the world’s hosts high species diversity and most of this diversity is also found biodiversity (Jenkins, Pimm, & Joppa, 2013; Lewin et al., 2016). A in mainland Guineo-Congolian forests (Jones, 1994). By contrast, number of studies support periods of climate-driven diversification the three oceanic islands are older (5–30 Myr), host lower overall (Bohoussou et al., 2015; Born et al., 2011; Dauby, Duminil, et al., species diversity, and most of this diversity is endemic (Jones, 2014; Duminil et al., 2015; Hassanin et al., 2015; Jacquet et al., 1994). This pattern indicates that the exposed land bridge likely 2015; Johnston & Anthony, 2012; Leache & Fujita, 2010; Nicolas enabled a sizeable proportion of Guineo-Congolian biodiversity to et al., 2011; Plana, 2004; Querouil, Verheyen, Dillen, & Colyn, 2003; colonize Bioko in a relatively short period of time whereas the Tosi, 2008) and cycles of vicariance and genetic exchange across the oceanic islands slowly accumulated endemic diversity via sweep- northern Congo Basin (Bowie, Fjelds a, Hackett, & Crowe, 2004; stakes overseas dispersal followed by in situ diversification. The Couvreur, Chatrou, Sosef, & Richardson, 2008; Fjelds a & Lovett, diversity of Hyperolius reed frogs in the Gulf of Guinea archipe- 1997; Kadu et al., 2011; Tosi, 2008) as key mechanisms shaping the lago mirrors this overall biogeographic pattern: the four species distribution of diversity in this region. Patterns of fine-scale phylo- that occur on the land-bridge island are shared with the adjacent geographic structure in rainforest taxa (Born et al., 2011; Hardy continental forests, whereas the three species that occur on the et al., 2013; Koffi, Hardy, Doumenge, Cruaud, & Heuertz, 2011; oceanic islands are single-island endemics. The most recent peri- Maley, 1996; Nicolas et al., 2011; Tosi, 2008) are consistent with ods of high connectivity between Bioko and continental Guineo- persistence in a central refuge in the west-central Congo Basin and Congolian forests coincided with the last glacial maximum (LGM; multiple smaller refugia throughout montane Cameroon and Gabon ~21 kya) and glacial period preceding the Last Interglacial (LIG; in western Central Africa (Maley, 1996); however, comparisons of ~120 kya). Yet, a range of molecular divergence (Barej et al., molecular divergence among sister species reveal a continuum of 2014; Butynski & Koster, 1994; Leache & Fujita, 2010; Melo, divergence times ranging from Late Miocene (Duminil et al., 2013; Warren, & Jones, 2011; Perez, Fa, Castroviejo, & Purroy, 1994) Holstein & Renner, 2011; Njabo, Bowie, & Sorenson, 2008; Zimkus indicates that for some taxa, populations on Bioko remained & Gvo zd ık, 2013; Zimkus et al., 2017) to Pleistocene (Bohoussou genetically isolated throughout this recent cycle of geographic et al., 2015; Duminil et al., 2015; Jacquet et al., 2015; Johnston & connectivity, whereas in others, gene flow during periods of geo- Anthony, 2012; Nicolas et al., 2011; Tosi, 2008). Likewise, rivers are graphic connectivity may have obscured population divergence barriers to dispersal in many mammal and understory bird species (Futuyma, 2010). across the Guineo-Congolian forest (Anthony et al., 2007; Bohous- Here, we focus on three largely sympatric species complexes of sou et al., 2015; Harcourt & Wood, 2012; Hassanin et al., 2015; Hyperolius reed frogs that inhabit rainforest, bushland (disturbed Huntley & Voelker, 2016; Jacquet et al., 2015; Mitchell, Locatelli, forest), and humid savannah habitats across the Lower Guineo-Con- Sesink Clee, Thomassen, & Gonder, 2015; Nicolas et al., 2011; golian forests. We quantify environmental niche, climatic stability Querouil et al., 2003; Telfer et al., 2003; Voelker et al., 2013), but and phylogeographic divergence within each species complex to appear to be less important for plants, reptiles and amphibians test whether differences in ecological breadth—environmental and (Freedman, Thomassen, Buermann, & Smith, 2010; Kindler et al., physiological tolerance (Bonier, Martin, & Wingfield, 2007; Lopez- 2016; Lowe, Harris, Dormontt, & Dawson, 2010; Zimkus et al., Uribe, Zamudio, Cardoso, & Danforth, 2014)—correspond with tem- 2017). These idiosyncratic patterns suggest that differences in poral and spatial heterogeneity in species’ responses to historical organismal traits (Heuertz, Duminil, Dauby, Savolainen, & Hardy, climatic fluctuations. Specifically, for the forest-dependent species 2014; Kindler et al., 2016; Ley et al., 2014), mediated by ecotones complexes (H. ocellatus and H. tuberculatus), we expect estimates of on small spatial scales (Freedman et al., 2010; Jacquet et al., 2015; environmental niche will reflect narrow ecological breadth and thus Smith, Wayne, Girman, & Bruford, 1997; Smith et al., 2011) and a history of fragmented climatic stability with centres of endemism environmental gradients on large spatial scales (Dauby, Hardy, Leal, concentrated in areas of high climatic stability. In the Hyperolius Breteler, & Stevart, 2014; Duminil et al., 2013; Kirschel et al., 2011; cinnamomeoventris complex, known from both forest and humid Smith et al., 2005), may explain why co-distributed species savannah habitats, we expect estimates of environmental niche will responded differently to a common landscape. reflect wide ecological breadth and consequently more contiguous The Lower Guinean forests extend from continental Africa habitat stability and population connectivity through time. Two of along the Gulf of Guinea archipelago. This chain of islands is com- our focal species complexes (H. ocellatus and H. tuberculatus) also prised of a land-bridge island (Bioko) currently separated from occur on Bioko Island, whereas the third (H. cinnamomeoventris)is mainland Africa by approximately 30 km of shallow sea, and three absent. We further predict that differences in the distribution of oceanic islands (Pr ıncipe, S~ao Tome and Annobon) that have never suitable habitat across the land bridge may also explain why some been connected to continental Africa. The volcanic peaks that com- reed frog species have colonized the island and others have not. prise Bioko Island formed 1–3 Mya (Deruelle et al., 1991; Marzoli Finally, we assess how genetic diversity, proposed biogeographic et al., 2000) and since then, cycles of rising and retreating sea barriers (climatic refugia and rivers) and regional phenotypic varia- levels resulted in several periods of isolation and connectivity tion (sexual dichromatism and colour polymorphism) correspond to between Bioko and the adjacent continent (Meyers, Rosendahl, our current understanding of taxonomic diversity within each spe- Harrison, & Ding, 1998). Relative to its size and age, Bioko Island cies complex. BELL ET AL. | 5227
Cornell University Museum of Vertebrates, the California Academy 2 | MATERIAL AND METHODS of Sciences, the North Carolina Museum of Natural Sciences, the U.S. National Museum of Natural History, the University of Texas at 2.1 | Focal species and sampling details El Paso Biodiversity Collections, the Museum of Comparative Zool- We identified each of the focal reed frog species complexes for this ogy at Harvard University, the Peabody Museum at Yale University, study by selecting the smallest unit of monophyletic lineages whose the Museum of Vertebrate Zoology at the University of California combined ranges encompass the Lower Guineo-Congolian forests. Berkeley, Museum fur€ Naturkunde in Berlin, the National Museum in Species of the Hyperolius cinnamomeoventris complex are the few Prague, the Institut National de Recherche sur les Sciences Exactes reed frogs that inhabit both bushland (disturbed forest) and humid et Naturelles and the Royal Belgian Institute of Natural Sciences in savannah habitats (Schiøtz, 1999). The vast majority of the H. cin- Brussels (Table S1). Samples without catalogue numbers are main- namomeoventris species complex range is attributed to the sub- tained in research collections of coauthors AC, MB, JK and SL. species H. c. cinnamomeoventris, and the rest of the complex is comprised of H. veithi from the Congo Basin (Schick et al., 2010), 2.2 | Ecological niche modelling the subspecies H. c. olivaceus in Gabon (Laurent, 1943; Schick et al., 2010) and three species (H. thomensis, H. molleri and H. drewesi) We compiled a database of vouchered occurrences from our own endemic to the oceanic islands of S~ao Tome and Pr ıncipe (Figure 2a; fieldwork and mappable museum specimen records in VERTNET (www. Bell, Drewes, & Zamudio, 2015; Bell, 2016). Coloration across most vertnet.org) that could confidently be attributed to each species of the species complex range is sexually dichromatic (males and complex based on locality, genetic and/or morphological examina- females differ in coloration); however, H. veithi and the island ende- tion. The final database included 65, 57 and 46 localities for the mics are sexually monochromatic (Figure 2d). For this study, we col- H. cinnamomeoventris, H. tuberculatus and H. ocellatus species com- lected 129 samples from 61 localities (Table S1) including type plexes, respectively (Table S1). Due to deep genetic divergence localities of H. cinnamomeoventris (Duque de Bragancßa, Angola), within the H. cinnamomeoventris complex dating to the mid-Miocene H. veithi (Salonga National Park, Democratic Republic of the Congo), (Bell, Drewes, Channing, et al., 2015; Schick et al., 2010) and poten- H. molleri and H. thomensis (S~ao Tome Island), H. drewesi (Pr ıncipe tial for niche divergence over this time period, we ran separate mod- Island) and H. c. olivaceus (Lambaren e, Gabon; Figure 2a). els for H. c. olivaceus (20 unique localities) and H. c. Species of the H. tuberculatus complex occur in bushland and cinnamomeoventris (45 unique localities). We excluded H. veithi and rainforest edge habitats where they can be found far from reproduc- the island endemics from these analyses because the limited geo- tive sites; consequently, frogs in this species complex are hypothe- graphic distribution of these four species precludes sufficient sam- sized to have high dispersal abilities and tolerance for habitat pling to generate individual models. The model for the disturbance (Amiet, 2012). Coloration, pattern and sexual dichroma- H. tuberculatus complex included all three recognized species (H. tu- tism vary widely across the range of the species complex (Figure 3d), berculatus, H. dintelmanni and H. hutsebauti) and that for the H. ocel- which includes H. tuberculatus in west Central Africa, H. hutsebauti latus complex included the two recognized subspecies (H. o. ocellatus north of the Congo River in eastern Central Africa (Laurent, 1956, and H. o. purpurescens). 1976; Schiøtz, 1999), and H. dintelmanni in montane south-western We estimated the niche and potential geographic distribution of Cameroon (Amiet, 2012; Lotters€ & Schmitz, 2004; Portik et al., H. c. cinnamomeoventris, H. c. olivaceus and the H. tuberculatus and 2016). We collected 104 samples from 47 localities (Table S1) H. ocellatus complexes under current climatic conditions using pres- including the type localities of H. tuberculatus (Lambaren e, Gabon) ence records and a maximum entropy algorithm (MAXENT v. 3.3.3k; and H. dintelmanni (Bakossi Mountains, Cameroon), sites near the Phillips, Anderson, & Schapire, 2006). MAXENT, a robust method for type locality of H. hutsebauti (Ibembo, Democratic Republic of the inferring ENMs when only presence data are available, estimates the Congo), and Bioko Island (Figure 3a). relationship between occurrences and environmental variables based Finally, species of the H. ocellatus complex occur in evergreen, on the environments sampled by occurrences relative to all available semi-deciduous and gallery forests and are sexually dichromatic environments in the study area. Following Soberon (2007), we speci- throughout the range with regional variation in female coloration fied a study area for the ENMs that represents the potential area (Figure 4d); morphotypes 1 and 2 occur north of the Sanaga River in the focal species can access given their dispersal or colonization abil- Cameroon (H. o. ocellatus), and morphotype 3 (H. o. purpurescens) ity. The genus Hyperolius is distributed throughout sub-Saharan occurs south of the Sanaga River and across the rest of the species Africa; thus, our study area included continental Africa and the Gulf complex range (Amiet, 2012; Laurent, 1943; Perret, 1975). We col- of Guinea islands, but we restricted the upper limit of the study area lected 137 samples from 39 localities (Table S1) including the three by the distribution of the Sahara Desert and the North Saharan regional female colour morphs, the type locality of H. ocellatus steppe and woodlands (Olson & Dinerstein, 2001). Given that our (Bioko Island, Equatorial Guinea) and the type locality of H. o. pur- focal taxa were not sampled randomly across the study area (sub- purescens (Kisangani, Democratic Republic of the Congo; Figure 4a). Saharan Africa), we applied the same sampling bias to the selection Tissue samples (toe clips, liver or muscle) were preserved in 95% of background points for the model (Elith et al., 2011) using a sam- ethanol or RNAlater and voucher specimens are deposited in the pling effort surface based on all mappable anuran specimen records 5228 | BELL ET AL.
(a) 0 km 400 km (b) (c)
H. c. olivaceus
Island endemics H. veithi
H. c. cinnamo- meoventris
Miocene1.0 Pliocene Pleistocene 0.10
(d)
H. c. olivaceus (female) H. molleri H. c. cinnamomeoventris H. veithi (male, female) (female) (male)
H. c. cinnamomeoventris H. c. olivaceus H. thomensis (male) H. drewesi (male) (male) (male) BELL ET AL. | 5229
FIGURE 2 (a) Distribution of Hyperolius cinnamomeoventris species complex sampling localities in Central Africa including the islands of S~ao Tome and Pr ıncipe. The approximate range is shown in green and the type localities for H. c. cinnamomeoventris, H. veithi, H. thomensis, H. molleri, H. drewesi and H. c. olivaceus are indicated with black arrows. Symbols without black borders reflect localities with only mtDNA sequence data. (b) Mitochondrial (16S and cytochrome b) phylogeography; posterior probabilities >.95 are denoted by black dots, probabilities >.85 are denoted by white dots, and 95% highest posterior density intervals for divergence time estimates on well-supported nodes are indicated. The axis indicates geological epochs (Miocene, Pliocene and Pleistocene) and time before present in increments of 1 million years. (c) Multilocus nuDNA networks generated using POFAD and SPLITSTREE. In all cases, samples are coded with shapes corresponding to mitochondrial lineages. The scale represents genetic distance equally weighted across loci. (d) Representative photographs of female and male coloration. Photograph credits: A. Stanbridge, D. Lin, B. Stuart, J. Kielgast
from sub-Saharan Africa in VERTNET. We used the number of unique were reasonable for our study system (Fig. S1). The potential distri- years that any anuran was collected at a site (resolution of 0.0083 butions of the taxa in historic time periods (see next section) are degrees or ~1 km at the Equator) as an indicator of sampling effort. based on the final versions of the contemporary distribution models. We aggregated this raster at a factor of 10 to produce a 10 9 10 km block, summed sampling effort within the aggregated 2.3 | Niche stability and connectivity through time block and disaggregated the results back to 1 km. For example, if two cells within the 10 9 10 km block each have 1 year of sampling To extend niche models to historic climates, we used climate esti- effort, following disaggregation all cells in the 10 9 10 km block mates for the last 120 kyr (Fuchs et al., 2013; Weber, VanDerWal, receive a sampling effort value of two. Sites with no anuran records Schmidt, McDonald, & Shoo, 2014) and assumed that the estimated were assigned a value of 0.01. The final sampling effort map is based contemporary niche remained constant within this time period on ~ 41,000 anuran records (Fig. S1), and we selected 10,000 back- (Wiens et al., 2010). The climate surfaces were based on snapshots ground points from the study area with probability proportional to of 1–4 kyr intervals using the Hadley Centre Climate model effort (sites with greater effort were more likely selected to describe (HadCM3; Singarayer & Valdes, 2010). Monthly temperature and the background area). precipitation anomalies were downscaled to 0.2 degrees using a For contemporary distribution models, we extracted values for bilinear spline and then to 0.0466667 degrees globally using a bicu- occurrence and background points for eight environmental variables bic spline. Anomalies were then applied to current monthly climates representative of means, extremes and seasonality of precipitation given 125 m lower sea levels (Hijmans et al., 2005). Past sea levels and temperature (WORLDCLIM V. 1.4-release 3; Hijmans, Cameron, were estimated as the consensus of several sources (Fleming et al., Parra, Jones, & Jarvis, 2005): mean annual temperature (bio1), tem- 1998; Fleming, 2000; Lea, Martin, Pak, & Spero, 2002; Milne, Long, perature seasonality (bio4), mean temperature of warmest (bio10) & Bassett, 2005; R. A. Rohde unpublished data; Global Warming Art and coldest quarter (bio11), total annual precipitation (bio12), precip- project http://en.wikipedia.org/wiki/File:Post-Glacial_Sea_Level.png, itation seasonality (bio15), precipitation of wettest (bio16) and driest and http://www.ncdc.noaa.gov/paleo/ctl/clisci100k.html#sea), and quarter (bio17). We selected this subset of variables to reflect envi- climate surfaces for each time period were clipped to only include ronmental variation that is biologically informative for amphibians, exposed land. We derived surfaces of mean annual temperature, while minimizing the number of correlated variables in the analysis temperature seasonality, mean temperature of the warmest and (Fig. S1). To evaluate model performance, we set aside 25% of the coldest quarters, mean annual precipitation, precipitation seasonality occurrences at random and measured the area under the curve and precipitation of the wettest and driest quarters for each time (AUC) of the receiver operating characteristic (ROC). The ROC plot slice. We deposited the geospatial bioclimatic layers for the current, reflects the fraction of correct presences (true positives) as a func- middle Holocene, LGM and LIG in Dryad (https://doi.org/10.5061/ tion of the fraction of incorrect absences (false negatives); however, dryad.vs76v). The full set of layers is available upon request from in data sets such as ours where absence data are unavailable, the coauthor JV. All analyses were conducted with the CLIMATES package fraction of correct presences is plotted against the fraction of the (VanDerWal, Beaumont, Zimmermann, & Lorch, 2011) in R v 2.9.0. study area predicted as present under each possible threshold. We estimated stability of suitable climate for each species over Although AUC estimates of performance can be misleading when the last 120 kyr using two approaches: static stability (Hugall, Mor- comparing different modelling algorithms (Peterson, Papesß,& itz, Moussalli, & Stanisic, 2002), which does not consider dispersal Soberon, 2008) or study areas (Lobo, Jimenez-Valverde, & Real, between pixels, and dynamic stability (Graham, VanDerWal, Phillips, 2008), it is a suitable performance metric for our data set (presence- Moritz, & Williams, 2010), which employs a graph theoretic approach only, same modelling algorithm and study area). If model evaluation with estimates of dispersal rate to model how organisms track suit- was positive (AUC > 0.8), we generated a final model using all avail- able climate conditions across time. With the dynamic stability able data. We evaluated the effects of changing the value of the approach a pixel is stable as long as pixels within the defined disper- regularization parameter and the allowed feature type classes on sal distance have suitable climate in adjacent time steps. For the sta- model performance using the ENMEVAL package (Muscarella et al., tic stability estimate, we summed the negative log of suitability
2014) and found that the default parameters suggested by MAXENT through time for each pixel, and for the dynamic stability estimate 5230 | BELL ET AL.
.1 (a) 0 km 400 km (b) H. dintelmanni (c)
H. hutsebauti
H. tuberculatus
Pliocene Pleistocene 0.10
(d)
H. dintelmanni (female) H. hutsebauti (female) H. tuberculatus (female)
H. dintelmanni H. tuberculatus (male) H. hutsebauti (male) (male) BELL ET AL. | 5231
FIGURE 3 (a) Distribution of Hyperolius tuberculatus species complex sampling localities in Central Africa including Bioko Island. The approximate range is shown in green, and the type localities for H. tuberculatus, H. dintelmanni and H. hutsebauti are indicated with black arrows. (b) Mitochondrial (16S and cytochrome b) phylogeography; posterior probabilities >.95 are denoted by black dots, probabilities >.85 are denoted by white dots, and 95% highest posterior density intervals for divergence time estimates on well-supported nodes are indicated. The axis indicates geological epochs (Pliocene and Pleistocene) and time before present in increments of 1 million years. (c) Multilocus nuDNA networks generated using POFAD and SPLITSTREE. In all cases, samples are coded with shapes corresponding to mitochondrial lineages. The scale represents genetic distance equally weighted across loci. (d) Representative photographs of female and male coloration. Photograph credits: D. Portik, E. Greenbaum, B. Stuart, R. Bell
we followed Graham et al. (2010) with a 10 m/year dispersal rate. Cleveland, OH, USA), and sequenced using a BigDye Terminator This dispersal rate is equivalent to 40 km or 8 pixels for 4 kya time Cycle Sequencing Kit v. 3.1 (Applied Biosystems, Foster City, CA, intervals, which is a conservative estimate based on lifetime dispersal USA) on an ABI Automated 3730xl Genetic Analyzer (Applied potential in frogs as well as large-scale biogeographic patterns in Biosystems). DNA sequences were edited using SEQUENCHER V. 5.0.1 African frogs. We calculated stability surfaces using modified R (Gene Codes Corp., Ann Arbor, MI, USA) and deposited in GenBank scripts provided by S. Phillips (personal communication). (Table S1). Sequences for 39 samples of the H. cinnamomeoventris The contemporary distributions of the Hyperolius tuberculatus complex (16S, cytochrome b, cmyc, POMC and RAG1) were collected and H. ocellatus species complexes include the land-bridge island in previous studies (Bell, Drewes, Channing, et al., 2015; Bell, Bioko, while the H. cinnamomeoventris species complex is absent Drewes, & Zamudio, 2015). from the island; thus, we considered historic changes in coastline that may have facilitated or restricted dispersal between continental 2.4.2 | Mitochondrial phylogeography Africa and Bioko. We classified “mainland” and “island” as the current exposed area on continental Africa or on Bioko within the We included sequences from closely related taxa (Table S1) in pre- window xmin = 7.59693°, xmax = 10.35984°, ymin = 2.650173°, liminary analyses to ensure that each species complex represents a ymax = 5.079467° and measured potential connectivity over the last monophyletic group based on the current understanding of phyloge-
120 kyr using PATHMATRIX (McRae, 2006; McRae, Dickson, Keitt, & netic relationships in Hyperolius (Portik, 2015). These analyses Shah, 2008) with the niche models for each time slice as conduc- revealed that a few samples included in Bell, Drewes, Channing, tance surfaces. PATHMATRIX implements circuit theory to calculate all et al. (2015; AC 3096–3097, PM 35 and VGCD 1273–1274) are possible movements between two areas (higher values of conduc- likely outside the H. cinnamomeoventris complex, and thus, they were tance correspond to higher potential dispersal), which better repre- not included in the present study. Sequences were aligned using sents connectivity among populations than the least-cost path CLUSTAL X v. 2.0.10 (Larkin et al., 2007). We used PARTITIONFINDER v. method (McRae & Beier, 2007). Effective conductance was scaled 1.1.0 (Lanfear, Calcott, Ho, & Guindon, 2012) to determine the best- relative to the current size of Bioko Island (99 pixels). fit nucleotide substitution models for each mtDNA locus (Table S3). For each species complex, we inferred the mtDNA phylogeny using
Bayesian phylogenetic analyses implemented in BEAST v. 1.8.0 (Drum- 2.4 | Genetic data collection and analysis mond, Suchard, Xie, & Rambaut, 2012). When we compared strict and uncorrelated lognormal relaxed molecular clock models in pre- 2.4.1 | Genetic data collection liminary analyses, we found low rate variation among branches (the We extracted total genomic DNA using DNeasy Blood & Tissue Kits standard deviation parameters [ucld.stdev] of the uncorrelated log- (Qiagen Inc., Valencia, CA, USA). Two specimens were formalin-fixed normal relaxed clock analyses were close to zero); therefore, we (YPM A8062 and A8086), and we extracted DNA from these sam- used the strict clock model for the final analyses. We chose a con- ples following a protocol developed by Cathy Dayton of the U.S. stant size coalescent tree prior and obtained posterior distributions Fish and Wildlife Service (Methods S1). We amplified and sequenced from two Markov chain Monte Carlo (MCMC) simulations, each run one or two mitochondrial (mtDNA) fragments (16S and cytochrome for 10 million generations and assessed convergence with TRACER v. b) and one to three nuclear (nuDNA) protein-coding genes (cmyc, 1.5 (Rambaut, Suchard, Xie, & Drummond, 2013). No fossils of POMC, RAG1) for each sample using published primers (Table S1– hyperoliid frogs exist with which to calibrate divergence times; S3). Polymerase chain reactions (PCRs) were carried out in a final therefore, we applied a sequence divergence rate for the most stable volume of 20 ll containing: 20 ng template DNA, 19 Buffer, 0.2 lM and informative reference locus in our data set—cytochrome b— of each primer, 0.4 mM dNTP mix and 0.125 units of Taq DNA poly- based on rates estimated for tropical bufonid frogs (0.80%–1.90% merase (Roche Diagnostics, Indianapolis, IN, USA). Amplification was per Myr; Sanguila, Siler, Diesmos, Nuneza,~ & Brown, 2011). We carried out with an initial denaturation for 5 min at 94°C, followed selected a rate prior with a mean of 1.4% and a normal distribution by 35 cycles (60 s denaturation at 94°C, 60 s annealing at 42–55°C (95% confidence interval of 0.8%–1.9%). The effective sample size [Table S2], 60 s extension at 72°C) and a final extension at 72°C for for each parameter was well above 200, and simulations were 5 min. PCR products were purified using ExoSAP-IT (USB Corp., repeated without sequence data to test the influence of priors on 5232 | BELL ET AL.
(a) 0 km 400 km
(b) 5 (c) 5 H. o. ocellatus
H. o. purpurescens
0.01 0.10 Miocene Pliocene Pleistocene 1.0 (d) H. o. ocellatus (male)
H. o. ocellatus “type 1” H. o. ocellatus “type 2” H. o. purpurescens “type 3” H. o. purpurescens (female) (female) (female) (female) BELL ET AL. | 5233
FIGURE 4 (a) Distribution of Hyperolius ocellatus species complex sampling localities in Central Africa including Bioko Island. The approximate range is shown in green, and the type localities for H. o. ocellatus and H. o. purpurescens are indicated with black arrows. (b) Mitochondrial (16S and cytochrome b) phylogeography; posterior probabilities >.95 are denoted by black dots, probabilities >.85 are denoted by white dots, and 95% highest posterior density intervals for divergence time estimates on well-supported nodes are indicated. The axis indicates geological epochs (Miocene, Pliocene and Pleistocene) and time before present in increments of 1 million years. (c) Multilocus nuDNA networks generated using POFAD and SPLITSTREE. In all cases, samples are coded with shapes corresponding to mitochondrial lineages. The scale represents genetic distance equally weighted across loci. (d) Representative photographs of distinct female colour morphs and male coloration. Photograph credits: D. Portik, P. McLaughlin, B. Stuart, E. Greenbaum
posterior distributions. We discarded the first 10% of trees as Kishino, & Yano, 1985). Using POFAD V. 1.03 (Joly & Bruneau, 2006), burn-in and combined the tree files from replicate runs using LOGCOM- we combined individual locus matrices into one, multilocus distance
BINER prior to summarizing the posterior distribution of trees using matrix (equally weighted across loci). Finally, we constructed a
TREEANNOTATOR. genetic network of the multilocus distance matrix in SPLITSTREE V. 4.6 (Huson & Bryant, 2006) using the NeighborNet algorithm (Bryant & Moulton, 2004). 2.4.3 | Differentiation at nuclear loci
Sequences were aligned using CLUSTAL X V. 2.0.10 (Larkin et al., 2007). 2.4.4 | Diversity and historical demography We verified the absence of recombination within nuclear loci using the sum of squares method implemented in TOPALI 2 (Milne et al., For the distinct lineages we identified in each species complex
2009), resolved haplotypes for heterozygous individuals using PHASE through our mtDNA and nuDNA analyses, we calculated the number v. 2.1 (Stephens, Smith, & Donnelly, 2001) implemented in DNASP v. of variable sites and nucleotide diversity (hs, hp)inARLEQUIN V. 3.1 5.1 (Librado & Rozas, 2009) and generated individual gene trees for (Excoffier, Laval, & Schneider, 2005). Using the combined mtDNA each locus using the neighbour-joining method in GENEIOUS V. 8.0.4 and nuDNA data set, we estimated changes in effective population (Figs S2–S4). size over time for the distinct lineages recovered within each species
We inferred population structure and assigned individuals to complex (based on mtDNA gene trees, the STRUCTURE analyses and genetic populations with the Bayesian assignment program STRUC- SPLITSTREE networks of nuDNA distances) with the Extended Bayesian
TURE V. 2.2.3 (Pritchard, Stephens, & Donnelly, 2000) using multilocus Skyline Plot approach (Heled & Drummond, 2008) implemented in genotypes (phased haplotypes with the linkage model among SNPs BEAST2 v 2.4.5 (Bouckaert et al., 2014). We used PARTITIONFINDER v. within a locus). For each species complex, we only included individu- 1.1.0 (Lanfear et al., 2012) to determine the best-fit nucleotide sub- als with data for at least two nuclear loci, which precluded including stitution models for the nuclear loci (Table S3) and applied the same H. dintelmanni in the H. tuberculatus species complex analysis (only mtDNA rate prior as described above. We obtained posterior distri- one sample met our missing data threshold). Preliminary analyses for butions from two MCMC simulations, each run for 100 million gen- the entire H. cinnamomeoventris species complex produced high vari- erations and assessed convergence and the influence of priors as ance in likelihood when K was greater than two. Thus, we split the described above. We plotted the median and 95% central posterior samples for subsequent analyses following the deep divergence density intervals of the demographic history of each lineage in R v within the species complex: we ran analyses with K from one 3.3.0 using BEAST2 scripts. through five in H. c. cinnamomeoventris + H. veithi, and K from one through four in H. c. olivaceus + the island endemics. Preliminary 2.4.5 | Species tree estimation analyses for the other two species complexes were more stable at higher values of K; consequently, we ran analyses with K from one Given that our focal species complexes have expansive geographic through four in the H. tuberculatus complex (H. tubercula- distributions and that current taxonomic designations do not neces- tus + H. hutsebauti), and K from one through five in the H. ocellatus sarily reflect true diversity or evolutionary relationships, we used the complex (H. o. ocellatus + H. o. purpurescens). For all analyses we ran multilocus coalescent model implemented in *BEAST (Heled & Drum- ten iterations of each value of K with a burn-in of 1 million steps, mond, 2010) to infer relationships within each species complex using
MCMC length of 3 million steps, correlated allele frequencies and the combined mtDNA and nuDNA data sets. The *BEAST approach the admixture ancestry model. To determine the optimal number of makes the assumption that incomplete lineage sorting is the main genetic clusters in each data set, we used the method described by source of inconsistency between gene trees and the underlying spe- Evanno, Regnaut, and Goudet (2005). cies tree and that there is no post-divergence gene flow between To visually represent overall divergence patterns for samples species. Thus, for each of the species complexes, we assigned indi- with complete sequence data for the three nuclear genes, we used a viduals to putative species for the analysis following the distinct lin- multilocus, individual-based network approach. We used PAUP V. 4.0b eages recovered in the mtDNA gene trees, the STRUCTURE analyses
(Swofford, 2003) to create genetic distance matrices between and SPLITSTREE networks of nuDNA distances (Figures 2–4, Fig. S7; phased haplotypes at each locus using the HKY85 model (Hasegawa, Table S3). We specified unlinked site, clock and tree models (the 5234 | BELL ET AL.
TABLE 1 Number of unique occurrences for Ecological Niche Models (ENM), model fit (AUC) and per cent contribution of Bioclim variables for full-parameter ENMs for the Hyperolius tuberculatus complex (Ht), H. ocellatus complex (Ho), H. c. olivaceus (Hco.) and H. c. cinnamomeoventris (Hcc) Ht Ho Hco Hcc
Unique localities (training/full) 43/57 35/46 15/20 34/45 AUC (train/test) 0.94/0.95 0.97/0.96 0.98/0.95 0.96/0.94 Variable (units) Annual precipitation bc12 (mm) 66.6 53.5 58.8 3.4 Precipitation of driest quarter bc17 (mm) 11.7 34.5 30.3 66.0 Temperature seasonality (SD*100) bc04 11.3 1.7 0.7 16.7 Mean temperature of coldest quarter bc11 (°C) 7.7 5.1 0.7 8.0 Precipitation seasonality (coefficient of variation) bc15 2.1 0.7 9.3 5.9 Annual mean temperature bc01 (°C) – 2.4 0.2 – Mean temperature of warmest quarter bc10 (°C) – 2.0 –– Precipitation of wettest quarter bc16 (mm) 0.6 ––– tree model for mtDNA loci was linked), a Yule process tree prior, (including Bioko Island, Fig. S6) and smaller pockets of stability in and a strict molecular clock model with the same mtDNA rate the Congolian forests (Figure 1). For H. c. olivaceus, the ENMs recov- prior as described above. We obtained posterior distributions ered suitable and stable climates in the Lower Guinean forests with from two MCMC simulations, each run for 200 million genera- the highest suitability centred in Gabon (Figure 1). Although stable tions, and assessed convergence and the influence of priors as climates were predicted farther north in coastal Cameroon, the pre- described above. We discarded the first 10% of trees as burn-in sent known range of H. c. olivaceus does not extend that far north. and combined the tree files from replicate runs using LOGCOMBINER By contrast, the ENMs recovered suitable and stable climates for prior to summarizing the posterior distribution of trees using H. c. cinnamomeoventris across much of the Congolian forests and
TREEANNOTATOR. stopping short of the Lower Guinean forests (Figure 1). Finally, maxi- mum connectivity for the H. tuberculatus and H. ocellatus complexes between Bioko Island and continental Africa occurred at the LGM 3 | RESULTS (21 kya and 22 kya, respectively) and remained high throughout most of the 15–72 kyr period (Figure 5, Fig. S7). By contrast, con- 3.1 | Ecological niche modelling, niche stability and nectivity for H. c. cinnamomeoventris and H. c. olivaceus remained connectivity through time low throughout the last 120 kyr (Figure 5, Fig. S7). Niche models for the Hyperolius tuberculatus complex, the H. ocella- tus complex, H. c. olivaceus and H. c. cinnamomeoventris had a rea- 3.2 | Mitochondrial phylogeography sonable performance with AUC values ranging from 0.94 (H. tuberculatus complex) to 0.98 (H. c. olivaceus; Figure 1, Table 1, The mtDNA phylogenies of the three species complexes revealed Fig. S3). Precipitation-related variables, specifically annual precipita- varying levels of divergence across the Lower Guineo-Congolian for- tion, precipitation seasonality and precipitation of the driest quarter, ests. In the Hyperolius cinnamomeoventris species complex, we recov- were critical in all four models (Table 1); relative occurrence rate ered Late Miocene divergence between H. c. olivaceus, the island was higher in areas with more than 1,500 mm of annual precipita- endemics, and a clade comprised of H. c. cinnamomeoventris and H. tion. Hyperolius c. cinnamomeoventris, however, occupies drier envi- veithi (Figure 2b). Hyperolius c. olivaceus is restricted to the Lower ronments (down to 500 mm of annual precipitation) and occurs in Guinean forests and is comprised of two lineages that diverged in areas with relatively high seasonality and a dry quarter with almost the early Pleistocene. By contrast, the H. c. cinnamomeoventris clade no rain, whereas the H. tuberculatus complex, H. ocellatus complex spans the Congolian forests and consists of five lineages (including and H. c. olivaceus occur in sites with lower seasonality in precipita- H. veithi) with episodes of further divergence throughout the Plio- tion and usually some rain in the driest quarter. cene and Pleistocene. In the H. tuberculatus complex, we recovered Estimates of climatic stability based on the static and dynamic three distinct lineages with divergence across the Lower Guineo- stability surfaces were qualitatively similar with the dynamic stability Congolian forests spanning the Pliocene-Pleistocene transition (Fig- models producing slightly larger and more contiguous stable areas ure 3b). We found that H. dintelmanni is distinct from adjacent popu- (Figure 1, Fig. S5). The ENMs identified similar predicted suitable lations of H. tuberculatus in Cameroon, H. hutsebauti north of the and stable climates in the H. tuberculatus and H. ocellatus complexes Congo River is nested within a lineage of eastern H. tuberculatus that included moderate-to-high suitability across the Congo Basin populations, and H. tuberculatus from Bioko Island (red triangles; Fig- with the greatest climatic stability in the Lower Guinean forests ure 3) is nested within a clade from coastal Cameroon and Gabon BELL ET AL. | 5235
(a) H. tuberculatus H. ocellatus H. c. olivaceus H. c. cinnamomeoventris Connectivity 0.0 0.01 0.02 0.03 0 20 40 60 80 100 120 Time (1000 years) (b)
5 H. tuberculatus H. ocellatus 21 kya 22 kya 4 FIGURE 5 (a) Estimates of potential connectivity between Bioko Island and the Longitude African continent for the Hyperolius tuberculatus complex, H. ocellatus complex, Suitability Suitability
3 0.92 0.89 H. c. olivaceus and H. c. cinnamomeoventris 0.78 0.62 0.47 0 estimated using PATHMATRIX with the niche models for each time slice as conductance 8910 Latitude surfaces. Higher values of conductance correspond to higher potential dispersal. H. c. olivaceus H. c. cinnamomeoventris Connectivity (effective conductance) is 56 kya 22 kya scaled relative to the contemporary size of Bioko Island (99 pixels) (b) Distribution of climatic suitability for the H. tuberculatus complex, H. ocellatus complex,H.c. olivaceus and H. c. cinnamomeoventris at the respective periods of peak potential connectivity between Bioko Island and the Suitability Suitability 0.83 0.38 African continent. Note the differences in 0.58 0.26 scale and time period between panels 0 0
(red squares; Figure 3). Finally, in the H. ocellatus complex we found H. cinnamomeoventris and H. tuberculatus species complexes, the extensive phylogeographic structure across the Lower Guineo-Con- STRUCTURE analyses and multilocus distance networks were largely golian forests with divergence throughout the Pliocene and Pleis- congruent with the mtDNA phylogenies. The H. cinnamomeoventris tocene. We identified three distinct mtDNA lineages in Cameroon species complex was comprised of five demes (H. c. olivaceus, the that correspond to the three regional morphotypes (Figure 4); how- island endemics, H. veithi, and distinct eastern and western popula- ever, populations attributed to H. o. purpurescens across the species tions of H. c. cinnamomeoventris; Fig. S7). In the H. tuberculatus spe- complex range consist of several distinct lineages that may be para- cies complex, H. dintelmanni was not included in the analyses due to phyletic with respect to H. o. ocellatus. Likewise, populations of H. o. small sample size and STRUCTURE recovered two demes corresponding ocellatus from Bioko Island (red triangles; Figure 4) form two distinct to western populations of H. tuberculatus and eastern populations of subclades that occur in sympatry on the island and are paraphyletic H. tuberculatus + H. hutsebauti (Fig. S3). In both species complexes, with respect to populations in southwest Cameroon (red squares; distinct mtDNA lineages that occur in near sympatry were also dif- Figure 4). ferentiated at nuclear loci (Figures 2 and 3, Figs S2 and S3). Although we recovered extensive mtDNA structure in the H. ocella-
tus species complex, the STRUCTURE analyses of nuDNA recovered 3.3 | Differentiation at nuclear loci two demes largely corresponding to H. o. ocellatus and H. o. pur- The vast majority of haplotypes were resolved computationally with purescens (Fig. S7). While the distinct mtDNA lineage of H. o. pur- probabilities of 0.51 or greater (>87%). Sets of haplotypes with low purescens in the eastern part of the range (yellow circles) was probabilities were due to singleton or low frequency SNPs for which differentiated from western lineages in the SPLITSTREE network, the phasing errors likely have very little influence on our estimates of remaining mtDNA lineages in the western part of the range (yellow genetic structure in the STRUCTURE analyses or genetic distance in the stars, hexagons, diamonds and triangles) were not differentiated
SPLITSTREE networks; thus, we retained the set of haplotypes with the from one another at nuclear loci (Figure 4, Fig. S4). Furthermore, the highest probability even if it was below 0.51. For the allopatric type 1 and type 2 morphotypes in Cameroon (red symbols; 5236 | BELL ET AL.
FIGURE 6 *BEAST species tree inference for combined mtDNA and nuDNA haplotypes collected from the Hyperolius cinnamomeoventris species complex, the H. tuberculatus species complex and the H. ocellatus species complex. Posterior probabilities >.95 are denoted by black dots, probabilities >.90 are denoted by white dots, and 95% highest posterior density intervals for divergence time estimates are indicated
Figure 4) and mtDNA subclades on Bioko Island (red triangles; Fig- single operational species unit. The species tree reconstruction ure 4) were undifferentiated at nuclear loci (Fig. S4). strongly supported H. c. cinnamomeoventris and H. veithi as sister lineages and H. c. olivaceus and the island endemics as sister lin- eages (Figure 6). For the H. tuberculatus complex, we assigned sam- 3.4 | Historical demography ples to three operational species units following the mtDNA and Given the small sample sizes for the geographically restricted lin- multilocus nuDNA analyses (H. dintelmanni,westernH. tuberculatus eages in two of the species complexes—H. veithi and the island and eastern H. tuberculatus + H. hutsebauti; distinct colours in Fig- endemics (H. cinnamomeoventris complex) and H. dintelmanni (H. tu- ure 3). The species tree reconstruction supported western H. tuber- berculatus complex)—we estimated changes in effective population culatus and H. dintelmanni as sister lineages in this recent radiation size over time for the primarily western versus eastern lineages (Figure 6). For the H. ocellatus complex, the multilocus nuDNA within each complex (Fig. S8). Across all three species complexes, analyses recovered two lineages (H. o. ocellatus and H. o. pur- the western lineages (H. c. olivaceus, western H. tuberculatus and H. purescens; distinct colours in Figure 4); thus, we assigned samples o. ocellatus) exhibited smaller and more stable population sizes to these two operational species units to obtain divergence time through time with moderate signatures of recent expansion in west- estimates that were comparable with the other species complexes ern H. tuberculatus and 95% HPD intervals in H. c. olivaceus and H. (Figure 6). Across all three species complexes, the point estimates o. ocellatus that include constant size (Fig. S8). By contrast, the east- of divergence in the species tree analyses were more recent than ern lineages (H. c. cinnamomeoventris, eastern H. tuberculatus + H. those estimated in the mtDNA analyses; however, divergence esti- hutsebauti and H. o. purpurescens) exhibited larger population sizes mates within the H. cinnamomeoventris complex predated those of with substantial increases and/or decreases in population size over the H. tuberculatus and H. ocellatus complexesinbothsetsofanal- time (Fig. S8). yses.
3.5 | Species tree estimation 4 | DISCUSSION
For the H. cinnamomeoventris species complex, we assigned samples 4.1 | Idiosyncratic responses to climate-driven from the mainland to operational species units following the mtDNA vicariance across Central Africa and multilocus nuDNA analyses (H. c. cinnamomeoventris, H. c. oli- vaceus and H. veithi). We grouped samples of the island endemic Our analyses of multilocus sequence data demonstrate both tempo- species (H. thomensis, H. molleri and H. drewesi) into a single opera- ral and spatial heterogeneity in patterns of diversification among tional species unit because few sites in our nuDNA data set were sympatric reed frog species across the Lower Guineo-Congolian for- variable between these recently diverged island species. Analyses ests. Factors such as environmental heterogeneity and ecological with H. c. cinnamomeoventris split into eastern and western lineages interactions can limit the geographical extent of species ranges, yet (blue and yellow symbols in Figure 2) did not reach convergence species distributions are often largely determined by the limits of a after 200 million generations so we grouped these samples into a species’ physiological tolerance and spatial variation in climate BELL ET AL. | 5237
(Chown et al., 2010). Our estimates of current climatic suitability high climatic stability in Lower Guinean forests (Born et al., 2011; and climatic stability through time indicate that differences in cli- Dauby, Duminil, et al., 2014; Faye, Deblauwe, Mariac, & Richard, matic niche may partly explain idiosyncratic patterns of diversifica- 2016; Leache & Fujita, 2010; Plana, 2004; Querouil et al., 2003), tion among our three focal species complexes. We acknowledge that versus fluctuating and more fragmented stability across the Congo our divergence date estimates rely on a number of assumptions (e.g., Basin (Jacquet et al., 2015; Johnston & Anthony, 2012; Nicolas substitution rate is equal to the mutation rate for within-species et al., 2011; Tosi, 2008). inferences) and that these approaches tend to overestimate diver- Although our modest molecular data set limits strong inferences gence dates between recently diverged lineages (Herman & Searle, based on divergence time estimates, we consistently recover more 2011). Even with these limitations, our divergence date estimates pronounced phylogeographic structure in the H. ocellatus complex based on the mtDNA-only data set and the combined mtDNA and than in the H. tuberculatus complex. Hyperolius ocellatus breeds in nuDNA data sets enable us to propose a general timeframe for small rainforest streams and swamps, and is rarely found outside of divergence events and demonstrate variation in divergence times forest habitats (Amiet, 2012), whereas frogs in the H. tuberculatus among the species complexes (e.g., non-overlapping 95% HPD inter- complex breed in forest clearings and swamps, and are more tolerant vals). of edge or open forest habitats (Amiet, 2012). These differences in The Hyperolius cinnamomeoventris species complex is broadly dis- ecological breadth between the species may buffer the effects of tributed in both forest and bushland habitats across Central Africa; habitat fragmentation in H. tuberculatus relative to H. ocellatus however, we recovered Late Miocene divergence between H. c. cin- (Rodr ıguez et al., 2015). The species complexes also differ in body namomeoventris and H. c. olivaceus across the boundary of the Lower size: frogs of the H. ocellatus complex are sexually dimorphic in size Guinean and Congolian forests in Gabon and Republic of the Congo (males ~20 mm snout-vent length [SVL], females ~30 mm SVL; (Figure 2). This region appears as largely contiguous lowland rain- Amiet, 2012), whereas males and females in the H. tuberculatus com- forest habitat, yet H. c. olivaceus is restricted to sites with low sea- plex are similar in size (~30 mm SVL; Amiet, 2012). Over ~90% of sonality, while H. c. cinnamomeoventris occupies drier environments frog species display sexual dimorphism in body size and this variation with relatively high seasonality. Consequently, the climatic stability is typically attributed to selection for increased fecundity when models demarcate distinct regions of climatic suitability for H. c. oli- females are larger (Salthe & Duellman, 1973) or strong sexual selec- vaceus and H. c. cinnamomeoventris throughout the variable climates tion when males are larger (Shine, 1979). Thus, differences in selec- of the last 120 kyr, and this demarcation appears to have persisted tion related to fecundity and/or mating system, as well as body size- since the Late Miocene. Thus, this pattern of allopatric divergence associated dispersal ability (Pabijan, Wollenberg, & Vences, 2012; coupled with marked divergence in environmental niche is consistent Wollenberg, Vieites, Glaw, & Vences, 2011), may also be contribut- with the “vanishing refuge” model of speciation wherein habitat frag- ing to the differences in phylogeographic structure we recovered. mentation and exposure to new environments result in closely Studies in co-distributed rainforest trees, understory plants and lia- related lineages occupying adjacent, contrasting habitats (Damas- nas in Lower Guinean forests also recover varied responses to pro- ceno, Strangas, Carnaval, Rodrigues, & Moritz, 2014; Vanzolini & posed barriers in this region (Budde, Gonz alez-Mart ınez, Hardy, & Williams, 1981). Furthermore, the estimated divergence between the Heuertz, 2013; Dauby, Duminil, et al., 2014; Faye et al., 2016; two lineages coincides with periods of aridification in the Miocene Heuertz et al., 2014; Ley et al., 2014). This growing literature indi- that are attributed with driving diversification of predominantly for- cates that accounting for species traits related to ecological breadth, est-restricted plant and animal lineages into drier savannah habitats dispersal and mating system will be essential for understanding (Holstein & Renner, 2011; Johnston & Anthony, 2012). shared phylogeographic patterns in this region. In addition, more Relative to the H. cinnamomeoventris species complex, diver- extensive geographic sampling in montane and lowland sites gences in the H. tuberculatus and H. ocellatus complexes are more throughout Cameroon and Gabon coupled with methods that explic- recent (Pliocene to Pleistocene), and these frogs are restricted to itly account for changes in population size and gene flow may more forest habitats with low seasonality. Correspondingly, the climatic rigorously detect differences in forest species’ responses to a shared suitability models recover similar suitable climates for these species climatic history (Bell et al., 2012; Dasmahapatra, Lamas, Simpson, & complexes with the highest stability centred in the Lower Guinean Mallet, 2010; Hickerson, Stahl, & Lessios, 2006; Leache, Crews, & forests. Although current distributions of both species complexes Hickerson, 2007). span the Congo Basin, our estimates of climatic suitability and stabil- Despite marked differences among the H. ocellatus, H. tubercula- ity over the last 120 kyr indicate that these populations likely experi- tus and H. cinnamomeoventris species complexes in the geographic enced several periods of contraction and expansion throughout the extent and temporal estimates of divergence, we recover a shared variable climates of the Pliocene and Pleistocene (Hardy et al., 2013; pattern of intermittent moderate-to-high climatic suitability across Plana, 2004). Genetic diversity in both the H. tuberculatus and the Congo Basin in the last ~120 kyr (Figure 1, Fig. S7) and mtDNA H. ocellatus complexes is concentrated in the Lower Guinean forests clades that extend across this expansive region originate at the Plio- and the Extended Bayesian Skyline Plots reveal smaller and more cene-Pleistocene transition (H. c. cinnamomeoventris, H. o. pur- stable populations in this region relative to the Congo Basin lineages. purescens and the H. tuberculatus complex; Figures 2–4). These These patterns are consistent with long-term lineage persistence and patterns are consistent with studies in rainforest trees (Couvreur 5238 | BELL ET AL. et al., 2008), rodents (Bryja et al., 2017; Nicolas et al., 2011) and pri- be difficult to interpret with respect to these coinciding biogeo- mates (Tosi, 2008), and support the hypothesis that genetic graphic features. exchange across the Congo Basin during more humid periods, fol- We recovered a wide range of divergence times and distinct lin- lowed by divergence during arid periods, has shaped patterns of eages among our focal taxa, many of which may reflect incipient regional diversity (Fjelds a & Lovett, 1997; Tosi, 2008). Populations species. In particular, the result that regional variation in female col- of H. c. cinnamomeoventris east and west of the Congo Basin are oration across the H. ocellatus complex corresponds to distinct strongly differentiated in both the mtDNA and nuDNA data sets mtDNA lineages is especially intriguing. Lineages that diverged in (blue and yellow symbols; Figure 2, Fig. S7), which is consistent with response to earlier climatic or geologic processes (e.g., Pliocene or allopatric divergence on either side of the Congo Basin. Further sam- Late Miocene) are expected to exhibit stronger postzygotic isolation pling across the northern Congo Basin is needed to confirm that than those that formed in response to more recent (Pleistocene) these lineages diverged in allopatry (versus simple isolation by dis- events (Avise, 2000; Singhal & Moritz, 2013; Weir & Price, 2011). tance) and if so, assess whether eastern and western lineages persist Likewise, divergence in phenotypic traits involved in social signalling or become obscured when they come into secondary sympatry in is also expected to scale with divergence in geographic isolation the Congo Basin (Holstein & Renner, 2011). Likewise, we find dis- (Winger & Bates, 2015). Although the function of colour polymor- tinct mtDNA lineages in the centre of the Congo Basin in the phism and sexual dichromatism in reed frogs is unknown, these traits H. ocellatus and H. tuberculatus complexes (yellow triangles, Fig- may be relevant for species recognition and/or mate choice. Alterna- ures 3 and 4) that coincide with the distribution of the recently tively, colour polymorphism may reflect varying selection processes described H. veithi in the H. cinnamomeoventris species complex (Fig- including differences in predation regimes, as occurs with poison- ure 2), but this pattern may also be an artefact of isolation-by-dis- dart frogs (Brown, Maan, Cummings, & Summers, 2010), or alternate tance and our discontinuous sampling. More extensive sampling reproductive strategies, as demonstrated in the rock-paper-scissors across the central Congo Basin will likely reveal additional cryptic system in side-blotched lizards (Sinervo & Lively, 1996). These selec- diversity within each species complex and provide greater insight tive mechanisms can result in rapid phenotypic evolution (Corl, into the temporal and spatial dynamics of lineage diversification in Davis, Kuchta, & Sinervo, 2010) and/or accelerated speciation rates this region. (Hugall & Stuart-Fox, 2012). Thus, the spectrum of divergence times across our focal taxa, regions of secondary contact between previ- ously isolated lineages, extensive colour polymorphism, and variation 4.2 | Rivers, lineage divergence and phenotypic in sexual size dimorphism and dichromatism present a rich compara- variation in Central African Hyperolius tive framework in which to investigate reproductive isolation and The distributions of our focal species complexes span three major phenotypic divergence in Central African Hyperolius. rivers (the Congo, the Ogooue and the Sanaga) that are barriers to dispersal in many mammal and understory bird species across the 4.3 | Marine incursions and population divergence Guineo-Congolian forest (Anthony et al., 2007; Hassanin et al., on the land-bridge island Bioko 2015; Huntley & Voelker, 2016; Mitchell et al., 2015; Nicolas et al., 2011; Querouil et al., 2003; Telfer et al., 2003). Although we recov- Although the H. cinnamomeoventris complex is widely distributed ered extensive phylogeographic structure in each species complex, across Central Africa, none of the lineages occur on Bioko Island, and these regional clades typically include localities on either side of the our estimates of connectivity indicate that low climatic suitability Sanaga, Ogooue and Congo rivers and these regional clades also between Bioko and the continent may have precluded colonization span two or more of the major river drainages. This pattern indicates when the land bridge was exposed (Figure 5). By contrast, H. tubercu- that broad-scale phylogeographic structure in these frogs does not latus and H. o. ocellatus occur on Bioko and niche models for the H. tu- correspond to Central African hydrobasins and that rivers are likely berculatus and H. ocellatus species complexes reveal high connectivity not significant barriers to dispersal. In particular, although the Congo between Bioko and the continent between 14 and 70 kya. As in previ- River is the proposed barrier between H. tuberculatus and H. hutse- ous phylogeographic studies of Bioko Island reptiles and amphibians bauti in the Democratic Republic of the Congo (Laurent, 1956, (Barej et al., 2014; Leache & Fujita, 2010; Leache, Fujita, Minin, & 1976), our results suggest that divergence in this species complex is Bouckaert, 2014), we find moderate genetic divergence between not delineated by the Congo River. Our data are consistent, how- mainland and island populations of H. o. ocellatus and H. tuberculatus, ever, with the Sanaga River serving as the boundary between H. o. indicating that gene flow between Bioko and the continent is not on- ocellatus and H. o. purpurescens in Cameroon (Laurent, 1943; Perret, going. In both species complexes, we recover Late Pleistocene origins 1975). Patterns of divergence across the Sanaga may be confounded for the Bioko Island populations, but with two distinct mtDNA clades by the historical presence of lowland rainforest climatic refugia on in H. o. ocellatus (TMRCA 30–290 kya, and 90–460 kya) and one in either side of the river (Dauby, Duminil, et al., 2014; Dauby, Hardy, H. tuberculatus (40–210 kya). Again, our modest molecular data set et al., 2014; Droissart et al., 2011). This region had high predicted limits strong inferences based on divergence time estimates; however, climatic suitability in our ENMs (Figure 1); consequently, patterns of the ages of the younger H. o. ocellatus clade and H. tuberculatus clade phylogeographic structure in the H. ocellatus species complex may are roughly coincident with the period of high connectivity we recover BELL ET AL. | 5239 in the historic climatic suitability models (Figure 5a) and indicate that recognizing each of these lineages as full species. The type locality dispersal across the land bridge has likely occurred multiple times in of H. olivaceus Buchholz & Peters, 1876 (Lambaren e, Gabon), which the island’s 3 Myr old history. In addition, in both H. o. ocellatus and was considered a subspecies of H. cinnamomeoventris by Laurent H. tuberculatus we find that adjacent continental populations are (1943), is among the localities for H. c. olivaceus in this study. Given nested within Bioko Island lineages (Figures 3 and 4), which may our results, we elevate H. c. olivaceus to H. olivaceus (i.e., from the reflect “reverse colonization” from Bioko to the continent during these rank of subspecies to species). Our results indicate that H. c. cin- periods of connectivity (Bellemain & Ricklefs, 2008). Reed frogs colo- namomeoventris exhibits substantial cryptic diversity and is para- nized two oceanic islands in the Gulf of Guinea (Pr ıncipe and S~ao phyletic with respect to H. veithi; however, we refrain from Tome) via rafting (Bell, Drewes, Channing, et al., 2015) and it is possi- taxonomic revision within this lineage pending additional sampling. ble that H. o. ocellatus and H. tuberculatus colonized Bioko Island via Given our concordant results among mtDNA and nuDNA analyses, similar sweepstakes colonization events; however, relatively high we refer the eastern lineages of H. tuberculatus to H. hutsebauti; genetic diversity of the island populations and recent (Late Pleis- however, resolving the geographic boundary between the two spe- tocene) population divergence from the mainland are more consistent cies will require more continuous sampling across the Congo Basin. with vicariance due to marine incursions rather than multiple overseas dispersal events. ACKNOWLEDGEMENTS Bioko harbours multiple mtDNA lineages of H. o. ocellatus that are not differentiated at nuDNA and a single mtDNA lineage of For fieldwork in Gabon we thank the CENAREST, ANPN and the H. tuberculatus, which mirrors the divergence pattern we recovered Direction de la Faune et des Aeres Proteg ees for permits, the Wild- in continental Africa and may further reflect how habitat specificity life Conservation Society Gabon Program and Organisation Eco- and dispersal ability shape species’ responses to shared biogeo- touristique du Lac Oguemoue for logistical support, and N. Emba- graphic barriers. The species complexes also differ with respect to Yao, F. Moiniyoko, B. Hylayre, E. Ekomy, A. Dibata, T. Ogombet, U. the geographic location of mainland populations most closely related Eyagui, P. Endazokou, for assistance in the field. For fieldwork in to those on Bioko Island. Hyperolius o. ocellatus on Bioko are most Equatorial Guinea we thank Universidad Nacional de Guinea Equato- closely related to mainland populations in southwestern Cameroon rial and Jose Manuel Esara Echube for permits, the Bioko Biodiver- (red squares, Figure 4), similar to the pattern recovered in African sity Protection Program, ExxonMobil Foundation, and Mobil forest geckos (Hemidactylus fasciatus; Leache & Fujita, 2010; Leache Equatorial Guinea Inc. for logistical support, and A. Fertig, B. Miles, et al., 2014). By contrast, Bioko populations of H. tuberculatus are and D. Matute for assistance in the field. For fieldwork in Cameroon, most closely related to mainland populations in coastal Cameroon we thank the Ministry of Scientific Research and Innovation and the and Gabon (red squares, Figure 3), which is similar to the pattern Ministry of Forestry and Wildlife for permits, and O. Kopecky, B. recovered in African clawed frogs (Xenopus calcaratus, X. allofraseri; Freiermuth, N.L. Gonwouo, G. Jongsma, M.T. Kouete and L. Schein- Evans et al., 2015). Differences in the timing of divergence between berg for assistance in the field. For fieldwork in the Republic of island and mainland populations of H. tuberculatus and H. o. ocellatus, Congo, we thank J. Gaugris of Flora Fauna & Man, Ecological Ser- as well as the geographic locations of the most closely related main- vices Ltd for logistical support and the Groupe d’Etude et de land populations, reinforce the conclusion that marine incursions are Recherche sur la Diversite Biologique for permits. Fieldwork by ZTN not the only factor restricting dispersal between Bioko and the con- in Democratic Republic of the Congo was supported by the Parc tinent. Instead, the composition of habitats connecting Bioko to the MarindesMangroves,ICCN(M.Collet)andtheBelgianNational rest of the continent when sea levels retreat and species-specific dif- Focal Point to the Global Taxonomy Initiative. Fieldwork in Angola ferences in dispersal potential through these habitats likely mediate was supported by P. vas Pinto. Fieldwork by EG in Democratic gene flow across the land-bridge for island residents and may pre- Republic of the Congo was supported by the Centre de Recherche clude island colonization for some continental species altogether. en Sciences Naturelles, ICCN, Mwenebatu M. Aristote and Wan- dege M. Muninga. For fieldwork in Rwanda, we thank the Rwan- dan Development Board for permits. We thank K. Jackson, D. 4.4 | Implications for reed frog taxonomy Mulcahy, L. Scheinberg, C. Spencer, J. Rosado, G. Watkins-Colwell, We recovered multiple distinct evolutionary lineages within the con- J. Vindum and A. Wynn for providing access to tissue samples in tinental distribution of the H. cinnamomeoventris complex that corre- their care, M. Hydeman for assistance in the molecular laboratory, spond to ecologically and phenotypically distinct populations. S.B. Reilly for computational resources, and H.W. Greene, R.G. Hyperolius veithi, H. c. cinnamomeoventris and H. c. olivaceus are dif- Harrison, J.I. Lovette and five anonymous reviewers for comments ferentiated in both mtDNA and nuDNA are strongly supported as that improved the manuscript. Funding was provided by the distinct lineages in our species tree analysis, and H. c. cinnamo- Explorer’s Club, American Philosophical Society, Sigma Xi, Society meoventris and H. c. olivaceus occupy different environmental niches. of Systematic Biologists, Mario Einaudi Center for International Furthermore, we find no evidence for introgression between these Studies, Cornell Graduate School, Andrew W. Mellon Foundation, three lineages even though we collected H. c. cinnamomeoventris EEB Paul P. Feeny Fund, EEB Paul Graduate Fellowship and a within 10 km of H. c. olivaceus and H. veithi; thus, we recommend University of California President’s Postdoctoral Fellowship (to 5240 | BELL ET AL.
RCB); Museum of Comparative Zoology Herpetology Division at Bell, R. C., Drewes, R. C., & Zamudio, K. R. (2015). Reed frog diversifica- Harvard University (to RCB & BLS); the BIOTA projects of the tion in the Gulf of Guinea: Overseas dispersal, the progression rule, and in situ speciation. Evolution, 69, 904–915. Federal Ministry of Education and Research (BMB+F, Germany to Bell, R. C., Mackenzie, J. B., Hickerson, M. J., Chavarr ıa, K. L., Cunning- SL and MOR); the US National Science Foundation (DEB-1309171 ham, M., Williams, S. E., & Mortiz, C. (2012). Comparative multi-locus to RCB, DEB-1202609 to DCB and DEB-1145459 to EG); the phylogeography confirms multiple vicariance events in co-distributed National Geographic Society (8556-08 to EG, 8868-10 to RCB); rainforest frogs. Proceedings of the Royal Society B: Biological Sciences, 279, 991–999. the California Academy of Sciences (to DCB); the Percy Sladen Bellemain, E., & Ricklefs, R. E. (2008). Are islands the end of the coloniza- Memorial Fund, IUCN/SSC Amphibian Specialist Group, K. Reed, tion road? Trends in Ecology and Evolution, 23, 461–468. M.D., the Department of Biology at Villanova University and the Bohoussou, K. H., Cornette, R., Akpatou, B., Colyn, M., Kerbis Peterhans, University of Texas at El Paso (to EG); the Czech Science Founda- J. C., Kennis, J., ... Nicolas, V. (2015). The phylogeography of the rodent genus Malacomys suggests multiple Afrotropical Pleistocene tion (GACR, project number 15-13415Y) and Ministry of Culture lowland forest refugia. Journal of Biogeography, 42, 2049–2061. of the Czech Republic (DKRVO 2017/15, National Museum, Bonier, F., Martin, P. R., & Wingfield, J. C. (2007). Urban birds have 00023272; to VG). broader environmental tolerance. Biology Letters, 3, 670–673. Born, C., Alvarez, N., McKey, D., Ossari, S., Wickings, E. J., Hossaert- McKey, M., & Chevallier, M. H. (2011). Insights into the biogeograph- DATA ACCESSIBILITY ical history of the Lower Guinea Forest Domain: Evidence for the role of refugia in the intraspecific differentiation of Aucoumea klai- DNA sequences: GenBank accession numbers in Supporting Informa- neana. Molecular Ecology, 20, 131–142. tion (Table S1). Bouckaert, R., Heled, J., Kuhnert,€ D., Vaughan, T., Wu, C.-H., Xie, D., ... Sampling localities and voucher information for molecular data Drummond, A. J. (2014). BEAST 2: A software platform for Bayesian evolutionary analysis. PLoS Computational Biology, 10, e1003537. analyses and environmental niche models in Supporting Information Bowie, R. C. K., Fjelds a, J., Hackett, S. J., & Crowe, T. M. (2004). Molecu- (Table S1). lar evolution in space and through time: mtDNA phylogeography of Geospatial bioclimatic layers for the current, middle Holocene, the Olive Sunbird (Nectarinia olivacea/obscura) throughout continental – last glacial maximum and last interglacial: Dryad https://doi.org/10. Africa. Molecular Phylogenetics and Evolution, 33,56 74. Brown, J. L., Maan, M. E., Cummings, M. E., & Summers, K. (2010). Evi- 5061/dryad.vs76v. dence for selection on coloration in a Panamanian poison frog: A coalescent-based approach. Journal of Biogeography, 37, 891–901. Bryant, D., & Moulton, V. (2004). Neighbor-Net: An agglomerative AUTHOR CONTRIBUTIONS method for the construction of phylogenetic networks. Molecular Biology and Evolution, 21, 255–265. R.C.B. designed the project; R.C.B., G.B., M.B., D.C.B., M.B, A.C., Bryja, J., Sumbera, R., Kerbis Peterhans, J. C., Aghov a, T., Bryjova, A., J.M.D., E.G., V.G., C.K., J.K., S.L., P.J.M., Z.T.N., M.-O.R., D.M.P., Mukula, O., ... Verheyen, E. (2017). Evolutionary history of the B.L.S. and A.G.Z.-B. collected field samples; J.V. contributed historical thicket rats (genus Grammomys) mirrors the evolution of African for- climate data; R.C.B. and J.L.P. collected and analysed the data; all ests since late Miocene. Journal of Biogeography, 44, 182–194. Budde, K. B., Gonz alez-Mart ınez, S. C., Hardy, O. J., & Heuertz, M. authors contributed funding to the project; R.C.B. wrote the manu- (2013). The ancient tropical rainforest tree Symphonia globulifera L. f. script with input from all authors. (Clusiaceae) was not restricted to postulated Pleistocene refugia in Atlantic Equatorial Africa. Heredity, 111,66–76. REFERENCES Butynski, T. M., & Koster, S. H. (1994). Distribution and conservation sta- tus of primates in Bioko Island, Equatorial Guinea. Biodiversity and Amiet, J.-L. (2012). Les rainettes du Cameroun (Amphibiens Anoures). Conservation, 3, 893–909. Nyons: Imprimerie Marque Depos ee. Chown, S. L., Hoffmann, A. A., Kristensen, T. N., Angilletta, M. J., Jr, Sten- Anthony, N. M., Johnson-Bawe, M., Jeffery, K., Clifford, S. L., Abernethy, seth, N. C., & Pertoldi, C. (2010). Adapting to climate change: A per- K. A., Tutin, C. E., ... Bruford, M. W. (2007). The role of Pleistocene spective from evolutionary physiology. Climate Research, 43,3–15. refugia and rivers in shaping gorilla genetic diversity in central Africa. Corl, A., Davis, A. R., Kuchta, S. R., & Sinervo, B. (2010). Selective loss of Proceedings of the National Academy of Sciences of the United States polymorphic mating types is associated with rapid phenotypic evolu- of America, 104, 20432–20436. tion during morphic speciation. Proceedings of the National Academy Avise, J. C. (2000). Phylogeography: The history and formation of species. of Sciences of the United States of America, 107, 4254–4259. Boston, MA: Harvard University Press. Couvreur, T. L., Chatrou, L. W., Sosef, M. S., & Richardson, J. E. (2008). Barej, M. F., Rodel,€ M. O., Loader, S. P., Menegon, M., Gonwouo, L. N., Molecular phylogenetics reveal multiple tertiary vicariance origins of Penner, J., ... Schmitz, A. (2014). Light shines through the spindrift— the African rain forest trees. BMC Biology, 6, 54. Phylogeny of African torrent frogs (Amphibia, Anura, Petropedetidae). Damasceno, R., Strangas, M. L., Carnaval, A. C., Rodrigues, M. T., & Mor- Molecular Phylogenetics and Evolution, 71, 261–273. itz, C. (2014). Revisiting the vanishing refuge model of diversification. Bell, R. C. (2016). A new species of Hyperolius (Amphibia: Hyperoliidae) Frontiers in Genetics, 5,1–12. from Pr ıncipe Island, Democratic Republic of Sao~ Tome and Pr ıncipe. Dasmahapatra, K. K., Lamas, G., Simpson, F., & Mallet, J. (2010). The Herpetologica, 72, 343–351. anatomy of a “suture zone” in Amazonian butterflies: A coalescent- Bell, R. C., Drewes, R. C., Channing, A., Gvozd ık, V., Kielgast, J., Lotters,€ based test for vicariant geographic divergence and speciation. Mole- S., ... Zamudio, K. R. (2015). Overseas dispersal of Hyperolius reed cular Ecology, 19, 4283–4301. frogs from Central Africa to the oceanic islands of S~ao Tome and Dauby, G., Duminil, J., Heuertz, M., Koffi, G. K., Stevart, T., & Hardy, O. Pr ıncipe. Journal of Biogeography, 42,65–75. J. (2014). Congruent phylogeographical patterns of eight tree species BELL ET AL. | 5241
in Atlantic Central Africa provide insights into the past dynamics of Graham, C. H., VanDerWal, J., Phillips, S. J., Moritz, C., & Williams, S. E. forest cover. Molecular Ecology, 23, 2299–2312. (2010). Dynamic refugia and species persistence: Tracking spatial Dauby, G., Hardy, O. J., Leal, M., Breteler, F., & Stevart, T. (2014). Drivers shifts in habitat through time. Ecography, 33, 1062–1069. of tree diversity in tropical rain forests: New insights from a compar- Harcourt, A. H., & Wood, M. A. (2012). Rivers as barriers to primate dis- ison between littoral and hilly landscapes of Central Africa. Journal of tributions in Africa. International Journal of Primatology, 33, 168–183. Biogeography, 41, 574–586. Hardy, O. J., Born, C., Budde, K. B., Da€ınou, K., Dauby, G., Duminil, J., Deruelle, B., Moreau, C., Nkoumbou, C., Kambou, R., Lissom, J., Njofang, ... Koffi, G. K. (2013). Comparative phylogeography of African rain E., & Nono, A. (1991). The Cameroon Volcanic Line: A review. In A. forest trees: A review of genetic signatures of vegetation history B. Kampunzu, & R. T. Lubala (Eds.), Magmatism in extensional struc- in the Guineo-Congolian region. Comptes Rendus Geoscience, 345, tural settings (pp. 274–327). Berlin: Springer. 284–296. Droissart, V., Sonke, B., Hardy, O. J., Simo, M., Taedoumg, H., Nguembou, Hasegawa, M., Kishino, H., & Yano, T. (1985). Dating of the human-ape C. K., & Stevart, T. (2011). Do plant families with contrasting func- splitting by a molecular clock of mitochondrial DNA. Journal of Molec- tional traits show similar patterns of endemism? A case study with ular Evolution, 22, 160–174. Central African Orchidaceae and Rubiaceae. Biodiversity and Conser- Hassanin, A., Khouider, S., Gembu, G.-C., Goodman, S. M., Kadjo, B., Nesi, vation, 20, 1507–1531. N., ... Bonillo, C. (2015). The comparative phylogeography of fruit Drummond, A. J., Suchard, M. A., Xie, D., & Rambaut, A. (2012). Bayesian bats of the tribe Scotonycterini (Chiroptera, Pteropodidae) reveals phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and cryptic species diversity related to African Pleistocene forest refugia. Evolution, 29, 1969–1973. Comptes Rendus Biologies, 338, 197–211. Duminil, J., Brown, R. P., Ewedj e E-E, B. K., Mardulyn, P., Doucet, J.-L., & Heled, J., & Drummond, A. J. (2008). Bayesian inference of population Hardy, O. J. (2013). Large-scale pattern of genetic differentiation size history from multiple loci. BMC Evolutionary Biology, 8, 289. within African rainforest trees: Insights on the roles of ecological gra- Heled, J., & Drummond, A. J. (2010). Bayesian inference of species trees dients and past climate changes on the evolution of Erythrophleum from multilocus data. Molecular Biology and Evolution, 27, 570–580. spp. (Fabaceae). BMC Evolutionary Biology, 13, 195. Herman, J. S., & Searle, J. B. (2011). Post-glacial partitioning of mitochon- Duminil, J., Mona, S., Mardulyn, P., Doumenge, C., Walmacq, F., Doucet, drial genetic variation in the field vole. Proceedings of the Royal Soci- J.-L., & Hardy, O. J. (2015). Late Pleistocene molecular dating of past ety B: Biological Sciences, 278, 3601–3607. population fragmentation and demographic changes in African rain Heuertz, M., Duminil, J., Dauby, G., Savolainen, V., & Hardy, O. J. (2014). forest tree species supports the forest refuge hypothesis. Journal of Comparative phylogeography in rainforest trees from Lower Guinea, Biogeography, 42, 1443–1454. Africa. PLoS One, 9, e84307. Elith, J., Phillips, S. J., Hastie, T., Dud ık, M., Chee, Y. E., & Yates, C. J. Hickerson, M. J., Stahl, E. A., & Lessios, H. A. (2006). Test for simultane- (2011). A statistical explanation of MaxEnt for ecologists. Diversity ous divergence using approximate Bayesian computation. Evolution, and Distributions, 17,43–57. 60, 2435–2453. Evanno, G., Regnaut, S., & Goudet, J. (2005). Detecting the number of Hijmans, R. J., Cameron, S. E., Parra, J. L., Jones, P. G., & Jarvis, A. clusters of individuals using the software structure: A simulation (2005). Very high resolution interpolated climate surfaces for global study. Molecular Ecology, 14, 2611–2620. land areas. International Journal of Climatology, 25, 1965–1978. Evans, B. J., Carter, T. F., Greenbaum, E., Gvozd ık, V., Kelley, D. B., Holstein, N., & Renner, S. S. (2011). A dated phylogeny and collection McLaughlin, P. J., ... Blackburn, D. C. (2015). Genetics, morphology, records reveal repeated biome shifts in the African genus Coccinia advertisement calls, and historical records distinguish six new poly- (Cucurbitaceae). BMC Evolutionary Biology, 11, 28. ploid species of African clawed frog (Xenopus, Pipidae) from West Hugall, A. F., Moritz, C., Moussalli, A., & Stanisic, J. (2002). Reconciling and Central Africa. PLoS One, 10, e0142823. paleodistribution models and comparative phylogeography in the Excoffier, L., Laval, G., & Schneider, S. (2005). Arlequin (version 3.0): An Wet Tropics rainforest land snail Gnarosophia bellendenkerensis (Bra- integrated software package for population genetics data analysis. zier 1875). Proceedings of the National Academy of Sciences of the Uni- Evolutionary Bioinformatics Online, 1, 47. ted States of America, 99, 6112–6117. Faye, A., Deblauwe, V., Mariac, C., & Richard, D. (2016). Phylogeography Hugall, A. F., & Stuart-Fox, D. M. (2012). Accelerated speciation in col- of the genus Podococcus (Palmae/Arecaceae) in Central African rain our-polymorphic birds. Nature, 485, 631–634. forests: Climate stability predicts unique genetic diversity. Molecular Huntley, J. W., & Voelker, G. (2016). Cryptic diversity in Afro-tropical Phylogenetics and Evolution, 105, 126–138. lowland forests: The systematics and biogeography of the avian Fjelds a, J., & Lovett, J. C. (1997). Geographical patterns of old and young genus Bleda. Molecular Phylogenetics and Evolution, 99, 297–308. species in African forest biota: The significance of specific montane Huson, D. H., & Bryant, D. (2006). Application of phylogenetic networks areas as evolutionary centres. Biodiversity and Conservation, 6, 325– in evolutionary studies. Molecular Biology and Evolution, 23, 254– 346. 267. Fleming, K. M. (2000). Glacial rebound and sea-level change: Constraints on Jacquet, F., Denys, C., Verheyen, E., Bryja, J., Hutterer, R., Kerbis Peter- the Greenland ice sheet. Canberra: Australian National University. hand, J. C., ... Nicolas, V. (2015). Phylogeography and evolutionary Fleming, K., Johnston, P., Zwartz, D., et al. (1998). Refining the eustatic history of the Crocidura olivieri complex (Mammalia, Soricomorpha): sea-level curve since the Last Glacial Maximum using far- and inter- From a forest origin to broad ecological expansion across Africa. mediate-field sites. Earth and Planetary Science Letters, 163, 327–342. BMC Evolutionary Biology, 15, 71. Freedman, A. H., Thomassen, H. A., Buermann, W., & Smith, T. B. (2010). Jenkins, C. N., Pimm, S. L., & Joppa, L. N. (2013). Global patterns of ter- Genomic signals of diversification along ecological gradients in a trop- restrial vertebrate diversity and conservation. Proceedings of the ical lizard. Molecular Ecology, 19, 3773–3788. National Academy of Sciences of the United States of America, 110, Fuchs, J., Parra, J. L., Goodman, S. M., Raherilalao, M. J., VanDerWal, J., E2602–E2610. & Bowie, R. C. K. (2013). Extending ecological niche models to the Johnston, A. R., & Anthony, N. M. (2012). A multi-locus species phy- past 120 000 years corroborates the lack of strong phylogeographic logeny of African forest duikers in the subfamily Cephalophinae: Evi- structure in the crested drongo (Dicrurus forficatus) on Madagascar. dence for a recent radiation in the Pleistocene. BMC Evolutionary Biological Journal of the Linnean Society, 108, 658–676. Biology, 12, 120. Futuyma, D. J. (2010). Evolutionary constraint and ecological conse- Joly, S., & Bruneau, A. (2006). Incorporating allelic variation for recon- quences. Evolution, 64, 1865–1884. structing the evolutionary history of organisms from multiple genes: 5242 | BELL ET AL.
An example from Rosa in North America. Systematic Biology, 55, 623– Lobo, J. M., Jimenez-Valverde, A., & Real, R. (2008). AUC: A misleading 636. measure of the performance of predictive distribution models. Global Jones, P. J. (1994). Biodiversity in the Gulf of Guinea: An overview. Biodi- Ecology and Biogeography, 17, 145–151. versity and Conservation, 3, 772–784. Lopez-Uribe, M. M., Zamudio, K. R., Cardoso, C. F., & Danforth, B. N. Kadu, C. A. C., Schueler, S., Konrad, H., Muluvi, G. M. M., Eyog-Matig, (2014). Climate, physiological tolerance and sex-biased dispersal O., Muchugi, A., ... Geburek, T. (2011). Phylogeography of the shape genetic structure of Neotropical orchid bees. Molecular Ecology, Afromontane Prunus africana reveals a former migration corridor 23, 1874–1890. between East and West African highlands. Molecular Ecology, 20, Lotters,€ S., & Schmitz, A. (2004). A new species of tree frog (Amphibia; 165–178. Hyperolius) from the Bakossi Mountains, South-West-Cameroon. Bon- Kindler, C., Moosig, M., Branch, W. R., Harvey, J., Kehlmaier, C., Nagy, Z. ner zoologische Beitrage€ , 52, 149–154. T., ... Fritz, U. (2016). Comparative phylogeographies of six species Lowe, A. J., Harris, D., Dormontt, E., & Dawson, I. K. (2010). Testing of hinged terrapins (Pelusios spp.) reveal discordant patterns and putative African tropical forest refugia using chloroplast and nuclear unexpected differentiation in the P. castaneus/P. chapini complex and DNA phylogeography. Tropical Plant Biology, 3,50–58. P. rhodesianus. Biological Journal of the Linnean Society, 117, 305–321. Maley, J. (1996). The African rain forest—Main characteristics of changes Kirschel, A. N. G., Slabbekoorn, H., Blumstein, D. T., Cohen, R. E., de in vegetation and climate from the Upper Cretaceous to the Quater- Kort, S. R., Buermann, W., & Smith, T. B. (2011). Testing alternative nary. Proceedings of the Royal Society of Edinburgh. Section B. Biologi- hypotheses for evolutionary diversification in an African songbird: cal Sciences, 104,31–73. Rainforest refugia versus ecological gradients. Evolution, 65, 3162– Marzoli, A., Piccirillo, E. M., Renne, P. R., Bellieni, G., Iacumin, M., 3174. Nyobe, J. B., & Tongwa, A. T. (2000). The Cameroon volcanic line Koffi, G. K., Hardy, O. J., Doumenge, C., Cruaud, C., & Heuertz, M. revisited: Petrogenesis of continental basaltic magmas from litho- (2011). Diversity gradients and phylogeographic patterns in Santiria spheric and asthenospheric mantle sources. Journal of Petrology, 41, trimera (Burseraceae), a widespread African tree typical of mature 87–109. rainforests. American Journal of Botany, 98, 254–264. Massatti, R., & Knowles, L. L. L. (2014). Microhabitat differences impact Lanfear, R., Calcott, B., Ho, S. Y. W., & Guindon, S. (2012). PartitionFin- phylogeographic concordance of codistributed species: Genomic evi- der: Combined selection of partitioning schemes and substitution dence in montane sedges (Carex L.) from the rocky mountains. Evolu- models for phylogenetic analyses. Molecular Biology and Evolution, 29, tion, 68, 2833–2846. 1695–1701. McRae, B. H. (2006). Isolation by resistance. Evolution, 60, 1551–1561. Larkin, M. A., Blackshields, G., Brown, N. P., Chenna, R., McGettigan, P. McRae, B. H., & Beier, P. (2007). Circuit theory predicts gene flow A., McWilliam, H., ... Higgin, D. G. (2007). Clustal W and Clustal X in plant and animal populations. Proceedings of the National Acad- version 2.0. Bioinformatics, 23, 2947–2948. emy of Sciences of the United States of America, 104, 19885– Laurent, R. F. (1943). Les Hyperolius (Batraciens) du Musee du Congo. 19890. Annales du Mus ee Royal du Congo Belge. Sciences Zoologiques. Tervu- McRae, B. H., Dickson, B. G., Keitt, T. H., & Shah, V. B. (2008). Using cir- ren, 4,61–140. cuit theory to model connectivity in ecology, evolution, and conser- Laurent, R. F. (1956). Notes herpetologiques africaines I. Revue de Zoolo- vation. Ecology, 89, 2712–2724. gie et de Botanique Africaines, 52, 229–256. Melo, M., Warren, B. H., & Jones, P. J. (2011). Rapid parallel evolution of Laurent, R. F. (1976). Nouveaux commentaires sur la superespece Hyper- aberrant traits in the diversification of the Gulf of Guinea white-eyes olius viridiflavus (Anura). Musee Royal d’Afrique Centrale Serie IN-8. (Aves, Zosteropidae). Molecular Ecology, 20, 4953–4967. Science Zoologique Meyers, J. B., Rosendahl, B. R., Harrison, C. G., & Ding, Z.-D. (1998). Lea, D. W., Martin, P. A., Pak, D. K., & Spero, H. J. (2002). Reconstructing Deep-imaging seismic and gravity results from the offshore Camer- a 350 ky history of sea level using planktonic Mg/Ca and oxygen iso- oon Volcanic Line, and speculation of African hotlines. Tectonophysics, tope records from a Cocos Ridge core. Quaternary Science Reviews, 284,31–63. 21, 283–293. Milne, I., Lindner, D., Bayer, M., Husmeier, D., McGuire, G., Marshall, D. Leache, A. D., Crews, S. C., & Hickerson, M. J. (2007). Two waves of F., & Wright, F. (2009). TOPALi v2: A rich graphical interface for evo- diversification in mammals and reptiles of Baja California revealed by lutionary analyses of multiple alignments on HPC clusters and multi- hierarchical Bayesian analysis. Biology Letters, 3, 646–650. core desktops. Bioinformatics, 25, 126–127. Leache, A. D., & Fujita, M. K. (2010). Bayesian species delimitation in Milne, G. A., Long, A. J., & Bassett, S. E. (2005). Modelling Holocene rela- West African forest geckos (Hemidactylus fasciatus). Proceedings of tive sea-level observations from the Caribbean and South America. the Royal Society B: Biological Sciences, 277, 3071–3077. Quaternary Science Reviews, 24, 1183–1202. Leache, A. D., Fujita, M. K., Minin, V. N., & Bouckaert, R. R. (2014). Spe- Mitchell, M. W., Locatelli, S., Sesink Clee, P. R., Thomassen, H. A., & Gon- cies delimitation using genome-wide SNP data. Systematic Biology, der, M. K. (2015). Environmental variation and rivers govern the 63, 534–542. structure of chimpanzee genetic diversity in a biodiversity hotspot. Lewin, A., Feldman, A., Bauer, A. M., Belmaker, J., Broadley, D. G., Chirio, BMC Evolutionary Biology, 15,1. L., ... Meiri, S. (2016). Patterns of species richness, endemism and Muscarella, R., Galante, P. J., Soley-Guardia, M., Boria, R. A., Kass, J. M., environmental gradients of African reptiles. Journal of Biogeography, Uriarte, M., & Anderson, R. P. (2014). ENMeval: An R package for 43, 2380–2390. conducting spatially independent evaluations and estimating optimal Ley, A. C., Dauby, G., Koehler, J., Wypior, C., Roeser, M., & Hardy, O. J. model complexity for Maxent ecological niche models. Methods in (2014). Comparative phylogeography of eight herbs and lianas Ecology and Evolution, 5, 1198–1205. (Marantaceae) in central African rainforests. Frontiers in Genetics, 5, Nicolas, V., Missoup, A. D., Denys, C., Kerbis Peterhans, J. C., Katuala, P., 403. Couloux, A., & Colyn, M. (2011). The roles of rivers and Pleistocene Librado, P., & Rozas, J. (2009). DnaSP v5: A software for comprehensive refugia in shaping genetic diversity in Praomys misonnei in tropical analysis of DNA polymorphism data. Bioinformatics, 25, 1451–1452. Africa. Journal of Biogeography, 38, 191–207. Linder, H. P., de Klerk, H. M., Born, J., Burgess, N. D., Fjelds a, J., & Rah- Njabo, K. Y., Bowie, R. C. K., & Sorenson, M. D. (2008). Phylogeny, bio- bek, C. (2012). The partitioning of Africa: Statistically defined biogeo- geography and taxonomy of the African wattle-eyes (Aves: Passeri- graphical regions in sub-Saharan Africa. Journal of Biogeography, 39, formes: Platysteiridae). Molecular Phylogenetics and Evolution, 48, 1189–1205. 136–149. BELL ET AL. | 5243
Olson, D. M., & Dinerstein, E. (2001). Terrestrial ecoregions of the world: Schiøtz, A. (1999). Treefrogs of Africa. Frankfurt am Main: Edition Chi- A new map of life on earth. BioScience, 51, 933–938. maira. Pabijan, M., Wollenberg, K. C., & Vences, M. (2012). Small body size Shine, R. (1979). Sexual selection and sexual dimorphism in the Amphibia. increases the regional differentiation of populations of tropical man- Copeia, 1979, 297–306. tellid frogs (Anura: Mantellidae). Journal of Evolutionary Biology, 25, Sinervo, B., & Lively, C. M. (1996). The rock-paper-scissors game 2310–2324. and the evolution of alternative male strategies. Nature, 30, 240– Papadopoulou, A., & Knowles, L. L. L. (2015). Species-specific responses 243. to island connectivity cycles: Refined models for testing phylogeo- Singarayer, J. S., & Valdes, P. J. (2010). High-latitude climate sensitivity graphic concordance across a Mediterranean Pleistocene Aggregate to ice-sheet forcing over the last 120 kyr. Quaternary Science Reviews, Island Complex. Molecular Ecology, 24, 4252–4268. 29,43–55. Papadopoulou, A., & Knowles, L. L. L. (2016). Toward a paradigm shift in Singhal, S., & Moritz, C. (2013). Reproductive isolation between phylo- comparative phylogeography driven by trait-based hypotheses. Pro- geographic lineages scales with divergence. Proceedings of the Royal ceedings of the National Academy of Sciences of the United States of Society B: Biological Sciences, 280, 20132246. America, 113, 8018–8024. Smith, T. B., Calsbeek, R., Wayne, R. K., Holder, K. H., Pires, D., & Barde- Paz, A., Iba nez,~ R., Lips, K. R., & Crawford, A. J. (2015). Testing the role leben, C. (2005). Testing alternative mechanisms of evolutionary of ecology and life history in structuring genetic variation across a divergence in an African rain forest passerine bird. Journal of Evolu- landscape: A trait-based phylogeographic approach. Molecular Ecology, tionary Biology, 18, 257–268. 24, 3723–3737. Smith, T. B., Thomassen, H. A., Freedman, A. H., Sehgal, R. N. M., Buer- Perez, J. D. V., Fa, J. E., Castroviejo, J., & Purroy, F. J. (1994). Species mann, W., Saatchi, S., ... Wayne, R. K. (2011). Patterns of divergence richness and endemism of birds in Bioko. Biodiversity and Conserva- in the olive sunbird Cyanomitra olivacea (Aves: Nectariniidae) across tion, 3, 868–892. the African rainforest–savanna ecotone. Biological Journal of the Lin- Perret, J. L. (1975). Les sous-especes d’Hyperolius ocellatus Gunther€ nean Society, 103, 821–835. (Amphibia, Salientia). Annales de la Facult e des Sciences du Cameroun. Smith, T. B., Wayne, R. K., Girman, D. J., & Bruford, M. W. (1997). A role Yaound e, 20,23–31. for ecotones in generating rainforest biodiversity. Science, 276, 1855– Peterson, A. T., Papesß, M., & Soberon, J. (2008). Rethinking receiver 1857. operating characteristic analysis applications in ecological niche mod- Soberon, J. (2007). Grinnellian and Eltonian niches and geographic distri- eling. Ecological Modelling, 213,63–72. butions of species. Ecology Letters, 10, 1115–1123. Phillips, S. J., Anderson, R. P., & Schapire, R. E. (2006). Maximum entropy Stephens, M., Smith, N. J., & Donnelly, P. (2001). A new statistical modeling of species geographic distributions. Ecological Modelling, method for haplotype reconstruction from population data. American 190, 231–259. Journal of Human Genetics, 68, 978–989. Plana, V. (2004). Mechanisms and tempo of evolution in the African Gui- Swofford, D. L. (2003). PAUP*: Phylogenetic analysis using parsimony neo-Congolian rainforest. Philosophical Transactions of the Royal Soci- (* and other methods) Software program, Version: 4.0. Sunderland ety B: Biological Sciences, 359, 1585–1594. MA: Sinauer Associates. Portik, D. M. (2015). Diversification of Afrobatrachian frogs and the her- Telfer, P. T., Souquiere, S., Clifford, S. L., Abernethy, K. A., Bruford, M. petofauna of the Arabian Peninsula. Berkeley: University of California. W., Disotell, T. R., ... Wickings, E. J. (2003). Molecular evidence for Portik, D. M., Jongsma, G. F. M., Kouete, M. T., Scheinberg, L. A., Freier- deep phylogenetic divergence in Mandrillus sphinx. Molecular Ecology, muth, B., Tapondjou, W. P., & Blackburn, D. C. (2016). A survey of 12, 2019–2024. amphibians and reptiles in the foothills of Mount Kupe, Cameroon. Tosi, A. J. (2008). Forest monkeys and Pleistocene refugia: A phylogeo- Amphibian and Reptile Conservation, 10,37–67. graphic window onto the disjunct distribution of the Chlorocebus Pritchard, J. K., Stephens, M., & Donnelly, P. (2000). Inference of popula- lhoesti species group. Zoological Journal of the Linnean Society, 154, tion structure using multilocus genotype data. Genetics, 155, 945– 408–418. 959. VanDerWal, J., Beaumont, L., Zimmermann, N. E., & Lorch, P. (2011). Cli- Querouil, S., Verheyen, E., Dillen, M., & Colyn, M. (2003). Patterns of mates: Methods for working with weather and climate. Retrieved diversification in two African forest shrews: Sylvisorex johnstoni from www rforge. and Sylvisorex ollula (Soricidae, Insectivora) in relation to paleo- Vanzolini, P. E., & Williams, E. E. (1981). The vanishing refuge: A mecha- environmental changes. Molecular Phylogenetics and Evolution, 28, nism for ecogeographic speciation. Pap eis Avulsos de Zoologia, 34, 24–37. 251–255. Rambaut, A., Suchard, M. A., Xie, D., & Drummond, A. J. (2013). Tracer. Voelker, G., Marks, B. D., Kahindo, C., A’genonga, U., Bapeamoni, F., Retrieved from http://beast.bio.ed.ac.uk/Tracer. Duffie, L. E., ... Light, J. E. (2013). River barriers and cryptic biodi- Rodr ıguez, A., Borner,€ M., Pabijan, M., Gehara, M., Haddad, C. F. B., & versity in an evolutionary museum. Ecology and Evolution, 3, 536– Vences, M. (2015). Genetic divergence in tropical anurans: Deeper 545. phylogeographic structure in forest specialists and in topographically Weber, L. C., VanDerWal, J., Schmidt, S., McDonald, W. J. F., & Shoo, L. complex regions. Evolutionary Ecology, 29, 765–785. P. (2014). Patterns of rain forest plant endemism in subtropical Aus- Salthe, S. N., & Duellman, W. E. (1973). Quantitative constraints associ- tralia relate to stable mesic refugia and species dispersal limitations. ated with reproductive mode in anurans. In J. L. Vial (Ed.), Evolution- Journal of Biogeography, 41, 222–238. ary biology of the anurans (pp. 229–249). Kansas City: University of Weir, J. T., & Price, T. D. (2011). Limits to speciation inferred from times Missouri Press. to secondary sympatry and ages of hybridizing species along a latitu- Sanguila, M. B., Siler, C. D., Diesmos, A. C., Nuneza,~ O., & Brown, R. M. dinal gradient. American Naturalist, 177, 462–469. (2011). Phylogeography, geographic structure, genetic variation, and Wiens, J. J., Ackerly, D. D., Allen, A. P., Anacker, B. L., Buckley, L. B., Cor- potential species boundaries in Philippine slender toads. Molecular nell, H. V., ... Stephens, P. R. (2010). Niche conservatism as an Phylogenetics and Evolution, 61, 333–350. emerging principle in ecology and conservation biology. Ecology Let- Schick, S., Kielgast, J., Rodder,€ D., Muchai, V., Burger, M., & Lotters,€ S. ters, 13, 1310–1324. (2010). New species of reed frog from the Congo basin with discus- Winger, B. M., & Bates, J. M. (2015). The tempo of trait divergence in sion of paraphyly in Cinnamon-belly reed frogs. Zootaxa, 2501,23– geographic isolation: Avian speciation across the Maranon~ Valley of 36. Peru. Evolution, 69, 772–787. 5244 | BELL ET AL.
Wollenberg, K. C., Vieites, D. R., Glaw, F., & Vences, M. (2011). Specia- SUPPORTING INFORMATION tion in little: The role of range and body size in the diversification of Malagasy mantellid frogs. BMC Evolutionary Biology, 11, 217. Additional Supporting Information may be found online in the sup- Zamudio, K. R., Bell, R. C., & Mason, N. A. (2016). Phenotypes in phylo- porting information tab for this article. geography: Species’ traits, environmental variation, and vertebrate diversification. Proceedings of the National Academy of Sciences of the United States of America, 113, 8041–8048. Zimkus, B. M., & Gvo zd ık, V. (2013). Sky Islands of the Cameroon Vol- How to cite this article: Bell RC, Parra JL, Badjedjea G, et al. canic Line: A diversification hot spot for puddle frogs (Phrynobatra- Idiosyncratic responses to climate-driven forest fragmentation chidae: Phrynobatrachus). Zoologica Scripta, 42, 591–611. and marine incursions in reed frogs from Central Africa and Zimkus, B. M., Lawson, L. P., Barej, M. F., Barratt, C. D., Channing, A., – Dash, K. M., ... Lotters,€ S. (2017). Leapfrogging into new territory: the Gulf of Guinea Islands. Mol Ecol. 2017;26:5223 5244. How Mascarene ridged frogs diversified across Africa and Madagas- https://doi.org/10.1111/mec.14260 car to maintain their ecological niche. Molecular Phylogenetics and Evolution, 106, 254–269.
Supplemental Information for: Idiosyncratic responses to climate-driven forest fragmentation and marine incursions in reed frogs from Central Africa and the Gulf of Guinea islands
Rayna C. Bell, Juan L. Parra, Gabriel Badjedjea, Michael F. Barej, David C. Blackburn, Marius Burger, Alan Channing, J. Maximilian Dehling, Eli Greenbaum, Václav Gvoždík, Jos Kielgast, Chifundera Kusamba, Stefan Lötters, Patrick J. McLaughlin, Zoltán T. Nagy, Mark-Oliver Rödel, Daniel M. Portik, Bryan L. Stuart, Jeremy VanDerWal, Ange Ghislain Zassi-Boulou, Kelly R. Zamudio
Table of Contents: Supplemental Methods Page 1 Formalin-fixed tissue DNA extraction Table S1 Pages 2–8 Sampling localities and voucher information for molecular data analyses and environmental niche models Table S2 Page 9 Primer sequences and amplification conditions for mitochondrial and nuclear sequences. Table S3 Pages 10–11 Summary statistics number of haplotypes (sequence length, variable sites, and nucleotide diversity) for mitochondrial (16s, cytochrome b) and nuclear (cmyc, POMC, Rag1) loci for distinct lineages in each species complex. Figure S1 Pages 12–13 Distribution of background sampling points for environmental niche models, spearman rank correlations among all 19 bioclimatic variables for the background study region, and changes in AICc and AUC for ENMs under different values of the regularization parameter and different feature type classes. Figure S2 Page 14 Hyperolius cinnamomeoventris species complex BEAST mtDNA phylogeny and neighbour-joining trees for cmyc, POMC, and RAG1. Figure S3 Page 15 Hyperolius tuberculatus species complex BEAST mtDNA phylogeny and neighbour-joining trees for cmyc, POMC, and RAG1.
1
Figure S4 Page 16 Hyperolius ocellatus species complex BEAST mtDNA phylogeny and neighbour- joining trees for cmyc, POMC, and RAG1. Figure S5 Page 17 Full-parameter modelled distribution of static stability estimated over the last 120 ky for H. tuberculatus, H. ocellatus, H. c. olivaceus and H. cinnamomeoventris. Figure S6 Page 18 Climatic stability on Bioko Island for Hyperolius tuberculatus and H. ocellatus. Figure S7 Page 19 Individual assignment probabilities from STRUCTURE analysis of nuclear loci (cmyc, POMC, and Rag1) are depicted for the most likely number of genetic clusters in each dataset based on log-likelihood scores (Evanno et al. 2005). Figure S8 Page 20 Extended Bayesian Skyline Plots (EBSP) based on SNPs from mitochondrial and nuclear loci for the “western” (red/dashed) and “eastern” (yellow/solid) lineages within each species complex.
2
Methods Formalin-fixed tissue DNA extraction Two specimens were formalin-fixed (A8062 and A8086; Yale Peabody Museum) and we extracted DNA from these samples following a protocol developed by Cathy Dayton of the U.S. Fish and Wildlife Service. We extracted DNA from these samples separately from our non-formalin fixed tissues. Briefly, we soaked ~20 mg of tissue in 10 mL of 1X GTE buffer (Shedlock et al. 1997) for 24 h and repeated the soak for a total of three washes over 72 h. We incubated the tissue at 55º C for 4 days in a 1.5 mL tube containing 500 µL PureGene Cell Lysis Solution (Gentra Systems, Inc.), 100 µl Proteinase K, and 20 µL 1mM DTT (dithiothreitol), adding 20µL of Proteinase K daily until tissue was digested. The samples were placed on ice for 5 min before adding 200 µl of PureGene Protein Precipitation Solution (Gentra Systems, Inc.). The samples were gently inverted 50 times and centrifuged at 14,000 x g for 3 min. The supernatant was poured into a 1.5 ml tube containing 600 µl of cold 100% isopropanol and 3 µl of PureGene glycogen solution (Gentra Systems, Inc.), inverted 50 times, incubated at -20º C for 48 h, and centrifuged at 14,000 g for 30 min at room temperature. We discarded the supernatant, added 200 µl of 70% ethanol to the sample, gently inverted the tube 50 times, and then centrifuged the sample at 14,000 g for 3 min. The tube was drained on absorbent paper and air-dried upside down for 6 h. Finally we resuspended the DNA with 40 µl of PureGene DNA Hydration Solution (Gentra Systems, Inc.) and incubated at room temperature for 12 h. We conducted the rest of the laboratory protocols for these samples along with our DNA extractions from non-formalin fixed tissues.
Shedlock, A. M., Haygood, M. G., Pietsch, T. W. & Bentzen,P. 1997 Enhanced DNA extraction and PCR amplification of mitochondrial genes from formalin-fixed museum specimens. Biotechniques 22, 394–400.
Table S1: Sampling localities, mitochondrial clade assignment, and voucher information. Abbreviations as follow: AOMA (Angola-Malanje), BICP (Burundi-Cibitoke), BIRP (Burundi- Rutana), CDBC (Dem Rep Congo-Bas-Congo), CDBP (Dem Rep Congo-Bandundu), CDKP (Dem Rep Congo-Katanga), CDNK (Dem Rep Congo-North Kivu), CDOR (Dem Rep Congo- Orientale), CDSK (Dem Rep Congo-South Kivu), CFSM (Central African Republic Sangha-Mbaéré), CMCE (Cameroon-Centre), CMEA (Cameroon-East), CMLI (Cameroon-Littoral), CMSO (Cameroon-South), CMSW (Cameroon-Southwest), CMWE (Cameroon-West), EGAB (Equatorial Guinea-Bioko-Arena Blanca), EGLL (Equatorial Guinea-Bioko-Lago Loretto), EGLU (Equatorial Guinea-Bioko-Luba), EGMO (Equatorial Guinea-Bioko-Moeri), EGMM (Equatorial Guinea-Bioko-Moka), EGPB (Equatorial Guinea-Bioko-Pico Basile), EGRI (Equatorial Guinea-Bioko-Riaba), EGWN (Equatorial Guinea-Rio Muni-Wele-Nzas), GAES (Gabon-Estuaire), GAMO (Gabon-Moyen-Ogooué), GANP (Gabon-Nyanga), GAOI (Gabon- Ogooué-Ivindo), GAOM (Gabon-Ogooué-Maritime), RCCO (Rep Congo-Cuvette-Ouest), RCLE (Rep Congo-Lekoumou), RCNI (Rep Congo-Niari), RWSO (Rwanda-Southern Province), UGWE (Uganda-Western Region). Cornell University Museum of Vertebrates (CU), the California Academy of Sciences (CAS), the North Carolina Museum of Natural Sciences (NCSM), the Smithsonian National Museum of Natural History (USNM), University of Texas El Paso (UTEP), Museum für Naturkunde, Berlin (ZMB), Yale Peabody Museum (YPM), Royal Belgian Institute of Natural Sciences (RBINS), the Institut National de Recherche sir les Sciences Exactes et Naturelles (IRSEN), and the National Museum in Prague (NMP). Samples without catalogue numbers (NA) are in personal collections: M. Barej (MB, MM) A. Channing (AC), J. Kielgast (JK), S. Lötters (SL).
Species Locality Catalog No. Field No. Lat Long Clade 16s Cyt b Cmyc Pomc Rag1 H. bobirensis GH UWBM 5648 ADL3954 6.2425 -0.5571 Hcinn outgroup KU166830 H. camerunensis CM CAS 253935 DMP875 4.8495 9.7713 Hocel outgroup KX671739 H. cinnamomeoventris AOMA N/A AC3008 -8.8954 16.0871 yellow diamond GU443990 MF376362 MF376278 H. cinnamomeoventris AOMA ZMB 83093 AC3017 -8.8954 16.0871 yellow diamond HM064461 MF376279 H. cinnamomeoventris CDBP RBINS:ZTN:A180 A180 -2.7531 20.3781 yellow triangle MF376177 MF376274 MF376405 MF376475 H. cinnamomeoventris CDBP RBINS:ZTN:A199 A199 -2.7531 20.3781 yellow triangle MF376178 MF376275 MF376406 H. cinnamomeoventris CDBP RBINS:ZTN:A300 A300 -2.7531 20.3781 yellow triangle MF376179 MF376277 MF376407 MF376477 H. cinnamomeoventris CDNK UTEP 20613 EBG2305 1.4007 28.5688 yellow circle MF376182 KJ865978 KJ866016 KJ865945 KJ865919 H. cinnamomeoventris CDNK UTEP 21442 EBG1884 0.5682 29.9178 yellow circle MF376180 MF376320 MF376408 MF376503 H. cinnamomeoventris CDNK UTEP 21443 EBG1885 0.5682 29.9178 yellow circle MF376181 MF376321 MF376409 MF376504 H. cinnamomeoventris CDOR RBINS:ZTN:A26 A26 0.8063 24.2839 yellow circle MF376183 MF376276 MF376410 MF376476 H. cinnamomeoventris CDOR RBINS:ZTN:CRT3646 CRT3646 0.8063 24.2839 yellow circle MF376184 MF376311 MF376411 MF376498 H. cinnamomeoventris CDOR RBINS:ZTN:CRT3702 CRT3702 0.8063 24.2839 yellow circle MF376185 MF376312 MF376412 MF376499 H. cinnamomeoventris CDOR RBINS:ZTN:CRT3760 CRT3760 2.0335 22.7869 yellow triangle MF376186 MF376313 MF376413 H. cinnamomeoventris CDOR RBINS:ZTN:CRT3771 CRT3771 2.0335 22.7869 yellow triangle MF376187 MF376314 MF376414 H. cinnamomeoventris CDOR RBINS:ZTN:CRT4059 CRT4059 1.2702 23.7311 yellow triangle MF376188 MF376315 MF376415 MF376500 H. cinnamomeoventris CDOR RBINS:ZTN:CRT4082 CRT4082 1.2702 23.7311 yellow circle MF376189 MF376316 MF376416 MF376501 H. cinnamomeoventris CDOR RBINS:ZTN:KG090 KG090 1.5270 25.2960 yellow circle MF376190 MF376330 MF376417 H. cinnamomeoventris CDOR RBINS:ZTN:KG118 KG118 1.5270 25.2960 yellow circle MF376191 MF376331 MF376418 H. cinnamomeoventris CDOR RBINS:ZTN:KG167 KG167 0.5390 25.9020 yellow circle MF376192 MF376332 MF376419 H. cinnamomeoventris CDOR RBINS:ZTN:KG168 KG168 0.5390 25.9020 yellow circle MF376193 MF376333 MF376420 H. cinnamomeoventris CDOR RBINS:ZTN:KG224 KG224 0.5390 25.9020 yellow circle MF376194 MF376334 MF376421 H. cinnamomeoventris CDSK UTEP 20994 EBG1306 -1.8744 28.4524 yellow circle MF376195 KJ865979 KJ866017 KJ865946 KJ865920 H. cinnamomeoventris CDSK UTEP 21444 EBG1503 -2.2078 28.6296 yellow circle MF376196 MF376319 MF376422 MF376502 H. cinnamomeoventris CDSK UTEP 21445 EBG2691 -3.0401 28.5050 yellow circle MF376197 MF376322 MF376423 H. cinnamomeoventris CDSK UTEP 21446 ELI438 -3.0368 28.4223 yellow circle MF376198 MF376323 MF376424 MF376505 H. cinnamomeoventris CFSM ZMB 81388 DS69 2.9250 16.2569 blue star GU443997 MF376390 MF376318 MF376425 H. cinnamomeoventris CM MVZ 234720 -- 2.9663 12.0816 ENM only H. cinnamomeoventris CMEA NMP6V 74716 VG10194 2.1000 15.3600 blue star MF376199 KJ865980 KJ866018 KJ865947 MF376511 H. cinnamomeoventris CMSO CAS 253475 DCB451 2.9399 11.9763 blue star MF376200 MF376317 H. cinnamomeoventris CMSO MVZ 234721 -- 2.9651 12.1054 blue star MF376201 H. cinnamomeoventris CMSO MVZ 234722 -- 2.9651 12.1054 blue star MF376202 H. cinnamomeoventris GAOI CU 14952 BLS13798 0.5112 12.8028 blue star MF376222 MF376363 MF376280 MF376426 MF376478 H. cinnamomeoventris GAOI CU 14953 BLS13799 0.5112 12.8028 blue star MF376223 MF376364 MF376281 MF376427 MF376479 H. cinnamomeoventris GAOI CU 14954 BLS13800 0.5112 12.8028 blue star MF376224 KJ865991 KJ866028 KJ865954 KJ865925 H. cinnamomeoventris GAOI CU 14955 BLS13801 0.5112 12.8028 blue star MF376225 KJ865992 KJ866029 KJ865955 KJ865926 H. cinnamomeoventris GAOI CU 15026 BLS14012 0.5112 12.8028 blue star KP137113 KP137229 MF376282 MF376428 MF376480 H. cinnamomeoventris GAOI CU 15027 BLS14014 0.5112 12.8028 blue star MF376226 MF376365 MF376283 MF376429 MF376481 H. cinnamomeoventris GAOI CU 15028 BLS14018 0.5112 12.8028 blue star MF376227 KJ865993 KJ866030 KJ865956 KJ865927 H. cinnamomeoventris GAOI CU 15029 BLS14019 0.5112 12.8028 blue star MF376228 MF376366 MF376284 MF376430 MF376482 H. cinnamomeoventris GAOI CU 15030 BLS14020 0.5112 12.8028 blue star MF376229 MF376367 MF376285 MF376431 MF376483 H. cinnamomeoventris GAOI CU 15031 BLS14021 0.5112 12.8028 blue star MF376230 MF376368 MF376286 MF376432 MF376484 H. cinnamomeoventris GAOI CU 15062 BLS14128 0.5112 12.8028 blue star MF376231 MF376369 MF376287 MF376433 MF376485 H. cinnamomeoventris GAOI CU 15063 BLS14129 0.5112 12.8028 blue star MF376232 MF376370 MF376288 MF376434 MF376486 H. cinnamomeoventris GAOI CU 15064 BLS14130 0.5112 12.8028 blue star MF376233 MF376371 MF376289 MF376435 MF376487 H. cinnamomeoventris GAOI CU 15065 BLS14131 0.5112 12.8028 blue star MF376234 MF376372 MF376290 MF376436 MF376488 H. cinnamomeoventris GAOI CU 15066 BLS14132 0.5112 12.8028 blue star MF376235 MF376373 MF376291 MF376437 MF376489 H. cinnamomeoventris GAOI CU 15067 BLS14133 0.5112 12.8028 blue star KP137114 KP137230 MF376292 MF376438 MF376490 H. cinnamomeoventris KE CAS 141577 -- 0.2778 34.7731 ENM only H. cinnamomeoventris KE CAS 141595 -- 0.1764 34.7792 ENM only H. cinnamomeoventris KE CAS 141645 -- 0.3750 34.8833 ENM only H. cinnamomeoventris KE CAS 152731 -- -0.0667 35.1333 ENM only H. cinnamomeoventris KE CAS 154511 -- -0.1000 35.1167 ENM only H. cinnamomeoventris KEKA MVZ 233830 -- 0.2663 34.8068 yellow circle MF376248 H. cinnamomeoventris KEKA MVZ 238701 -- 0.0422 34.7469 yellow circle MF376249 H. cinnamomeoventris RCCO NMP6V 77514/1 VGCG12092 0.0700 14.2400 blue star MF376250 KJ866019 KJ865948 MF376512 H. cinnamomeoventris RCCO NMP6V 77514/2 VGCG12093 0.0700 14.2400 blue star MF376251 MF376402 KJ866020 KJ865949 MF376513 H. cinnamomeoventris RCNI IRSEN 0312 MBUR03556 -2.2245 12.8195 blue star MF376259 MF376339 MF376465 Species Locality Catalog No. Field No. Lat Long Clade 16s Cyt b Cmyc Pomc Rag1 H. cinnamomeoventris RCNI IRSEN 0314 MBUR03746 -2.2245 12.8195 blue star MF376265 MF376345 MF376471 H. cinnamomeoventris RCNI ZMB 83930 MBUR03547 -2.2919 12.8081 blue star MF376255 MF376335 MF376461 H. cinnamomeoventris RCNI ZMB 83931 MBUR03548 -2.2919 12.8081 blue star MF376256 MF376336 MF376462 H. cinnamomeoventris RCNI ZMB 83932 MBUR03550 -2.2919 12.8081 blue star MF376257 MF376337 MF376463 H. cinnamomeoventris RCNI ZMB 83933 MBUR03551 -2.2919 12.8081 blue star MF376258 MF376338 MF376464 MF376509 H. cinnamomeoventris RCNI ZMB 83934 MBUR03557 -2.2245 12.8195 blue star MF376260 MF376340 MF376466 MF376510 H. cinnamomeoventris RCNI ZMB 83935 MBUR03670 -2.4013 12.7502 blue star MF376263 MF376343 MF376469 H. cinnamomeoventris RCNI ZMB 83936 MBUR03715 -2.2245 12.8195 blue star MF376264 MF376344 MF376470 H. cinnamomeoventris RCNI ZMB 83937 MBUR03989 -2.4164 12.8699 blue star MF376267 MF376347 MF376473 H. cinnamomeoventris RWSO ZMB 77533 JMD651 -2.6011 29.7372 blue star MF376268 KJ866003 KJ866043 KJ865965 KJ865934 H. cinnamomeoventris UG CAS 180131 -- -1.0685 29.7483 ENM only H. cinnamomeoventris UG CAS 180132 -- -0.8855 29.7295 ENM only H. cinnamomeoventris UG CAS 180713 -- -1.0098 29.6207 ENM only H. cinnamomeoventris UG CAS 180720 -- -0.9869 29.6343 ENM only H. cinnamomeoventris UG CAS 202511 -- 0.3559 31.8774 ENM only H. cinnamomeoventris UG CAS 204508 -- 0.5667 30.3500 ENM only H. cinnamomeoventris UG CAS 204512 -- 0.4333 30.4000 ENM only H. cinnamomeoventris UG CAS 204615 -- 0.5000 30.4167 ENM only H. cinnamomeoventris UG MCZ A-17252 -- 0.3167 32.5833 ENM only H. cinnamomeoventris UG MVZ 233827 -- 0.3704 32.5888 ENM only H. cinnamomeoventris UG ROM 11517 -- 1.7500 31.5833 ENM only H. cinnamomeoventris UGWE N/A SL326 0.8000 31.0667 yellow circle GU443995 MF376359 MF376474 H. dintelmanni CMLI CAS 253991 DMP925 4.8498 9.7718 blue diamond MF377045 MF377145 H. dintelmanni CMSW CAS 254135 DMP1083 4.9606 9.6537 blue diamond MF377036 MF377136 H. dintelmanni CMSW CAS 254136 DMP1084 4.9606 9.6537 blue diamond MF377037 MF377137 H. dintelmanni CMSW CAS 254137 DMP1085 4.9606 9.6537 blue diamond MF377038 MF377138 H. dintelmanni CMSW CAS 254156 DMP1105 4.9596 9.6524 blue diamond MF377039 MF377139 H. dintelmanni CMSW CAS 254157 DMP1106 4.9596 9.6524 blue diamond MF377040 MF377140 H. dintelmanni CMSW CAS 254158 DMP1107 4.9596 9.6524 blue diamond MF377041 MF377141 H. dintelmanni CMWE MVZ 234791 -- 5.2817 9.9760 blue diamond MF377074 MF377223 MF377174 MF377297 MF377337 H. drewesi PRBA CAS 219203 -- 1.6009 7.3531 red diamond KP137119 KJ866004 KJ866044 KJ865966 KJ865935 H. drewesi PRCD CAS 219128 -- 1.6521 7.4161 red diamond KP137130 KJ866005 KJ866045 KJ865967 KJ865936 H. drewesi PRPR CAS 233492 -- 1.6259 7.4166 red diamond KP137135 KJ866006 KJ866046 KJ865968 KJ865937 H. hutsebauti BIRP UTEP 21452 ELI994 -4.0108 30.1468 yellow circle MF377059 MF377159 MF377290 MF377331 H. hutsebauti BIRP UTEP 21453 ELI997 -4.0108 30.1468 yellow circle MF377060 MF377160 MF377291 MF377332 H. hutsebauti CD CAS 196165 -- 1.3992 28.5817 ENM only H. hutsebauti CD MCZ A-26330 -- 2.5000 27.6167 ENM only H. hutsebauti CD RBINS ZTN:KG383 -- -0.2150 25.8040 ENM only H. hutsebauti CDBP RBINS:ZTN:A358 A358 -2.7531 20.3781 yellow triangle MF376995 MF377094 MF377243 MF377305 H. hutsebauti CDBP RBINS:ZTN:A415 A415 -2.7531 20.3781 yellow triangle MF376996 MF377095 MF377244 MF377306 H. hutsebauti CDKP UTEP 21454 EBG2922 -7.7149 29.7696 yellow circle MF377053 MF377153 MF377284 MF377326 H. hutsebauti CDKP UTEP 21455 EBG2973 -7.7149 29.7696 yellow circle MF377054 MF377154 MF377285 MF377327 H. hutsebauti CDKP UTEP 21456 ELI169 -8.7190 27.4227 yellow circle MF377057 MF377157 MF377288 MF377330 H. hutsebauti CDOR RBINS:ZTN:CRT3574 CRT3574 0.8063 24.2839 yellow circle MF377031 MF377131 MF377274 MF377319 H. hutsebauti CDOR RBINS:ZTN:CRT3832 CRT3832 2.0335 22.7869 yellow circle MF377032 MF377132 MF377275 MF377320 H. hutsebauti CDOR RBINS:ZTN:CRT4073 CRT4073 1.2702 23.7311 yellow circle MF377033 MF377133 MF377276 MF377321 H. hutsebauti CDOR RBINS:ZTN:KG125 KG125 1.5270 25.2960 yellow circle MF377061 MF377161 MF377292 MF377333 H. hutsebauti CDOR RBINS:ZTN:KG158 KG158 0.5186 25.2603 yellow circle MF377062 MF377162 H. hutsebauti CDOR RBINS:ZTN:KG286 KG286 0.5390 25.9020 yellow circle MF377063 MF377163 MF377293 H. hutsebauti CDOR RBINS:ZTN:KG287 KG287 0.5390 25.9020 yellow circle MF377064 MF377164 H. hutsebauti CDOR UTEP 21457 EBG2318 1.2433 29.6933 yellow circle MF377050 MF377150 MF377281 H. hutsebauti CDOR UTEP 21458 EBG2526 1.4007 28.5688 yellow circle MF377051 MF377151 MF377282 MF377325 H. hutsebauti CDSK UTEP 21451 EBG1996 -2.8667 29.0362 yellow circle MF377049 MF377149 MF377280 MF377324 H. hutsebauti CDSK UTEP 21459 CFS1504 -3.1258 28.4150 yellow circle MF377030 MF377130 MF377273 H. hutsebauti CDSK UTEP 21460 EBG1506 -2.2078 28.6296 yellow circle MF377046 MF377146 MF377277 MF377322 H. hutsebauti CDSK UTEP 21461 EBG1657 -3.3734 28.6431 yellow circle MF377047 MF377147 MF377278 MF377323 H. hutsebauti CDSK UTEP 21462 EBG1678 -3.4039 28.5866 yellow circle MF377048 MF377148 MF377279 H. hutsebauti CDSK UTEP 21463 EBG2736 -3.0288 28.2826 yellow circle MF377052 MF377152 MF377283 H. hutsebauti CDSK UTEP 21464 ELI1305 -4.1078 29.0972 yellow circle MF377055 MF377155 MF377286 MF377328 H. hutsebauti CDSK UTEP 21465 ELI1438 -4.0901 28.1531 yellow circle MF377056 MF377156 MF377287 MF377329 H. hutsebauti CDSK UTEP 21466 ELI569 -3.0368 28.4223 yellow circle MF377058 MF377158 MF377289 H. molleri STBS CAS 219055 -- 0.2885 6.6031 red circle KP137146 KJ866007 KJ866047 KJ865969 KJ865938 H. molleri STCA CAS 218850 -- 0.2980 6.7304 red circle KP137157 KJ866008 KJ866048 KJ865970 KJ865939 H. molleri STJA CAS 218974 -- 0.2611 6.6509 red circle KP137174 KJ866009 KJ866049 KJ865971 KJ865940 H. ocellatus CD CAS 145289 -- -3.3377 23.2755 ENM only H. ocellatus CD CAS 196183 -- 1.3983 28.5736 ENM only H. ocellatus CDBP N/A JK713 -2.5854 20.1057 yellow triangle MF376571 MF376707 MF376890 Species Locality Catalog No. Field No. Lat Long Clade 16s Cyt b Cmyc Pomc Rag1 H. ocellatus CDBP N/A JK716 -2.5854 20.1057 yellow triangle MF376572 MF376708 MF376891 H. ocellatus CDBP N/A JK717 -2.5854 20.1057 yellow triangle MF376573 MF376709 MF376892 H. ocellatus CDOR CU 15079 -- 0.6326 25.2573 yellow circle MF376556 MF376827 MF376692 H. ocellatus CDOR CU 15080 -- 0.6326 25.2573 yellow circle MF376557 MF376828 MF376693 H. ocellatus CDOR CU 15081 -- 0.6326 25.2573 yellow circle MF376558 MF376829 MF376694 H. ocellatus CDOR CU 15082 -- 0.6326 25.2573 yellow circle MF376559 MF376830 MF376695 MF376886 MF376959 H. ocellatus CDOR CU 15083 -- 0.6326 25.2573 yellow circle MF376560 MF376831 MF376696 H. ocellatus CDOR CU 15087 -- 0.6326 25.2573 yellow circle MF376561 MF376832 MF376697 MF376887 MF376960 H. ocellatus CDOR RBINS:ZTN:KG104 KG104 1.5270 25.2960 yellow circle MF376574 MF376710 MF376893 MF376964 H. ocellatus CDOR RBINS:ZTN:KG105 KG105 1.5270 25.2960 yellow circle MF376575 MF376711 MF376894 MF376965 H. ocellatus CDOR RBINS:ZTN:KG106 KG106 1.5270 25.2960 yellow circle MF376576 MF376712 MF376895 H. ocellatus CDOR RBINS:ZTN:KG166 KG166 0.5390 25.9020 yellow circle MF376577 MF376896 H. ocellatus CDOR RBINS:ZTN:KG248 KG248 0.5390 25.9020 yellow circle MF376578 MF376713 MF376897 MF376966 H. ocellatus CDOR RBINS:ZTN:KG249 KG249 0.5390 25.9020 yellow circle MF376579 MF376714 MF376898 MF376967 H. ocellatus CDOR RBINS:ZTN:KG361 KG361 -0.2150 25.8040 yellow circle MF376580 MF376715 MF376899 MF376968 H. ocellatus CDOR RBINS:ZTN:KG362 KG362 -0.2150 25.8040 yellow circle MF376581 MF376716 MF376900 MF376969 H. ocellatus CDOR RBINS:ZTN:KG363 KG363 -0.2150 25.8040 yellow circle MF376582 MF376717 MF376901 H. ocellatus CDOR UTEP 21447 EBG2496 1.4007 28.5688 yellow circle MF376568 MF376834 MF376704 H. ocellatus CDOR UTEP 21448 EBG2597 1.2455 28.3434 yellow circle MF376569 MF376705 MF376888 MF376962 H. ocellatus CDSK UTEP 21449 EBG1318 -1.8873 28.4495 yellow circle MF376567 MF376833 MF376703 MF376961 H. ocellatus CDSK UTEP 21450 EBG2763 -3.0229 28.2803 yellow circle MF376570 MF376706 MF376889 MF376963 H. ocellatus CM MCZ A-2644 -- 2.7830 10.5330 ENM only H. ocellatus CM MCZ A-2649 -- 3.0170 12.3670 ENM only H. ocellatus CM MVZ 234784 -- 5.3374 9.4739 ENM only H. ocellatus CMCE CAS 249970 BJE3085 4.6116 12.2254 yellow star MF376554 MF376825 MF376690 MF376957 H. ocellatus CMCE CAS 249971 -- 4.6116 12.2254 yellow star MF376555 MF376826 MF376691 MF376885 MF376958 H. ocellatus CMEA CAS 199210 -- 3.1906 12.8117 yellow diamond MF376548 MF376819 MF376684 H. ocellatus CMEA NMP6V 74560 VG09047 3.0900 13.8300 yellow star MF376628 MF376764 MF376915 MF376990 H. ocellatus CMEA NMP6V 74700 VG10142 2.4400 15.4300 yellow star MF376629 MF376765 MF376916 MF376991 H. ocellatus CMLI CAS 254057 DMP1003 4.8498 9.7718 red star MF376564 MF376700 H. ocellatus CMLI CAS 254058 DMP1004 4.8498 9.7718 red star MF376565 MF376701 H. ocellatus CMLI CAS 254075 DMP1016 4.8498 9.7718 red star MF376566 MF376702 H. ocellatus CMLI N/A MB365 4.8397 9.9303 red star MF376589 MF376724 H. ocellatus CMLI N/A MB366 4.8397 9.9303 red star MF376590 MF376836 MF376725 H. ocellatus CMLI N/A MB367 4.8397 9.9303 red star MF376591 MF376726 H. ocellatus CMLI N/A MB368 4.8397 9.9303 red star MF376592 MF376727 H. ocellatus CMLI N/A MB369 4.8397 9.9303 red star MF376593 MF376728 H. ocellatus CMLI N/A MB370 4.8397 9.9303 red star MF376594 MF376729 H. ocellatus CMLI N/A MB371 4.8397 9.9303 red star MF376595 MF376730 H. ocellatus CMLI N/A MB374 4.9172 9.9892 red star MF376596 MF376731 H. ocellatus CMLI N/A MB375 4.9172 9.9892 red star MF376597 MF376732 H. ocellatus CMLI N/A MB376 4.9172 9.9892 red star MF376598 MF376733 H. ocellatus CMLI N/A MB377 4.9172 9.9892 red star MF376599 MF376734 H. ocellatus CMSO CAS 253588 DCB571 3.1733 12.5290 yellow diamond MF376562 MF376698 H. ocellatus CMSO CAS 253592 DCB575 3.1733 12.5290 yellow diamond MF376563 MF376699 H. ocellatus CMSO N/A MB349 2.3972 10.0452 yellow star MF376583 MF376718 H. ocellatus CMSO N/A MB350 2.3972 10.0452 yellow star MF376584 MF376835 MF376719 H. ocellatus CMSO N/A MB352 2.3972 10.0452 yellow star MF376585 MF376720 H. ocellatus CMSO N/A MB354 2.3972 10.0452 yellow star MF376586 MF376721 H. ocellatus CMSO N/A MB355 2.3972 10.0452 yellow star MF376587 MF376722 H. ocellatus CMSO N/A MB357 2.3972 10.0452 yellow star MF376588 MF376723 H. ocellatus CMSW MCZ 136833 -- 5.6200 9.9200 red square MF376605 MF376837 MF376740 MF376905 MF376974 H. ocellatus CMSW N/A MM030 5.7284 9.2939 red square MF376606 MF376838 MF376741 H. ocellatus CMWE MVZ 234777 -- 5.0080 10.1789 red star MF376607 MF376839 MF376742 MF376906 MF376975 H. ocellatus CMWE MVZ 234779 -- 5.0080 10.1789 red star MF376608 MF376840 MF376743 MF376907 MF376976 H. ocellatus CMWE MVZ 234780 -- 5.0080 10.1789 red star MF376609 MF376841 MF376744 H. ocellatus CMWE MVZ 234781 -- 5.0080 10.1789 red star MF376610 MF376745 H. ocellatus CMWE MVZ 234782 -- 5.0080 10.1789 red star MF376611 MF376842 MF376746 MF376908 MF376977 H. ocellatus EGAB CU 15978 RCB0413 3.5258 8.5809 red triangle MF376622 MF376758 H. ocellatus EGAB CU 15979 RCB0414 3.5258 8.5809 red triangle MF376623 MF376759 H. ocellatus EGAB CU 15980 RCB0415 3.5258 8.5809 red triangle MF376624 MF376760 MF376911 MF376986 H. ocellatus EGAB CU 15981 RCB0416 3.5258 8.5809 red triangle MF376625 MF376761 MF376912 MF376987 H. ocellatus EGAB CU 15982 RCB0417 3.5258 8.5809 red triangle MF376626 MF376762 MF376913 MF376988 H. ocellatus EGAB CU 15983 RCB0418 3.5258 8.5809 red triangle MF376627 MF376763 MF376914 MF376989 H. ocellatus EGLU CAS 207784 -- 3.4830 8.5820 red triangle MF376549 MF376820 MF376685 MF376880 MF376953 H. ocellatus EGLU CAS 207785 -- 3.4830 8.5820 red triangle MF376550 MF376821 MF376686 MF376881 MF376954 H. ocellatus EGLU CAS 207794 -- 3.4830 8.5820 red triangle MF376551 MF376822 MF376687 MF376882 MF376955 H. ocellatus EGLU CAS 207795 -- 3.4830 8.5820 red triangle MF376552 MF376823 MF376688 MF376883 MF376956 Species Locality Catalog No. Field No. Lat Long Clade 16s Cyt b Cmyc Pomc Rag1 H. ocellatus EGMM CU 15231 RCB0197 3.4673 8.6411 red triangle MF376616 MF376848 MF376752 MF376910 MF376981 H. ocellatus EGMO CU 15204 RCB0167 3.4673 8.6411 red triangle MF376612 MF376843 MF376747 H. ocellatus EGMO CU 15205 RCB0168 3.4673 8.6411 red triangle MF376613 MF376844 MF376748 H. ocellatus EGMO CU 15206 RCB0169 3.4673 8.6411 red triangle MF376614 MF376845 MF376749 MF376978 H. ocellatus EGMO CU 15207 RCB0170 3.4673 8.6411 red triangle MF376846 MF376750 MF376979 H. ocellatus EGMO CU 15208 RCB0171 3.4673 8.6411 red triangle MF376615 MF376847 MF376751 MF376909 MF376980 H. ocellatus EGMO CU 15244 RCB0214 3.4673 8.6411 red triangle MF376617 MF376849 MF376753 MF376982 H. ocellatus EGMO CU 15245 RCB0217 3.4673 8.6411 red triangle MF376618 MF376850 MF376754 MF376983 H. ocellatus EGMO CU 15246 RCB0218 3.4673 8.6411 red triangle MF376619 MF376851 MF376755 MF376984 H. ocellatus EGMO CU 15248 RCB0221 3.4673 8.6411 red triangle MF376620 MF376852 MF376756 MF376985 H. ocellatus EGPB CAS 207829 -- 3.7052 8.8794 red triangle MF376553 MF376824 MF376689 MF376884 H. ocellatus EGRI CU 15957 RCB0396 3.3917 8.7625 red triangle MF376621 MF376757 H. ocellatus EGWN YPM-A8086 -- 1.1708 11.1284 yellow triangle MF376514 MF376633 H. ocellatus GA CU 14892 -- 0.4999 12.8018 ENM only H. ocellatus GA CU 14897 -- 0.4536 10.2781 ENM only H. ocellatus GA CU 14939 -- 0.5037 12.7941 ENM only H. ocellatus GA CU 14989 -- 0.5112 12.8028 ENM only H. ocellatus GA CU 14993 -- 0.2938 12.5662 ENM only H. ocellatus GA CU 15032 -- 0.2927 12.5739 ENM only H. ocellatus GA CU 15528 -- -0.0426 12.2983 ENM only H. ocellatus GA CU 15537 -- 12.3132 0.0565 ENM only H. ocellatus GAES CU 14895 BLS13532 0.4536 10.2781 yellow star MF376515 MF376769 MF376634 MF376856 MF376920 H. ocellatus GAES CU 14896 BLS13533 0.4536 10.2781 yellow star MF376516 MF376770 MF376635 MF376857 MF376921 H. ocellatus GAES CU 14904 BLS13590 0.4536 10.2781 yellow star MF376517 MF376771 MF376636 MF376858 MF376922 H. ocellatus GAOI CU 14891 BLS13826 0.4999 12.8018 yellow triangle MF376524 MF376779 MF376644 MF376863 MF376929 H. ocellatus GAOI CU 14892 BLS13827 0.4999 12.8018 yellow triangle MF376525 MF376780 MF376645 MF376864 MF376930 H. ocellatus GAOI CU 14893 BLS13828 0.4999 12.8018 yellow triangle MF376526 MF376781 MF376646 MF376865 MF376931 H. ocellatus GAOI CU 14939 BLS13771 0.5037 12.7941 yellow triangle MF376778 MF376643 H. ocellatus GAOI CU 14956 BLS13756 0.5162 12.7946 yellow triangle MF376518 MF376772 MF376637 MF376859 MF376923 H. ocellatus GAOI CU 14957 BLS13757 0.5162 12.7946 yellow triangle MF376519 MF376773 MF376638 MF376860 MF376924 H. ocellatus GAOI CU 14958 BLS13758 0.5162 12.7946 yellow triangle MF376520 MF376774 MF376639 MF376861 MF376925 H. ocellatus GAOI CU 14959 BLS13759 0.5162 12.7946 yellow triangle MF376521 MF376775 MF376640 MF376862 MF376926 H. ocellatus GAOI CU 14960 BLS13760 0.5162 12.7946 yellow triangle MF376522 MF376776 MF376641 MF376927 H. ocellatus GAOI CU 14961 BLS13761 0.5162 12.7946 yellow triangle MF376523 MF376777 MF376642 MF376928 H. ocellatus GAOI CU 14989 BLS14127 0.5112 12.8028 yellow triangle MF376543 MF376814 MF376679 MF376876 MF376949 H. ocellatus GAOI CU 14993 BLS14045 0.2938 12.5662 yellow triangle MF376782 MF376647 H. ocellatus GAOI CU 14994 BLS14046 0.2938 12.5662 yellow triangle MF376783 MF376648 H. ocellatus GAOI CU 14995 BLS14047 0.2938 12.5662 yellow star MF376784 MF376649 H. ocellatus GAOI CU 14996 BLS14048 0.2938 12.5662 yellow triangle MF376527 MF376785 MF376650 MF376932 H. ocellatus GAOI CU 14998 BLS14050 0.2938 12.5662 yellow star MF376528 MF376786 MF376651 MF376933 H. ocellatus GAOI CU 14999 BLS14051 0.2938 12.5662 yellow triangle MF376787 MF376652 H. ocellatus GAOI CU 15000 BLS14052 0.2938 12.5662 yellow star MF376788 MF376653 H. ocellatus GAOI CU 15001 BLS14053 0.2938 12.5662 yellow star MF376789 MF376654 H. ocellatus GAOI CU 15002 BLS14054 0.2938 12.5662 yellow triangle MF376529 MF376790 MF376655 MF376866 MF376934 H. ocellatus GAOI CU 15003 BLS14055 0.2938 12.5662 yellow triangle MF376791 MF376656 MF376935 H. ocellatus GAOI CU 15004 BLS14056 0.2938 12.5662 yellow star MF376530 MF376792 MF376657 MF376867 MF376936 H. ocellatus GAOI CU 15005 BLS14057 0.2938 12.5662 yellow triangle MF376793 MF376658 H. ocellatus GAOI CU 15006 BLS14058 0.2938 12.5662 yellow star MF376531 MF376794 MF376659 MF376937 H. ocellatus GAOI CU 15007 BLS14059 0.2938 12.5662 yellow triangle MF376795 MF376660 H. ocellatus GAOI CU 15008 BLS14060 0.2938 12.5662 yellow triangle MF376796 MF376661 MF376938 H. ocellatus GAOI CU 15010 BLS14062 0.2938 12.5662 yellow star MF376797 MF376662 H. ocellatus GAOI CU 15011 BLS14064 0.2938 12.5662 yellow star MF376532 MF376798 MF376663 MF376939 H. ocellatus GAOI CU 15013 BLS14067 0.2938 12.5662 yellow triangle MF376533 MF376799 MF376664 MF376868 MF376940 H. ocellatus GAOI CU 15014 BLS14068 0.2938 12.5662 yellow triangle MF376534 MF376800 MF376665 MF376941 H. ocellatus GAOI CU 15015 BLS14070 0.2938 12.5662 yellow triangle MF376801 MF376666 H. ocellatus GAOI CU 15016 BLS14071 0.2938 12.5662 yellow triangle MF376535 MF376802 MF376667 MF376869 MF376942 H. ocellatus GAOI CU 15017 BLS14073 0.2938 12.5662 yellow star MF376803 MF376668 H. ocellatus GAOI CU 15018 BLS14074 0.2938 12.5662 yellow triangle MF376536 MF376804 MF376669 MF376870 MF376943 H. ocellatus GAOI CU 15019 BLS14076 0.2938 12.5662 yellow triangle MF376537 MF376805 MF376670 H. ocellatus GAOI CU 15020 BLS14077 0.2938 12.5662 yellow star MF376538 MF376806 MF376671 MF376871 MF376944 H. ocellatus GAOI CU 15032 BLS14091 0.2927 12.5739 yellow triangle MF376539 MF376807 MF376672 MF376872 MF376945 H. ocellatus GAOI CU 15033 BLS14092 0.2927 12.5739 yellow triangle MF376540 MF376808 MF376673 MF376873 MF376946 H. ocellatus GAOI CU 15034 BLS14093 0.2927 12.5739 yellow triangle MF376809 MF376674 H. ocellatus GAOI CU 15035 BLS14094 0.2927 12.5739 yellow triangle MF376541 MF376810 MF376675 MF376874 MF376947 H. ocellatus GAOI CU 15036 BLS14095 0.2927 12.5739 yellow triangle MF376542 MF376811 MF376676 MF376875 MF376948 H. ocellatus GAOI CU 15037 BLS14096 0.2927 12.5739 yellow triangle MF376812 MF376677 H. ocellatus GAOI CU 15038 BLS14097 0.2927 12.5739 yellow triangle MF376813 MF376678 H. ocellatus GAOI CU 15528 BLS14770 -0.0426 12.2983 yellow triangle MF376544 MF376815 MF376680 MF376877 MF376950 Species Locality Catalog No. Field No. Lat Long Clade 16s Cyt b Cmyc Pomc Rag1 H. ocellatus GAOI CU 15535 BLS14786 -0.0426 12.2983 yellow triangle MF376545 MF376816 MF376681 MF376878 MF376951 H. ocellatus GAOI CU 15536 BLS14787 -0.0426 12.2983 yellow triangle MF376546 MF376817 MF376682 MF376879 MF376952 H. ocellatus GAOI CU 15897 BLS14842 -0.0426 12.2983 yellow triangle MF376547 MF376818 MF376683 H. ocellatus RCCO NMP6V 77515/1 VGCG12096 0.0600 14.2400 yellow triangle MF376630 MF376853 MF376766 MF376917 MF376992 H. ocellatus RCCO NMP6V 77515/2 VGCG12097 0.0600 14.2400 yellow triangle MF376631 MF376854 MF376767 MF376918 MF376993 H. ocellatus RCCO NMP6V 77515/5 VGCG12100 0.0600 14.2400 yellow triangle MF376632 MF376855 MF376768 MF376919 MF376994 H. ocellatus RCNI IRSEN 0313 MBUR03624 -2.2106 12.8324 yellow hexagon MF376601 MF376736 MF376903 MF376971 H. ocellatus RCNI IRSEN 0316 MBUR03932 -2.4061 12.8398 yellow hexagon MF376604 MF376739 H. ocellatus RCNI ZMB 83941 MBUR03623 -2.2106 12.8324 yellow hexagon MF376600 MF376735 MF376902 MF376970 H. ocellatus RCNI ZMB 83942 MBUR03716 -2.3267 12.7924 yellow hexagon MF376602 MF376737 MF376904 MF376972 H. ocellatus RCNI ZMB 83943 MBUR03717 -2.3267 12.7924 yellow hexagon MF376603 MF376738 MF376973 H. olivaceus CDBC RBINS:ZTN:PM043 PM043 -5.9200 12.3500 orange hexagon MF376171 MF376397 MF376354 H. olivaceus CDBC RBINS:ZTN:PM044 PM044 -5.9200 12.3500 orange hexagon MF376172 MF376398 MF376355 H. olivaceus CDBC RBINS:ZTN:PM055 PM055 -5.8900 12.7700 orange hexagon MF376399 MF376356 MF376403 H. olivaceus CDBC RBINS:ZTN:PM056 PM056 -5.8900 12.7700 orange hexagon MF376173 KJ865999 KJ866039 KJ865963 H. olivaceus CDBC RBINS:ZTN:PM057 PM057 -5.8900 12.7700 orange hexagon MF376174 MF376400 MF376357 H. olivaceus CDBC RBINS:ZTN:PM058 PM058 -5.8900 12.7700 orange hexagon MF376175 KJ866000 KJ866040 KJ865964 KJ865933 H. olivaceus CDBC RBINS:ZTN:PM059 PM059 -5.8900 12.7700 orange hexagon MF376176 MF376401 MF376358 MF376404 H. olivaceus GA CU 15545 -- -0.0426 12.2983 ENM only H. olivaceus GA CU 15572 -- -0.1956 12.1960 ENM only H. olivaceus GA CU 15573 -- -0.0955 12.3212 ENM only H. olivaceus GAES USNM 578128 -- 0.6030 9.3373 orange star MF376203 KJ865983 KJ866021 H. olivaceus GAES USNM 578129 -- 0.6030 9.3373 orange star KJ865984 KJ866022 H. olivaceus GAES USNM 578131 -- 0.6030 9.3373 orange star MF376204 MF376392 MF376349 H. olivaceus GAES USNM 578134 -- 0.6030 9.3373 orange star MF376205 MF376393 MF376350 H. olivaceus GAES USNM 578138 -- 0.5736 9.3384 orange star MF376206 MF376396 MF376353 H. olivaceus GAMO CAS 254482 BLS16217 -0.6861 10.2281 orange star MF376381 MF376302 H. olivaceus GAMO CAS 254490 BLS16229 -1.1237 10.0283 orange star MF376209 KJ866026 KJ865953 KJ865924 H. olivaceus GAMO CAS 254491 BLS16230 -1.1237 10.0283 orange star MF376210 MF376382 MF376303 MF376448 H. olivaceus GAMO CAS 254547 BLS16306 -1.1154 10.0235 orange star MF376211 MF376383 MF376304 MF376449 H. olivaceus GAMO CAS 254589 BLS16372 -1.1086 10.0303 orange star MF376215 MF376387 MF376308 MF376453 MF376495 H. olivaceus GAMO CAS 254605 BLS16393 -1.1001 10.0276 orange star MF376216 MF376388 MF376309 MF376454 MF376496 H. olivaceus GAMO CAS 254606 BLS16394 -1.1001 10.0276 orange star MF376217 MF376389 MF376310 MF376455 MF376497 H. olivaceus GAMO NCSM 81280 BLS16215 -0.6861 10.2281 orange star KJ865985 KJ866023 KJ865950 KJ865921 H. olivaceus GAMO NCSM 81281 BLS16216 -0.6861 10.2281 orange star MF376207 KJ865986 KJ866024 KJ865951 KJ865922 H. olivaceus GAMO NCSM 81282 BLS16228 -1.1237 10.0283 orange star MF376208 KJ865987 KJ866025 KJ865952 KJ865923 H. olivaceus GAMO NCSM 81287 BLS16358 -0.6861 10.2281 orange star MF376212 MF376384 MF376305 MF376450 H. olivaceus GAMO NCSM 81288 BLS16370 -1.1086 10.0303 orange star MF376213 MF376385 MF376306 MF376451 H. olivaceus GAMO NCSM 81289 BLS16371 -1.1086 10.0303 orange star MF376214 MF376386 MF376307 MF376452 H. olivaceus GANP USNM 578115 -- -2.7868 10.0455 orange star MF376218 KJ865989 KJ866034 H. olivaceus GANP USNM 578116 -- -2.7868 10.0455 orange star MF376219 KJ865990 KJ866035 H. olivaceus GANP USNM 578117 -- -2.7868 10.0455 orange star MF376220 MF405234 MF376348 H. olivaceus GANP USNM 578136 -- -2.7868 10.0455 orange star MF376221 MF376394 MF376351 H. olivaceus GANP USNM 578137 -- -2.7356 10.0144 orange star MF376395 MF376352 H. olivaceus GAOI CU 15495 BLS14714 -0.2095 12.2905 orange star MF376236 KJ865994 KJ866031 KJ865957 KJ865928 H. olivaceus GAOI CU 15496 BLS14715 -0.2095 12.2905 orange star KP137115 KP137231 MF376293 MF376439 MF376491 H. olivaceus GAOI CU 15498 BLS14717 -0.2095 12.2905 orange star MF376237 KJ865995 KJ866032 KJ865958 KJ865929 H. olivaceus GAOI CU 15499 BLS14718 -0.2095 12.2905 orange star MF376238 MF376374 MF376294 MF376440 H. olivaceus GAOI CU 15514 BLS14740 -0.2095 12.2889 orange star KP137116 KP137232 MF376295 MF376441 MF376492 H. olivaceus GAOI CU 15515 BLS14741 -0.1956 12.1960 orange star MF376239 MF376375 MF376296 MF376442 H. olivaceus GAOI CU 15516 BLS14742 -0.1956 12.1960 orange star MF376240 MF376376 MF376297 MF376443 H. olivaceus GAOI CU 15517 BLS14743 -0.1956 12.1960 orange star MF376241 MF376377 MF376298 MF376444 H. olivaceus GAOI CU 15518 BLS14744 -0.1956 12.1960 orange star MF376242 KJ865996 KJ866033 KJ865959 KJ865930 H. olivaceus GAOI CU 15545 BLS14796 -0.0955 12.3212 orange star MF376243 MF376378 MF376299 MF376445 H. olivaceus GAOI CU 15553 BLS14830 -0.1956 12.1960 orange star MF376244 MF376379 MF376300 MF376446 MF376493 H. olivaceus GAOI CU 15554 BLS14831 -0.0426 12.2983 orange star MF376380 MF376301 MF376447 MF376494 H. olivaceus GAOM CU 15088 BLS14227 -1.8098 9.3877 orange star MF376245 H. olivaceus GAOM CU 15089 BLS14228 -1.8098 9.3877 orange star MF376246 H. olivaceus GAOM CU 15092 BLS14236 -1.8140 9.3556 orange star KP137117 KJ865997 KJ866027 KJ865960 H. olivaceus GAOM CU 15105 BLS14257 -1.8914 9.5682 orange star MF376247 KJ865998 KJ866036 KJ865961 H. olivaceus RCLE USNM 584159 FSKJ246971 -2.7942 13.5350 orange hexagon MF376252 KJ866001 KJ866041 H. olivaceus RCLE USNM 584160 FSKJ246979 -2.7942 13.5350 orange hexagon MF376253 KJ866002 KJ866042 H. olivaceus RCLE USNM 584161 FSKJ246989 -2.7942 13.5350 orange hexagon MF376254 MF376391 MF376324 H. olivaceus RCNI ZMB 83938 MBUR03663 -2.4013 12.7502 orange hexagon MF376261 MF376341 MF376467 H. olivaceus RCNI ZMB 83939 MBUR03666 -2.4013 12.7502 orange hexagon MF376262 MF376342 MF376468 H. olivaceus RCNI ZMB 83940 MBUR03867 -2.5134 12.9252 orange hexagon MF376266 MF376346 MF376472 H. riggenbachi CM MVZ 234752 -- 6.0674 10.4581 Hocel outgroup GU443976 H. thomensis STBS CAS 218929 -- 0.2761 6.6056 red triangle KP137219 KJ866012 KJ866050 KJ865972 Species Locality Catalog No. Field No. Lat Long Clade 16s Cyt b Cmyc Pomc Rag1 H. thomensis STBS CAS 218934 -- 0.2761 6.6056 red triangle KP137212 KJ866010 KJ866051 KJ865973 H. thomensis STBS CAS 233475 -- 0.2761 6.6056 red triangle KP137215 KJ866011 KJ866052 KJ865974 KJ865941 H. tuberculatus CM CAS 199192 -- 3.1906 12.8117 ENM only H. tuberculatus CM MVZ 234788 -- 2.9663 12.0816 ENM only H. tuberculatus CMCE CAS 249988 -- 4.6041 12.2045 red star MF377028 MF377222 MF377128 MF377271 MF377317 H. tuberculatus CMCE CAS 249989 -- 4.6041 12.2045 red star MF377029 MF377129 MF377272 MF377318 H. tuberculatus CMCE NMP6V 74561 VG09060 3.3913 11.4663 red star MF377093 MF377197 H. tuberculatus CMEA CAS 253397 DCB372 2.5946 14.0283 red star MF377034 MF377134 H. tuberculatus CMEA CAS 253400 DCB375 2.5946 14.0283 red star MF377035 MF377135 H. tuberculatus CMSO N/A MB378 2.3972 10.0452 red star MF377065 MF377165 H. tuberculatus CMSO N/A MB379 2.3972 10.0452 red star MF377066 MF377166 H. tuberculatus CMSO N/A MB380 2.3972 10.0452 red star MF377067 MF377167 H. tuberculatus CMSO N/A MB381 2.3972 10.0452 red star MF377068 MF377168 H. tuberculatus CMSO N/A MB382 2.3972 10.0452 red square MF377069 MF377169 H. tuberculatus CMSO N/A MB384 2.3972 10.0452 red square MF377070 MF377170 H. tuberculatus CMWE CAS 254219 DMP1174 5.1530 10.5325 red star MF377042 MF377142 H. tuberculatus CMWE CAS 254220 DMP1175 5.1530 10.5325 red star MF377043 MF377143 H. tuberculatus CMWE CAS 254224 DMP1179 5.1530 10.5325 red star MF377044 MF377144 H. tuberculatus EG CU 15911 -- 3.3524 8.6374 ENM only H. tuberculatus EG CU 15917 -- 3.3528 8.6300 ENM only H. tuberculatus EG CU 15928 -- 3.3525 8.6328 ENM only H. tuberculatus EG CU 15973 -- 3.4037 8.6698 ENM only H. tuberculatus EGLB CAS 207703 -- 3.3554 8.6215 red triangle MF377017 MF377211 MF377117 MF377260 H. tuberculatus EGLB CAS 207704 -- 3.3554 8.6215 red triangle MF377018 MF377212 MF377118 MF377261 MF377314 H. tuberculatus EGLB CAS 207713 -- 3.3554 8.6215 red triangle MF377019 MF377213 MF377119 MF377262 MF377315 H. tuberculatus EGLB CAS 207714 -- 3.3554 8.6215 red triangle MF377020 MF377214 MF377120 MF377263 MF377316 H. tuberculatus EGLB CAS 207715 -- 3.3554 8.6215 red triangle MF377021 MF377215 MF377121 MF377264 H. tuberculatus EGLB CAS 207716 -- 3.3554 8.6215 red triangle MF377022 MF377216 MF377122 MF377265 H. tuberculatus EGLB CAS 207717 -- 3.3554 8.6215 red triangle MF377023 MF377217 MF377123 MF377266 H. tuberculatus EGLB CAS 207736 -- 3.3554 8.6215 red triangle MF377024 MF377218 MF377124 MF377267 H. tuberculatus EGLB CAS 207737 -- 3.3554 8.6215 red triangle MF377025 MF377219 MF377125 MF377268 H. tuberculatus EGLB CAS 207738 -- 3.3554 8.6215 red triangle MF377026 MF377220 MF377126 MF377269 H. tuberculatus EGLB CAS 207739 -- 3.3554 8.6215 red triangle MF377027 MF377221 MF377127 MF377270 H. tuberculatus EGLB CU 15144 RCB0016 3.3525 8.6370 red triangle MF377083 MF377234 MF377186 MF377299 H. tuberculatus EGLB CU 15147 RCB0017 3.3531 8.6307 red triangle MF377084 MF377235 MF377187 MF377300 MF377340 H. tuberculatus EGLB CU 15148 RCB0018 3.3531 8.6307 red triangle MF377236 MF377188 MF377301 MF377341 H. tuberculatus EGLB CU 15149 RCB0019 3.3531 8.6307 red triangle MF377237 MF377189 H. tuberculatus EGLB CU 15150 RCB0020 3.3531 8.6307 red triangle MF377085 MF377238 MF377190 MF377302 MF377342 H. tuberculatus EGLB CU 15154 RCB0024 3.3531 8.6307 red triangle MF377086 MF377239 MF377191 MF377303 MF377343 H. tuberculatus EGLL CU 15145 RCB0026 3.3521 8.6340 red triangle MF377087 MF377240 MF377192 H. tuberculatus EGLL CU 15146 RCB0027 3.3525 8.6363 red triangle MF377088 MF377241 MF377193 H. tuberculatus EGLL CU 15969 RCB0404 3.4033 8.6674 red triangle MF377089 MF377194 H. tuberculatus EGLL CU 15970 RCB0405 3.4033 8.6674 red triangle MF377090 MF377195 H. tuberculatus EGLL CU 15972 RCB0407 3.4038 8.6691 red triangle MF377091 MF377196 MF377304 H. tuberculatus EGMM CU 15136 RCB0006 3.3639 8.6580 red triangle MF377082 MF377233 MF377185 MF377298 MF377339 H. tuberculatus EGWN YPM-A8062 -- 1.1708 11.1284 red star MF376997 MF377096 H. tuberculatus GA NCSM 81294 -- -1.1433 10.0047 ENM only H. tuberculatus GAES CU 14908 BLS13669 0.6212 10.4076 red star MF376998 MF377198 MF377097 MF377245 MF377307 H. tuberculatus GAES CU 14909 BLS13670 0.6212 10.4076 red star MF376999 MF377199 MF377098 MF377246 H. tuberculatus GAES CU 14910 BLS13671 0.6212 10.4076 red star MF377000 MF377200 MF377099 MF377247 H. tuberculatus GAES CU 14911 BLS13672 0.6212 10.4076 red star MF377201 MF377100 H. tuberculatus GAES CU 14912 BLS13673 0.6212 10.4076 red star MF377001 MF377202 MF377101 MF377248 H. tuberculatus GAES CU 14913 BLS13674 0.6212 10.4076 red star MF377002 MF377203 MF377102 MF377249 MF377308 H. tuberculatus GAES CU 14914 BLS13675 0.6212 10.4076 red star MF377003 MF377204 MF377103 H. tuberculatus GAES CU 14915 BLS13676 0.6212 10.4076 red star MF377004 MF377205 MF377104 MF377250 H. tuberculatus GAES USNM 578127 -- 0.6030 9.3373 red star MF377078 MF377227 MF377178 H. tuberculatus GAES USNM 578183 -- 0.6030 9.3373 red star MF377079 MF377179 H. tuberculatus GAES USNM 578184 -- 0.6030 9.3373 red star MF377080 MF377228 MF377180 H. tuberculatus GAES USNM 578185 -- 0.6030 9.3373 red star MF377229 MF377181 H. tuberculatus GAES USNM 578186 -- 0.6030 9.3373 red star MF377230 MF377182 H. tuberculatus GAES USNM 578187 -- 0.6030 9.3373 red star MF377081 MF377231 MF377183 H. tuberculatus GAES USNM 578188 -- 0.6030 9.3373 red star MF377232 MF377184 H. tuberculatus GALO CAS 254546 BLS16305 -1.1078 10.0269 red square MF377010 MF377110 MF377255 MF377312 H. tuberculatus GALO CAS 254559 BLS16327 -1.1100 10.0278 red square MF377012 MF377112 MF377256 MF377313 H. tuberculatus GALO CAS 254569 BLS16341 -1.1403 10.0081 red square MF377013 MF377113 MF377257 H. tuberculatus GALO CAS 254570 BLS16342 -1.1403 10.0081 red square MF377014 MF377114 MF377258 H. tuberculatus GALO CAS 254599 BLS16386 -1.1078 10.0269 red square MF377015 MF377115 H. tuberculatus GALO CAS 254600 BLS16387 -1.1078 10.0269 red square MF377016 MF377116 MF377259 Species Locality Catalog No. Field No. Lat Long Clade 16s Cyt b Cmyc Pomc Rag1 H. tuberculatus GALO NCSM 81292 BLS16326 -1.1100 10.0278 red square MF377011 MF377111 H. tuberculatus GAOI CU 15513 BLS14739 -0.1956 12.1960 red star MF377005 MF377206 MF377105 MF377251 MF377309 H. tuberculatus GAOI CU 15538 BLS14789 -0.1956 12.1960 red star MF377006 MF377207 MF377106 MF377252 MF377310 H. tuberculatus GAOI CU 15539 BLS14790 -0.1956 12.1960 red star MF377007 MF377208 MF377107 MF377253 H. tuberculatus GAOI CU 15540 BLS14791 -0.1956 12.1960 red star MF377008 MF377209 MF377108 MF377254 MF377311 H. tuberculatus GAOI CU 15577 BLS14849 -0.1956 12.1960 red star MF377009 MF377210 MF377109 H. tuberculatus GAOM USNM 578122 -- -2.7296 10.0188 red star MF377075 MF377224 MF377175 H. tuberculatus GAOM USNM 578123 -- -2.7296 10.0188 red star MF377076 MF377225 MF377176 MF377338 H. tuberculatus GAOM USNM 578126 -- -2.7296 10.0188 red star MF377077 MF377226 MF377177 H. tuberculatus RCNI IRSEN 0315 MBUR03882 -2.5134 12.9252 red star MF377072 MF377172 MF377295 MF377335 H. tuberculatus RCNI ZMB 83944 MBUR03868 -2.5134 12.9252 red star MF377071 MF377171 MF377294 MF377334 H. tuberculatus RCNI ZMB 83945 MBUR03991 -2.4164 12.8699 red star MF377073 MF377173 MF377296 MF377336 H. veithi CDBP N/A JK692 -2.7531 20.3781 green square MF376270 MF376325 MF376456 H. veithi CDBP N/A JK693 -2.7531 20.3781 green square MF376269 MF376326 MF376457 MF376506 H. veithi CDBP N/A JK702 -2.7531 20.3781 green square MF376271 MF376327 MF376458 H. veithi CDBP N/A JK703 -2.7931 20.3568 green square MF376272 MF376328 MF376459 MF376507 H. veithi CDBP N/A JK704 -2.7931 20.3568 green square MF376273 MF376329 MF376460 MF376508 H. veithi CDBP RBINS:ZTN:A519 A519 -2.8800 20.4100 green square GU443988 MF376360 H. veithi CDBP RBINS:ZTN:A520 A520 -2.8800 20.4100 green square GU443975 MF376361 H. viridiflavus KE N/A SL0250 0.3522 34.8647 Htube outgroup MF377092 MF377242 Table S2 Primer sequences and amplification conditions for mitochondrial and nuclear sequences collected from Hyperolius cinnamomeoventris, H. ocellatus, and H. tuberculatus. PCR Annealing Temperature Primer Sequence Hc Ho Ht Reference 16S A-L 5’ CGC CTG TTT ATC AAA AAC AT 3’ 50* 50* 50* (Palumbi et al. 1991) 16S B-H 5’ CCC GTC TGA ACT CAG ATC ACG T 3’ (Palumbi et al. 1991) MVZ15 5' GAA CTA ATG GCC CAC ACW WTA CG 3' 43* 43* 43* (Moritz et al. 1992) MVZ16 5' AAA TAG GAA RTA TCA YTC TGG TTT RAT 3' (Moritz et al. 1992) CMYC 1U 5' GAG GAC ATC TGG AAR AAR TT 3' 48 48 48 (Crawford 2003) CMYC ex2dR 5' TCA TTC AAT GGG TAA GGG AAG ACC 3' (Wiens et al. 2005) POMC1 5' GAA TGT ATY AAA GMM TGC AAG ATG GWC CT 3' 55* 54* 54* (Wiens et al. 2005) POMC2 5' TAY TGR CCC TTY TTG TGG GCR TT 3' (Wiens et al. 2005) RAG1 F 5' GCC AGA TCT TTC ARC CAC TC 3' 55* 54** 55** (Lawson et al. 2015) RAG1 R 5' TGA TCT CTG GAA CRT GGG CTA 3' (Lawson et al. 2015) * indicates 0.3 µl of additional MgCl per reaction ** indicates 0.2 µl of additional MgCl per reaction
Crawford AJ (2003) Huge populations and old species of Costa Rican and Panamanian dirt frogs inferred from mitochondrial and nuclear gene sequences. Molecular Ecology, 12, 2525–2540. Lawson LP, Bates JM, Menegon M, Loader SP (2015) Divergence at the edges: peripatric isolation in the montane spiny throated reed frog complex. BMC Evolutionary Biology, 15, 1–15. Moritz C, Schneider CJ, Wake DB (1992) Evolutionary relationships within the Ensatina eschscholtzii complex confirm the ring species interpretation. Systematic Biology, 41, 273–291 Palumbi SR, Martin A, Romano S et al. (1991) The Simple Fool's Guide to PCR, Version 2.0. Privately Published, University of Hawaii. Wiens JJ, Fetzner J Jr., Parkinson CL, Reeder TW (2005) Hylid Frog Phylogeny and Sampling Strategies for Speciose Clades. Systematic Biology, 54, 778–807. Table S3 A-C: Substitution models and summary statistics for mitochondrial and nuclear loci by species complex. Abbreviations as follow: number of individuals sampled (N), sequence length (bp), variable sites (S), and nucleotide diversity (θs, θπ). None of the best-fit substitution models are partitioned by codon position.
A. Hyperolius cinnamomeoventris complex. H. c. cinnamomeoventris (Hcc), H. c. olivaceus (Hco), H. veithi (Hv), H. thomensis, H. molleri and H. drewesi (island).
Locus Model Lineage N bp S θs/bp θπ /bp Hcc 61 476 76 0.013 0.021 Hco 46 476 17 0.006 0.008 16s TrN +G Hv 7 476 1 0.001 0.001 Island 9 476 10 0.007 0.009 Hcc 23 616 70 0.031 0.027 Hco 46 616 46 0.017 0.024 cyt b HKY + I Hv 2 616 2 0.003 0.003 Island 9 616 24 0.016 0.018 Hcc 57 434 20 0.009 0.005 Hco 50 434 11 0.005 0.004 cmyc JC Hv 5 434 9 0.007 0.007 Island 9 434 2 0.001 0.001 Hcc 54 518 29 0.011 0.008 Hco 33 518 28 0.011 0.008 POMC TrN Hv 5 518 3 0.002 0.001 Island 9 518 2 0.001 0.001 Hcc 35 686 11 0.003 0.001 Hco 15 686 12 0.004 0.005 RAG1 K80 Hv 3 686 8 0.005 0.005 Island 7 686 1 0.000 0.001
B. Hyperolius tuberculatus complex. H. dintelmanni (Hd), eastern H. tuberculatus + H. hutsebauti (Hh), western H. tuberculatus (Ht).
Locus Model Lineage N bp S θs/bp θπ /bp Hd 8 483 4 0.003 0.002 16s HKY + I Hh 25 483 28 0.015 0.016 Ht 65 483 33 0.014 0.019 Hd 1 677 0 0.000 0.000 cyt b HKY Hh 0 ------Ht 43 677 34 0.012 0.020 Hd 8 434 3 0.002 0.002 cmyc K80 + I Hh 25 434 31 0.016 0.010 Ht 71 434 32 0.013 0.006 Hd 1 624 1 0.002 0.002 POMC TrN + G Hh 23 624 43 0.016 0.015 Ht 38 624 22 0.007 0.005 Hd 1 846 1 0.001 0.001 RAG1 TrN + I Hh 17 846 47 0.014 0.013 Ht 21 846 23 0.006 0.006 C. Hyperolius ocellatus complex. H. o. ocellatus (Hoo), H. o. purpurescens (Hop).
Locus Model Lineage N bp S θs/bp θπ /bp Hoo 42 504 23 0.011 0.018 16s HKY + I Hop 77 504 43 0.017 0.016 Hoo 22 654 79 0.033 0.044 cyt b HKY + G Hop 65 654 123 0.040 0.049 Hoo 43 434 26 0.012 0.005 cmyc JC Hop 93 434 48 0.019 0.006 Hoo 15 475 11 0.006 0.007 POMC TrN Hop 49 475 37 0.015 0.009 Hoo 20 598 7 0.004 0.005 RAG1 JC Hop 55 598 23 0.009 0.004
Figure S1: (A) Distribution of spatial sampling bias within the study area. Each polygon indicates the amount of sampling efort within that area. The environments present in areas with greater sampling efort are given more weight in the environmental niche models than environments in areas with little sampling efort, (B) Spearman rank correlations among all 19 bioclimatic variables for the background data used to train all models. Upper diagonal circles represent the magnitude (size) and the sign (color) of the correlation; positive correla- tions in blue and negative in red. Lower diagonal numbers are the correlation coefcients. (A) (B)
20 1 bio1 ● ● ● ● ● ● ● ● ● ● ● 0.95 bio10 ● ● ● ● ● ● ● ● ● ● 0.8 0.84 0.68 bio11 ● ● ● ● ● ● ● ● ● ● ● ●
−0.23 −0.4 0.2 bio12 ● ● ● 0.6 −0.07 −0.21 0.27 0.89 bio13 ● ● ● ● ● ● ● ●
−0.33 −0.44 −0.02 0.6 0.32 bio14 ● ● ● ● 0.4 0 0.48 0.59 0.09 −0.65 −0.31 −0.85 bio15 ● ● count ● 60 −0.13 −0.26 0.22 0.91 0.98 0.34 −0.36 bio16 ● ● ● ● ● ● 0.2 40 −0.38 −0.5 −0.02 0.66 0.38 0.95 −0.9 0.4 bio17 ● ● ●
0 Latitude −0.47 −0.59 −0.14 0.78 0.62 0.59 −0.64 0.66 0.65 bio18 ● ● 20 −0.21 −0.34 0.19 0.69 0.49 0.78 −0.78 0.51 0.83 0.49 bio19 ● ● ● ● 0.03 0.21 −0.3 −0.64 −0.51 −0.56 0.56 −0.5 −0.6 −0.55 −0.58 bio2 ● ● −0.2 −0.09 −0.33 0.35 0.65 0.47 0.58 −0.59 0.45 0.63 0.5 0.62 −0.52 bio3 ● ● ● −20 −0.4 0.06 0.32 −0.41 −0.76 −0.6 −0.57 0.61 −0.59 −0.63 −0.56 −0.64 0.67 −0.94 bio4 ● ● ● 0.84 0.93 0.53 −0.52 −0.3 −0.58 0.69 −0.34 −0.63 −0.7 −0.45 0.48 −0.43 0.45 bio5 ● −0.6 0.64 0.45 0.89 0.39 0.37 0.28 −0.2 0.32 0.29 0.12 0.42 −0.63 0.55 −0.6 0.22 bio6 ● 0.07 0.31 −0.35 −0.73 −0.56 −0.65 0.65 −0.55 −0.7 −0.61 −0.67 0.91 −0.81 0.89 0.54 −0.67 bio7 ● ● −0.8 0.89 0.95 0.6 −0.46 −0.31 −0.41 0.57 −0.36 −0.47 −0.51 −0.39 0.18 −0.31 0.32 0.86 0.42 0.27 bio8
0.87 0.73 0.94 0.1 0.17 −0.03 0.14 0.12 −0.05 −0.26 0.18 −0.25 0.25 −0.29 0.59 0.84 −0.27 0.64 bio9 −1 −20 0 20 40 Longitude Figure S1: (C) we evaluated the efects of changing the value of the regularization parameter and the allowed feature type classes (the shapes of the response curves) on the performance of the models for each species using the ENMeval package (Muscarella et al. 2014). For each species, we applied a checker- board pattern to divide the occurences into training and testing sets of points using an aggregation factor of fve. We evaluated the following values of regularization parameter: 0.5, 1.0, 1.5, and 2.0. We also evaluated increasingly complex shapes of the responses curves: Linear only (L), Linear + Quadratic (LQ), Hinge (H), Linear + Quadratic + Hinge (LQH), and Linear + Quadratic + Hinge + Product (LQHP). All fnal models in the study allowed for Linear + Quadratic + Hinge features. The following fgures show changes in corrected Akaike Information Criterion (AICc) and Area under the curve (AUC) for diferent feature type classes and values of the regulariation parameters. Smaller values of AICc indicate a relatively better model and higher values of AUC indicate a better performance in terms of predicting presences.
Hyperolius tuberculatus species complex Hyperolius ocellatus species complex Simpler models (i.e. only using linear features) have Some simpler models (i.e. only using linear features) lower AICc and the performance of these models in have higher AICc and lower AUC values. terms of AUC values is lower.