Reinforcing and expanding the predictions of the disturbance vicariance hypothesis in Amazonian harlequin frogs: a molecular phylogenetic and climate envelope modelling approach
- First Online:
- Cite this article as:
- Lötters, S., van der Meijden, A., Rödder, D. et al. Biodivers Conserv (2010) 19: 2125. doi:10.1007/s10531-010-9869-y
- 885 Downloads
The disturbance vicariance hypothesis (DV) has been proposed to explain speciation in Amazonia, especially its edge regions, e.g. in eastern Guiana Shield harlequin frogs (Atelopus) which are suggested to have derived from a cool-adapted Andean ancestor. In concordance with DV predictions we studied that (i) these amphibians display a natural distribution gap in central Amazonia; (ii) east of this gap they constitute a monophyletic lineage which is nested within a pre-Andean/western clade; (iii) climate envelopes of Atelopus west and east of the distribution gap show some macroclimatic divergence due to a regional climate envelope shift; (iv) geographic distributions of climate envelopes of western and eastern Atelopus range into central Amazonia but with limited spatial overlap. We tested if presence and apparent absence data points of Atelopus were homogenously distributed with Ripley’s K function. A molecular phylogeny (mitochondrial 16S rRNA gene) was reconstructed using Maximum Likelihood and Bayesian Inference to study if Guianan Atelopus constitute a clade nested within a larger genus phylogeny. We focused on climate envelope divergence and geographic distribution by computing climatic envelope models with MaxEnt based on macroscale bioclimatic parameters and testing them by using Schoener’s index and modified Hellinger distance. We corroborated existing DV predictions and, for the first time, formulated new DV predictions aiming on species’ climate envelope change. Our results suggest that cool-adapted Andean Atelopus ancestors had dispersed into the Amazon basin and further onto the eastern Guiana Shield where, under warm conditions, they were forced to change climate envelopes.
KeywordsAtelopus Bayesian inference Bioclimate MaxEnt Maximum likelihood Ripley’s K function South America
Area under the receiver operating characteristic curve
Disturbance vicariance hypothesis
Harlequin frogs (Atelopus) are a species-rich bufonid genus of Andean origin, with more than 100 species occurring in forest or paramo habitats in the Andes (Lötters 1996; La Marca et al. 2005). In this paper we focus on the less than 10 Atelopus (depending on the taxonomy applied; see Lötters et al. 2002) occurring exclusively in forest habitats in the Amazon basin and on the eastern Guiana Shield. In an earlier molecular genetic study, Noonan and Gaucher (2005) showed that the five nominal species of the eastern Guiana Shield harlequin frogs are genetically little differentiated and that they apparently interbreed in nature. Supported by divergence time estimates, these authors advocated that the observed phylogeographical patterns in Atelopus fit DV predictions, i.e. that a single Andean ancestor had invaded the eastern Guiana Shield (likely in late Miocene, as also suggested for other anuran amphibians; Santos et al. 2008) and has started speciation there in the Pleistocene due to the alteration of glacial and interglacial phases (as illustrated in Fig. 1a–d).
To their molecular phylogeny, Noonan and Gaucher (2005) added only four Atelopus species from outside the eastern Guiana Shield. As a result, the validity of their study is pending on additional corroboration. This is especially significant because our knowledge on the current distribution of harlequin frogs in central Amazonia is poorly understood. Lescure and Gasc (1986), with providing data, proposed a continuous distribution of harlequin frogs from the Andes to the eastern Guiana Shield. In contrast, Lötters et al. (2002), in a taxonomic study, were unable to trace Atelopus material in scientific collections from a large part of central Amazonia, casting some doubt on a continuous distribution. Such a hiatus could be well explained by DV predictions, since the recolonisation of central Amazonia, either from the western Amazonian lowlands or from the eastern Guiana Shield plus vicinities, would be impossible during the current postglacial. From a phylogenetic point of view, according to DV predictions and the findings of Noonan and Gaucher (2005), we expect that harlequin frogs from east of this suspected distribution gap in central Amazonia constitute one clade nested within those from the Andes and Amazonian lowlands adjacent to them (Fig. 1d) if more species were included from more of the genus’ entire geographic range than available to Noonan and Gaucher (2005).
Species can respond to climate change in two ways. One is change of geographic range (i.e. increase, decrease down to extinction, shift) and maintenance of the specific climate envelope, termed niche conservancy (e.g. Peterson et al. 1999; Wiens and Graham 2005). In the other the geographical range is retained, necessarily accompanied by climate envelope shift (e.g. Broennimann et al. 2007; Pearman et al. 2007; Rödder et al. 2009). Species climate envelope predictions have never been formulated with regard to DV. According to our understanding of DV, we largely expect climate conservancy in Amazonian and Guianan Atelopus as, under DV, species change their geographic ranges as a response to a changing climate (Fig. 1a–d). Vertical range shift of cool-adapted species along the Andean versant was up to 800 m (Bush 1994). However, maximum altitudes found on the eastern Guiana Shield have been about 300 m above today’s sea level only. As niche shift is facilitated in small populations pushed to their margin of environmental tolerance (Holt and Gomulkiewicz 2004; Holt et al. 2005; Jakob et al. 2010), it may be assumed that within the eastern glacial forest fragment (Fig. 1c) climate envelopes have shifted in those cool-adapted species which have survived warmer periods. As a consequence, when comparing current-day Atelopus populations from the western and eastern Amazonian (including the eastern Guiana Shield) lowlands (Fig. 1c) their climate envelopes under today’s macroclimate are predicted to show some divergence.
The contemporary postglacial was warmest about 8,000–4,500 years BP and temperature has decreased since then. According to DV, harlequin frog species should currently be able to re-expand their distributions into lower areas. When mapping climate envelopes of current-day Atelopus populations from both western and eastern Amazonia under macroclimatic conditions into geographic space, they should range into central Amazonia. However, because of the expected climate envelope shift in eastern Amazonian Atelopus, mapped climate envelopes (which can be understood as species’ potential distributions) are predicted to be rather allopatric than sympatric.
In this paper we combined different methodological approaches to study (i) if extant harlequin frogs display a central Amazonian distribution gap; (ii) if eastern Amazonian Atelopus constitute a single clade nested in a phylogeny comprising an enlarged data set from the Andes and adjacent lowlands; (iii) if climate envelopes of western versus eastern Amazonian populations (i.e., geographically well delimitated by a natural central Amazonian distribution gap) are divergent under today’s macroclimate; (iv) if allopatry is the result rather than sympatry when mapping these climate envelopes into geographic space. We discuss in how far our result reinforce and expand DV predictions.
A central Amazonian distribution gap
In addition, 42 data points of apparent absence of harlequin frogs, illustrated in Fig. 2 (see Appendix), were obtained from published references and expert interviews as described above. We only included data points at elevations ≤800 m above sea level and situated in an area defined through a Minimum Convex Polygon (MCP) for all presence data, created with DIVA-GIS 5.4.
We are aware that absence is nearly impossible to prove and should be handled with caution; therefore, we independently analysed presence and absence information. For this, Ripley’s K function, a multi-distance spatial cluster analysis, was used to independently study spatial dependence in both data sets (Fig. 2) by comparison to a random pattern, which follows a Poisson distribution (Ripley 1977; Haase 1995). If the K function of the data differs significantly from that of the random distribution, data points under study are clustered (i.e. aggregated, when above that of the random distribution) or highly dispersed (i.e. when below random expectation). Analysis was performed with the Spatial Statistics (confidence envelope: 99 permutations) tool box of ArcGIS Desktop 9.2 (ESRI; http://www.esri.com).
Nested monophyly of eastern Amazonian Atelopus
List of species used for the molecular analysis, their GenBank accession numbers and origin
Ecuador: near Zhund
French Guiana: Lac des Americains
Atelopus flavescens (‘barbotini’)
French Guiana: near Saül
Ecuador: near Plan de Milagro
French Guiana: Monts Bakra
Ecuador, Mindo region
Peru: Tarapoto region
Guiana: Mabura Hill region
French Guiana: near Saül
Colombia: Bahía Solano
Peru: Alto Mayo
Atelopus cf. spumarius
Peru: Iquitos region
Bolivia: Yungas de La Paz
Costa Rica: near Las Alturas
Panama: Las Filipinas
Peru: Oxapampa region
Atelopus sp. ‘cusco’
Peru: near Puente Fortaleza
Atelopus sp. ‘cocha’
Colombia: Laguna Cocha
Osornophryne sp. 1
Osornophryne sp. 2
Eleutherodactylus cf. johnstonei
DNA was extracted from toe clips. Tissue samples (stored in 99% ethanol) were digested using proteinase K (final concentration 1 mg/mL), homogenised and subsequently purified following a high-salt extraction protocol (Bruford et al. 1992). Polymerase chain reaction (PCR) primers for the fragment of the 16S rRNA gene were 16SA-L and 16SB-H of Palumbi et al. (1991), used as in Van der Meijden et al. (2007). PCR products were purified via spin columns (Qiagen). Sequencing was performed directly using the corresponding PCR primers. New sequences were combined with existing sequences taken from GenBank in the final dataset containing 27 taxa including bufonid and non-bufonid outgroups (Table 1). Sequences were aligned using ClustalW (Thompson et al. 1994) and subsequently edited by hand. The final alignment contained a total of 570 positions of which 219 were variable and 136 were parsimony-informative.
Phylogeny reconstruction was performed using Maximum Likelihood (ML) and Bayesian Inference (BI) methods. Gaps were treated as unknown characters. The best fitting models of sequence evolution were determined by the AIC criterion as implemented in Modeltest 3.06 (Posada and Crandall 1998). ML tree searches were performed using PhyML, version 2.4.4 (Guindon and Gascuel 2003). Bootstrap branch support values were calculated with 200 replicates. The Bayesian analyses of the combined and separate datasets was conducted with MrBayes 2.0 (Huelsenbeck and Ronquist 2001) with 2 million generations, sampling trees every 10th generation and calculating a consensus tree after omitting the first 200,000 trees as burn-in determined with the Tracer 1.4 (http://beast.bio.ed.ac.uk/Tracer). No well supported topological differences were found between the BI and ML trees; the ML tree was used in the subsequent analysis.
Divergence in climate envelopes and allopatry
AUC values per model, climate envelope overlap in terms of I and D values and assessment of their similarity and equivalency via randomization tests (see text)
Annual mean temperature
Mean monthly temperature range
Maximum temperature of the warmest month
Minimum temperature of the coldest month
Precipitation of the wettest month
Precipitation of the driest month
We quantitatively compared climate envelopes of western and eastern Amazonian Atelopus with Schoener’s index (D) and Hellinger distance (I) as modified by Warren et al. (2008). Both indices allow for testing climate envelope similarity between two probability distributions of (e.g. climate envelope) distributions over geographic space, whereby D and I values range from 0 to 1 (i.e. models have no to entire overlap). We evaluated the significance of D and I values with null models regarding climate envelope similarity and equivalency representing two extremes within the spectrum of niche conservatism (Warren et al. 2008). Tests were performed separately for each bioclimatic parameter in the manner of Rödder and Lötters (2009). Moreover, for climate envelope equivalency, we applied a randomization test as proposed by Warren et al. (2008) which relies on the metrics D and I. For western and eastern Amazonian harlequin frog occurrences 100 pseudoreplicate datasets were created by randomly partitioning the combined number of western and eastern occurrences into sets of the same size of the original of western and eastern datasets. Climate envelope models were built from each pseudoreplicate in order to generate null distributions. The overlap between models computed with the original data sets were compared to the percentiles of these null distributions in a one-tailed test to evaluate the hypothesis that climate envelope models for western and eastern records were not significantly different. This test allows for an assessment of climate envelope maintenance (i.e. niche conservancy) in a strict sense, i.e. the effective equivalency of the climate envelope in the western and eastern geographic ranges. It is expected to be only met if western and eastern harlequin frogs tolerate exactly the same set of climatic conditions and have the same set of environmental conditions available to them.
In order to assess climate envelope similarity, we again used a randomization test of Warren et al. (2008). It compares the actual similarity of climate envelopes in terms of D and I values to the distribution of similarities obtained by comparing them to a climate envelope model created through randomly choosing cells from among the cells in the study area. The same procedure was performed in both directions (western to eastern records vice versa) 100 times to construct an expected distribution of D and I values between a climate envelope model generated using actual occurrences and another one generated from random background data points extracted within a MCP enclosing one set of records. These null distributions served as a two-tailed test to assess the null hypothesis that measured climate envelope overlap between western and eastern Amazonian Atelopus is explained by regional similarities or differences in available habitat (‘background effects’). This hypothesis is rejected if the actual similarity falls outside the 95% confidence limits of the null distribution suggesting active habitat choice. Significantly higher values suggest that climate envelopes are more similar than expected by chance and lower values indicate greater differences. Computations of D, I, climate envelope similarity and equivalency were performed with a Perl script developed by Warren et al. (2008).
Results and discussion
A central Amazonian distribution gap
The existence of a natural distribution gap is expectable under DV (Fig. 1c) and therefore reinforces our hypothesis of Amazonian harlequin frog historical biogeography. However, it needs to be noted that this explanation for the observed geographic pattern is a single possibility out of many possible causes. A gap alone leaves also space for other explanations than DV.
Nested monophyly of eastern Amazonian Atelopus
Atelopus species from the Venezuelan Andes and the Caribbean coastal range, i.e. proximate to the Guiana Shield, show osteological and external morphological characters suggesting a closer relationship to Colombian Andean taxa (McDiarmid 1971). However, we lack other characters, such as those from molecular phylogenetics studies, to validate or dispose this view.
Divergence in climate envelopes and allopatry
Prediction accuracy of MaxEnt climate envelope models was high as suggested by ‘excellent’ AUC values (western Amazonian Atelopus: test 0.955, training 0.980; eastern Amazonian Atelopus: test 0.979, training 0.985) following the AUC classification accuracy of Swets (1988).
We provide molecular phylogenetic evidence that all Amazonian Atelopus constitute a monophyletic group and find support that a natural distribution gap in central Amazonia for these amphibians exists. Harlequin frogs from east of this gap are a monophyletic subset, suggesting that they have derived from a single ancestral stock which subsequently has started vicariant speciation. Our findings corroborate the results of Noonan and Gaucher (2005). These authors advocated that DV predictions are met in Amazonian and in particular eastern Guiana Shield Atelopus. We here demonstrate that DV predictions are also met when genetic sampling is expanded by inclusion of more species from the entire genus’ distribution.
The justified spatial breakup into western and eastern Amazonian groups afforded us for the first time to derive DV predictions regarding climate envelope change in taxa of Andean origin. These predictions were met, as we were able to show that climate envelopes of both groups were similar regarding some parameters but that other parameters significantly differed. These different parameters result in allopatric potential distributions of western and eastern Amazonian Atelopus. Geographic range shift does not strictly result in climate envelope change, as commonly species tend to change their distributions with changing climate being bound to physiological constrains hampering climate envelope shifts regarding some parameters (e.g. Parmesan 2006). Because of the limited elevational range in the eastern Guiana Shield, cool-adapted taxa facing extinction risk were forced with a strong selective pressure to change their climate envelopes. We suggest that this is a prediction which is generally applicable to Andean species under DV.
We are grateful to all collaborators who supported us with their knowledge on amphibian communities in Amazonia and the Guiana region (see Appendix), as well as to curators of scientific collections reviewed (E. Ahlander, W. Böhme, B.T. Clarke, J.H. Córdova, W.E. Duellman, L. Ford, J.D. Lynch, I. Sazima, H. Zaher). This project benefited from grants by the Wilhelm-Peters-Fonds of the Deutsche Gesellschaft für Herpetologie und Terrarienkunde (DGHT) to S. Lötters and M. Veith and by the Graduiertenförderung des Landes Nordrhein-Westfalen to D. Rödder. C.F.B. Haddad thanks FAPESP and CNPq for financial supports. For tissue samples processed in this paper, we thank D. Bernauer, M. Blanc, R. Boistel, L.A. Coloma, I. De la Riva, R. Ernst and E. Lehr. A. van der Meijden was supported by FCT postdoctoral grant SFRH/BPD/48042/2008. Special thanks to B.P. Noonan for helpful comments on a previous version of this paper.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.