Molecular Ecology (2010) 19, 427–430
NEWS AND VIEWS
PERSPECTIVE
Testing evolutionary hypotheses for phenotypic divergence using landscape genetics W . C H R I S F U N K and M E L A N I E A . M U R P H Y Department of Biology, Colorado State University, 1878 Campus Delivery, Fort Collins, CO 80523-1878, USA Understanding the evolutionary causes of phenotypic variation among populations has long been a central theme in evolutionary biology. Several factors can influence phenotypic divergence, including geographic isolation, genetic drift, divergent natural or sexual selection, and phenotypic plasticity. But the relative importance of these factors in generating phenotypic divergence in nature is still a tantalizing and unresolved problem in evolutionary biology. The origin and maintenance of phenotypic divergence is also at the root of many ongoing debates in evolutionary biology, such as the extent to which gene flow constrains adaptive divergence (Garant et al. 2007) and the relative importance of genetic drift, natural selection, and sexual selection in initiating reproductive isolation and speciation (Coyne & Orr 2004). In this issue, Wang & Summers (2010) test the causes of one of the most fantastic examples of phenotypic divergence in nature: colour pattern divergence among populations of the strawberry poison frog (Dendrobates pumilio) in Panama and Costa Rica (Fig. 1). This study provides a beautiful example of the use of the emerging field of landscape genetics to differentiate among hypotheses for phenotypic divergence. Using landscape genetic analyses, Wang & Summers were able to reject the hypotheses that colour pattern divergence is due to isolation-by-distance (IBD) or landscape resistance. Instead, the hypothesis left standing is that colour divergence is due to divergent selection, in turn driving reproductive isolation among populations with different colour morphs. More generally, this study provides a wonderful example of how the emerging field of landscape genetics, which has primarily been applied to questions in conservation and ecology, now plays an essential role in evolutionary research. Keywords: colour morph, Dendrobates pumilio, gene flow, landscape genetics, phenotypic divergence, speciation Received 4 November 2009; revision received 16 November 2009; accepted 17 November 2009
Correspondence: W. Chris Funk, Fax: (970) 491 0649; E-mail:
[email protected]
2010 Blackwell Publishing Ltd
Dendrobates pumilio is a rainforest frog species found on the Caribbean slope of Central America from Nicaragua to Panama. Populations in the Bocas del Toro region of northwestern Panama exhibit the most striking colour divergence among populations, ranging from red to blue to orange to green, with or without spots (Fig. 1). In contrast, populations in Costa Rica vary relatively little in colour, all essentially variations of the red morph. Importantly, this colour variation is heritable (Summers et al. 2004), thus it is not simply a plastic response to environmental variation. Many of the Bocas del Toro populations are isolated on oceanic islands off the coast of Panama and even the mainland populations are separated by potential barriers such as swamps and hills, whereas the Costa Rican populations occupy fairly continuous rainforest habitat. Thus Wang & Summers set out to test the relative importance of IBD, landscape constraints (such as topography, physical barriers such as intervening water bodies, and habitat suitability), and phenotypic divergence in causing genetic divergence among populations. To this end, they genotyped 705 individuals from 19 populations from Costa Rica and the Bocas del Toro region at 15 microsatellite loci, and conducted a variety of population genetic and landscape genetic analyses to test the relationship of the above variables to genetic divergence. Their initial analyses revealed a striking pattern of genetic structure: pairwise genetic distances were much higher among the phenotypically divergent and geographically isolated Bocas del Toro populations than among the less phenotypically differentiated and less isolated Costa Rican populations. Two general hypotheses could explain this pattern (Fig. 2): (1) landscape resistance has restricted gene flow, allowing drift and ⁄ or divergent selection to generate divergence in colour patterns; or (2) divergent selection caused colour pattern divergence and selection against immigrants with different colour morphs has resulted in the accumulation of genetic differences between populations with different colour patterns. Wang & Summers used a landscape genetic analysis known as causal modelling (Legendre 1993) to test 12 specific hypotheses regarding the landscape and phenotypic variables that best explain genetic divergence among D. pumilio populations. In essence, causal modelling uses a series of partial Mantel tests to test several predictions stemming from each hypothesis. Wang & Summers’ hypotheses included straight-line geographic distance, least cost path distances (incorporating the effects of habitat variation and topography), dorsal coloration, or ventral coloration as predictor variables. The results of this analysis were quite interesting. The hypothesis that genetic divergence is associated with differences in dorsal coloration was the only one that was not rejected. In contrast, some of the
428 N E W S A N D V I E W S : P E R S P E C T I V E Fig. 1 Divergent colour morphs observed among populations of the strawberry poison frog, Dendrobates pumilio. Frogs are from San Cristobal (upper left), Cerro Brujo (upper right), Bastimentos (lower right), and Agua (lower left).
Fig. 2 Two general hypotheses that can explain a positive relationship between genetic and phenotypic divergence. These hypotheses are not mutually exclusive; both may be involved in generating this relationship. Moreover, isolation-by-distance (IBD) caused by dispersal-limited gene flow can contribute to genetic and phenotypic divergence.
predictions of the IBD, least cost path, and ventral coloration hypotheses were rejected, indicating that these variables are not as strongly associated with genetic
divergence. Thus, these results support the second hypothesis in Fig. 2, that divergent selection has driven colour pattern divergence and that selection against immigrants 2010 Blackwell Publishing Ltd
NEWS AND VIEWS: PERSPECTIVE 429 with different colour patterns has led to genetic divergence between populations with different colour morphs. But is there any evidence that immigrants with different colour morphs are selected against in the wild? Thanks to several field and laboratory experiments conducted on D. pumilio and related species, there is evidence for both sexual and natural selection against immigrants with different colour patterns. In particular, female preference experiments show that female D. pumilio from the Bocas del Toro region prefer mates with colour patterns similar to their own (Summers et al. 1999; Reynolds & Fitzpatrick 2007). Moreover, predation experiments show that immigrant phenotypes have a greater risk of predator attack in D. tinctorius (Noonan & Comeault 2009). Thus multiple forms of selection may drive genetic divergence between populations with different colour morphs. In addition to their impressive sample sizes (in terms of number of individuals, localities, and loci), Wang & Summers’ study is exemplary for several reasons. First, they combined different classes of factors, landscape and phenotypic, in their analysis to test alternative hypotheses. By far the majority of landscape genetic studies only use landscape variables, limiting the types of questions that can be addressed. Second, they tested clearly defined and independent hypotheses. Third, they derived their least cost path distances in an empirical fashion rather than relying on ‘expert opinion’, a priori values chosen by the authors, or extensive model fitting exercises. Finally, they interpret their results in the context of field and lab experiments that provide independent evidence corroborating their results. As with any study in a rapidly expanding field, however, their analysis also has some limitations. One is the use of a single ecological cost model based on a suite of landscape variables. The limitation of this approach is that variables used to estimate landscape resistance are not independently tested to determine their statistical significance or relative importance in relation to limiting or facilitating gene flow. Another limitation is the reliance of causal modelling on partial Mantel tests which have been the subject of a debate regarding how to determine their statistical significance (Raufaste & Rousset 2001; Castellano & Balletto 2002). Finally, an ideal analysis would quantify the relative importance of different hypotheses and their respective explanatory variables rather than only provide a binary answer of ‘support’ or ‘no support’ for each hypothesis. For example, Murphy et al. (2010) applied Random Forests to quantify the percent variation explained by each of three hypotheses and to rank the importance of variables comprising each hypothesis. Nonetheless, the strengths of this paper outweigh its limitations, and Wang & Summers have raised the bar in landscape genetics. This study has important implications for our understanding of phenotypic divergence and speciation. It is one of only a handful of case studies showing phenotypic divergence, divergent selection in the field, and resulting genetic divergence (Boul et al. 2007; Funk et al. 2009). In the context of speciation, it is important to show that phenotypic divergence is correlated with genetic divergence,
2010 Blackwell Publishing Ltd
because without this, it remains unknown whether divergent selection actually restricts gene flow, completing the process of speciation. This study shows this nicely. It is also clear from studies such as these that geographic barriers are not necessary for the evolution of reproductive isolation. In addition, this paper has important implications for the field of landscape genetics. To date, the vast majority of landscape genetic studies have focused on trying to understand how landscape structure affects patterns of genetic variation in the context of conservation and ecology. This is and will remain an important application of landscape genetics. However, this study is one of a growing number of studies that use landscape genetics as a tool to address fundamental questions in evolutionary biology regarding phenotypic variation, adaptation, and speciation, another use of landscape genetics promoted by the pioneers of the field (Manel et al. 2003). Addressing these types of questions is made possible by combining multiple classes of factors in the analysis, and by combining landscape genetics with essential field experiments. The next step, and an exciting frontier in the field, will be to include the genes underlying the traits of interest in a landscape genetics context which will allow direct testing of their association with traits and environmental variables and of whether they are under selection (Holderegger & Wagner 2008).
References Boul KE, Funk WC, Darst CR, Cannatella DC, Ryan MJ (2007) Sexual selection drives speciation in an Amazonian frog. Proceedings of the Royal Society B: Biological Sciences, 274, 399–406. Castellano S, Balletto E (2002) Is the partial Mantel test inadequate? Evolution, 56, 1871–1873. Coyne JA, Orr HA (2004) Speciation. Sinauer Associates, Inc., Sunderland, Massachusetts. Funk WC, Cannatella DC, Ryan MJ (2009) Genetic divergence is more tightly related to call variation than landscape features in the Amazonian frogs Physalaemus petersi and P. freibergi. Journal of Evolutionary Biology, 22, 1839–1853. Garant D, Forde SE, Hendry AP (2007) The multifarious effects of dispersal and gene flow on contemporary adaptation. Functional Ecology, 21, 434–443. Holderegger R, Wagner HH (2008) Landscape genetics. BioScience, 58, 199–207. Legendre P (1993) Spatial autocorrelation: trouble or new paradigm? Ecology, 74, 1659–1673. Manel S, Schwartz MK, Luikart G, Taberlet P (2003) Landscape genetics: combining landscape ecology and population genetics. Trends in Ecology and Evolution, 18, 189–197. Murphy MA, Evans JS, Storfer A (2010) Quantifying Bufo boreas connectivity in Yellowstone National Park with landscape genetics. Ecology, 91, 252–261. Noonan BP, Comeault AA (2009) The role of predator selection on polymorphic aposematic poison frogs. Biology Letters, 5, 51–54. Raufaste N, Rousset F (2001) Are partial Mantel tests adequate? Evolution, 55, 1703–1705. Reynolds RG, Fitzpatrick BM (2007) Assortative mating in poisondart frogs based on an ecologically important trait. Evolution, 61, 2253–2259.
430 N E W S A N D V I E W S : P E R S P E C T I V E Summers K, Symula R, Clough M, Cronin T (1999) Visual mate choice in poison frogs. Proceedings of the Royal Society B: Biological Sciences, 266, 2141–2145. Summers K, Cronin TW, Kennedy T (2004) Cross-breeding of distinct color morphs of the strawberry poison frog (Dendrobates pumilio) from the Bocas del Toro Archipelago, Panama. Journal of Herpetology, 38, 1–8. Wang IJ, Summers K (2010) Genetic structure is correlated with phenotypic divergence rather than geographic isolation in the highly polymorphic strawberry poison-dart frog. Molecular Ecology, 19, 447–458.
Chris Funk is an evolutionary ecologist who studies speciation, interactions between adaptive divergence and gene flow, and conservation genetics of amphibians, fish, and birds. Melanie Murphy is a Postdoctoral Fellow in the Funk lab focusing on landscape genetics and conservation genetics of amphibians and large mammals.
doi: 10.1111/j.1365-294X.2009.04466.x
2010 Blackwell Publishing Ltd
