Frogs reveal the global importance and interplay of colour-based protective and thermoregulatory functions

doi:10.21203/rs.3.rs-2717887/v1

Recent small-scale and experimental studies have shown that colour lightness variation can have important physiological implications, with darker species having greater heating rates, as well as protection against pathogens and photooxidative damage. Using data for 41% (3,059) of all frog and toad species (Anura) from across the world, we reveal ubiquitous and strong clines of decreasing colour lightness towards colder regions and regions with higher pathogen pressure and UVB radiation. The relative importance of pathogen resistance was higher in the tropics and that of thermoregulation was higher in temperate regions. These functions strongly influenced colour lightness evolution in anurans and filtered for more similarly coloured species under climatic extremes, while concurrent mechanisms resulted in high within-assemblage variation in productive regions. The findings establish three fundamentally important functions of colour lightness that were previously unknown in anurans, and they substantially broaden support for colour lightness–environment relationships in ectotherms.

Biological sciences/Ecology/Biodiversity

Biological sciences/Ecology/Ecophysiology

Biological sciences/Zoology/Animal physiology

Biological sciences/Ecology/Macroecology

Biological sciences/Ecology/Freshwater ecology

In the face of climate change and biodiversity decline, trait-based inferences of the processes that underpin species’ distributions are becoming increasingly important because of their potential for understanding and forecasting biological responses^1-3. The most fundamental and general explanations for reaction norms across taxa, biogeographical realms, and spatial scales – so-called ecophysiological rules – address morphological features with broad impacts on species’ development, activity, and reproduction^4,5. Ecophysiological research was inspired by evidence for size-based thermoregulation in mammals and birds⁶. Consequently, body size is by far the most commonly used morphological predictor for determining species’ responses to climate change^2,3. However, size-based thermoregulation is less important in species that rely on external rather than metabolically produced heat (i.e., ectotherms)^7,8, which constitute 99.9% of all animal species⁹. In addition, despite the crucial role of species’ thermal requirements in models of range shifts and extinction risk, the effects of body size therein are generally weak and inconclusive^2,3.

Recent small-scale studies and experiments suggest that colour lightness has multiple physiological functions in ectotherms with important implications for species’ ecological dynamics^10-12. For instance, the biophysical principle that darker objects heat up faster than lighter ones confers a thermoregulatory advantage to lighter coloured species in warmer environments and to darker coloured species in colder environments (thermal melanism hypothesis)^10,13-15. In addition, dark-coloured pigments, melanins, are known to have protective functions. On the one hand, darker colours provide greater protection against UVB radiation¹⁶. On the other hand, darker species tend to be more resistant to pathogens and should thus have an advantage under warm and wet conditions where pathogens thrive particularly well (Gloger’s rule^12,17). Understanding these functions of colour lightness is relevant not only in the context of climatic changes but also in the face of massive population declines and local extinctions caused by pathogens such as pandemic chytrid disease¹⁸. However, despite growing evidence for their ecological significance, support for the functions of colour lightness is generally limited in both taxonomic and spatial extent¹⁰, and their interplay in structuring communities remains unknown.

We investigated whether contemporary species distributions and the functional composition of assemblages are driven by colour-based thermoregulation, pathogen resistance and UVB protection using data for 41% of all species of frogs and toads (Order: Anura) from across the world. Next, we assessed the extent to which these functions shape the phylogenetic signal underlying contemporary colour lightness variation. Because regionally dominant environmental drivers might favour specific functions of colour lightness, we also explore their independent contributions across biogeographical realms as well as their impact on filtering for species more similar in colour lightness (i.e., functional diversity). Biological information for anurans is uniquely rich and complete compared to that for other ectothermic taxa, providing the potential to not only rigorously test our predictions but also identify mechanisms that – from a physiological point of view – are readily applicable to most animal taxa.

Overall assemblage-level analyses

In models of the average colour lightness of assemblages (co-occurring species) and environmental factors, colour lightness increased with increasing temperature and elevation as well as decreasing productivity and UVB radiation. Temperature and productivity were the most important predictors of this pattern (Table 1, Fig. 1). All variables together explained 59% of the variation in colour lightness in models that included a geographical trend surface term to account for latent spatially autocorrelated variables (Table 1).

Pagel’s lambda-based tests for phylogenetic signals in both colour lightness and averages of the environmental factors across species’ ranges showed a strong impact of phylogenetic relatedness in our data (λ_Colour = 0.49, λ_Temperature = 0.88, λ_Elevation = 0.86, λ_Productivity = 0.85, λ_UVB = 0.88). To assess the extent to which the putative function influenced the species’ distributions as well as the biogeographical pattern in colour lightness, we decomposed the raw colour lightness of anuran species into the part predicted by the phylogeny (the phylogenetic signal) and its phylogenetically independent deviation from this prediction (species-specific signal) using Lynch’s comparative method¹⁹. Separate models for the phylogenetic and species-specific components of colour lightness of assemblages showed that the phylogenetic component was mainly predicted by temperature and UVB radiation (colour-based thermoregulation and UVB protection), whereas the phylogenetically independent component was mainly driven by productivity (pathogen resistance; Table 1). All variables together explained 70% of the variation in the phylogenetically predicted part of colour lightness in models that included a geographical trend surface term to account for latent spatially autocorrelated variables (Table 1).

Differences in the independent contributions of functions across biogeographic realms

When accounting for spatial autocorrelation, the effects of environmental drivers were generally weaker but remained robust and did not change in their direction or relative importance (Table 1). The trend surface term of longitude and latitude in these models explained a rather large additional proportion of the variance in colour lightness compared to spatially naive models (59% > 24%), suggesting that these global models lacked a spatially structured latent variable. However, interaction terms of the environmental variables and the biogeographical realm as predictors explained 52% of the variation in colour lightness of anuran assemblages (Fig. 2). Hence, accounting for idiosyncrasies of the responses among realms with different biogeographical histories largely removed the effect of latent spatially autocorrelated variables.

Across biogeographical realms, the effects of environmental variables on the average colour lightness of anuran assemblages were largely consistent but differed in magnitude (Fig. 2). Of the 13 biogeographical regions with more than 50 grid cells covered by our data, eight showed positive effects of temperature (mean annual temperature or elevation). Productivity negatively affected colour lightness in six realms (Fig. 2). Hierarchical partitioning analyses of the independent effects of environmental predictors on colour lightness showed that proxies for temperature were the dominant predictors of the variation in assemblage-level colour lightness in five of six temperate realms as well as Madagascar and sub-Saharan Africa. The mean annual enhanced vegetation index (EVI) was most important in three of the seven tropical realms, and mean annual UVB radiation was most important in northern South America, the Australis, and the Western Palearctic.

At the global scale, the colour lightness variation among co-occurring species, i.e., colour lightness diversity (see Methods), decreased with increasing mean annual UVB radiation and increased with decreasing mean annual temperature and elevation (Table 1). Environmental predictors together explained 35% of the variation in colour lightness diversity in models that included a geographical trend surface term to account for latent spatially autocorrelated variables. The relative and overall importance of environmental predictors differed between biogeographical realms. All variables together explained between 5% and 67% of the variation in colour lightness diversity among the 13 investigated biogeographical realms (Fig. 3). Proxies for temperature were the most important drivers of colour lightness diversity in three of six temperate and five of seven tropical realms. Mean annual UVB radiation and temperature were similarly important in the Western Palearctic. Annual UVB radiation was the most important predictor in North America, northern South America and southern South America. Productivity was the main predictor of colour lightness diversity in Central Tropical Africa.

Species-level analyses

Because frogs are mainly active at dusk or dawn and because thermoregulation is thought to provide only developmental rather than activity benefits for nocturnal species, we analysed a subset of 820 African, European, and North American anurans with activity data. In line with previous findings for moths²⁰, the effects of all environmental factors, including proxies for temperature, were consistent among species with different activity patterns. Activity explained only a minor portion (2%, P = 0.001) of additional variance in colour lightness (Extended Data Table 1). At the species level, we explored the drivers of colour lightness with models that included interactions between environmental drivers and anuran families. In these models, species of families with fewer than 10 species were excluded, which reduced the data to 2,984 species from 35 of 50 families (Extended Data Fig. 4). Of these families, 14 showed supporting effects of temperature (positive mean annual temperature or negative elevation effects, R² = 0.14 and 0.06, respectively (both P = 0.001)). Productivity negatively affected colour lightness in 18 families, and UVB radiation negatively affected colour lightness in 10 families (R² = 0.16 and 0.12, respectively; both P = 0.001). Temperature variables, productivity and UVB radiation explained 29% of the colour lightness variation among species. Using the latitudinal distribution centre of each family (absolute values averaged across all species) showed that the relative importance of temperature and UVB increased with increasing latitude, while the importance of productivity showed a U-shaped relationship with a flatter end at low latitudes (Extended Data Fig. 5).

The origin and diversity of morphological variation in animals have fascinated biologists since the beginning of natural history research^21,22. Frogs and toads show a remarkable spectrum of colours, and a rich body of research provides explanations for this phenomenon from the perspective of biotic interactions²³. Many anuran species resemble vegetation and soil in colour to reduce predation risk²⁴. Other taxa, such as the Dendrobatidae, have bright warning colours to signal toxicity to predators²⁵. However, these functions are limited to certain taxa and are only of local importance²³. Our findings reveal three crucial functions of colour that are currently unrecognized in anurans¹² and provide uniquely rigorous support for the general functional significance of colour lightness in ectotherms.

We documented strong and globally consistent clines in colour lightness along gradients of temperature, productivity and UVB radiation that broadly expands our understanding of the physiology of anurans. We show that species are generally darker coloured in realms with higher productivity and higher UVB radiation. Thus, corresponding to the predictions of Gloger’s rule and the UVB protection hypothesis^12,16,17, the enhanced structural integrity and greater UV absorbance of melanized cells seem to provide an advantage to darker species in realms with high pathogen pressure and high UVB levels. Both mechanisms are well supported for plumage colouration in birds¹⁷. Together with evidence for selected ectotherm taxa and realms^16,26, our findings suggest much broader implications of the protective functions of colour in animals. In particular, the role of pathogen resistance has likely been underestimated in previous studies due to a geographical bias towards the North American and European faunas¹⁷. Specifically, we demonstrated that productivity was much more important for the colour lightness of anuran assemblages in tropical realms than in temperate realms and hence in realms where pathogen pressure is expected to be highest.

Our results not only stress the significance of the protective functions of colour lightness but also reveal the central importance of colour lightness for thermoregulation in ectotherms. Most ectotherms depend on ambient environmental temperatures to control their body temperature. Size-based thermoregulation sensu lato Bergmann has dominated the discussion of body size variation. In ectotherms, however, comprehensive studies on body size variation across species highlight that body size clines typically suggest a converse pattern^27,28. Our results provide convincing support for the prediction of the thermal melanism hypothesis, with species being on average darker in colder environments. Therefore, and in the context of weak and inconclusive support for size-based thermoregulation but growing evidence for thermal melanism¹⁰, colour lightness currently represents the most generally supported thermal adaptation in ectotherms. Nevertheless, our analysis of the relative importance of the three investigated functions also uncovers a possible trade-off between colour-based thermoregulation and pathogen protection. Thus, a large part of the colour lightness variation at the warm end of the colour lightness–temperature relationship was explained by pathogen resistance. Similarly, the species-level analysis highlights that families tend to either follow the expectation under thermal melanism or that from the protective functions of colour lightness (Extended Data Fig. 4). These patterns point to divergent selection pressures on certain functions associated with idiosyncrasies in the biogeographical history of regions and taxa.

Our general and strong support for colour lightness-based thermoregulation, pathogen resistance and UVB protection implies significant gains in the predictive power and accuracy of mechanistic and biophysical models^1,29 of ectothermic species when incorporating colour lightness. Moreover, accounting for the covariance of environmental drivers and colour lightness will help understand, forecast, and ultimately develop strategies to mitigate the biological consequences of environmental changes. Across European dragonflies, for instance, range shifts in accordance with colour-based thermoregulation changed the community composition towards lighter coloured species on average³⁰. However, our results suggest that temperature is one but not the only driver of such changes and consequent threats to species. Multiple strong colour lightness–environment interactions and evolutionary constraints contribute to complexity in the spatial variation in colour lightness. On the one hand, darker anuran species in temperate regions are likely more vulnerable to extinction because they are already at the cold end of their tolerance limits (mountain tops or northernmost boundaries). On the other hand, lighter-coloured species may become more threatened because of their poorer resistance to pathogens. Temperature and UVB radiation were strong filters of the colour lightness of anurans with exceptionally high importance in temperate realms and families (Fig. 2 and Extended Data Fig. 4). Therefore, lineages with a predominantly temperate distribution might be susceptible to novel pathogens because they are incapable of overcoming phylogenetically conserved cold and UVB adaptations. Together, these results suggest that multiple, partly competing functions of colour lightness can result in the loss of a distinct part of the overall functional and phylogenetic diversity. Since threats to biodiversity, including pathogens and climate change, interact³¹, we also urge researchers to account for the multifunctional nature of traits when making mechanistic predictions of species’ responses.

DATA AVAILABILITY

Data that supports the findings of the study are available on figshare (DOI: https://doi.org/10.6084/m9.figshare.22303480).

ACKNOWLEDGEMENTS

We acknowledge support in assessing colour lightness and data validation by Antje Schmidt. R.L. compiled the data and led the analyses. S.P. and R.B. conceived the study. R.L. wrote the first draft. S.P. assisted in writing, data processing and data analysis. M.-O.R. provided additional data. All authors contributed substantially to subsequent revisions.

Buckley, L. B. & Kingsolver, J. G. Functional and phylogenetic approaches to forecasting species' responses to climate change. Annu. Rev. Ecol. Evol. Syst. 43, 205–226 (2012).
Chichorro, F., Juslén, A. & Cardoso, P. A review of the relation between species traits and extinction risk. Biol. Conserv. 237, 220–229 (2019).
MacLean, S. A. & Beissinger, S. R. Species' traits as predictors of range shifts under contemporary climate change: a review and meta-analysis. Glob. Chang. Biol. 23, 4094–4105 (2017).
Gillooly, J. F., Brown, J. H., West, G. B., Savage, V. M. & Charnov, E. L. Effects of size and temperature on metabolic rate. Science 293, 2248–2251 (2001).
Gillooly, J. F. Effect of body size and temperature on generation time in zooplankton. J. Plankton Res. 22, 241–251 (2000).
Brown, J. H., Gillooly, J. F., Allen, A. P., Savage, V. M. & West, G. B. Toward a metabolic theory of ecology. Ecology 85, 1771–1789 (2004).
Shelomi, M. Where are we now? Bergmann's rule sensu lato in insects. Am. Nat. 180, 511–519 (2012).
Watt, C., Mitchell, S. & Salewski, V. Bergmann's rule; a concept cluster? Oikos 119, 89–100 (2010).
Atkinson, D. & Sibly, R. M. Why are organisms usually bigger in colder environments? Making sense of a life history puzzle. Trends Ecol. Evol. 12, 235–239 (1997).
Pinkert, S. & Zeuss, D. Thermal biology: melanin-based energy harvesting across the tree of life. Curr. Biol. 28, R887–R889 (2018).
Pinkert, S. et al. Mobility costs and energy uptake mediate the effects of morphological traits on species' distribution and abundance. Ecology 101, e03121 (2020).
McNamara, M. E. et al. Decoding the evolution of melanin in vertebrates. Trends Ecol. Evol. 36, 430–443 (2021).
Bogert, C. M. Thermoregulation in reptiles, a factor in evolution. Evolution 3, 195–211 (1949).
Clusella-Trullas, S., Terblanche, J. S., Blackburn, T. M. & Chown, S. L. Testing the thermal melanism hypothesis: a macrophysiological approach. Funct. Ecol. 22, 232–238 (2008).
Gates, D. M. Biophysical Ecology (Springer, 1980).
Bishop, T. R. et al. Ant assemblages have darker and larger members in cold environments. Glob. Ecol. Biogeogr. 25, 1489–1499 (2016).
Delhey, K. A review of Gloger's rule, an ecogeographical rule of colour: definitions, interpretations and evidence. Biol. Rev. 94, 1294–1316 (2019).
Scheele, B. C. et al. Amphibian fungal panzootic causes catastrophic and ongoing loss of biodiversity. Science 363, 1459–1463 (2019).
Lynch, M. Methods for the analysis of comparative data in evolutionary biology. Evolution 45, 1065–1080 (1991).
Heidrich, L. et al. The dark side of Lepidoptera: colour lightness of geometrid moths decreases with increasing latitude. Glob. Ecol. Biogeogr. 27, 407–416 (2018).
Darwin, C. On the Origin of Species by Means of Natural Selection, or the Preservation of Favoured Races in the Struggle for Life (John Murray, Albemarle Street, 2010).
Wallace, A. R. Natural Selection and Tropical Nature: Essays on Descriptive and Theoretical Biology (Macmillan, 1891).
Toledo, L. F. & Haddad, C. F. B. Colors and some morphological traits as defensive mechanisms in anurans. Int. J. Zool. 2009, 1–12 (2009).
Cuthill, I. C. et al. The biology of color. Science 357, eaan0221 (2017).
Myers, C. W. & Daly, J. W. Dart-poison frogs. Sci. Am. 248, 120–133 (1983).
Stelbrink, P. et al. Colour lightness of butterfly assemblages across North America and Europe. Sci. Rep. 9, 1760 (2019).
Mähn, L., Hof, C., Brandl, R. & Pinkert, S. Beyond latitude: temperature, productivity, and thermal niche conservatism drive body size variation in Odonata. Glob. Ecol. Biogeogr. (2023). https://doi.org/10.1111/geb.13661.
Slavenko, A. et al. Global patterns of body size evolution in squamate reptiles are not driven by climate. Glob. Ecol. Biogeogr. 28, 471–483 (2019).
Briscoe, N. J. et al. Mechanistic forecasts of species responses to climate change: the promise of biophysical ecology. Glob. Change Biol. 29, 1451–1470 (2022).
Zeuss, D., Brandl, R., Brändle, M., Rahbek, C. & Brunzel, S. Global warming favours light-coloured insects in Europe. Nat. Commun. 5, 3874 (2014).
Hof, C., Araújo, M. B., Jetz, W. & Rahbek, C. Additive threats from pathogens, climate and land-use change for global amphibian diversity. Nature 480, 516–519 (2011).

Colour lightness estimation

The colour lightness of 3,386 anuran species was estimated based on photos and illustrations from 27 amphibian field guides and monographs (for sources and details, see Supplementary Table 1). These literature sources were selected because they cover the entire amphibian fauna of the respective regions. Due to missing distribution maps for several species, our final dataset covered 3059 species. We focused on anurans instead of all amphibian species because the two remaining orders of amphibians differ from frogs and toads in their body shape, distribution, and ecology. Specifically, Caudata (9%) are mostly absent from tropical regions and comprise several semi-fossorial and aquatic species. Gymnophiona (3%) are exclusively distributed in the tropics and are mainly fossorial and aquatic. In contrast, except for Antarctica, anurans occur on all continents and mainly comprise semi-aquatic, terrestrial and arboreal taxa.

To estimate the colour lightness of the dorsal surface of individuals (i.e., the part of the body that is exposed to sunlight) from images, we used a standardised colour wheel with 72 colours (Extended Data Fig. 1). The wheel comprises 18 rays, each consisting of four colours with different saturation levels (S: 10, 40, 70, 100%). In addition to assigning a colour per individual, we also visually estimated the coverage of the colours in 5% steps. To obtain a proxy for melanisation, we calculated the coverage-weighted mean of the red, green, and blue colour channels of the colours of individuals (Extended Data Fig. 1). This average colour lightness value can range from 0 (black) to 255 (white). We did not distinguish between sexes, as they were typically not indicated. Instead, colour lightness values of all available individuals were averaged (morphs and sexes). We chose a visual assessment of colour lightness instead of a digital image analysis to facilitate the exclusion of reflections (wet skin) and shaded areas. Another advantage of visual estimation is that it allows the integration of multiple available sources of in situ images despite differences in the camera–object angle and light conditions (e.g., due to the use of flash). This is particularly important for amphibians, as they fade in colour after their death, impeding the use of the only comparatively rich source of images — material from museum collections — for colour assessments. The main colour assessment was performed by a single person (Ricarda Laumeier). However, a subset of species was also assessed by a second person (Antje Schmidt), and analyses of these data showed high agreement between the assessments (R² = 0.65, slope ± SE 0.60 ± 0.04, intercept: 24.45, P < 0.001, n = 107; Extended Data Fig. 2). We also acknowledge the systematic observer bias documented by this comparison, which urges the need to align estimates based on species overlap when data assessed by different observers should be integrated.

Distribution data

Vector distribution maps³² for anuran species were reassigned to an equal-area grid (MGRS, cell size of approximately 100 km × 100 km) with functions provided in the R package sf³³. Species were scored as present if their ranges covered more than 50% of the cell area. Grid cells that contained more than 50% water were excluded. Distribution data were transformed and mapped to a Mollweide equal-area projection with functions of the R package sp³⁴.

Environmental data

Based on predictions of the thermal melanism hypothesis (colour-based thermoregulation), Gloger’s rule and the UVB protection hypothesis, four environmental proxies for temperature (mean annual temperature (v2³⁵), elevation³⁶), productivity (mean annual EVI³⁷), and UVB radiation³⁸ were derived from global high-resolution data. Values for each of these variables were averaged across species’ ranges for species-level analyses and across grid cells of our equal-area grid for assemblage-level analyses.

Phylogenetic autocorrelation

To construct a comprehensive phylogeny for our subset of species, 230 species were added to the currently most complete molecular information³⁹ at the respective genus level, and the intra-genus relationships were randomly resolved with functions of the R package phytools⁴⁰. Subsequently, species without colour lightness data were pruned from this tree for phylogenetic comparative analyses. We tested for a phylogenetic signal (Pagel’s lambda⁴¹) in the colour lightness variation of anuran species using functions of the R package phytools⁴⁰. To evaluate the phylogenetic uncertainty introduced by the inclusion of species that were added to the original phylogeny, we also calculated the phylogenetic signal based on the original phylogeny only. Because the strength of the phylogenetic signal in colour lightness was similar between the original tree (λ = 0.51) and 10 alternatives of the extended tree [median λ = 0.50], one phylogeny from the set of extended trees was selected for subsequent analyses. Finally, the phylogenetic signal for colour lightness and environmental aggregates were calculated based on one randomly selected tree to test for phylogenetic niche conservatism. Because of this strong impact of the phylogenetic relationship of species, colour lightness was decomposed into its phylogenetically predicted part and the deviation from this prediction using Lynch’s comparative method¹⁹. The advantage of this method, compared to merely accounting for phylogenetic autocorrelation, is that it allows separate analyses of the long-term evolutionary (P-component) and short-term (species-specific, S-component) significance of trait–environment relationships.

Spatial autocorrelation

Macroecological patterns typically show spatial autocorrelation (i.e., locations closer together have more similar values than those farther apart), which reduces the effective sample size and thereby leads to false model estimates⁴². To account for spatial autocorrelation in our models, we repeated assemblage-level analyses with generalized additive models (GAMs) of colour lightness and environmental factors that included a smoothed (trend surface) term of the geographical coordinates of each assemblage using functions of the R package mgcv.

Regression models

Residuals of all models were checked, and if necessary, predictors were transformed to meet the assumption of normality. All analyses were performed with the statistical software R⁴³. Trait–environment relationships were studied at the species and assemblage levels to examine differences in the importance of thermoregulation, UV protection and pathogen resistance between clades and regions in taxonomically and spatially specific contexts, respectively.

Assemblage-level analyses

In assemblage-level analyses, we fitted multiple regressions for the relationships of environmental predictors with ‘raw’ colour lightness, its phylogenetically predicted component, and its species-specific component as well as colour lightness diversity. Assemblages for which colour lightness data were available for less than 33% of species were removed to reduce the effects of incomplete representation (Extended Data Fig. 3), leaving 16,686 of the initial 17,170 assemblages. Colour lightness diversity was calculated as the standardized effect sizes of mean pairwise distances of co-occurring species minus a null model of 1,000 alternative species sets with the same number of species randomly drawn from the species pool of the respective biogeographical realm⁴⁴. This measure quantifies colour lightness diversity as underdispersed, overdispersed, or randomly dispersed for assemblages with more than one species (n = 14,715). At our coarse grain, underdispersion (negative values) indicates filtering of species by a dominant driver (in our case, a function of colour), whereas overdispersion indicates the presence of multiple concurrent drivers⁴⁵. For the main assemblage-level analyses, we fitted both multiple linear regressions and a GAM that included a trend surface term to account for spatial autocorrelation. In addition, we assessed regional differences in the relative importance of presumed environmental drivers of the colour lightness and colour lightness diversity of anuran assemblages by repeating our main assemblage-level analysis for the biogeographical realms of amphibians, as identified by Holt et al.⁴⁴ For these analyses, only assemblages of biogeographical realms with more than 50 assemblages were considered, leaving 13 of 19 realms.

Species-level analyses

To understand differences in the importance of different environmental drivers among major lineages, we repeated the analysis of raw colour lightness, including interactions between all pairs of the environmental predictors and anuran families. In these models, species belonging to families with fewer than 10 species were excluded, which reduced the dataset to 2,984 species (Extended Data Fig. 4). In addition, for each family, the relative contribution of environmental drivers of colour lightness from hierarchical partitioning analysis was plotted against the average absolute latitude of species to assess latitudinal changes in the dominant driver (Extended Data Fig. 5). To test for an effect of activity, we fitted a single regression of colour lightness with the interaction terms of each environmental variable and activity time (diurnal, nocturnal or diurnal and nocturnal) (Extended Data Table 1) for 820 North American and African anuran species.

ONLINE METHOD REFERENCES

19. Lynch, M. Methods for the analysis of comparative data in evolutionary biology. Evolution 45, 1065–1080 (1991).

32. IUCN. The IUCN red list of threatened species. Version 2021-3. https://www.iucnredlist.org (2021).

33. Pebesma, E. Simple features for R: standardized support for spatial vector data. R J. 10, 439 (2018).

34. Bivand, R., Pebesma, E. & Gomez-Rubio, V. Applied Spatial Data Analysis with R (Springer, 2013).

35. Karger, D. N. et al. Climatologies at high resolution for the earth’s land surface areas. Sci. Data 4, 170122 (2017).

36. Robinson, N., Regetz, J. & Guralnick, R. P. EarthEnv-DEM90: a nearly-global, void-free, multi-scale smoothed, 90m digital elevation model from fused ASTER and SRTM data. ISPRS J. Photogramm. Remote Sens. 87, 57–67 (2014).

37. Tuanmu, M. N. & Jetz, W. A global, remote sensing-based characterization of terrestrial habitat heterogeneity for biodiversity and ecosystem modelling. Glob. Ecol. Biogeogr. 24, 1329–1339 (2015).

38. Beckmann, M. et al. glUV: a global UV-B radiation data set for macroecological studies. Methods Ecol. Evol. 5, 372–383 (2014).

39. Jetz, W. & Pyron, R. A. The interplay of past diversification and evolutionary isolation with present imperilment across the amphibian tree of life. Nat. Ecol. Evol. 2, 850–858 (2018).

40. Revell, L. J. phytools: an R package for phylogenetic comparative biology (and other things). Methods Ecol. Evol. 3, 217–223 (2012).

41. Pagel, M. Inferring the historical patterns of biological evolution. Nature 401, 877–884 (1999).

42. Dormann, C. F. et al. Methods to account for spatial autocorrelation in the analysis of species distributional data: a review. Ecography 30, 609–628 (2007).

43. R CoreTeam. R: a language and environment for statistical computing. R foundation for statistical computing. https://www.r-project.org/ (2022).

44. Holt, B. G. et al. An update of wallace’s zoogeographic regions of the world. Science 339, 74–78 (2013).

45. Poos, M. S., Walker, S. C. & Jackson, D. A. Functional-diversity indices can be driven by methodological choices and species richness. Ecology 90, 341–347 (2009).

Table 1. Environmental drivers of the colour lightness of anuran assemblages. Linear (lm) and generalized additive models (GAM) of environmental predictors of the raw colour lightness, its phylogenetic component (P), and its species-specific component (S) (n = 16,686) as well as the colour lightness diversity of anuran assemblages (n = 14,436). The effect of species richness was controlled for in the colour lightness diversity calculation (see Methods). Models of colour lightness diversity include fewer assemblages because singletons are invariant. GAMs additionally include a smoothed (trend surface) term of assemblage coordinates to account for spatial autocorrelation.

Variable		Predictor	Linear model				Generalized additive model
Variable		Predictor	slope±SE	t	P	R²	slope±SE	t	P	R²
Average colour lightness	Raw	MAT	19.47±0.46	42.76	<0.001	0.24	9.89±0.79	12.57	<0.001	0.59
		Elevation	4.28±0.17	24.58	<0.001		2.10±0.23	9.05	<0.001
		EVI	–5.12±0.14	–35.48	<0.001		–2.63±0.18	–14.52	<0.001
		UVB	–7.84±0.39	–20.35	<0.001		–8.44±0.66	–12.70	<0.001
	P component	MAT	1.59±0.16	10.22	<0.001	0.08	2.70±0.21	13.09	<0.001	0.70
		Elevation	–0.47±0.06	–7.80	<0.001		0.32±0.06	5.19	<0.001
		EVI	–1.07±0.05	–21.63	<0.001		–0.31±0.05	–6.52	<0.001
		UVB	0.10±0.13	0.76	0.447		–2.95±0.17	–16.94	<0.001
	S component	MAT	17.88±0.36	49.28	<0.001	0.27	7.20±0.69	10.39	<0.001	0.52
		Elevation	4.75±0.14	34.19	<0.001		1.79±0.20	8.74	<0.001
		EVI	–4.06±0.12	–35.24	<0.001		–2.32±0.16	–14.57	<0.001
		UVB	–7.94±0.31	–25.86	<0.001		–5.49±0.58	–9.40	<0.001
Colour lightness diversity		MAT	0.01±0.03	0.41	0.683	0.09	0.44±0.07	6.69	<0.001	0.35
		Elevation	0.14±0.01	11.32	<0.001		0.20±0.02	10.62	<0.001
		EVI	–0.12±0.01	–12.77	<0.001		0.02±0.01	1.76	0.078
		UVB	–0.24±0.03	–8.84	<0.001		–0.52±0.05	–9.82	<0.001

Note: MAT = Mean annual temperature, EVI = Mean annual EVI, UVB = Mean annual UVB. Predictors with the strongest effect per model are highlighted.

There is NO Competing Interest.

SuppInfo.docx
SUPPLEMENTARY INFORMATION
EXTENDEDDATA.docx

Frogs reveal the global importance and interplay of colour-based protective and thermoregulatory functions

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results

Overall assemblage-level analyses

Differences in the independent contributions of functions across biogeographic realms

Species-level analyses

Discussion

Declarations

References

Online Methods

Table

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1