Global patterns of colouration complexity in the Paridae: Effects of climate and species characteristics across body regions

J Anim Ecol. 2025;94:1461–1473. | 1461wileyonlinelibrary.com/journal/jane Received: 24 March 2024 | Accepted: 15 May 2025 DOI: 10.1111/1365-2656.70077 RESEARCH ARTICLE Global patterns of colouration complexity in the Paridae: Effects of climate and species characteristics across body regions David LópezIdiáquez1,2,3 | Claire Doutrelant2 | Peter B. Pearman1,4,5 This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. © 2025 The Author(s). Journal of Animal Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society. 1Department of Plant Biology and Ecology, Faculty of Sciences and Technology, University of the Basque Country (UPV/ EHU), Leioa, Bizkaia, Spain 2CEFE, Université de Montpellier, CNRS, EPHE, IRD, Montpellier, France 3Department of Biology, Edward Grey Institute, University of Oxford, Oxford, UK 4IKERBASQUE Basque Foundation for Science, Bilbao, Spain 5BC3 Basque Center for Climate Change, Scientific Campus of the University of the Basque Country, Leioa, Bizkaia, Spain Correspondence David LópezIdiáquez Email: david.lopezidiaquez@biology. ox.ac.uk; [email protected] Funding information European Commission, Grant/Award Number: GBTAFTA3027 20; Basque Country Government, Grant/Award N u m b e r : P O S - 2 0 1 9 - 1 - 0 0 2 6 Handling Editor: Beatriz Willink Abstract 1. Avian plumage colouration is an iconic example of trait variability among species. Sexual, social and natural selection, and the environmental variables modulating them are the main drivers of this variability. 2. So far, most research exploring environmental effects on the variability of plumage colouration has focused on the variation in overall plumage darkness. Research on other aspects of colour variation, such as the diversity of colours exhibited by a species (i.e. colour complexity), is limited and has produced inconsistent results. Furthermore, colour complexity has mostly been analysed at the wholeplumage level, despite the possibility that the colour complexity of different plumage patches may be sensitive to different environmental factors. 3. Here, we quantify male and female colouration in 58 species of the family Paridae, and use multipredictor Bayesian phylogenetic mixed models to estimate the relationship of colouration with biotic and climatic variables that quantify environment, and with several speciesspecific characteristics. We consider both the colouration of the whole plumage and the colouration of four separate colour patches (head, chest, back and wing). 4. We find that Paridae species in climates with intermediate temperatures present more complex colouration than do species in warmer/colder climates. In addition, males, relatively small species, and species with relatively greater sexual dichromatism have more complex plumage colouration. We find that the numbers of predators and sympatric conspecifics are more associated with female colouration than with male colouration. Finally, the strength of the associations with colour complexity is specific to each plumage region: species recognition, beak size and climate variables related to competition for reproductive resources (i.e. precipitation seasonality) are more strongly associated with colouration complexity of the head and breast than with that of the back and wing. 5. Overall, our results illustrate the importance of climatic and social variables, the link between colour complexity and dichromatism in both sexes, and the analysis 1462 | LÓPEZ-IDIÁQUEZ et al. 1 | INTRODUCTION Understanding the factors that promote interspecific phenotypic diversification is a major aim of ecology and evolution. Avian plumage colouration is an iconic example of trait variability, with some species being dull and homogeneous and others colourful, varied and complex (Cooney et al., 2022; Dale et al., 2015). Plumage colouration exhibits diverse functions, including signalling (Terrill & Shultz, 2022), camouflage (Koskenpato et al., 2020) and thermoregulation (Rogalla et al., 2022), which are variously influenced by natural, sexual and social selection (Delhey, Valcu, Muck, et al., 2023b; Gomez & Théry, 2007). The specific functions and associated evolutionary processes vary among plumage patches (Terrill & Shultz, 2022). For instance, head and chest colours are usually involved in communication and are probably more strongly influenced by sexual selection than is back colouration, which is often involved in camouflage and likely evolves under natural selection (Gomez & Théry, 2007; Terrill & Shultz, 2022). However, our knowledge of the associations between the colouration of different plumage areas and environmental conditions is limited. Thus, understanding the processes that promote species differences in colouration requires examination of how environmental factors and species traits are associated with characteristics of both the entire plumage and distinct plumage areas. Research exploring environmental effects on the variability of plumage colouration has mostly focused on the association between different abiotic and biotic environmental variables and the expression of a certain colour component, like brightness or hue (Delhey, Valcu, Muck, et al., 2023b; Shultz & Burns, 2013). For instance, many studies have analysed the validity of Gloger's rule, usually finding that the plumage of avian species in wetter and warmer areas tends to be darker than that of species in drier and colder areas (Delhey, 2019; Passarotto et al., 2022). Some research has also been conducted on the evolution of carotenoidbased colours, showing that they are more likely to evolve in areas with higher primary productivity (Delhey, Valcu, Dale, et al., 2023a). However, the role of environmental factors in shaping variation in the diversity of colours exhibited by a species, that is its colour complexity, has received less attention. Variation in temperature, precipitation and their seasonality are likely associated with colour complexity through their effects on resource availability and competition for reproductive resources. For instance, resource availability in seasonal environments usually occurs in pulses, limiting the duration of the breeding season and enhancing competition for mates and territories (Botero & Rubenstein, 2012; MacíasOrdóñez et al., 2013). These conditions imply that sexual selection should be stronger in seasonal than in aseasonal environments (Barber et al., 2024) and, thus, species inhabiting seasonal environments should exhibit more complex colouration (Macedo et al., 2022). In addition to the effects of climate on resource availability and variability, climate can also impose direct constraints on colouration through the physiological demands of thermoregulation and the advantages of camouflage (Delhey et al., 2019; Galván et al., 2018). Colouration complexity, for example, could be limited in cold and arid climates, where camouflage against snowy and/or brown/monochromatic surroundings and avoidance of overheating favour light or homogeneous colouration, leading to a nonlinear relationship between colour complexity and temperature (Delhey et al., 2019; Galván et al., 2018). Studies testing the association between climate and colour complexity have, however, presented mixed results when examining wholeplumage colouration. Temperature, for instance, is positively associated with colour complexity in acanthizid thornbills (Friedman & Remeš, 2017), in all passerines (Cooney et al., 2022), and in female, but not male, antbirds (Macedo et al., 2022). Precipitation is positively linked to wholeplumage colour complexity in male, but not female, antbirds (Macedo et al., 2022), in thornbills (Friedman & Remeš, 2017) and in all passerines (Cooney et al., 2022). Finally, climate seasonality is positively associated with more complex, wholeplumage colouration in thornbills and female antbirds, but not in male antbirds or in honeyeaters (Friedman & Remeš, 2017; Macedo et al., 2022). The lack of consistent results in the few available studies highlights the potential for variation in the links between climate and plumage colouration complexity. Further clarification of the relationship between climate and colour complexity could benefit from more nuanced quantification of climate variation, an allowance for nonlinear associations between climate variables and colour complexity, and examination of the association between climatic variables and colour complexity in distinct plumage areas. In addition to climate, the diversity of habitats used by a species is likely involved in the evolution of colour complexity. Selection can act to maximise camouflage in a particular habitat or act to increase the conspicuousness of certain colour patches given needs for communication and the habitat context (Gomez & Théry, 2007). For instance, forestdwelling passerine birds usually exhibit more complex colouration than do species in open habitats of distinct plumage areas for understanding global patterns of colouration complexity and the processes that promote them. KEYWORDS bird colouration, climate, colour complexity, Paridae, sexual dichromatism, sexual selection, signalling, thermoregulation 13652656, 2025, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2656.70077 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [12/11/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 1463 LÓPEZ-IDIÁQUEZ et al. (Cooney et al., 2022). Still, the influence of the diversity of habitats used by birds on their colouration complexity remains little explored. Although some bird, reptile and moth species with broad habitat use have greater interspecific colour complexity than do habitat specialists (Forsman & Åberg, 2008; Forsman et al., 2020; Passarotto et al., 2018), more research is needed to confirm the generality of this pattern. In addition to environmental influences, a variety of biotic factors, including sexual selection, competition and predation and species characteristics, such as body and beak size, are likely involved in the evolution of plumage colouration complexity (Table 1). For example, colouration plays an important role in signalling quality to potential mates, and sexual dichromatism, which can partly reflect sexual selection intensity, is associated with greater colour complexity in males (Cooney et al., 2022). Further, divergence in colour traits among species can contribute to the maintenance of reproductive isolation (WestEberhard, 1983). Species that cooccur with relatively many confamilial species exhibit more complex colouration than those cooccurring with fewer confamilials, likely because the former species face greater risk of heterospecific mating than do the latter (Doutrelant et al., 2016; Simpson et al., 2021). In addition, several studies document that high predation pressure is associated with reduced colour complexity in birds (Bliard et al., 2020). If smaller avian species face weaker predation pressure than do larger species, we expect a negative association between colour complexity and body size, due to constraints placed by the need for camouflage (Cooney et al., 2022, but see Dale et al., 2015). Here, we analyse plumage colouration complexity in the family Paridae (Aves: Passeriformes), a widely distributed taxon, the species of which have developed ample variation in colouration complexity (Figures 1 and 2). We use the Paridae to examine the association of climate, life history, habitat use, sexual dichromatism and biotic interactions with colouration complexity. We use data on the whole plumage and on four specific areas of the plumage (head, chest, back and wing) in modelling sexspecific associations for the effects of sympatry, predation and social and sexual selection, as the associations may differ between males and females (Cooney et al., 2022; Doutrelant et al., 2020; Fargevieille et al., 2023). We expect to find greater colouration complexity in smaller, more dichromatic species that live in seasonal climates, while encountering low predation pressure and relatively many sympatric confamilial species (Table 1). Further, we expect (a) head and chest colour complexity to be associated with variables linked to sexual/social selection, due to their role in communication and (b) the complexity of the colouration of the back and wing to be associated with factors linked to natural selection, as these plumage areas likely function in providing camouflage and/or thermoregulation. We examine these associations by using data on environment and species ranges from public sources and data on bird traits and behaviour from published sources and measurement of museum specimens. The analyses are conducted in a causal framework using a directed acyclical graph to inform the selection of appropriate covariates, and phylogenetic mixed models to account for the nonindependence of confamilial species. 2 | METHODS 2.1 | Ethical note Since this study did not involve the use of live specimens, ethical approval was not required. 2.2 | Spectrometry and avian visual models We measured the plumage colouration of 58 out of the 63 Paridae species (Johansson et al., 2018) at the Natural History Museum at Tring, United Kingdom, using a spectrometer (AVASPEC2048, Avantes, Apeldoorn, Netherlands) with a deuteriumhalogen light source (AVALIGHTDHS lamp), covering a spectral range of 300 to 700 nm and a 200 μm optical probe (FCR7UV200245ME). We measured three males and three females of each species, except in a few instances where fewer were available (Table S1). We made three replicate measurements of 22 predefined plumage patches (Figure S1) in wellpreserved adult specimens. When a specimen had a patch including more than one colour, we measured all colours (range: 22 to 27 patches, see Table S1 for further detail). All specimens were measured by a single person (DLI). We used the average of the three spectra taken in each patch of each specimen to compute colour complexity of each specimen using pavo (v. 2.7.1; Maia et al., 2019). We calculated relative colour volumes respective to the avian tetrahedral space as the convex hull that included all patches in the cloud of points. We used equal illuminances across wavelengths and the blue tit (Cyanistes caeruleus) visual system (UV sensitive) to reconstruct how colours are perceived by conspecifics (Fargevieille et al., 2023). Colour volume is regularly used as a proxy for the diversity of colours in avian plumage, and larger colour volumes indicate greater colour complexity (Cooney et al., 2022; Doutrelant et al., 2016; Fargevieille et al., 2023; Macedo et al., 2022). Finally, we used the mean chromatic distance (dS) between values for male and female patches within a species as a measure of sexual dichromatism (Price et al., 2024). 2.3 | Climate information We obtained climate data within the range of each Paridae species from the bioclimatic variables (BIO) available from WorldClim (v. 2.1; Fick & Hijmans, 2017) at a 2.5arcminute resolution using the package raster (v. 3.5–29; Hijmans, 2022). We based our geographic analyses on a dataset of Paridae range maps produced by Birdlife International (2021). We resolved the avian taxonomic and distribution differences between the Johansson tree (Johansson et al., 2018) 13652656, 2025, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2656.70077 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [12/11/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 1464 | LÓPEZ-IDIÁQUEZ et al. and those in the BirdLife maps by manually editing the distribution range maps. We focused our environmental analysis on the breeding ranges and regions inhabited (or probably inhabited) by our focal species, as done previously (Cooney et al., 2022). We extracted from within each range annual mean temperature (BIO1), isothermality (BIO3), temperature seasonality (BIO4), temperature annual range (BIO7), annual precipitation (BIO12) and precipitation seasonality (BIO15). In addition, we calculated annual precipitation range as precipitation in the wettest month (BIO16) minus precipitation in the driest month (BIO17). We reduced the dimensions of the climatic variables associated with temperature and precipitation variability using principal component analysis (PCA). We included the variables isothermality (BIO3), temperature seasonality (BIO4) and temperature annual range (BIO7) to represent temperature variability. We retained the first axis of the PCA, explaining 95.0% of the variance, in which higher values of the axis represented more seasonal temperatures. We successfully used precipitation seasonality and precipitation annual range to represent precipitation variability and conducted a separate PCA. We retained the first axis of this PCA, which explained 74.4% of the variance. Higher values on this axis represent greater precipitation seasonality (see Supporting Information S2 for further details on the PCAs). 2.4 | Predation pressure We used Birdlife International species range maps and the R package raster (Hijmans, 2022) to estimate predation pressure for each Paridae species. We estimated predation pressure as the number of sympatric avian predator species across its range, a metric that has been successfully used as an index of predation pressure (Bliard et al., 2020; Buckley & Jetz, 2007; Valcu et al., 2014, 2023). Predator densities or estimates of predatordriven mortalities in our studied species would provide more specific information on the pressure experienced by a species; however, this information is not available for such a TABLE 1 Explanatory variables included in our models and their expected associations with colour complexity. Fixed effects are the variables included in the model to test each prediction. Variable Prediction Fixed effects Annual mean temperature (t) Species living in areas with intermediate temperatures within the Paridae ranges will exhibit more complex colouration than species in colder/warmer areas (1–3) t + t2 + sex Annual precipitation (p) Species living in areas with higher precipitation will exhibit more complex colouration (4–6) p + t + sex Temperature seasonality (ts) Species in more seasonal areas will exhibit more complex colourations (7–10)ts + t + sex Precipitation seasonality (ps) ps + ts + p + sex Body size (bos) Smaller species will exhibit more complex colourations (5, 11, 12, but see 27)bos + t + sex Beak size (bes) Beak size has been linked to different ecological and behavioural aspects of the species that can impose constraints and pressures on bird colouration, but we do not have a strong expectation about the direction of this association (13–17) bes + bos + t + sex Predation (pr) Species under stronger predation pressure will have less complex colourations, and the effects will be stronger in males than in females (18–21) pr + bos + sex + pr × sex Sympatry (sy) Species inhabiting areas with more confamilials will display more complex colouration, with stronger effects in males than in females. (22–26) sy + bos + sex + sy × sex Sex Males will display more complex colouration than females (5, 27, 28)sex Habitat breadth (hb) Species with greater breadth of habitat use will display more complex colouration (29–31) hb + ps + ts + sex Migration (m) Migrants will display less complex colouration than residents (32, 33)m + ts + ps + sex Sexual dichromatism (sd) More dichromatic species will have greater complexity than monochromatic ones, with stronger effects in males than in females (5, 28, 34) sd + ts + ps + sex + sd × sex Note: 1. K. Delhey, Ecol Lett. 22, 726–736 (2019). 2. N. R. Friedman, Global Ecology and Biogeography. 26, 261–274 (2017). 3. I. Galván, Funct Ecol. 32, 1531–1540 (2018). 4. G. E. Maurer, Ecol Lett. 23, 527–536 (2020). 5. C. R. Cooney, Nature Ecology & Evolution. 6, 622–629 (2022). 6. K. Delhey, Journal of Animal Ecology. 92, 66–77 (2023). 7. B. J. Stutchbury, Behavioural Ecology and Sociobiology. 29, 297–306 (1991). 8. C. A. Botero, Current Biology. 19, 1151–1155 (2009). 9. C. A. Botero, D. R. Rubenstein, PLoS ONE. 7, e32311 (2012). 10. R. MacíasOrdóñez, in Sexual Selection pp. 1–32 (2013) 11. R. A. Kiltie, Functional Ecology. 14, 226–234 (2000). 12. I. Galván, Acta ornithologica. 48, 65–80 (2013). 13. P. R. Grant, Animal Behaviour. 29, 785–793 (1981). 14. P. R. Grant, Science. 313, 224–226 (2006). 15. T. Price, The Journal of Animal Ecology. 60, 643 (1991). 16. N. R. Friedman, Evolution. 71, 2120–2129 (2017). 17. A. G. Gosler, Ibis. 129, 451–476 (1987). 18. E. Huhta, Ecology. 84, 1793–1799 (2003). 19. A. P. Møller, 60, 227–233 (2006). 20. J. F. Husak, Ethology. 112, 572–580 (2006). 21. L. Bliard, Biology Letters. 16, 20,200,002 (2020). 22. P. R. Martin, The American Naturalist. 185, 443–451 (2015). 23. P. R. Martin, Evolution. 64, 336–347 (2010). 24. C. Doutrelant, Ecology Letters. 19, 537–545 (2016). 25. R. K. Simpson, Proceedings of the Royal Society B. 288, 20,202,804 (2021). 26. A. B. Luro, Journal of Evolutionary Biology. 35, 1558–1567 (2022). 27. J. Dale, Nature. 527, 367–370 (2015). 28. C. R. Cooney., Nat Commun. 10, 1773 (2019). 29. A. Forsman, Ecology. 89, 1201–1207 (2008). 30. A. Forsman, Ecography. 43, 823–833 (2020). 31. K. Delhey, J. Evol. Biol. 26, 1559–1568 (2013). 32. T. Alerstam, Oikos. 103, 247–260 (2003). 33. R. K. Simpson, Proc. R. Soc. B. 282, 20,150,375 (2015). 34. A. J. Shultz, Evolution. 71, 1061–1074 (2017). 13652656, 2025, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2656.70077 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [12/11/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 1465 LÓPEZ-IDIÁQUEZ et al. largescale study. Avian predators of adult birds included species in the orders Accipitriformes and Falconiformes. Nocturnal avian predators (Strigiformes) and mammals were not included in the study because we did not expect them to impose strong selection pressures on colouration. The retina of Strigiformes species eyes contains mostly rod cells (Harmening & Wagner, 2011), resulting in weak colour vision. Mammalian predators are mostly blind to ultraviolet and have poor colour vision, as they only have two types of photoreceptors (vs. four in birds; Bowmaker, 2008; Guilford & Harvey, 1998). Finally, although there are several snake species that feed on birds, we did not include snakes in our analyses, since the great majority of them predate on eggs and chicks in nests (Barends & Maritz, 2022; Weatherhead & BlouinDemers, 2004) and, thus, are unlikely to impose selection pressure on avian colouration. To control for potential size mismatches between predator and prey species (i.e. counting predators with much larger or smaller prey), we computed predator–prey mass allometry relationships, then excluded those Accipitriformes and Falconiformes when size mismatches with potential prey were detected, as previously done by Valcu et al. (2014) and Bliard et al. (2020; see Supporting Information S3 for further details). 2.5 | Sympatry with other Paridae Using the range maps of the Paridae and the package raster (Hijmans, 2022), we calculated the degree of sympatry with other FIGURE 1 Variation in plumage colour complexity across the Paridae family. Orange and blue bars represent the average colour volumes of males and females, respectively. Larger volumes indicate greater colour complexity. We plotted the raw values but note that in the analyses we have used a logtransformed version of this variable (see Section 2). In Poecile hypermelaena, the value is the same in males and females as the sex of measured individuals was not determined and in Poecile davidi there is no value for females as we only measured males. The red dots by tip names indicate the depicted species. Colour complexity values represent the average of the three measures taken in each male and female (see Section 2). Species illustrations by Hilary Burn ©Lynx Editions. 13652656, 2025, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2656.70077 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [12/11/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 1466 | LÓPEZ-IDIÁQUEZ et al. Paridae species as the number of confamilial species with ranges overlapping the range of a focal species. 2.6 | Morphology We extracted data on tarsus, wing and tail lengths, and body mass of Paridae species from the AVONET dataset (Tobias et al., 2022) and the Handbook of the Birds of the World (Billerman et al., 2022). We conducted PCA of these data to obtain a proxy of body size (hereafter body size). Specifically, we retained the species scores on the first PCA axis, which explained 71.5% of the variance in these data. Additionally, one of us (DLI) measured beak length and width (to the nearest 0.1 mm) from the museum specimens using a digital calliper. We used PCA to summarise these two variables as a single proxy of beak size (here after beak size). We retained species scores on the first axis of this PCA, which explained 81.0% of the data variation (see Supporting Information S2 for further information on the PCAs). 2.7 | Habitat breadth We computed habitat breadth by compiling information about the general habitat (e.g. forest) and subhabitat (e.g. temperate or boreal) from the Habitat Classification Scheme of the IUCN (IUCN, 2022). Briefly, we calculated habitat breadth using an integer value corresponding to the number of general habitats a species occupies and a decimal figure of the number of subhabitats it occupies (for further details see Supporting Information S4 and Estrada et al., 2018). Thus, higher values of this variable represent species inhabiting a more diverse set of environments. 2.8 | Migration We obtained information about the migratory behaviour of our focal species from the IUCN (IUCN, 2022). Specifically, we classified species into two categories based on whether they exhibit altitudinal migration or not, as the species within the Paridae family do not perform longdistance migratory movements. 2.9 | Statistical analyses To explore the association between colouration complexity and our predictor variables (Table 1) we fitted multipredictor Bayesian phylogenetic mixed models using the brms package (v. 2.34; Bürkner, 2017) in R (v. 4.2.0, R Core Team, 2022). The models included a phylogenetic random effect to control for phylogenetic nonindependence of the species within the Paridae family, based on the Paridae tree published by Johansson et al. (2018). This tree was constructed using Bayesian inference and was strongly supported, with most nodes achieving probabilities of 1.00 (for further details, see Johansson et al., 2018). We also included species as a random effect to control for the nonindependence of the measures taken in different specimens of the same species. We fitted separate models for the colouration complexity of the entire plumage and each plumage area (i.e. head, chest, back FIGURE 2 Maps illustrating the geographical distribution of (a) Paridae mean colouration complexity (scaled), (b) Paridae species richness, (c) mean annual temperature (BIO1) and (d) temperature seasonality (BIO4) within the Paridae distribution range. Annual mean temperature and seasonality represent the information obtained from WorldClim (see methods). In A, we have plotted the raw colour complexity values but note that in the analyses we have used a logtransformed version of this variable (see Section 2). 13652656, 2025, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2656.70077 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [12/11/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 1467 LÓPEZ-IDIÁQUEZ et al. and wing; Figure S1). The dependent variable in these models was the colour volume, which was logtransformed to achieve normality (Fargevieille et al., 2023). To select the fixed effect structure of the models, we built a direct acyclical graph (DAG) using the package DAGitty (v. 0.3–4, Textor et al., 2016). A DAG is a diagram of the causal relationships between the variables in a study and allows identification of the minimal set of variables for inclusion in a model for inferring causal relationships with the dependent variable (Textor, 2023; for further information on the causal relationship between our explanatory variables see Supporting Information S5). All explanatory variables were standardised to a mean of zero and standard deviation of one to facilitate comparison of effect sizes (see Supporting Information S6 for further information on the variability of the explanatory variables). In all models, we included sex as a covariate. The sex of the two measured Poecile hypermelaena individuals was unknown. To include this species in the analysis, we randomly assigned the sex of these two specimens and repeated the analyses with all possible sex combinations to control for the potential biases of our assignment. The results of all models are consistent, showing that our random assignation of sex did not bias any of our results (see Supporting Information S7). All models included default priors, with the exception of the fixed effects, which included a Normal (0, 1) regularising prior (Macedo et al., 2022). The models were run for 10,000 iterations across four chains, with a warmup of 2000 iterations and a thin of 10, except the model of the association between head colour complexity and temperature and precipitation seasonality, which was run for 100,000 iterations across four chains, with a warmup of 2000 iterations and a thin of 100. In each case, the values were selected to ensure model convergence and good effective sample sizes, which were assessed by inspection of trace plots and the GelmanRubin convergence diagnostic. We estimated phylogenetic signal (λ) from models including sex as a predictor as the phylogenetic heritability (Lynch, 1991), λ = σ2Phylo/(σ2Phylo + σ2 Res), where σ2Phylo and σ2 Res are the variances due to phylogeny and unmodelled sources (residual variance), respectively (De Framond et al., 2025; de Villemereuil & Nakagawa, 2014; Lynch, 1991; Macedo et al., 2022). 3 | RESULTS We find that wholebody plumage colour complexity shows a nonlinear association with annual mean temperature, a negative association with body size and a positive association with sexual dichromatism, presenting a stronger relationship in males than in females (Table 2, Figure 3a). Females exhibit less complex wholebody colouration than males do. Further, although the 95% credible intervals (CI) include zero, the posterior distribution of beak size strongly gravitates towards positive values of association (i.e. >90% of the posterior distribution was positive; Figure 3a). Similarly, the posterior distributions for predation and sympatry strongly gravitate towards negative values of association for females, indicating lower TABLE 2 Posterior means and 95% credible intervals (in brackets) from models linking explanatory terms with plumage colour complexity of the entire body and of five separate plumage patches. Mean annual temperature2Annual precipitation Temperature seasonality Precipitation seasonality Habitat breadth Predationfem Predationmal Sympatryfem Whole body −0.212 [−0.395, −0.023] 0.151 [−0.092, 0.399] 0.105 [−0.318, 0.523] 0.085 [−0.248, 0.430] 0.024 [−0.238, 0.292] −0.251 [−0.613, 0.121] −0.240 [−0.614, 0.134] −0.203 [−0.428, 0.027] Head −0.046 [−0.286, 0.201] 0.166 [−0.157, 0.483] 0.083 [−0.410, 0.587] 0.293 [−0.070, 0.643] 0.092 [−0.200, 0.390] −0.273 [−0.704, 0.158] −0.282 [−0.726, 0.169] −0.317 [−0.604, −0.006] Chest −0.226 [−0.529, 0.088] 0.158 [−0.229, 0.544] 0.311 [−0.375, 1.044] −0.124 [−0.614, 0.441] 0.231 [−0.175, 0.609] −0.556 [−1.150, 0.049] −0.297 [−0.917, 0.334] −0.164 [−0.520, 0.278] Back −0.304 [−0.573, −0.036] 0.174 [−0.179, 0.521] 0.034 [−0.569, 0.648] 0.032 [−0.419, 0.477] −0.084 [−0.469, 0.326] −0.196 [−0.736, 0.326] −0.106 [−0.644, 0.448] −0.164 [−0.520, 0.278] Wing −0.155 [−0.501, 0.177] 0.135 [−0.272, 0.550] 0.103 [−0.560, 0.776] 0.107 [−0.445, 0.719] 0.036 [−0.415, 0.502] −0.022 [−0.675, 0.646] 0.066 [−0.609, 0.753] −0.005 [−0.447, 0.403] Sympatrymal Migrantyes Beak size Body size Sexual dichromatismfem Sexual dichromatismmal Sexfem λ Whole body −0.092 [−0.319, 0.139] −0.135 [−0.886, 0.594] 0.109 [−0.017, 0.239] −0.212 [−0.398, −0.010] 0.692 [0.489, 0.901] 0.865 [0.665, 1.072] −0.185 [−0.262, −0.109] 45.46% [26.16, 62.84] Head −0.164 [−0.502, 0.286] −0.851 [−1.662, 0.040] 0.254 [0.031, 0.488] −0.080 [−0.348, 0.193] 0.526 [0.271, 0.775] 0.811 [0.557, 1.064] −0.057 [−0.197, 0.085] 18.20% [4.95, 32.82] Chest −0.117 [−0.511, 0.262] 0.030 [−0.967, 1.000] 0.276 [−0.047, 0.601] 0.013 [−0.402, 0.450] 0.675 [0.233, 1.124] 0.978 [0.557, 1.441] −0.180 [−0.381, 0.017] 29.59% [15.30, 42.63] Back −0.198 [−0.527, 0.145] 0.150 [−0.831, 1.087] 0.173 [−0.094, 0.445] −0.435 [−0.751, −0.116] 0.725 [0.352, 1.110] 0.904 [0.530, 1.289] −0.237 [−0.411, −0.074] 21.11% [7.74, 35.53] Wing −0.136 [−0.544, 0.275] −0.390 [−1.500, 0.704] 0.159 [−0.224, 0.527] −0.504 [−0.922, −0.080] 0.698 [0.184, 1.190] 0.589 [0.075, 1.083] −0.219 [−0.466, 0.023] 23.09% [12.27, 35.44] Note: λ is the estimated phylogenetic signal (see Section 2). 13652656, 2025, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2656.70077 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [12/11/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 1468 | LÓPEZ-IDIÁQUEZ et al. female colour complexity in species that cooccur with relatively many confamilial species and predators (Figure 3a, see Tables S1– S10 for further details on model results). The colour complexity of distinct plumage patches presents a variety of associations with the environment, species morphology and biotic interactions. Head colour complexity shows a positive association with variation in beak size and sexual dichromatism, which is stronger in males than in females (Table 2, Figure 3b). The colour complexity of the head of females is negatively associated with the number of sympatric species, but head colour complexity of males is not (Table 2, Figure 3b). Further, the posterior distribution of precipitation seasonality strongly tends towards a positive association with colour complexity, while the posterior distributions of migration tend towards negative associations, although this association is estimated with high uncertainty (Figure 3b; see Tables S11–S20 for further details on model results). For the chest patch, we find that colour complexity shows a positive association with sexual dichromatism for both males and females (Table 2, Figure 3c). The posterior distribution for the quadratic component of mean annual temperature tends towards negative values, suggesting higher chest colour complexity at intermediate than at colder/warmer temperatures (Figure 3c). The posterior distribution of temperature variability strongly tends towards positive values, while the posterior distributions of both predation and sex in females tend towards negative values of association (Figure 3c; see Tables S21–S30 for further details on model results). The colour complexity of the back presents a nonlinear association with mean annual temperature (i.e. higher colour complexity at intermediate temperature, as shown by a negative quadratic term), a negative association with body size and a positive association with sexual dichromatism, which is stronger in females than in males (Table 2, Figure 3d; see Tables S31–S40 for further details on model results). The colour complexity of the wing covaries negatively with body size and positively with sexual dichromatism (Table 2, Figure 3e). For sex, although the 95% CIs include zero, the posterior distribution suggests that wings of females tend to have less complex colouration than do those of males (Figure 3e; see Tables S41–S50 for further details on model results). 4 | DISCUSSION Our findings show that variation in plumage colouration complexity among Paridae species is associated with climatic and ecological variation, as well as with factors that are intrinsic to the species. Our results also show that the relationship of colour complexity with the environmental variables we included depends on the plumage patch we considered and underscores the importance of considering distinct plumage areas separately, in order better to understand the relationship between environmental variation and avian colour complexity. FIGURE 3 Associations between the colour complexity of the whole plumage (a), head (b), chest (c), back (d) and wing (e) with the scaled predictors. Black dots and whiskers represent the mean ± 95% credible intervals (CI), and the density curve indicates the posteriors distributions for each predictor. Density plots with solid colours represent associations in which the 95% CIs do not include zero. Density plots with intermediate shading denote associations in which the 95% CIs includes zero but strongly gravitate (>90% of the posterior distributions) towards positive or negative values. Density plots with transparent colours denote nonsupported associations. Purple, blue and dark orange shades represent climatic, biotic and intrinsic variables, respectively. Grey represents sex. (a) (b) (c) (d) (e) 13652656, 2025, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2656.70077 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [12/11/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 1469 LÓPEZ-IDIÁQUEZ et al. 4.1 | Whole plumage Research exploring the association between plumage colouration complexity and temperature variation has yielded contrasting results as some studies report significant positive associations (Macedo et al., 2022) while other studies do not (Cooney et al., 2022; Friedman & Remeš, 2017). Until now, the association between temperature and colour complexity has been analysed assuming only linear trends may exist. Our findings, in contrast, suggest that the association is likely more complex than previously thought, in that colour complexity is higher in Paridae species in regions with intermediate temperatures compared to those species in warmer and colder areas within the Paridae distribution ranges. This pattern resembles that of previous results on plumage darkness, in which passerine species that experience extreme temperatures within their ranges typically exhibit lighter colouration than other passerines (Delhey et al., 2019). One explanation for this is that lighter plumages are favoured in such areas due to their role in providing camouflage and aiding in thermoregulation (Delhey et al., 2019; Galván et al., 2018). We believe that a comparable phenomenon may explain our results on colour complexity. For instance, selection favouring uniform colouration that enhances concealment against snowy surfaces may constrain the evolution of complex colouration at high latitudes. A complementary explanation is that the resources required to exhibit complex colours may be limited in extremely warm or cold areas. For example, carotenoids are likely less available in areas with extreme temperatures due to limited plant productivity (Zhang et al., 2017). Beyond being the primary pigments for flashy avian colours (Thomas et al., 2014), carotenoids also serve as crucial immuneenhancers and antioxidants (Weaver et al., 2018). Natural selection in areas with extreme temperatures may favour allocation of carotenoids to maintaining robust antioxidative and immune functions, rather than to developing colourful plumage pigmentation. The relevance of sexual selection as a driver of plumage colouration complexity is highlighted by our results showing that greater dichromatism is associated with more complex colouration. This aligns with previous work on passerines (Cooney et al., 2022) and tanagers (Shultz & Burns, 2017) in highlighting the role of sexual selection as an evolutionary driver of colouration complexity. However, while these studies report significant results in males and not females, our results for the Paridae support the association between sexual dichromatism and plumage colour complexity in both sexes. Thus, sexual selection in the Paridae may also explain amongspecies variation in female colour complexity rather than it being simply a byproduct of selection on males. The sensitivity of female colouration complexity to sexual selection suggests that female colouration, rather than being a genetic correlate of colouration in males, plays a signalling role, something supported by studies of species in the Paridae (Doutrelant et al., 2020; Thys et al., 2020) and in other bird species (LópezIdiáquez et al., 2016). In the typically cavitynesting Paridae, there could be little pressure for the evolution of camouflage in females and bipaternal care may favour the evolution of female traits under sexual selection (Fargevieille et al., 2023). Our results also show that females in the Paridae display less complex colouration than males, and despite model support, we observe little difference in plumage colour complexity between males and females. This lack of strong difference between sexes can be explained by Paridae biology. As a taxon of socially monogamous species with moderate levels of extrapair paternities (Brouwer & Griffith, 2019), similar selective pressure may act on ornamental traits in both sexes. At a more general level, the divergence between our results and those reported for the Order Passeriformes (Cooney et al., 2022), the encompassing taxon of the Paridae, reveals the importance of exploring these evolutionary patterns at different taxonomic scales. Results from global analyses including thousands of species, despite being appropriate for inferring broad patterns, can be biased towards trends in numerically dominant taxa. This may conceal relationships in less diverse taxa and, thus, limit understanding of processes explaining variation at lower taxonomical levels. In line with our prediction, we found that smaller species exhibit more complex colouration. These results align with those of Cooney et al. (2022) for the whole plumage in the passerines. The negative association between colour complexity and body size could be explained by a physiological limitation on the expression of complex colours in larger birds (Galván et al., 2013) as, for instance, larger bird species have lower circulating carotenoid concentrations than smaller species (Tella et al., 2004). An alternative hypothesis is that there may be differing communication requirements associated with body size differences (Kiltie, 2000). For example, selection may favour the presence of a higher number of differently coloured patches in smaller species, which communicate at shorter distances than in larger species in which a larger, uniformly coloured patch could be more effective. Our results also show that species with larger beaks exhibit more complex colourations. As dietary niches and preferences are linked both to the tempo and mode of evolution of cranial morphology and beak size (Felice et al., 2019; Olsen, 2017; Pigot et al., 2020), this association suggests a link between colour complexity and diet. For instance, species with larger beaks could have access to carotenoidrich resources unavailable for species with smaller beaks. Alternatively, variation in beak size and shape has also been linked to nondietary factors, such as song production and thermoregulation (Friedman et al., 2019; Navalón et al., 2019), which were not considered in our study and could be linked to colouration. Finally, our results suggest a negative relationship between colour complexity and both predation intensity and sympatry with confamilials. For predation, the results partially align with our prediction of less complex colouration in species facing stronger predation pressure, as suggested by previous studies on the role of plumage colouration as a modulator of predation risk (Bliard et al., 2020; Huhta et al., 2003). While predation can explain amongspecies differences in plumage brightness (Huhta et al., 2003) and differences in colour complexity between island 13652656, 2025, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2656.70077 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [12/11/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License