Birds communicate using multiple sensory channels, most commonly through vocalisations and colourful plumage patches. Such colourful plumage has evolved through a complex interplay of processes, acting not only on the ability of a patch to convey information but also in response to physiological and environmental factors. Although much research on inter-specific variation in bird plumage has concentrated on sexual selection, much less work has considered the role of non-sexual selection and how it is affected by joint effects of avian viewing conditions and receiver vision. Here we combined taxonomically diverse databases of avian plumage measurements with habitat use and behavioural data to test the effect of three factors which effect viewing conditions - habitat openness, migratory preference, and daily phenology - on avian plumage contrast, accounting for shared evolutionary history and variation in avian visual systems. We find that habitat structure and migratory preference predicted plumage visual contrast, especially for females. Our study demonstrates the important role of non-sexually selected traits and viewing conditions in shaping avian plumage contrast.

The database contains taxonomic information on birds, eye-type (U or V) and measures of plumage contrast collected using a spectrophotometer. The database is compiled from several sources (all cited in the accompanying paper), and compiles data used in other publications.

We accessed reflection data from published and unpublished sources (including Fargevieille et al. 2023; Doutrelant et al. 2016; Eaton 2005, 2007; Cardoso and Mota 2008; Cardoso and Mota 2022; Gomes, Sorenson, and Cardoso 2016; Shultz and Burns 2017; Igic, D’Alba, and Shawkey 2018; Dunning et al. 2023; Stoddard and Prum 2008), which have been compiled into a single database, the ‘Bird Colour Base’, by Thanh-Lan Gluckman and Peter Dunn (made available here - https://github.com/BirdColorBase/home). 

Spectra were in the form of calibrated reflectance every 2nm from 300 nm to 700nm collected from 2,610 species, using similar methods and equipment, representing 35 orders and 170 families and 984 genera.  On average there were 6.8±2.8 (mean±SD) birds scanned per species and 6.2±2.6 different plumage patches scanned per bird per species. The database contains 352,155 spectra of which 174,794 are from adult males, 161,952 are from adult females, and 15,409 from individuals with unrecorded sex/maturity.  Given the high rates of sexual dichromatism and concomitant separate evolutionary drivers of sex-related plumage in adult birds, we omitted data from specimens of unknown sex or recorded as immature. We did not select any patch type specifically, opting instead to encompass the full spectrum of avian plumage contrast. However, to avoid taxonomic imbalance in our study we omitted a subset of some well-represented taxonomic groups – from studies that collected a large amount of data on a single group, for example the Estrildid finches (Gomes et al. 2016). Note though, that all data in the Bird Colour Base are included in the supplementary spectrophotometry dataset of for future use.

Because we were interested in the possible relationships between environment and plumage colour, particularly in visual communication, we measured plumage contrast using two approaches (A) bird vision models including the effects of ambient light, plumage reflectance and bird vision (following Endler & Mielke 2005) when seen under open/cloudy light environments; open/cloudy light environments occur in all habitats, including at night when it is cloudy, or any time that there is no canopy (Endler 1993); and, (B) bird vision models under the most likely light environment (Endler 1993) based on a given bird species' main habitat and diel activity. For both approaches we set the species eye type (Ultraviolet U or Violet V, following Endler & Mielke 2005) based on published data (Bloch 2015; Endler & Mielke 2005; Ödeen & Håstad 2003, 2010, 2013; Lind et al. 2014) and phylogenetic relationships (Endler & Mielke 2005; Olssen et al. 2021) using Jetz et al. (2012).  Eye models follow that of Vorobyev and Osorio (1998) as in Endler & Mielke (2005); see Renoult et al. (2015) for a detailed discussion.

we estimated several measures of within-bird contrast using all adult spectra for a given species (i.e. across different studies/patches). all are related to contrast in bird plumages and thus, conspicuousness: Luminance:  maximum (MaxL), minimum (MinL), difference (Diff), geometric mean (gMeanL) and geometric standard deviation (gSDL; luminance is log-normally distributed); Chroma: mean (MeanCr) and SD (SDCr), and hue circular SD (SDHue), see table 1. Following Dunning et al. (2023), we suggest that maxL may be predicted from the environment, but the other measures of within-bird contrast may also be related to the environment if they are important in intraspecific communication.  For example, maxL is one measure of visual contrast.  This contrast can either be within a given bird's plumage or relative to the background or both.  The difference between the lightest ('brighest") and darkest patch (Diff) also affects contrast, and gmeanL and gSDL could also affect conspicuousness.  Chroma and hue also affect conspicuousness and SD is a measure of the variation across the bird. The gmeanL and meanC are potential measures of possible contrast with the background depending upon average background colouration.