Many falcons (Falco spp.) exhibit a distinct dark plumage patch below the eye, termed the malar stripe. This stripe is hypothesised to reduce the amount of solar glare reflected into the eyes while foraging, thereby increasing hunting efficiency in bright conditions. Here, we use a novel, global-scale correlative approach to test this “solar glare hypothesis” in peregrine falcons (Falco peregrinus), the most widespread falcon species, using web-sourced photographs from across the species’ global range. We found that the size and prominence of the malar stripe were positively associated with average annual solar radiation, but not with other environmental variables, such as temperature and rainfall. Our results provide the first published evidence for the hypothesis that this plumage feature functions to reduce the amount of solar glare reflected into the falcon’s eyes, thereby improving the ability to pinpoint and target agile prey in bright conditions.

Photograph metadata for user-submitted photographs of peregrine falcons (including the name of the photographer, date and location at which the photograph was taken, GPS coordinates, photographer comments, and sex and subspecies identification) were downloaded from the Macaulay Library (https://www.macaulaylibrary.org/) and iNaturalist (https://www.inaturalist.org/) online databases, using the websites' export tools. All photographs were then downloaded using the list of image URLs and filtered to include only adult birds correctly identified as peregrine falcons. For each country represented in the photograph metadata (and for each state or province, in the case of the USA and Canada), a random selection of 50 photographs was then chosen for analysis. This selection process was repeated until the maximum number of 50 usable photographs for each country or state was reached, or until there were no more photographs available. Data on average annual solar radiation (W/m²), average annual rainfall (mm) and average minimum daily temperature (ºC) at each photograph's GPS location were then extracted from the TerraClimate dataset (http://www.climatologylab.org/terraclimate.html) using Google Earth Engine. For photographs which did not have user-submitted GPS coordinates available, the GPS coordinates for the general area were sourced from Google Maps. All values represent 30-year (1988-2018) averages taken within a 20km buffer of the photograph's GPS location. Photograph metadata and climate data were collected between March and October 2019.

For each individual bird in each photograph, a single observer then scored the characteristics of the malar stripe according to a ten-point visual scale constructed for this purpose. Five aspects of the malar stripe were quantified: width, contiguity with the hood (extent of dark plumage between the malar area and the hood), prominence (darkness), length, and elongation (ratio of length score to width score). The same observer also scored the angle and position of the bird's head in each photograph according to a visual scale (see ReadMe for details). The dataset was then filtered to include observations of uniform photograph quality and angle, and from these, a representative sample of 213 observations was selected. Using ImageJ (https://imagej.nih.gov/), the same observer defined the malar stripe area on each photograph using the freehand selection tool, and converted these selections to binary image masks (Edit>Selection>Create Mask, Process>Binary>Make Binary). All masks were standardized to a width of 40px (Image>Adjust>Size), and the area and perimeter in pixels, circularity, and roundness of the thresholded area, as well as the major and minor axis lengths and aspect ratio of the ellipse fitted to each selection, were calculated using ImageJ's shape measurement tools (Analyze>Measure). The same menu tool was also used to calculate the mean and median grey values of the initial freehand selection for each image.

See the ReadMe file for descriptions of the variables used in this dataset and package installation requirements. For detail regarding the malar stripe and head position scores and how these were calculated, see the supplementary material of the associated publication. 

All statistical analysis for this study was performed using R version 4.0.3. Instructions for running the code and descriptions of outputs are provided in the descriptive text within the R code file, and in the associated publication.

A list of all photographs analysed in this study is included as a set of appendices accompanying the associated publication.