Carotenoid-dependent ornaments can reflect animals’ diet and foraging behaviors. However, this association should be spatially flexible and variable among populations to account for geographic variation in optimal foraging behaviors. We tested this hypothesis using populations of a marine predator (the brown booby, Sula leucogaster) that forage across a gradient in ocean depth in and near the Gulf of California. Specifically, we quantified green chroma for two skin traits (foot and gular color) and their relationship to foraging location and diet of males, as measured via GPS tracking and stable carbon isotope analysis of blood plasma. Our three focal colonies varied in which foraging attributes were linked to carotenoid-rich ornaments. For gular skin, our data showed a shift from a benthic prey-green skin association in the shallow waters in the north to a pelagic prey-green skin association in the deepest waters to the south. Mean foraging trip duration and distance of the foraging site from the coast also predicted skin coloration in some colonies. Finally, brown booby colonies varied in which trait (foot vs. gular skin color) was associated with foraging metrics. Overall, our results indicate that male ornaments reflect the quality of diet and foraging – information that may help females select mates who are adapted to local foraging conditions and therefore, are likely to provide better parental care. More broadly, our results stress that diet-dependent ornaments are intimately linked to animals’ environments and that we cannot assume ornaments or ornament signal content are ubiquitous within species, even when ornaments appear similar among populations. 

We studied three colonies of brown booby from Islas Marietas (n=16; July to August 2016), Isla San Pedro Mártir (n=44; January to February 2018), and Isla San Jorge (n=17; December 2016 - January 2017) Mexico, during the incubation and chick-rearing phases of breeding. To measure foraging behavior and habitat use at sea, we tracked brown boobies from each colony for 6–10 days using miniature GPS tags. We used tracking data to 1) quantify foraging trip duration and total distance traveled on unique foraging trips (as indicators of foraging behavior), and 2) measure ocean depth and distance to coast at foraging locations (two physical variables describing the environment). The mean values of tracking metrics per individual were used for the analyses.

To evaluate skin color, we measured the reflectance of two brown booby skin traits  (gular and feet) using a handheld spectrophotometer (MINOLTA CM 2600d, Osaka, Japan), which collected reflectance spectra at 10 nm intervals from 360 to 740 nm (details in Supplementary Materials). Three sequential measurements of nonoverlapping patches of the skin were taken from each patch and the average of these three measurements was automatically calculated by the device and was used to obtain the spectra data to calculate a green chroma variable (sum of reflectance from 480-550 nm/sum of reflectance from 360 to 700 nm) for the gular and the foot regions.

For a subsample of tracked brown boobies, we analyzed the stable carbon isotopic composition of blood plasma, which was collected from birds upon retrieval of GPS tracking devices (San Jorge n=17, San Pedro Mártir n= 20, Marietas n=14). Stable carbon isotope values in brown booby blood plasma likely reflect diet over approximately 10 days (Hobson and Clark 1993) and therefore match the period of tracking.

Statistical Analyses

We used independent linear models (LM) to first, assess differences between colonies on foraging parameters, diet (estimated by plasma d¹³C), and gular and foot color and second, to evaluate the relationship between skin color (gular and foot color) and stable isotope data and tracking metrics. All variables were standardized within the colony (standardized value = X – μ / σ) before the analyses to satisfy the linearity and homoscedasticity criteria, which were tested using the Shapiro-Wilk normality test and the studentized Breusch-Pagan test, respectively. To facilitate interpretation, we converted d¹³C values to positive numbers (by adding 18.9, the absolute value of the lowest d¹³C value) before standardization. Because data from San Pedro Mártir included males during the incubation and chick-rearing period, gular and foot color were standardized according to the breeding stage. All statistical analyses were conducted in R version 4.2.1 (R Core Team 2022).