Sexual dichromatism in the fur of a bat: An exploration of color differences and potential signaling functions

Sex differences in body color (i.e., sexual dichromatism) are rare in bats and, more broadly, in mammals. The eastern red bat (Lasiurus borealis) is a common tree-roosting bat that occupies much of North America and has long been described as sexually dichromatic. However, previous research on this species found that absolute body size and collection year were better predictors of fur color in preserved specimens than sex. We revisited this issue and photographed 82 live eastern red bats under standardized conditions, then used image analysis to quantify pelage hue, saturation, and value. We used an information theoretic approach to evaluate four competing hypotheses about the principal drivers of color differences in the fur of eastern red bats. Our analyses demonstrated that sex was a better predictor of pelage color than body size; males had redder, more saturated, and lighter pelages than females. Additionally, fur color of juvenile vs. adult bats differed somewhat as juveniles were darker than adults. In general, absolute body size (i.e., forearm length in bats) was a poor predictor of color in live eastern red bats. In an exploratory post-hoc analysis, we confirm that fur color is related to body condition (i.e., mass in bats), suggesting color might serve as a sexually-selected signal of mate quality in this partially-diurnal species. Future work should investigate the functional role of sexual dichromatism in this species, which could be related to signaling or possibly thermoregulation. 

Methods
Bat Capture
We followed the guidelines of the American Society of Mammalogists for the use of wild mammals in research (Sikes et al. 2019), institutional animal care and use protocols (University of Illinois Urbana-Champaign, #21022), and all white-nose syndrome decontamination recommendations (White-nose Syndrome Disease Management Working Group 2020). We held appropriate Indiana and USFWS permits.

Photography
We performed all photography in the field. We held each bat in a standardized position (wings closed, forearms held against body and uropatagium spread, with bats’ ventral surface down) and photographed them in a portable LED-lit photo studio box (EMART International Inc., Westminster, CA, USA) using a digital single-lens reflex camera (EOS Rebel T6) with an EFS 18-55 mm lens (Canon, Tokyo, Japan). To achieve homogeneous lighting conditions, we did not use flash to photograph bats but relied only on the illumination produced by the two LED strips within the photo studio box. We photographed the bats from a standardized distance (35.56 cm) with camera settings at IOS 800, a shutter speed of 1/800th of a second, and f/5.6 (10.9 mm). We photographed bats against a neutral gray background and included a color correction card within each photograph. We photographed and stored all images in large format. 

Color Analyses
We performed all color analyses in ImageJ version 1.8.0 (Schneider et al. 2012; Rasband 2018). For this study, we used hue, saturation, and value scores to describe fur color. Hue describes colors in their purest forms (e.g., red, blue, etc.). Hue scores range from 0–359°, with shades of red, orange, and yellow occurring from 0–30°. Saturation describes the purity or intensity of any color, expressed as a percent. Lower saturation scores indicate less vibrant colors. Value describes how close a color is to white or black (i.e., brightness) and is also expressed as a percent. Lower value scores indicate darker colors. To quantify the color of bat fur, we selected the furred dorsal surface of each bat’s exposed uropatagium (the membrane that extends between the hind limbs) to measure (hereafter, “selected area”). For each bat, we quantified the average hue, saturation, and value scores across the selected area. We also quantified color across the whole of each bat’s exposed dorsal surface. Ultimately, the color scores of the two surface types were highly correlated (Pearson’s correlation coefficients >0.90; data not shown), and we opted to use the uropatagium scores to facilitate comparisons with earlier work by Davis and Castleberry (2010). To assess whether variation in the size of the selected area influenced our analysis, we quantified the number of pixels within each selected area. Ultimately, the selected area size was not related to hue, saturation, or value scores (linear models, F1,78 = 0.80, 0.04, and 0.19; p = 0.373, 0.838, and 0.661, respectively), and we did not include it as a covariate in our models. Overall, these scoring techniques produced reliable measurements that could be consistently replicated (for details on reliability analyses, see Appendix S1, Supplementary Text).