Characterizing Color Diversity in Ants Using Databases and Image Analysis


Meeting Abstract

P1-296  Friday, Jan. 4 15:30 – 17:30  Characterizing Color Diversity in Ants Using Databases and Image Analysis IDEC, JH*; FISHER, BL; Hendrix College, Conway ; California Academy of Sciences, San Francisco idecjj@hendrix.edu

Ants (Order Hymenoptera, Family Formicidae) are a widespread and diverse group of insects. Despite many years of research in ant ecology and evolution, the diversity of color in ants as a group has not yet been closely examined or explained. In doing so we used a dataset consisting of ~50k images from the ant specimen database AntWeb. Images of ant heads were first segmented from their backgrounds using an active contour algorithm. The RGB colors of pixels in the segmented region were averaged and assigned to that specimen. We then tested for the effects of phylogeny, caste, microhabitat, and environment on these colors. Color metrics included HSL lightness and proxies for pigment saturation devised from Euclidean distance in the HSV color space. We hypothesized that darker and more heavily pigmented ants would be found in habitats prone to higher UV exposures or colder temperatures due to the UV-shielding and heat-absorbing properties of melanin. We found that genera, primary ant castes, and ants living in different microhabitats showed statistically significant differences in lightness and pigment saturation. Despite a strong signal for color at the genus level, ants as a group showed on average decreased lightness and greater pigmentation under conditions where they would be more likely to be exposed to sunlight. Queens and males, which often mate in sun-exposed locations, were darker on average than workers. Ants collected from trees were also darker than those from the litter or underground. Caste and microhabitat effects compared between genera and possible global effects of UV and temperature are still under investigation at the time of writing.

the Society for
Integrative &
Comparative
Biology