67.5 Friday, Jan. 6 How well do orbit dimensions predict diel activity in sciurid rodents? SMITH, S.M.*; ANGIELCZYK, K.D.; SCHMITZ, L.; WANG, S.C.; Johns Hopkins University; Field Museum of Natural History; University of California, Davis; Swarthmore College email@example.com
Eye dimensions are good predictors of diel activity patterns because they relate directly to light sensitivity of the eye. Osteological characters such as scleral ring dimensions also are reliable proxies, but orbit dimensions have proven difficult to use because soft tissues other than the eye affect orbit size and shape. Nevertheless, it would be useful if orbit dimensions could be used as proxies because it would allow predictions of activity patterns for fossil taxa such as non-mammalian synapsids, which infrequently preserve scleral rings and have poorly ossified braincases. We measured four orbit dimensions and six body size proxies in 429 sciurid specimens from 51 species. No individual measurement differed significantly among nocturnal, diurnal, and crepuscular species, but a MANOVA indicated that the three groups had significantly different orbit dimensions. Nocturnal species also had a significantly higher intercept than diurnal and crepsucular species for a regression of orbit length on skull length, although the three groups had similar slopes. The intercepts of the nocturnal taxa remained significantly higher when the regression was run using phylogenetic independent contrasts. To simulate the conditions of a fossil data set, we used logistic regression and classification trees to determine whether we could classify specimens without a priori binning. We obtained overall correct classification rates as high as 95%. The analyses based only on orbit dimensions had consistent difficulty correctly identifying nocturnal specimens, but we obtained better results when additional skull measurements were included. This raises the possibility that such a combined data set might be useful in inferring activity patterns in fossils.