Meeting Abstract

75.3  Tuesday, Jan. 6  Orthogonal Regression and Phylogenetic Correction Applied to Mammalian Metabolic Allometry SIEG, A*; OCONNOR, M; AGOSTA, S; MCNAIR, J; GRANT, B; DUNHAM, A; Drexel Univ; Drexel Univ; Univ of Pennsylvania; Academy of Natural Sciences; Widener Univ; Univ of Pennsylvania aes48@drexel.edu

A contentious issue in metabolic ecology is the slope of the relationship between body size and metabolic rate. Slopes in metabolic allometry are often estimated via ordinary least squares regression (OLS) without considering trait variation and error variance in the independent variable. Allometric exponent estimation is also often performed without reference to phylogeny. We apply an orthogonal regression technique with variance-oriented residuals (LSVOR) to estimate the mammalian basal metabolic rate (BMR) allometric exponent. This regression technique is less biased than OLS when both physiological variables have random variation. We culled BMRs from the literature. Our database includes all available multiple unique measurements of body mass and BMR for each of 680 mammal species. To phylogenetically correct the data, we used the fullest available mammal phylogeny (Bininda-Emonds 2007). We estimate the slope of mammal metabolic allometry with and without phylogenetic correction and outliers and use re-sampling to generate confidence intervals. Analysis of 440 species yields a distribution of slope estimates that, with phylogenetic correction, does not include 0.75. Without phylogenetic correction or removal of outliers, the OLS slope=0.69 (0.66-0.72) and LSVOR=0.73 (0.70-0.75). With phylogenetic correction and removal of outliers, the OLS slope=0.73 (0.72-0.74) and LSVOR=0.77 (0.76-0.78). Our technique and our rich dataset allow us to control for issues currently confounding physiological allometry: the preponderance of certain clades (e.g. rodents) in mammal data and performing regressions on two variables with differing error variances.