BEAUPRE, S.J.; University of Arkansas: On the representation and analysis of specific dynamic action (SDA).

Specific Dynamic Action (SDA, or the bioenergetic cost of digestion and assimilation, has been studied in a variety of animals to understand the physiological, ecological and evolutionary determinants of digestive function. In comparative studies, researchers often extrapolate their SDA data to common body mass using mass-specific SDA (kJ kg^-1) or to common food mass using the SDA coefficient (SDA kJ expressed as % of ingested kJ). In both cases, raw data are converted to a ratio, which assumes that the underlying relationship between component variables is isometric (linear, slope = 1, intercept = 0). Departures from isometry (i.e. allometry) between ratio components are known to result in estimation errors, statistical errors, and spurious correlations. By re-analyzing data from two early and influential publications, and by considering my own SDA data on the timber rattlesnake (Crotalus horridus), I demonstrate that these relationships are actually allometric. By comparing mass-specific estimates (kJ kg^-1) to regression-based scaling relationships, I show that large errors are possible (up to 35%) when mass-specific ratios are used. I also demonstrate that the relationship between SDA and ingested energy is often allometric, and that the SDA coefficient is affected by body size. Therefore, the SDA coefficient should not be used to standardize SDA data for groups of organisms that differ in body size or meal size. As suggested by previous authors in other contexts, I recommend the abandonment of ratios, such as mass-specific SDA and SDA coefficient, in favor of more robust statistical techniques (ANCOVA, and allometric scaling relationships) that directly analyze raw SDA data.