Physiologically Inspired Regression Models for Estimating and Predicting Nutrient Stores and Their Composition in Birds

Publisher Website
Google Scholar
Abstract
Regression models are presented that can be helpful in estimating and predicting the amount and fat and lean composition of nutrient stores in birds, when carcass data on body mass, fat mass, and body size are available. It is assumed that birds will show a breakpoint in composition during starvation, because stores, the nutrients accumulated in anticipation of shortage, have a different composition from that of the structural part of the body. The models differ only in the assumptions on the sources of variation in the data. One model, for example, assumes that all variation is due to unexplained variation in the mass of the structural part of the body. Another model assumes that the only variation is in the composition of the stores. It is shown that these models may yield completely different results. Examples suggest that the most reliable assumption on the source of variation differs among bird species. The need for ancillary information to verify the assumptions is emphasized. Prevailing methods, such as regressing fat mass on body mass and size, or regressing fat mass on condition indices such as body mass divided by size, are shown to be based on questionable assumptions, or to lack any theoretical justification, respectively. Their use should be abandoned.

This publication has 0 references indexed in Scilit: