Introduction
External measures of animal size, e.g. body mass, wing chord, foot length, and so on are often used by ecologists to develop body condition indices, e.g. body mass/length. Body condition indices are thought to reflect variation in diverse aspects of organismal quality, e.g. health, nutritional status, fat content and even Darwinian fitness (Krebs & Singleton, 1993; Brown, 1996; Viggers et al., 1998). Body condition indices are generally easy to compute, so if they are highly correlated with variables, such as fatness or health, that are difficult to measure, they may be useful to ecologists for at least two reasons. First, they may be indicators of variables, e.g. fat content, that are difficult to measure accurately without harming an animal. Secondly, they may be more efficient or experimentally simpler to measure than variables that are hard to quantify, such as health or Darwinian fitness.
Ecologists seeking rapid non-destructive methods for estimating body condition have used two basic approaches for estimating body condition from external morphology. These two approaches are based on the construction of ratio variables, e.g. body mass divided by length, and the generation of residual variables, e.g. residuals from the regression of mass on length. Use of both kinds of condition indices requires considerable care to prevent errors in inferring the biology of interest, i.e. to prevent drawing erroneous conclusions about the biology being studied. One serious problem with the use of condition indices is that apparently subtle differences in the method used to compute the condition index may lead to vastly different conclusions about an animal's condition.