Evaluating the validity of using unverified indices of body condition

• Published in: Journal of avian biology Vol. 40; no. 1; pp. 49 - 56
• Main Authors: Schamber, Jason L., Esler, Daniel, Flint, Paul L.
• Format: Journal Article
• Language: English
• Published: Oxford, UK Oxford, UK : Blackwell Publishing Ltd 2009; Blackwell Publishing Ltd; Blackwell Publishing; John Wiley & Sons, Inc
• Subjects: Animal physiology; Average linear density; Body condition; Body fat; Body mass index; Lipids; Modeling; Parametric models; Tarsus; Waterfowl
• ISSN: 0908-8857; 1600-048X
• DOI: 10.1111/j.1600-048X.2008.04462.x

Condition indices are commonly used in an attempt to link body condition of birds to ecological variables of interest, including demographic attributes such as survival and reproduction. Most indices are based on body mass adjusted for structural body size, calculated as simple ratios or residuals from regressions. However, condition indices are often applied without confirming their predictive value (i.e., without being validated against measured values of fat and protein), which we term 'unverified' use. We evaluated the ability of a number of unverified indices frequently found in the literature to predict absolute and proportional levels of fat and protein across five species of waterfowl. Among indices we considered, those accounting for body size never predicted absolute protein more precisely than body mass, however, some indices improved predictability of fat, although the form of the best index varied by species. Further, the gain in precision by using a condition index to predict either absolute or percent fat was minimal (rise in r²[less-than or equal to]0.13), and in many cases model fit was actually reduced. Our data agrees with previous assertions that the assumption that indices provide more precise indicators of body condition than body mass alone is often invalid. We strongly discourage the use of unverified indices, because subjectively selecting indices likely does little to improve precision and might in fact decrease predictability relative to using body mass alone.