An improved body mass dataset for the study of marsupial brain size evolution.

data is expected to be small, as the brain sizes of several adult specimens and their body masses are typically recorded, and thus the body mass range of individualbased body mass data is expected to approximate that of the averages of the literature. It is therefore not surprising that studies of body mass versus brain sizes have successfully used both literature-based [Isler and Van Schaik, 2009] and individual-based body mass data [e.g. Eisenberg and Wilson, 1981]. In a recent study on marsupial endocranial volumes (considered to proxy brain weight), Ashwell [2008] preferred the use of individual body mass data which were, however, mostly unavailable from museum collections. Instead, he estimated individual body mass based on a frequently used body mass estimation formula specifically for marsupials, which is based on mandible length [Myers, 2001]. In this note, we point out that although many of the body mass estimates from Ashwell [2008] provide a good fit to individual museum records, there are some issues that suggest to us that body mass averages from the literature reflect specimen body mass more accurately. In particular, mandibular length estimates appear to have underestimated body mass in many species, particularly those with a feeding ecology that favours short snouts. In a few instances, the body masses of comThe scaling of vertebrate (particularly mammalian and avian) brain size relative to body mass has received much attention and its precise nature has been the subject of long-standing debate [Jerison, 1973; van Dongen, 1998; Weisbecker and Goswami, 2011]. Virtually all studies of vertebrate brain size evolution are conducted in the framework of assessing brain size relative to body mass [but see Deaner et al., 2007], mostly through the use of brain size residuals on a logarithmic regression of brain size against body mass. While body mass data are essential for the meaningful analysis of brain size, the choice of an appropriate proxy of body mass is not straightforward. Two avenues, each with their merits, have been used in the past: first, the recording of body mass along with brain measurement of specimens, and second, the use of literaturebased mass averages. The fact that brain size/body mass scaling differs within as well as between species [e.g. Kruska, 2005] might suggest that the use of individualbased brain sizes would give greater precision. However, it can also be argued that the use of literature-based body mass better acknowledges the error inherent to the use of a single species value, as it represents the statistical mean of a species-specific ‘cloud’ of data points. In most cases, the difference between literature averages and individual Published online: April 24, 2013