Do lizard communities partition the trophic niche? A worldwide meta-analysis using null models

Lizards have been widely used as model organisms in community ecology over the past 30 years. I have reviewed more than 50 studies from the literature on the community ecology of lizards worldwide, with a focus on studies from 1990-2007. I determined if there is support for the hypotheses that many lizard communities 1) are non-randomly organized along the trophic niche dimension, and 2) partition the available food resource to minimize interspecific competition. I used data on number of prey items and percentage of prey volume, to calculate dietary overlap among species. I compared true datasets to randomly generate new datasets produced by 3x10⁴ Monte Carlo permutations using two algorithms (RA2 and RA3). The great majority (more than 80%) of the communities were randomly organized along the trophic niche dimension using RA2 or RA3; the frequency of occurrence of detecting a non-random structure in lizard community dietary studies did not differ significantly between datasets based on either number of prey items or prey volume. Thus, lizard communities usually do not partition the trophic niche axis. Concerning the number of prey items, logistic regression models showed that the presence of a structure in the dataset did not depend on number of species, method employed to obtain dietary data, matrix size, or location (continent), but instead significantly depended on whether a community was tropical. Concerning prey volume, presence of a structure in the dataset also did not depend on number of species, method employed to obtain dietary data, and tropical vs non-tropical origin, but was marginally dependent on continent (South America was favoured for identifying a structure in the dataset) and matrix size. In general, a structure in the dataset was more often uncovered by using RA2 than RA3. Overall, I conclude that 1) lizard communities are unlikely to partition available food resources, and 2) as the method for identifying dietary items does not allow for accurate prey identification (at least at the genus level), there is a risk of obtaining false null results about community structure. This is because structure along the trophic niche axis is genuinely rare in lizards and also because inappropriate methods may erroneously lead to this conclusion. Therefore, conclusions from studies utilizing stomach dissection as a source of diet data (still common in the literature) should be cautiously considered for ethical reasons and because it is difficult to find any difference between stomach-dissection and fecal-pellet data when assessing the presence of a structure in datasets. My data also showed there are minor differences in the probability of detecting a structure in datasets using prey item number or prey volume data for lizards. The biases, both statistical and biological, associated with this meta-analysis are discussed.