Niche shifts in Greater Antillean Anolis communities: effects of niche metric and biological resolution on null model tests.

Using published data, a Monte Carlo approach was used to determine if niche shifts in communities of Anolis lizards have occurred within two islands in the Greater Antilles. Communities at 4 sites on Puerto Rico and 4 sites on Jamaica were randomly assembled using resource-use data from any of the 4 sites with the constraint that the species list of the random community was identical to that observed for each site. The observed niche structure and that of the randomly assembled community were compared at each site. The null hypothesis tested was that one of several functions of niche overlap did not produce lower values in the observed communities compared to those randomly assembled. The sensitivity of this method to details of the analysis was investigated by using 5 variations on the procedure for quantifying the niche structure of the communities. I investigated the effects of (1) using Pianka's versus Levins' metric of niche overlap, (2) using complete resource-use data versus the marginals of the distributions, (3) asymmetric responses to overlap due to size differences between overlapping species, (4) using estimated electivities versus the observed proportional use of resource states, and (5) methods of niche analysis based on different degrees of biological resolution (e.g., community-wide statistics versus individual species responses). The results of the simulations and statistical comparisons did not provide a clear, unambiguous answer to the question of the occurrence of niche shifts. Statistical evidence for niche shifts was more frequently detected using Levins' metric compared to Pianka's, and shifts were easier to detect using high biological resolution (examination of individual species) compared to community-wide measures of niche structure response.