Phenetic Structure and Species Richness in North and Central American Bat Faunas

Thirty—one North and Central American bat faunas were examined with multivariate morphometric techniques to determine if faunas of bats found in the tropics are more tightly packed in morphological attribute space than those found in temperate latitudes. Eighteen morphological features for each of 103 species of insectivorous bats wee measured. The density of morphological packing was calculated from nearest neighbor Euclidean distances between species within each fauna. Nearest neighbor distances were also computed for species' projections on principal component axes for each fauna. For these different character spaces, substantially different patterns emerged, resulting from the correlation of many, morphological characters. Morphological packing in these faunas was also compared to levels of packing predicted from two different neutral models in which species composition was generated by Monte Carlo sampling methods. In one neutral model, species were drawn from the total pool of insectivorous bats, including 11 species which were statistically significant morphological outliers. Frequency distributions of nearest neighbor distances of real and randomly assembled faunas with equal species richness were compared by Kolmogorov—Smirnov tests. Real faunas which did not contain the most significant outliers did not differ from random faunas. Seven faunas, which include outlier species with geographic ranges restricted to the tropics, differed from random communities in the majority of large numbers of Monte Carlo samples. In an alternate neutral model, outliers were excluded in the real and random faunas. Morphological packing was compared between real and random faunas by comparing regressions of average nearest neighbor distance on faunal species richness. In this analysis, levels of packing in real faunas of varying species richness were not significantly different from that predicted under a null hypothesis. Mean nearest neighbor distance decreased with increasing species richness in both real and randomly constructed faunas. Indicating that measures of species packing based on nearest neighbor distance values are sample size dependent. Both types of neutral models suggest that bat faunas are not more tightly packed in a morphological attribute space than would be predicted from chance alone. Tropical faunas, though, may accommodate more species than predicted from a neutral model when species can be added to the extreme peripheral limits of morphological variation.