2005). In a separate analysis, we examined the relationship between this website population density and likelihood of drastic population decline, among all species. We defined drastic population decline as possessing a sampled distribution in which at least 90% of individuals were captured in uninvaded plots (taking the average among sites for species that occurred at multiple sites). This level of inferred population reduction, while somewhat

arbitrary, identifies those species that are arguably the most likely to experience local extinction. We grouped species, both rare and non-rare, by successively larger population density categories, such that evenness was maximized among all but the lowest density category (in terms of number of species included) for both endemic and introduced species. We then calculated the percentage of species exhibiting click here patterns of drastic population decline in each density category. Because the likelihood of obtaining a highly skewed sampling distribution purely by chance is much higher among small populations, we also calculated the percentage of species expected to exhibit patterns consistent with drastic population decline, through random sampling alone, for each population density category. We did this by (1) calculating the probability of obtaining 90% or more of sampled

individuals in uninvaded plots for each observed population size, under the assumption that each individual had equal probability of existing in an invaded versus uninvaded plot, (2) multiplying these probabilities by the number of species that occurred at each population size, and (3) summing over population sizes and dividing by the total number of species, within each density category. Finally, we calculated a chance-corrected likelihood of drastic population decline for each density category by subtracting the percentage

of species expected to exhibit patterns of drastic decline due solely to chance from the observed percentage of species exhibiting this pattern. To examine variability in the inferred response to ant invasion, both Racecadotril within and among species, we tabulated species responses within each order, using the entire dataset including multiple incidences of species occurrence. Species were classified according to the identity and consistency of their responses. For non-rare species, we designated four categories: species whose responses were always strongly negative (impact scores ≤ −0.5 at all sites), always weakly interacting (between −0.5 and 0.5 at all sites), always strongly positive (≥0.5 at all sites), or variable (including scores in more than one of the categories at different sites). Rare species were classified into three categories: those that were absent in invaded plots at all sites, those that were present in invaded plots at all sites, and those that had variable responses among sites.