Strength of Habitat and Landscape Metrics in Predicting Golden-Headed Lion Tamarin Presence or Absence in Forest Patches in Southern Bahia, Brazil

We investigated the effects of forest fragmentation on golden-headed lion tamarins (Leontopithecus chrysomelas) by qualitatively and quantitatively characterizing the landscape throughout the species range, conducting surveys, and exploring predictive models of presence and absence. We identified 784 forest patches that varied in size, shape, core area, habitat composition, elevation, and distance to neighboring patches and towns. We conducted 284 interviews with local residents and 133 playback experiments in 98 patches. Results indicated a reduction in the western portions of the former species range. We tested whether L. chrysomelas presence or absence was related to the aforementioned fragmentation indices using Monte Carlo logistic regression techniques. The analysis yielded a majority of iterations with a one-term final model of which Core Area Index (percent of total area that is core) was the only significant type. Model concordance ranged between 65 and 90 percent. Area was highlighted for its potential predictive ability. Although final models for area lacked significance, their failure to reach significance was marginal and we discuss potential confounding factors weakening the term's predictive ability. We conclude that lower Core Area Index scores are useful indicators of forest patches at risk for not supporting L. chrysomelas. Taken together, our analyses of the landscape, survey results, and logistic regression modeling indicated that the L. chrysomelas metapopulation is facing substantial threat. The limited vagility of lion tamarins in nonforest matrix may lead to increasingly smaller and inbred populations subject to significant impact from edge effects and small population size. Local extinction is imminent in many forest patches in the L. chrysomelas range.