Content area
Full Text
Southeastern Brazil has seen dramatic landscape modifications in recent decades, due to expansion of agriculture and urban areas; these changes have influenced the distribution and abundance of vertebrates. We developed predictive models of ecological and spatial distributions of capybaras (Hydrochoerus hydrochaeris) using ecological niche modeling. Most occurrences of capybaras were in flat areas with water bodies surrounded by sugarcane and pasture. More than 75% of the Piracicaba River basin was estimated as potentially habitable by capybara. The models had low omission error (2.3-3.4%), but higher commission error (91.0-98.5%); these "model failures" seem to be more related to local habitat characteristics than to spatial ones. The potential distribution of capybaras in the basin is associated with anthropogenic habitats, particularly with intensive land use for agriculture.
Key words: Brazil, ecological niche models, genetic algorithm for rule-set prediction (GARP), geographic distribution, Hydrochoerus hydrochaeris, Piracicaba River basin
Habitat fragmentation is one of the more dramatic processes of landscape modification (Forman and Godron 1986) and causes of biodiversity loss (Bisbal 1988; Collinge 1996; Hannah et al. 1995; Lacher et al. 1999; Noss and Csuti 1994). Southeastern Brazilian habitats have suffered extensive fragmentation due to agriculture and urbanization in recent decades. Deforestation in São Paulo State, Brazil, totaled ~4.1 × 10^sup 6^ ha during 1962-1992 (Kronka 1994). Indeed, the Piracicaba River basin in Sao Paulo State ranks among the most developed and impacted regions of Brazil, due to both urban and agricultural expansion (Lara et al. 2001), with 70% of its area occupied by pasture and sugarcane (Ballester 2001). Habitat fragmentation that frequently accompanies deforestation further influences species' distributions and abundances (Kalkhoven 1993; Savard et al. 2000; Wiens 1996).
Ecological niche modeling allows identification of environmental factors affecting species' distributions and abundances (Morrison et al. 1998). Ecological niche modeling is receiving considerable attention from the scientific community because it permits objective characterization of species-habitat relationships on broad scales (Jackson et al. 2000; Leemans 1999). Ecological niche models developed using the genetic algorithm for rule-set prediction (GARP - Stockwell and Noble 1992) have proven useful in diverse applications (Feria and Peterson 2002; Illoldi et al. 2004; Levine et al. 2004; Martínez-Meyer et al. 2004, 2006; Nakazawa et al. 2004; Peterson 2006a; Peterson et al. 2002, 2006; Peterson and Vieglais 2001; Wiley...