Biogeographical influence on bird species sensitivity to habitat fragmentation / Influência de fatores biogeográficos sobre a sensibilidade das espécies de aves à fragmentação do habitat

AUTOR(ES)
DATA DE PUBLICAÇÃO

2009

RESUMO

Species distribution models (SDM) is a quite new technique that has been used in many fields of natural sciences. Nevertheless, as problem inherent of a new technique, there is still a lot of questions about the most appropriated manner of its use and application. As the SDM consists a powerful tool with high potential to be used use in many fields of research, it is extremely useful that tests are carried out, to improve and refine this technique. To help in the development of the SDM, this study consist a methodological test that aims to compare the performance of different algorithms used in SDM. We used points of occurrence of 21 bird species along an area of nearly 1,200,000 km2 in southeastern Brazil. The data were obtained through a rough bibliographic revision. Environmental layers used in the modeling procedure consisted of 19 bioclimatic variables, besides altitude and slope of the terrain. The algorithms used were Maxent, GARP and SVM. As a performance parameter of the algorithms we used the area under de ROC curve (AUC), generated with points of occurrence independent of the ones used for the modeling procedure. Based on the AUC values, we observed a higher prevision capacity of the models generated with Maxent, followed by the ones generated with SVM, and at least, the ones generated with GARP. The pattern of the probability distributions of the models generated with Maxent and SVM had a higher biological meaning, besides agreeing more with the species real area of occurrence. As conclusion of this study we consider that Maxent is the most appropriate algorithm to be used in species distribution models, that are made under the same conditions of the ones used in the present study. Biogeographical factors, specially the location of populations in relation to the occurrence area of the species, could mediate interpopulational variations in species response to habitat fragmentation. The present study aimed to verify if the biogeographical position of populations can predict bird species sensitivity to habitat spatial distribution, measured as the forest cover and configuration in fragmented landscapes. Considering 21 species present in at least 30 fragments, we established regression models relating the abundance of the species with patch cover and configuration of the patches, and with the population biogeographical position. This position was obtained through species distribution models, with the maximum entropy algorithm (Maxent). We used the Akaike Information Criteria (AIC) to compare the models and verify the relative importance of each variable. Most of the species (n= 10) presented a similar response pattern, being influenced by landscape configuration and also by biogeographical position parameter. Six other species were influenced only by landscape structure, while two species were influenced only by biogeographical position. There were also three other species that were not influenced by any of the used variables. The response pattern to the independent variables seems to be independent of the biological characteristic of the species. The present study showed that the biogeographical position of the populations could be, with the landscape structure, an important factor to determine the abundance of major part of bird species in a fragmented landscape. Such results have implications on the comprehension of bird species persistence in fragmented landscapes, as well to driven effort for conservation plans in regional scales.

ASSUNTO(S)

biogeography birds fragmentação do habitat biogeografia habitat fragmentation aves

ACESSO AO ARTIGO

Documentos Relacionados