![]() ![]() Compared with regression models, ecological niche models only need the animals "presence points" and do not need "non-presence points” data. With field observations, the absence of animal traces in a certain place does not mean that the animals have never appeared there. However, mechanism models have some limitations because they do not consider the accessibility of the habitat, and are subjective in the classification and weight determination of factors. Mechanism models do not need species distribution point data, but establish the corresponding evaluation criteria according to the influence of habitat factors on species distribution, so as to simulate the suitable habitat of a species. Recently, with the development of GIS, Remote Sensing, and GPS techniques, multiple models have been used to assess suitable habitat distribution, including mechanism models 5, regression models 6, and ecological niche models 7, 8, 9, 10. Many fields of research rely on predictive models for assessing patterns of species distribution 3, 4. In recent years, several statistical and computer-based methods have been utilized to map biological and ecological data and to spatially interpolate species distributions and other bio-spatial variables of interest. Knowledge of habitat preference and geographical distribution is essential for the conservation of threatened species 1, 2. Habitat is the type of natural environment in which a particular species lives or can find food, shelter, protection, and mates for reproduction. ![]() For remote and inaccessible regions, the proposed methods are promising tools for wildlife management and conservation, deserving further popularization. We suggest that proper management should be given to the overlapping habitats in the buffer zone. The black bear showed higher habitat selectivity than red panda. Maxent outperformed GARP in terms of habitat suitability modeling. Of the suitable habitat, both models indicated forest as the most preferred land cover for both species (63.7% for black bear and 61.6% for red panda from Maxent 59.9% black bear and 58.8% for red panda from GARP). The results of land cover exhibited that barren land covered the highest percentage of area in MBNP (36.0%) followed by forest (32.6%). Maxent predicted that the overlapping area was 83% of the red panda habitat and 40% of the black bear habitat, while GARP estimated 88% of the red panda habitat and 58% of the black bear habitat overlapped. The suitable habitat estimated by Maxent for black bear and red panda was 716 km 2 and 343 km 2 respectively, while the suitable area determined by GARP was 1074 km 2 and 714 km 2 respectively. The habitat suitability modeling accuracy, characterized by the mean area under curve, was moderate for both species when GARP was used (0.791 for black bear and 0.786 for red panda), but was moderate for black bear (0.857), and high for red panda (0.920) when Maxent was used. Analysis of the regularized training gain showed that the three most important environmental variables for habitat suitability were distance to settlement, elevation, and mean annual temperature. The modeled results were validated by using an independent dataset. All of the predictor variables were extracted from freely available remote sensing and publicly shared government data resources. With the resulting ecological niche models, the suitable habitat for asiatic black bear ( Ursus thibetanus) and red panda ( Ailurus fulgens) in Nepal Makalu Barun National Park (MBNP) was predicted. Species presence points collected through field GPS observations, in conjunction with 13 different topographic, vegetation related, anthropogenic, and bioclimatic variables, as well as a land cover map with seven classification categories created by support vector machine (SVM) were used to implement Maxent and GARP ecological niche models. These models can help us to study habitat suitability at the scale of the species range, and are particularly useful for examining the overlapping habitat between sympatric species. With the rise of innovative powerful statistical techniques in partnership with Remote Sensing, GIS and GPS techniques, spatially explicit species distribution modeling (SDM) has rapidly grown in conservation biology. Habitat evaluation is essential for managing wildlife populations and formulating conservation policies. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |