RIVERA Luis Osvaldo
congresos y reuniones científicas
Performance of novel remotely-sensed variables in predictive species distribution models for Argentinean forest birds
OLAH ASHLEY; RADELOFF VOLKER; SILVEIRA, EDUARDA M.O.; MARTINUZZI SEBASTIÁN; MARTINEZ PASTUR GUILLERMO; POLITI NATALIA; RIVERA LUIS; PIDGEON ANNA
Congreso; Annual Meeting AOS 2023; 2023
American Ornithological Society
Halting biodiversity loss is a major conservation goal, and requires understanding the environmental correlates of biodiversity patterns. One way to assess these relationships is through species distribution modelling. In such models, commonly used environmental variables include land cover classes and climate metrics. However, these may fall short of adequately characterizing factors that influence species distributions. Recently developed remotely sensed measures may offer additional predictive power, complementarity, or even replace commonly used environmental variables in species distribution models. We modeled species distributions for 181 forest-affiliated bird species in Argentina, using three sets of environmental variables: 1) remotely sensed variables characterizing forest structure, phenoclusters, spatial-temporal variability in greenness and land surface temperature, 2) commonly used environmental variables characterizing elevation, land cover, soil type and precipitation 3) the combination of all remotely sensed and commonly-used variables. Within the remotely sensed variable set, spatial heterogeneity in winter land surface temperature, phenoclusters, and mean forest height were important for many species. Within the ´commonly used´ variables, precipitation of the driest quarter was important for the most species. When all variables were combined, phenoclusters and precipitation of the wettest quarter were the most important variables in models. The best models using remotely-sensed, non-remotely sensed, and all variables in combination performed similarly. In areas where land cover classes are not well delineated, the use of remotely sensed variables expands options for modeling species´ distributions.