Predicting bird species presence in urban areas with NDVI: An assessment within and between cities

Abstract

The increasing urbanization within cities correlates with the loss of green cover, affecting the distribution of bird species. Therefore, to foresee possible effects induced by green cover loss, it is essential to have spatial data of species occurrence. Remote sensors produce spatial data of green cover, such as the Normalized Difference Vegetation Index (NDVI), which can be used to model bird species occurrence within and between cities. The objectives of this study were: 1) to construct and evaluate predictive models between NDVI and bird species distribution in Mar del Plata city, Argentina; and 2) to extrapolate these models to two other cities of the region, Balcarce and Miramar. Most of the 21 species analyzed showed positive associations with NDVI. However, the minimum NDVI values at which species started to be present were dissimilar, showing different habitat requirements. The models of nine species had r-squared values ≥ 0.40, and eight of these species showed an adequate prediction of occurrence in additional areas of Mar del Plata. However, only the occurrence of the Tropical Kingbird (Tyrannus melancholicus) was adequately extrapolated to Miramar and Balcarce. Results showed that NDVI is successful to predict bird species occurrence within a city, but the extrapolation to other cities is not totally adequate. Thus, bird-habitat suitabilities for urban species are considered to be specific for each city.