Remotely sensed indicators and open-access biodiversity data to assess bird diversity patterns in Mediterranean rural landscapes.

Abstract

Changes in ecosystem area are often used to assess human impacts on habitats and estimate biodiversity change. However, because species respond to structural changes at fine spatial scales the use of area alone may not capture all relevant changes. Operational costs limit the assessment of biodiversity change at a simultaneously fine spatial resolution and large scales. The development of cost-effective and expedite methods to monitor biodiversity change is therefore required. We use open access satellite imagery and biodiversity data to investigate the importance of variables of habitat extent and structure in explaining species richness and community dissimilarity of forest and open-land birds at the regional scale. Moreover, because Mediterranean landscapes are subject to seasonal dynamics, we explore the indicator value of remotely sensed variables measured in spring and summer. A large-scale dataset of bird occurrence data, including 8042 observations and 78 species, distributed by 40 landscape-sized cells, was assembled from GBIF after controlling for data quality. We found that summer satellite imagery, when the green perennial vegetation is more apparent, is particularly suited to model the diversity patterns of forest species, because distribution of tree cover in the landscape is well captured. Summer data is also useful to monitor the perennial elements that shape landscape structure and the habitat of open-land species. Specifically, mean NDVI and a second-order NDVI texture variable, were found to be good indicators of forest and open-land habitats, respectively. The use of spring imagery appears to be useful to monitor habitat structure within open-land habitat patches. Overall, NDVI texture measures were found to be good predictors of bird diversity patterns at large scales. Also, we were able to successfully conduct a regional scale analysis using open-access data, which illustrates their potential to inform large scale biodiversity monitoring.