Vegetation canopy height shapes bats’ occupancy: a remote sensing approach

Abstract

Listen

Translate article

ABSTRACT Anthropogenic activities have significantly altered land cover on a global scale. These changes often have a negative effect on biodiversity limiting the distribution of species. The extent of the effect on species’ distribution depends on the landscape composition and configuration at a local and landscape level. To better understand this effect on a large scale, we evaluated how land cover and vegetation structure shape bat species’ occurrence while considering species’ imperfect detection. We hypothesize that intensification of anthropogenic activities in agriculture, for example, reduces heterogeneity of land cover and vegetation structure, and thereby, limits bat occurrence. To investigate this, we conducted acoustic bat sampling across 59 locations in southern Portugal, each with three spatial replicates. We derived fine-scale vegetation structural metrics by combining spaceborne LiDAR (GEDI) and synthetic aperture radar data (Sentinel-1 and ALOS/PALSAR-2). Additionally, we included land cover metrics and high-resolution climate data from CHELSA. Our findings revealed an important relationship between bat species’ occupancy and vegetation structure, particularly with vegetation canopy height. Moreover, forest and shrubland proportions were the main land cover types influencing bat species responses. All species’ best-ranking occupancy models included at least one climatic variable (temperature, humidity, or potential evapotranspiration), demonstrating the importance of climate when predicting bat distribution. Our acoustic surveys had a species’ detection probability varying from 0.19 to 0.86, and it was influenced by night conditions. These findings underscore the importance of modeling imperfect detection, especially for highly vagile and elusive organisms like bats. Our results demonstrate the effectiveness of using vegetation and landscape metrics derived from high-resolution remote sensing data to model species distribution in the context of biodiversity monitoring and conservation.