Heterogeneity does matter: estimating relationships between anuran diversity and the environment at different spatial scales in a protected páramo of Colombia

The impact of Environmental Heterogeneity (EH) on assemblage diversity remains unclear due to biases in landscape context, taxa studied, and spatial scales typically analyzed. We explore how EH components influence anuran diversity at local and landscape scales in a protected páramo ecosystem in Chingaza National Park, Colombia. We implemented a multi-scale approach using habitat focal patches (HFPs) with 50, 100, and 200 m buffers derived from drone photography. Land cover mosaics were generated for 42 types, and landscape metrics were calculated. The most significant 10 land cover types were further analyzed with three class metrics each, identifying the scale of effect on species richness, total abundance, and assemblage structure. Multivariate DISTLM and Poisson models were applied to determine predictors at each spatial scale. Species richness showed a slight negative correlation with local-scale land cover diversity, while total abundance was strongly influenced by rosette cover and relative humidity, reflecting the importance of microhabitat features at smaller scales. Assemblage structure was shaped by exposed soil cover and Chusquea spp., indicating the role of these variables in assemblage turnover. At the landscape scale, class-level metrics were consistently stronger predictors than landscape composition, with the 200 m buffer emerging as the most influential scale, particularly through water bodies and dense scrub. These findings highlight that landscape configuration, rather than composition, plays a key role in shaping anuran diversity across scales. The habitat amount hypothesis and the scale of effect are more complex than expected in naturally heterogeneous landscapes. Structurally complex land cover types, such as dense scrub and water bodies availability, were critical in explaining anuran diversity. The use of drones allowed us to capture ecologically meaningful spatial scales at 4 cm/píxel resolution, offering valuable insights into how landscape configuration influences amphibian diversity patterns.