Fecha de publicación: -- Fuente: PubMed "hive" Insects. 2026 Sep 7;17(9):937. doi: 10.3390/insects17090937.ABSTRACTThe stingless bee Tetragonisca angustula (Apidae: Meliponini) is ecologically and socioeconomically important in northern Peru, yet the factors shaping its habitat suitability remain poorly quantified. We modeled current suitability across Amazonas and San Martín using 28 occurrence records: 17 field-georeferenced hive locations and 11 records from the Global Biodiversity Information Facility (GBIF). An initial set of 28 climatic, vegetation, hydrological, topographic, and edaphic predictors was reduced to 15 variables after repeated variance inflation factor analyses. Maximum Entropy models were calibrated with linear, quadratic, and hinge features, four regularization multipliers, 20 bootstrap replicates, and a 75:25 training-testing partition. A multiplier of 1.5 was selected for its higher minimum test AUC across replicates (0.72), indicating more consistent worst-case performance (training AUC = 0.94 ± 0.02; test AUC = 0.84 ± 0.07). Forest cover contributed most to model fit (31.7%), whereas predictions were most sensitive to slope (27.2% permutation importance), precipitation of the coldest quarter (14.2%) and soil organic carbon (13.3%). Suitability peaked at intermediate forest cover (~50%) and gentle slopes (~4°). Applying the 10th-percentile training-presence threshold, 1,190,582 ha (13.2%) were classified as suitable, concentrated in mixed forest-agricultural landscapes rather than continuous lowland forest. Vegetation structure and terrain, not climate alone, shaped predicted suitability; given uneven sampling, these patterns require field validation.PMID:42783304 | DOI:10.3390/insects17090937