High-resolution topsoil property maps for the agricultural zones of the Galápagos islands

The agricultural highlands of the Galápagos Islands are critical for local food production but exhibit high spatial variability in soil fertility due to differences in geological age, climate, and land-use history. This study provides high-resolution (1 arc-second) soil property maps for these areas, utilizing point samples taken from 130 farms across the islands Isabela, Santa Cruz, San Cristóbal, and Floreana. In a Digital Soil Mapping (DSM) approach, we calibrated Quantile Regression Forest (QRF) models to spatially predict 51 physical and chemical topsoil properties based on gridded environmental covariates representing climate, vegetation, relief, and parent material. Model performance varied by soil property, with best metrices for total nitrogen, water holding capacity (15 bar), aggregate stability, cation exchange capacity, and several macro- and micronutrients (Lin’s Concordance Correlation Coefficient CCC up to 0.90; R 2 up to 0.83). Lower predictive performance was observed for CaCO 3 , phosphorus (Mehlich-3), and PO 4 retention. Spatial uncertainty was visualized via maps of 95 % prediction interval widths, highlighting areas of greater model uncertainty. Predictor importance analysis revealed distance from the volcanic hotspot, precipitation- and temperature-related variables, and lithology as key drivers of the observed spatial patterns. The resulting soil maps provide continuous, up-to-date information to guide sustainable land management, agricultural planning, and conservation in Galápagos. By offering locally calibrated, multi-property soil maps at high spatial resolution, this work overcomes limitations of globally calibrated soil mapping products, thereby supporting informed decision-making in one of the world’s most ecologically sensitive agricultural landscapes.