Soil heterogeneity and acidity are major constraints to Coffea arabica production in the Amazonian soils of Peru. This study developed a spatial predictive framework that integrates a weighted Soil Quality Index (SQIw) and geostatistical modelling (Regression–Kriging and Ordinary Kriging) to estimate lime requirements (LRs) and delineate management zones. A total of 69 coffee-cultivated soil samples were analysed, and spectral information (NDVI) was incorporated to estimate relative yield (RR). Multivariate analysis defined a Minimum Data Set (MDS) composed of exchangeable Na, available P, pH and silt percentage; the highest weights were assigned to P (Wi = 0.292) and pH (Wi = 0.276). SQIw exhibited wide variability (0.01–0.87; CV = 51.8%) and was grouped into five classes, with low (43.5%)- and very low (21.7%)-quality classes predominating. SQIw showed a strong relationship with RR (r = 0.64). Geostatistical models performed differently between localities: in Nuevo Huancabamba, Regression–Kriging improved prediction accuracy (SQIw: R2 = 0.58; LR: R2 = 0.396), whereas in San José de Sisa, Ordinary Kriging provided better fits only for LRs (R2 = 0.32). Nuevo Huancabamba is dominated by moderate-to-high-quality soils (87.29%; SQIw > 0.6) and low lime requirements (74.94%; <0.84 t ha−1), in contrast with San José de Sisa, where low-quality soils prevail (89.45%; SQIw < 0.4) alongside high LRs (75.26%; 2.54–7.13 t ha−1). The resulting maps enable targeted interventions—precision liming and focused P fertilisation—to correct acidity and phosphorus deficiency, thereby improving input-use efficiency and enhancing the sustainability of Amazonian coffee systems.
Díaz-Chuquizuta et al. (Wed,) studied this question.