Soil-driven mechanisms of organic fertilizer ratios regulating blueberry fruit quality: A comprehensive evaluation based on interpretable machine learning

This study evaluated the effects of organic fertilizer type and application rate on blueberry fruit quality and identified key rhizosphere and bulk soil variables associated with fruit-quality variation. A controlled pot experiment was conducted using blueberry cultivar ‘42’ ( Vaccinium corymbosum L. ) with fourteen fertilization treatments (pig manure, rabbit manure, compound fertilizer, and a control). Ten replicates per treatment were established, and three biological replicates were sampled at fruit maturity. An integrated framework combining Random Forest-SHAP analysis, TOPSIS-based multi-trait evaluation, and structural equation modeling (SEM) was applied. Organic fertilization significantly influenced fruit quality, with higher application rates (200 g pot −1 ) showing consistently better performance. Pig manure (L3) achieved the highest comprehensive quality score. Anthocyanin content, soluble solids, and titratable acidity were the main contributors to fruit-quality variation. SHAP analysis revealed clear spatial heterogeneity between rhizosphere and bulk soils. In the rhizosphere, fruit quality was strongly associated with soil pH and moisture, with a model-derived response transition around pH 4.3–4.4. In bulk soil, nitrate–N exhibited nonlinear associations, with negative SHAP values at higher levels (≈38–40 mg kg −1 ). SEM further indicated statistically structured associations among bulk soil conditions, rhizosphere properties, and fruit quality. Fruit quality improvement was associated with higher organic inputs. Variation in fruit quality was jointly associated with rhizosphere and bulk soil conditions, highlighting the importance of integrated soil environmental regulation in blueberry production systems.