Document Type : Original Research
Authors
1
college of agriculture , v c farm mandya
2
Department of Soil Science and Agricultural Chemistry, College of Agriculture, V.C. Farm, Mandya, University of Agricultural Sciences, Bangalore, Karnataka, India.
3
Department of Soil Science and Agricultural Chemistry, College of Agriculture, V.C. Farm, Mandya, University of Agricultural Sciences, Bangalore, Karnataka, India
4
Dean (Agri.), College of Agriculture, V.C. Farm, Mandya, University of Agricultural Sciences, Mandya, Karnataka, India.
5
Department of Agricultural Engineering, College of Agriculture, V.C. Farm, Mandya, University of Agricultural Sciences, Bangalore, Karnataka, India
Abstract
Soil quality strongly influences banana yields, but few field studies have linked soil properties to actual farmer-managed production, particularly in India’s diverse banana regions. This work assessed the physical, chemical and micronutrient status of soils under banana in five villages of Nanjangud taluk, Mysore district, Karnataka, and related these traits to recorded yields. Samples and yield data were gathered from 150 geo-referenced farmers’ fields. The mean observed banana yield was 46.53 ± 10.14 t ha⁻¹, ranging from 30.00 to 77.60 t ha⁻¹. Principal Component Analysis singled out iron, porosity, pH, available nitrogen, available phosphorus and zinc as the main soil-quality indicators, these were combined into a Soil Quality Index (SQI). Village-mean SQI ranged from 0.401 in Halepura to 0.523 in Kanenuru, with Huskuru and Karya at 0.490 and Hosaveedu at 0.479. Across individual fields, 22 % of soils ranked low, 72 % moderate and 6 % high in quality. The Random Forest model showed promising predictive performance, with a squared correlation of 0.765 between observed and predicted yields, a test-set R² of 0.700, an RMSE of 5.32 t ha⁻¹ and an MAE of 4.30 t ha⁻¹.Banana variety proved the strongest driver, followed by zinc, iron, latitude, organic carbon, available potassium and nitrogen. By linking soil-quality indices with machine-learning yield models, we can identify which soil and crop factors most strongly influence banana yield. This, in turn, provides a basis for developing site-specific nutrient management strategies in the region.
Keywords
Subjects