Statistical Assessment of Sustainability of Finger Millet Yield through Rainfall and Soil Fertility Variables using Regression and Principal Component Models in Rainfed Semi-Arid Alfisol

Abstract

ABSTRACTField experiments were conducted in farmyard manure (FYM) and maize-residue (MR) blocks during 1984–2011 with the objective of identifying a superior treatment for attaining maximum finger millet yield and soil fertility at Bangalore. The treatments tested in the FYM block were Control; FYM@10tha−1; FYM@10tha−1 + 50% nitrogen, phosphorus, and potassium (NPK); FYM@10tha−1 + 100%NPK; 100%NPK; and in the MR block, Control; MR@5tha−1; MR@5t ha−1 + 50%NPK; MR@5tha−1 + 100%NPK; 100%NPK. FYM@10t/ha + 100%NPK gave a maximum mean yield of 3207 kg/ha in the FYM block, while MR@5t/ha + 100%NPK gave 2548 kg/ha in the MR block. Regression and principal component (PC) models of yield were developed through soil fertility and rainfall variables to assess the treatments. Maximum yield predictability of 60% and 65% under regression, and 76% and 75% under the PC model were observed for the FYM and MR blocks, respectively. FYM@10t/ha + 50%NPK was superior, with maximum gross returns of Rs.41286/ha and benefit–cost ratio (BCR) of 2.27 in the FYM block compared to MR@5t/ha + 100%NPK (gross returns of Rs.34530/ha and BCR of 2.09) in the MR block with maximum soil fertility, and are recommended for adoption under semi-arid Alfisols.