Trend and stability analysis to interpret results of long-term effects of application of fertilizers and manure to cotton grown on rainfed vertisols

A field experiment was conducted from 1985-1986 to 2002-2003 on Vertisols under rainfed conditions to evaluate the effect of cropping systems and application of fertilizers and manure on seed cotton yield. To determine the long-term effects, trends and stability analyses were performed. Soil samples (0-0.15 and 0.15-0.30 m) were collected at the end of year 18 and analysed for available P and AB-DTPA extractable Zn. Among cropping systems, Asiatic diploid cotton (Gossypium arboreum) yielded 233 kg ha(-1) more seed cotton than the upland tetraploid cotton (Gossypium hirsutum). Yield trend was positive for G. arboreum compared with G. hirsutum. However, the slope was not significant. Stability analyses indicated overall higher yield stability for G. arboreum than G. hirsutum. Compared with monocropping G. hirsutum, G. hirsutum-sorghum (Sorghum bicolor L.) (H-S) rotation was significantly more stable. Soil samples (0-0.30 m) of the manure-amended plots had significantly greater P and Zn content (above the critical limit) compared to those receiving inorganic fertilizers alone. With regard to nutrient management practices, seed cotton yield was the highest for the integrated nutrient management (INM) treatment receiving a combination of organic and inorganic fertilizers. Among primary nutrients, the effect of P was significant while that of K was not. Balanced fertilizer application was significantly better than treatments receiving N and NK. Yield trends were, in general, not significant. However, a positive trend was noticed for treatments receiving manure compared to fertilizer alone. Stability analysis, on the other hand, indicated that the slopes were, in general, significant. Among the nutrients, mean yield response with and without P was 1007 and 672 kg ha(-1) respectively. Combined application of manure and fertilizer (INM) resulted in the highest mean yield response (1218 kg ha(-1)) and the slope was highly significant (P < 0.004). In the manure-amended plots, a better nutrient status probably imparted a greater degree of yield stability. The present study suggests that compared to trend analysis, stability analysis being sensitive as it recognizes the treatment x environment interaction, is a better option to interpret results from long-term agronomic experiments.