| Abstrakt: |
General fertilizer response models are useful to derive robust recommendations for farmers who face different circumstances. This paper shows that a flexible functional form such as the transcendental can be applied to develop both agronomically and economically sound recommendation tables using a set of soil test, climate, and management variables. Data from thirty‐eight experiments on wheat response to N and P in the Humid Pampa, Argentina, were used to build a general response model that incorporates the initial levels of soil test measures (NO3‐N and Bray‐1 P), rainfall from sowing to maturity, and previous crop. The economic analysis was performed for relevant nutrient/grain price ratios ro= 8.0 and rp= 18.3 for N and P, respectively. A table of fertilizer recommendations for wheat was derived for different combinations of rainfall expectations, previous crop, and soil test values. The following results were obtained: (i) the estimated model not only accounted for a significant percentage of the total variability in the dependent variable (InY), with R2=0.56, P= 0.01, but it also gave coefficients with signs in conformity with agronomic expectations; (ii) the economic optima for N (N*) and P (P*) are compatible with the range of optima computed with per‐site economic analyses; (iii) N* and P* increased as rainfall increased; (iv) N* was lower for wheat after soybeans than for wheat after maize; (v) N* was higher for smaller values of soil N, while P* was practically unaffected by soil N; and (vi) P* was higher for smaller values of soil P, while N* was affected by soil P, with a pattern that depended on both previous crop and rainfall. Similar tables can be derived for groups of farmers under different economic circumstances (i.e., different input/output price ratios). |
Plný text