Planning, Designing and Analysis of Experiments relating to AICRP on Soil Test Crop Response Correlations

Abstract

The balanced application of fertilizer nutrients particularly the major nutrients, N P and K in optimum quantity, based on soil test and crop requirement is one of the most vital aspects for sustaining higher agricultural production. This requires the application of optimally balanced quantity of fertilizers in right proportion through correct method and time of application for a specific soil-crop-climate situation. It ensures increased quantity of produce, maintenance of soil productivity and the most efficient and judicious use of applied fertilizers. Thus in this situation the soil fertility evolution and refined fertilizer prescription for sustained agricultural production is of great importance to any country of the world in general and farming community in particular. Hence the soils have to be tested precisely for their available nutrient status for making fertilizer recommendation based on crop response and economic circumstances. Soil tests can provide a valuable piece of information and as such should be used in conjunction with such other information that is available for the estimation of fertilizer requirements. Soil test crop-response studies has been going on for a quite a long period of time both in India and abroad. The All India Coordinated Research Project on Soil Test Crop Response Correlations (STCR) was initiated during the year 1967-68. Currently, STCR project is having 17 cooperating centres. Earlier, the STCR project used multiple regression approach to calculate the dose of nutrient (s) required to obtain the maximum yield of crops under given set of experimental conditions. It can further be used to calculate the economic dose of fertilizer nutrients by incorporating a constant factor i.e. per unit cost of input (fertilizer) in the original equation. In this approach yield is regressed with soil nutrients, fertilizer nutrients, their quadratic terms and the interaction term of soil and fertilizer nutrients. For this the following criteria should be fulfilled (Annual report, AICRP on STCR, 1993-98) (a) Soil test crop response calibration for economic yield of a crop is possible only when the response to added nutrients follow the law of diminishing returns. i.e. the signs of partial regression coefficients of linear, quadratic terms of nutrients and their interaction with available soil nutrients should in general be positive, negative and negative ( + , _ , _ ) respectively. (b) The coefficient of determination (R2) should be high. (c) The partial regression coefficients should be statistically significant. (d) The experiment should have sufficient design points i.e. the number of treatments should be at least two or more than the number of variables in the model. The above criterions are seldom fulfilled using multiple regression. In such cases the optimum values of the nutrients cannot be derived or if they could be derived, they are either too high or too low. Keeping in view of the above problems and for better analysis of data, their interpretation and improvement in soil test calibration, the projector coordinator (STCR) Indian Institute of Soil Science, Bhopal, formally approached IASRI, New Delhi for collaboration. Consequently a project entitled “Planning, designing and analysis of experiments relating to AICRP on soil test crop response correlations” was under taken at IASRI w.e.f.1st march 2000 and its report was published in 2003. Thereafter in the annual workshop of STCR on 1st January, 2005 at IISS, Bhopal, a new design structure which was suggested in the earlier report, was accepted for experimentation. Since 2005, experiments with new treatment structure involving organic manures and major nutrients N, P and K along, in 24 design points are being conducted at all the centres of AICRP on STCR using the design suggested by the Institute and the data of conducted experiments are being analyzed at IASRI. New Delhi. As a consequence of that this present project was launched with effect from. March 01, 2007 at IASRI with collaboration of STCR (IISS), Bhopal with the following objectives: (1) To develop suitable methodology for the analysis of data of past experiments conducted under STCR; (2) To plan, design and analyze the data of experiments relating to AICRP on Soil test crop response correlations (STCR). In this report, the first two chapters contain introduction and review of literature. In the third chapter the methodology used has been discussed In order to analyze the data, at first, it is examined whether the fertility gradient has been created. For this analysis of variance was carried out using the soil nutrients, SN, SP and SK separately as dependent variables.. Then the following types of analysis were performed: (1) Evaluation of responses to middle doses of N, P and K, (2) Analysis of variance with and without covariates SN , SP and SK, (3) Fitting of response surfaces at various levels of organic manure and also combined over all levels, (4) Testing of Homogeneity of the regression equations, (5) Exploration of response surface in the vicinity of the stationary point, (6) Estimating the optimal values of N, P and K to be applied and (7) Targeted yield equations. In chapter four, results of all the analyzed experiments have been presented with interpretation. Although we have received the data from 16 centres but due to pending query for discrepancies, only data of 13 centres (about 37 experiments) have been discussed in detail. To summarize the results, it is observed that out of 13 centres, the fertility gradient was created in respect of SN, SP and SK over all the field of experimentation in 9 centres. In the remaining centres it is not created for either in respect of SN or SP or SK. While checking the creation of fertility gradient, Organic manure level wise, it was observed that in 6 centres the fertility gradient was created for SN, SP and SK. In the remaining centres, it was not created for either in respect of SN or SP or SK. When analysis of variance was carried out for Treatment, FYM and their interaction, it was observed that in 6 centres all the effects were significant. In the remaining centres, it was observed that in most of the cases, the interaction effect is not significant. However when Analysis of Covariance was carried out taking the soil available nutrients SN ,SP and SK, it was observed that in most of the cases, the interaction effect was also significant along with considerable reduction in coefficient of variation. In Pantnagar it was observed that for all the 5 crops, the Treatment, FYM and their interaction effects are non-significant for both Analysis of variance as well as Analysis of Covariance. When analysis of variance was carried out for Treatment, FYM and Strips (within FYM levels), it was observed that at almost all the centres, all the effects were significant. In some cases we find that the effects of the Strips (within FYM levels) were not significant. To evaluate the optimal doses of N, P and K, the STCR project uses Fertilizer adjustment equations for a fixed targeted yield when multiple regression equations do not provide any solution. The equations thus generated although provide good results at the follow up trials but are not statistically sound. Therefore, there is variation in the coefficients from year to year and so these cannot be pooled. In order to give a statistical backing to the whole process, a method has been worked out at IASRI to get the desired results by combining the method of fertilizer adjustment equations with that of response surface methodology. The basic assumption in the targeted yield approach is that the plant nutrient uptake from the control plots and treated plots is same. Therefore it was felt that the doses of FN, FP and FK be worked out through Response Surface Methodology by exploring the response surface in the vicinity of stationery point. The stationery point is a point of a maximum, minimum or a saddle point (which neither maximum nor minimum). This method is applicable when the stationery point lies within the experimental region. If it is not within the experimental region, then also it is possible to find out the different combination of doses of FN, FP and FK with the help of canonical analysis of the response surface and ridge analysis. In this method, the optimal values of N, P and K fertilizer nutrients can be derived if the soil test values for a particular site are available. The optimal values of the fertilizer nutrients N, P and K obtained by Response Surface Methodology, has been found to be closely related to that obtained by Targeted yield approach(fertilizer adjustment equations) adopted by the STCR project. Thus one could advocate for the adoption of the Targeted yield approach as has been tested by sound statistical system of Response Surface Methodology.