A N EVALUATION OF DRlS BASED O N LEAF ANALYSIS FOR SUGARCANE

In recent years increasing attention has been given to the Diagnosis and Recommedation lntegrated System (DRIS) for diagnosing the nutrient requirement of sugarcane and other crops. Sugarcane yield data and third leaf analysis from 96 fertilizer trials have been used to establish whether DRlS can be used to improve the quality of the fertilizer advisory service offered by the Experiment Station. In general it was found that predictions of a yield response to applied N, P and K were more reliable when DRlS was used than when the nutrient threshold approach was used at an early rather than a late stage of crop development. The results suggest that imbalances of N, P and K can be detected four to six weeks earlier by using DRlS than can be accomplished by using the threshold approach. DRlS can be used fairly reliably to indicate N, P and K deficiencies in order of decreasing importance. It will now be possible to check the adequacy of fertilizer programmes at an early stage of crop development to recommend additional treatment where necessary for the benefit of the crop being sampled. Sometimes unsuspected problems related to P fixation and early losses of N due to leaching or denitrification may also now be identified when the crop is very young. DRIS soil norms are currently being investigated. Introduction The Experiment Station of the South African Sugar Association has successfully conducted a fertilizer advisory service, based on soil and leaf analysis, for almost 27 years. While the demand for soil tests has always exceeded that for foliar diagnosis, the records show that the gap has steadily closed during the past decade, largely due to the wider acceptance by growers of whole cycle recommendations. Foliar diagnosis has the important advantage over soil testing that by determining the nutrients the plant has actually taken up doubts about effective rooting depth of the crop and the choice of suitable chemical extractants for estimating plant-available nutrients are to a large extent eliminated. The su,garcane plant with its extensive root system will always provide a more representative sample under the highly variable soil conditions that occur in the sugar industry than will a man with an auger and somewhat arbitrary nutrient extraction procedures. Despite the considerable merits of foliar diagnosis, it must be conceded that with its introduction a number of variables such as crop age, month and season of sampling and bioclimatic regions become important, and that these factors have little effect on soil testing. The effects of crop age and time of year on foliar N are well known, and although a standard sampling procedure and the use of a comprehensive set of threshold values may minimize these effects, it is not always possible or convenient to ensure that sugarcane is sampled at a standard physiological stage. The major drawback to leaf analysis, however, is that the results can seldom be used to the benefit of the crop from which the sample was taken because the growing season, or most of it, has already passed. Because of these difficulties, and because current methods can be used neither to classify nutrient excesses nor to estabish the order of importance of nutrient deficiencies, interest has arisen in ratios of the amounts of nutrient elements in cane leaves. The DRlS technique (Diagnosis and Recommendati,on Integrated System) of Beaufils"~ is based on nutrient ratios and it is being evaluated for sugarcane not only in South Africa but also in Brazil (Zambello20), Florida (Gascho") and Hawaii (Jones'". Interest in this approach for sugarcane arose in 1972 at the Experiment Statimon when it was found that the P requirement of sugarcane grown on high P-fixing soils could be assessed most reliably from the N: P and K: P ratios in the TVD (Top visible dewlap) leaf, using a DRlS chart (Anon'). A subsequent study of the results of twenty four 4Nx2Px3K Regional Fertilizer trials (Meyer16) indicated that the nutrient status of the crop with respect to both N and P could be predicted more reliably by DRIS than by the conventional nutrient threshold approach, although it appeared unlikely that the same DRIS norms could be used in all circumstances. The use of DRlS tor sugarcane was studied subsequently in a three-year project conducted by the Soil Science Department of the University of Natal. A number of diagnostic norms were developed for establishing the order of importance of both macro and micro nutrient deficiencies. Tentative leaf N, P, K, Ca and Mg norms were first developed from yield data obtained from South African cane growers' files in association with analytical results provided by the SASA Experiment Station's Fertilizer Advisory Service (FAS). A yield of five tons cane per hectare per month was used to discriminate between high-yielding and lowyielding crops. Because these data did not represent a sufficiently wide range of sampling conditions, a second set of norms was developed for diagnosing both macro and micro nutrient imbalances in leaf and soil samples. In this exercise a yield of seven tons cane per hectare per month was used to separate high yielding from low yielding cane. The results indicated that the order in which nutrients limit cane yield could be established at any stage of crop development (Beaufils and Sumner6). Similar findings have been reported for crops such as maize, rubber and cotton (Beaufils2p 3, wheat and soya beans (SumnerlT1 la), potatoes (MeldalJohnstonI5) and tea (Leal4). Although these findings were promising, the testing of the newly-derived norms was unfortunately restricted to the results of a small number of trials. Crop yield data and leaf analysis from 96 fertilizer trials conducted throughout the South African sugar industry were therefore used to provide a more vigorous test of the system, and this paper concerns some of the more important conclusions that could be reached. 170 Proceedings of The South African Sugar Technologists' Association June 1981 Procedure The evaluation process was divided into four phases : (i) selecting suitable DRIS norms and retrieving yield data and leaf and soil analysis in forms suitable for processing by computer into DRIS indices. (ii) using the computed data to compare the relative effectiveness of nutrient threshold values and DRIS in predicting responses to applied nutrients at various stages of crop development and for cane grown in the coast lowlands, midlands and lowveld regions of the industry. (iii) testing whether DRIS norms correctly predict the order of importance of nutrient deficiencies, and whether the results can be related to actual fertilizer requirements. (iv) establishing how DRIS could best be implemented for advisory purposes. Although the main object of this investigation was to test the norms supplied by the University of Natal it was considered worthwhile to include the set developed previously by the SASA Experiment Station, and a further set from the Florida Sugar Experiment Station (Gascholl). Five sets of N P K norms, designated A, B, C. D and E, were thus selected and details are given in Table 1. The general principles used in establishing a data bank and selectiqg parameters with significant co-variance ratios have been outlined previously by Beaufils4. Since the DRIS approach is intended to include as many nutrient elements as possible, a sixth set of norms, developed from the University data bank for minor elements, was also tested and the average values are shown in Table 2. These are subsequently referred to as the "general" set of norms. Considering that the data to establish the various n m s come from such widely divergent sources, the norms shorn in Table I are surprisingly similar. The N/P and KIP norms established in Florida deviated most from the overall mean values. These norms were used to convert approximately 1 200 sets of leaf analyses, #obtained from the results of 96 fertilizer trials, into N, P, K, Ca, Mg and Zn indices using the general formula developed by Beaufils4. The equation used for calculating the N index is given by : f(N/P) -If(N/K) f(Mg/N) f(Ca/N) N index = 4 where f(N/P) = 100(N/P -1) 10/CV when NIP > n/p nl P or f(N/P) = 100 (1 n/p) 10/CV when N/P < n/p NIP N/P is the value of the ratio (N% oven dry third leaf tissue t P% oven dry third leaf tissue) for a particular sample, and n/p is the value of the norm for this ratio given in Table 1. CV is the coefficient of variation for the population of high yielding plants. The indices for the other nutrients are calculated similarly. The DRIS indices may be positive or negative but their sum is always zero because they measure the balance among the nutrients N, P, K, Ca and Mg. The more negative the value of an index, the greater the probability of the nutrient concerned being deficient and the crop yields being reduced. Conversely, the more positive the value of an index the smaller is the likelihood of the nutrient under consideration being deficient. An index close to zero indicates that the nutrient concerned is in adequate supply. From their indices for a given leaf sample, the nutrients can be classified in order of deficiency, adequacy and excess, the most negative index indicating the nutrient most required and the most positive index the nutrient least required. The details concerning the sources of data used for the assessment of DRIS discussed in this paper, according to the main physiographic regions, are given in Table 3. Data from only two trials in the midlands were used for evaluating N and K indices for this area, but otherwise the distribution of trials represents fairly adequately the soil and climatic conditions which occur in the sugar industry. Responses to treatment with nitrogen in the 72 crops examined varied from 12 to 62 tons cane per hectare, the average being 33,O tons cane per hectare. TABLE 1 Details of the five sources of data, and the average third leaf norms for N, P and K I I I I I. I Norms