DRIS norms for ‘ Valencia ’ sweet orange on three rootstocks

Diagnosis and Recommendation Integrated System (DRIS) applies nutrient ratios instead of the isolated concentration values of each nutrient in interpretation of tissue analysis. The objectives of this research were to establish adequate DRIS norms for ‘Valencia’ sweet orange irrigated commercial groves budded on three rootstocks and correlate indexes of nutrition balance with yield. Experiments were conducted in São Paulo State, Brazil. Rootstocks Rangpur lime, Caipira sweet orange, and Poncirus trifoliata, with more than six years old and yield above 40 ton ha were utilized. Data referred to yield, tree spacing, rootstock and foliar concentrations of N, P, K, Ca, Mg, S, Fe, Mn, Cu, Zn, and B in non fruiting terminals for each grove were processed for the years 1994 through 1998. DRIS indexes were calculated by Nick criterion for choosing the ratio order of the nutrients and Jones calculation method of the ratio functions. Indexes of nutritional balance calculated from DRIS norms presented high correlation with yield for the three scion/rootstock combinations. DRIS norms defined in this research are valid, since leaf sampling is done on non fruiting terminals and the grove is irrigated.


Introduction
The usual methods for leaf analysis interpretation are based on the comparison of the nutrient concentration with critical reference values (sufficiency range approaches).Concentration values above or below reference values are associated with decrease in vegetative growth, yield, and quality.These methods consider the association of isolated concentration values with deficiency or excess, without considering the nutritional balance.Moreover, investigations related to this subject indicate great difficulty in the establishment of consistent critical values and its relationship with high yields, especially because the nutritional status varies with leaf tissue maturation, and also due to sink and concentration effects in high or low yield years.Therefore, sampling is an essential step for better efficiency of these methods.

DRIS (Diagnosis and Recommendation Integrated
System) applies nutrient ratios instead of absolute concentration values of each one for leaf analysis interpretation.DRIS norms have been developed for several field, forest, and horticultural crops, and have been applied as an additional tool for nutritional status diagnosis in the United States, Canada, and China (Lopes, 1998;Hallmark & Beverly, 1991).Beverly (1987) suggested three modifications in DRIS method and proposed two new methods of diagnosis in 'Valencia' sweet orange, already defined in a previous investigation (Beverly et al., 1984).The author emphasized that the logarithmic transformation, population parameters, and use of a single calculation method may reduce systematic errors, simplify the diagnostic method, and extend its applicability.The two new proposed methods use individual nutrient concentration values, instead of ratios.
Investigations by Woods & Villiers (1992), in South Africa, pointed out that DRIS can be successfully applied in nutrient diagnosis of 'Valencia' sweet orange groves.The authors correlated yield (kg per plant) and quality (fruit mass) with DRIS indexes, working in a database with more than 1,700 observations.DRIS norms were also evaluated in field fertilization trials, and successfully associated with increases in yield and fruit quality.Cerda et al. (1995)  In Brazil, investigations about DRIS in citrus are rare.Creste (1996), in 'Siciliano' lemon, organized a databank with leaf analysis in fruiting terminals, from plants with different ages, rootstocks and harvest years.Standard populations were grouped according to yield above 80 ton ha -1 .After calculation of DRIS norms, the method was evaluated in field conditions.DRIS was considered an efficient method, especially because it takes into account deficient or excess nutrients in an order of importance.Santos (1997) utilized a databank of leaf analysis from a N, P, K-fertilization field trial network and commercial groves in São Paulo State to evaluate DRIS.Among three DRIS index calculation methods, the one proposed by Jones (1981) showed more advantages.
Citrus nutritional status can be affected by numerous factors, such as soil and climatic effects, scion/rootstock combination, depth of root system, pests and diseases.
This work focused on the establishment of adequate DRIS norms for 'Valencia' sweet orange on three rootstocks, and on the correlation between them and the yield.

Material and Methods
This study was carried out with data from irrigated 'Valencia' sweet orange groves budded on Rangpur lime (Citrus limonia L. Osbeck), 'Caipira' sweet orange (Citrus sinensis L. Osbeck), and Poncirus trifoliata, over six years old and yield above 40 ton ha -1 .Groves were located at a commercial citrus farm, in Mogi Guaçú, SP, Brazil.Data referred to yield, tree spacing, rootstock and leaf N, P, K, Ca, Mg, S, Fe, Mn, Cu, Zn, and B concentrations in non fruiting terminals of each grove were processed from 1994 through 1998.Leaf sampling was proceeded according to Hanlon et al. (1995).
Leaf analysis and yield data were organized in a large database, classified by grove referential number, rootstock, plant density (number of plants per hectare), leaf sampling and harvest year.
The nutrient ratio order criterion used was described by Nick (1998).The ratio functions of nutrients were calculated according to Jones (1981).The index of nutritional balance (INB) was calculated by the average of all DRIS indexes (including I DM ) irrespective of sign (Nick, 1998).
Population was classified, for each rootstock, according to yield in a decreasing order.Data was analyzed for each year or a combination of years, considering the influence of non nutritional factors affecting yield, such as diseases and the effect of climate on flowering (Mourão Filho et al., 2002).Simulations varying standard (high-yield) population size were performed.DRIS norms and indexes were calculated in each simulation.Adjusted equation, R 2 , for each regression analysis was registered.Standard population selected was that which induced the highest R 2 in regression analysis.

Results and Discussion
Adjusted equation of INB versus yield of 'Valencia' sweet orange on Rangpur lime resulted in calculated R 2 of 0.62 (Figure 1A).The study included 12 groves of this scion/rootstock combination with data referred to 1995 leaf sampling and harvest year (Table 1).Yield of the selected population of 'Valencia' sweet orange on Rangpur lime varied from 40.5 to 71.5 ton ha -1 , and the reference population presented 62.2 to 71.5 ton ha -1 (numbers 1 through 12).Plant density did not vary much among population tested (222 to 247 plants per hectare), as well as plant age, which for most groves, varied between 22 and 33 years.Only grove referential number 12 was 11-year-old at the time of study.In general, groves were considered in full production after six years from planting.Considering that all groves were irrigated, and kept under toping and hedging pruning, it is very likely that the root system of an 11-year-old grove already explores all available soil around, for the same spacing, in a similar way as in an older grove.Therefore, these groves were analyzed within the same population.
Selection of the standard population of 'Valencia' sweet orange on Rangpur lime took into consideration the yield potential of the scion/ rootstock combination.All groves included in the low-yield population were considered with potential to increase yield levels up to those presented by the standard population.Those groves with yield bellow 40 ton ha -1 were not included in DRIS analysis because non-nutritional factors might have influenced fruit production.Considering the fact that Rangpur lime is very susceptible to Citrus Blight (Castle et al., 1993), the high incidence of this disorder may have played an important role in decreasing yield, especially in those groves older than 20 years.
DRIS norms for 'Valencia' sweet orange on Rangpur lime were calculated and are valid for this scion/rootstock combination, since leaf samples are collected in non fruiting terminals and the groves are irrigated.
Correlation coefficient values (r) between each pair of nutrient ratios (A/B or B/A) and yield are also reported herein (Table 2).These values were determined from the analyzed populations (standard and low-yield populations), and are useful to verify the influence of each pair of nutrients on yield; the theoretical basis of the r value criterion according to Nick (1998), was applied in this research.The pairs of nutrients with r values over 0.5, irrespective to sign, are 1/Ca, 1/S, Ca/N, N/S, P/S, Ca/K, K/S, K/Zn, Mg/Ca, 1/Zn, B/Ca, Fe/Ca, Mn/Ca, Mg/S, Mg/Zn, B/S, Cu/S, Fe/S, Mn/S, B/Zn, and Mn/Zn.Adjusted equation of INB versus yield of 'Valencia' sweet orange on Caipira sweet orange resulted in calculated R 2 of 0.74 (Figure 1B).The study included 19 groves of this scion/rootstock combination with data referred to 1994 and 1995 leaf sampling and harvest years (Table 1).Yield of the selected population varied from 42.8 to 94.9 ton ha -1 , and the standard population was grouped by those groves with yield between 77.5 and 94.9 ton ha -1 (numbers 13 through 31).Plant density varied more intensively in this population (222 to 370 plants per hectare), as well as plant age (9 to 31 years).Caipira rootstock, as any other sweet orange variety, is very susceptible to Phytophthora (Castle et al., 1993), and this fact may have played an important role in decreasing yield.Younger groves were also those with higher plant densities and presented higher yields.However, as well as already discussed for 'Valencia' on Rangpur lime, plant age probably is not an important factor.Considering that the groves are irrigated and kept under pruning, it is likely that a nine-year-old grove already explore all available soil around, for a similar plant spacing, as well as older groves.
Moreover, the cut-off limit of the standard population selected for Caipira rootstock is different from the one chosen for 'Valencia' sweet orange on Rangpur lime; in the first yield varied from 77.5 to 94.9 ton ha -1 , whereas in Rangpur lime, the maximum yield of the standard population was 71.5 ton ha -1 .DRIS norms for 'Valencia' sweet orange on Caipira sweet orange, as well as correlation coefficient values (r) between each pair of nutrient ratios (A/B or B/A) and yield are also reported (Table 3).The pairs of nutrients with r values over 0.5, irrespective to sign, are 1/S, N/S, Cu, Mn, Cu/N, Mn/N, Cu/P, Mn/P, K/Mg, S/K, Cu/K, Mn/K, Ca/Cu, Mn/Ca, Cu/Mg, Mg/Fe, Mn/Mg, S/Fe, Mn/S, Mn/B, Cu/Fe, Cu/Zn, Mn/Fe, and Mn/Zn (  , Brazil. 1994, Brazil. , 1995. .cia' sweet orange on Poncirus trifoliata resulted in calculated R 2 of 0.83 (Figure 1C).The study included 16 groves of this scion/rootstock combination with data referred to 1994 sampling year and harvest year (Table 1).Yield of the selected population varied from 42.6 to 67.8 ton ha -1 and the standard population presented 62.6 to 67.8 ton ha -1 (numbers 32 through 47).Among scion/rootstock combinations tested, this was the one that presented higher variation in plant density (222 to 556 plants per hectare), even though the very high plant density was observed only in the grove referential number 39. Average grove age also varied (6 to 32 years), and this fact seemed not to be relevant for DRIS norms calculation.It must be pointed out that the younger groves are also those with higher plant densities.Therefore, considering the conditions of this study, younger high-density groves may have the same yield potential as older low-density groves.
Considering the effect of the scion/rootstock combination, selected cut-off limit of standard population of 'Valencia' on Poncirus trifoliata was different from the other combinations (Table 1).In this case, yield of standard population varied from 62.6 to 67.8 ton ha -1 , much lower than the average yield in 'Valencia' on 'Caipira' sweet orange (77.5 to 94.9 ton ha -1 ), and in 'Valencia' on Rangpur lime (62.2 to 71.5 ton ha -1 ).All groves included in the low-yield population presented yield above 40 ton ha -1 , and were considered with potential to increase yield.
It is important to point out that, for all three rootstochs in spite of different regression equations, INB was highly correlated with yield, which decreased with the increase of nutritional imbalance (Figure 1A, 1B e 1C).This is a strong evidence that the method used to calculate DRIS indices and, therefore, INB, took into consideration the right criteria, especially the selection of specific standard populations for each scion/rootstoch combination.
According to previous investigations by other authors, calculated norms can be as much representative as more specific high-and low-yield populations are selected.Databases to be used for calculation of DRIS norms may have variable size, according to the method (Letzsch & Sumner, 1984), and may not be directly related to the quality of the standards (Walworth et al., 1988).Possibly, more general DRIS norms may result in lower diagnosis efficiency.The high quality of the observations must be the main characteristic in searching and choosing the database, as oppose as quantity.
In this work, 'Valencia' sweet orange, older than six years and with yield above 40 ton ha -1 , was the selected population to optimize DRIS efficiency.The establishment of DRIS norms was based on restricted databases, but extremely uniform and with high quality, varying from 12 to 19 observations.
Considering the significant influence of the soil and climatic conditions, scion and rootstock variety, this work shows that DRIS norms for citrus must be calculated for specific conditions, when higher correlation between INB and yield is achieved.These results were also found by Nick (1998), calculating DRIS norms in coffee.Therefore, general DRIS norms (Rodriguez et al., 1997;Santos, 1997), in spite of being more extended, may present less application when applied in specific conditions such as those of the present work.
The establishment of DRIS norms was made within specific populations, with segregation by rootstock, leaf sampling and harvest year.Simulation of DRIS norms involving populations grouped by several leaf sampling and harvest years (1994 through 1998) did not result in high correlation between INB and yield (Mourão Filho et al., 2002).The most reasonable explanation for this fact could be the great influence of climate on flower bud induction and the different degree of incidence of diseases, such as Postbloom Fruit Drop (Colletotrichum gloesporioides) throughout the study period, affecting yield more significantly than nutritional factors.Separation of the populations of

Table 1 .
Yield, macro and micronutrient concentrations in leaf samples from non fruiting terminals of 'Valencia' sweet orange groves budded on the rootstocks (RT) Rangpur lime (RL), Caipira sweet orange (CP), Poncirus trifoliata (TR) utilized for the establishment of DRIS norms according to the grove referential number (NM), plant density in number of plants per hectare (DEN), grove age in years (GA) and leaf sampling and harvest year (HY).MogiGuaçú, SP, Brazil.1994, 1995.

Table 2 .
Mean and standard deviation (SD) of DRIS norms for 'Valencia' sweet orange on Rangpur lime and correlation coefficient (r) between each nutrient ratio pair (A/B or B/A) and yield.Mogi Guaçú,SP, Brazil.1995.

Table 3 .
Mean and standard deviation (SD) of DRIS norms for 'Valencia' sweet orange on Caipira sweet orange and correlation coefficient (r) between each nutrient ratio pair (A/B or B/A) and yield.Mogi Guaçú, SP