Diagnosis of the Nutritional Status of Garlic Crops

Univariate methods for diagnosing nutritional status such as the sufficiency range and the critical level for garlic crops are very susceptible to the effects of dilution and accumulation of nutrients. Therefore, this study aimed to establish bivariate and multivariate norms for this crop using the Diagnosis and Recommendation Integrated System (DRIS) and Nutritional Composition Diagnosis (CND), respectively. The criteria used were nutritional status and the sufficiency range, and then the diagnoses were compared. The study was performed in the region of Alto Paranaíba, MG, Brazil, during the crop seasons 2012 and 2013. Samples comprised 99 commercial fields of garlic, cultivated with the cultivar “Ito” and mostly established in Latossolo Vermelho-Amarelo Distrófico (Oxisol). Copper and K were the nutrients with the highest number of fields diagnosed as limiting by lack (LF) and limiting by excess (LE), respectively. The DRIS method presented greater tendency to diagnose LF, while the CND tended towards LE. The sufficiency range of both methods presented narrow ranges in relation to those suggested by the literature. Moreover, all ranges produced by the CND method provided narrower ranges than the DRIS method. The CND method showed better performance than DRIS in distinguishing crop yield covered by different diagnoses. Turning to the criterion of evaluation, the study found that nutritional status gave a better performance than sufficiency range in terms of distinguishing diagnoses regarding yield.


INTRODUCTION
Leaf analysis allows us, inter alia, to monitor, evaluate and adjust agricultural fertilization programs (Menesatti et al., 2010;Tomio et al., 2015).The use of leaves to evaluate nutritional status of plants is based on the relations between the leaf content, nutrient uptake and plant yield (Fageria et al., 2009).
Because of the ease of interpretation, the sufficiency range method is the one most often used to evaluate the values found by leaf analysis (Wadt et al., 2013).In this method, the nutritional contents are compared individually with reference values, and require very similar soil and climatic conditions among evaluated crops and places where the calibration experiments were carried out (Serra et al., 2010a).
The Diagnosis and Recommendation Integrated System (DRIS) (Beaufils, 1973) was developed to overcome several limitations inherent in the sufficiency range.This method employs the DRIS index calculation by using the ratio between one nutrient and another.The main advantage is the greater constancy of the ratios among the nutrients in crops with the same nutritional status compared with individual contents.
Another alternative method to the sufficient range method is Compositional Nutrient Diagnosis -CND (Parent and Dafir, 1992).In this method, the nutrient balance is based on all nutrients, rather than two, as in the DRIS method.The CND method considers all possible interactions among the nutrients and the non-mineral dry matter of the plant, which allows more accurate diagnosis, as the concentration and dilution effects of the nutrients in dry matter are minimized.In addition, CND has all the advantages inherent in DRIS and makes calculations easier (Wadt et al., 2013).
Currently, norms created by the sufficiency range method are in use and there is an established critical level for the nutritional assessment of garlic crops.Besides the limitations presented by the univariate character of both methods (where the levels are individually compared), the most recent sufficiency ranges are those proposed by Trani and Raij (1997) and the critical levels established by Malavolta et al. (1997).
Some studies have aimed to establish norms for assessing the nutritional status of garlic plants as regards N (Backes et al., 2008).The reference values found were based not only on the nutritional content but also on the SPAD index (Soil Plant Analysis Development).However, attempts to improve the nutritional diagnosis of garlic have been limited to N.
In this context, the hypothesis was that the method multivariate (CND) may be more effective in the bivariate (DRIS) for nutritional diagnosis of garlic crop.This study aimed to generate norm and nutrient ranges by using modern methods for foliar diagnosis of garlic crops, to compare the methods and to determine the most limiting nutrient in terms of yield.

MATERIALS AND METHODS
The database of this study comprised the Alto Paranaíba region of Minas Gerais, Brazil, during the 2012 and 2013 cropping seasons.Data collection was carried out over two years in order to increase the database representation and perform diagnosis with a two-season average.The regional climate is classified as Cwa, following the Köppen-Geiger system.The altitude of the areas ranged from 900 to 1,200 m.The vast majority of the garlic fields under study have soils classified as Latossolo Vermelho-Amarelo and a smaller number as Latossolo Vermelho and Latossolo Amarelo (Santos et al., 2013b).
The norms for the nutritional diagnosis of garlic crops were given by the Diagnosis and Recommendation Integrated System (Beaufils, 1973) and the Compositional Nutrient Diagnosis methods (Parent and Dafir, 1992), both using the nutritional status (NS) (Silva et al., 2004) and the sufficiency range (SR) criteria.
The database of this study was made up of 142 commercial fields.Leaves and bulbs were sampled.All the fields evaluated were planted to the "Ito" garlic variety, belonging to the noble group, with a late cycle and purple coloring.The cloves went through the process of vernalization before planting, with temperatures from 2 °C to 5 °C for a period of 45 to 60 days.All the areas were under center pivot irrigation, except for two fields that were irrigated by conventional sprinkling.The mean values and the standard deviations of the fertilizations were 222±30 kg ha -1 of N; 371±39 of P, and 417± 62 of K.For liming, it was used the saturation method bases, aiming to achieve 80 to 85 % of the expected saturation.
The index leaf was taken after the beginning of clove differentiation by sampling 15 leaves at four distinct points, resulting in 60 samples per field.The younger and completely expanded leaf of garlic was considered the index leaf at the beginning of bulbification (Trani and Raij, 1997;Rosen et al., 2008).The bulbs were collected to determine the yield of each planting field.Fifteen sequential plants were collected at the same four points marked per field for leaf index collection, forming a sample of 60 bulbs.The leaves were dried in an air circulation laboratory oven at 70 °C for 72 h, followed by grinding in a Willey-type mill and sent for chemical analysis according to the method described by Santos et al. (2009).
For N determination, sulfuric digestion was carried out, followed by Kjeldahl distillation.The other nutrients were subjected to nitro-perchloric digestion and analyzed by spectrophotometry (P and B), flame photometry (K), turbidimetry (S), and atomic absorption spectrophotometry (Ca, Mg, Cu, Fe, Mn, and Zn).To determine the yield of each field, the bulbs collected were weighed, after natural drying for 30 days, and the value obtained was extrapolated to one hectare.
The fields were classified and grouped into high and low yield subpopulations for the application of both methods.The high yield fields were those with yield above 18,500 kg ha -1 and low yield was below this value.The limit yield (18,500 kg ha -1 ) was the average yield plus two-thirds of the standard deviation of the sampled fields, as proposed by Urano et al. (2007) and Kurihara et al. (2013).
For the calculation of DRIS method, first the ratios were established between the content of each nutrient and the contents of others.The mean and standard deviation of the ratios of high yield subpopulation composed the DRIS norms.Further, the dual relationships of all fields were transformed into normal reduced variables [Z(A/N)], through the DRIS values, which were rounded to integers by the adjustment factor "c" = 10.The arithmetical average of the normal reduced variables, direct [Z(A/N)] and inverse [Z(N/A)] relations of each nutrient were defined as DRIS index (I A ) of this element (Alvarez V and Leite, 1999): where Z (A/N) = normal reduced variable for the dual relationship between the nutrient contents of the sample; A/N = dual relationship between the nutrient contents of the sample; a/n = dual relationship between the nutrient contents in the high yield subpopulation; c = adjustment factor; s = standard deviation of the dual relationship between the nutrient contents for the high yield subpopulation; and n = number of nutrients involved.
For the CND method, the nutrient contents of each field were fitted to the same unit (dag kg -1 ).Then, the content of dry matter components was calculated, except for the macro-and micronutrients (R).Then, the geometric average was calculated between the contents of dry matter (G) constituents for further correction regarding the content of each nutrient, resulting in a multinutrient variable (Vi): n+1)   where x i = contents of nutrients in the leaves; and n = number of nutrients in the evaluation.
Average and standard deviation of the high yield subpopulation multinutrient variables composed the CND norms.From these norms, for each nutrient and each field, the multinutrient variable (I Vi ) index was calculated: where I Vi = multinutrient variable index; Vi = multinutrient variable of the sample; vi = average of the multinutrient variables in the subpopulation of high yield; and si = standard deviation of the multinutrient variables in the subpopulation of high yield.
The index normality was evaluated, for the two methods, by the Shapiro-Wilk test at 0.05 level.When the indexes did not cater for the normality criterion test, the data were transformed through a Napierian logarithm.For both methods, the sum of the DRIS index modules or multinutrient variable indices of all nutrients resulted in the Nutritional Balance Index (IEN).The ratio between the IEN and the number of nutrients involved (n) resulted in the medium Nutritional Balance Index (IEN m ).
The results obtained for DRIS and CND methods were evaluated using the nutritional status criterion.To that end, the DRIS indices (I A ) or multinutrient variable indices (Ivi) of each nutrient were related to their medium nutritional balance indices (IEN m ), being classified as deficient, balanced or excessive (Table 1).For each nutrient the nutritional status diagnosis obtained by the two methods was compared.
Based on the principle of null values of DRIS indices (I A ) or multinutrient variable indices (Ivi) consistent with nutritional equilibrium conditions, the optimum contents of nutrients were estimated through fitting linear models of correlation between the contents and DRIS or CND indices for the high yield subpopulation.After the indices were obtained, the sufficiency ranges were defined through the standard deviation range of -6.66 s to +6.66 s for the DRIS index and -0.66 s to +0.66 s for the multinutrient variable index (Ivi).The intervals of DRIS and CND ranges were compared with each other and with the ranges established in the literature.
Finally, the capacity of the DRIS and CND methods was evaluated through nutritional status and sufficient range criteria to distinguish by yield the diagnostics of deficiency and toxicity regarding nutritional balance.For the sufficient range, nomenclature equivalent to that specified for the nutritional status criterion was determined.Thus, the NL fields were taken as the fields where the contents were within the established ranges and LF or LE as those where the contents were lower or higher than the respective ranges.
After the diagnostic produced by the two methods and the criteria were obtained, garlic fields with the same ratings were grouped to form the NL, LF and LE groups.Within each method and criterion, the average yield of the NL group was compared with that of LF and FL groups individually through 't' test at 5 %.Finally, the number of distinctions in yield and individual distinction capacity for each nutrient of all methods and criteria were compared.

RESULTS AND DISCUSSION
Twenty-five fields of the 99 sampled presented yield higher than 18,500 kg ha -1 , forming the high yield subpopulation.The remaining 74 fields were grouped into the low yield subpopulation.
The dual relationships between nutrients were transformed by the Napierian logarithm in order to capture the normality premise of the DRIS method (Alvarez V and Leite, 1999).Before data transformation, only 59.1 % of the ratios between nutrients presented normality according to the Shapiro-Wilk test, and 92.7 % after transformation, except for N/P, N/S, P/N, P/B, Ca/Cu, S/N, B/P and Cu/Ca relationships.Thus, the DRIS norms developed (mean and standard deviation) were based on the transformed relationships (Table 2).The normality deviations was reduced through transformation by the Napierian logarithm as proposed by Serra et al. (2010a).After this procedure, the authors obtained, respectively, 86.4,94.5 and 95.5 % of bivariate relationships with normal distribution.Even considering 92.7 % of the normal relationships of this study close of previous studies, the average between direct and inverse relationships was used in order to minimize the effects of those that deviated from the distribution (Alvarez V and Leite, 1999).
The multinutrient variables (Vn) of all nutrients showed normality according to the Shapiro-Wilk test.Thus, data transformation was not performed and the norms of the CND method (mean and standard deviation) were based on the original values (Table 3).At this point, the use of CND regarding the data found in this study could be advantageous because, unlike DRIS, none of its norms interfere with normality.
Copper was the nutrient with the highest percentage of fields diagnosed as LF, exceeding by 50 % the DRIS method and 40 % the CND method (Table 4).Potassium was, for most of the fields, ranked as LE, 55.4% of the total for DRIS and 58.1 % for CND.
The K excess, observed in more than half of the fields for both methods, is owed to the high doses applied in the areas, with an average of 417 kg ha -1 of K, while the extraction was approximately 120 kg ha -1 (data not shown).Thus, the K supply was excessive and influenced the balance of this nutrient in the soil-plant system.Whereas interactions occur in the absorption of ions such as K + , NH + 4 , Ca 2+ e Mg 2+ (Marschner, 2012), imbalance in the soil can occur because of the high doses of K employed.
The micronutrients with the highest percentage of fields diagnosed as LF were Cu> Mn> Fe.In contrast, B and Zn fitted within the micronutrients with the highest number of LE diagnostics.In part this result reflects the focus of research and fertilization with micronutrients on only B and Zn needs for garlic.In the official publications of Trani et al. (1997) for São Paulo state, within micronutrients there are only criteria for B and Zn recommendations.Only B supply is considered in Souza et al. (1999) for Minas Gerais state.
There was large percentage of the fields diagnosed as LF for P.This result is unexpected, considering that, on average, the fields received 371 kg ha -1 of P and the average P extraction was 30 kg ha -1 .Thus, it is questionable whether P application in garlic crops is being performed efficiently.In the cultivation areas, phosphate fertilizer was broadcast and incorporated about 0.20 m deep.This application promotes the adsorption of phosphate ions in the soil because of the Fe/Al oxyhydroxides present, leading to non-labile P formation (Leite et al., 2006).   (Yield lower than 18,500 kg ha -1 , in 74 fields. (2) According to Silva et al. (2004).DRIS: Diagnosis and Recommendation Integrated System; CND: Compositional Nutrient Diagnosis.
Localized P application provided higher accumulation of P in most of the development stages of wheat plants, compared with P broadcast and incorporated, certainly because of the lower contact of P with soil, which minimizes P non-labile formation (Barbieri et al., 2014).Similarly, for soybean crop, Arad rock phosphate and triple superphosphate were applied together in different proportions and in localized form provided positive linear correlation for yield; however, for the broadcast application, there was no increase in yield from 50 % of relative solubility (Oliveira Júnior et al., 2011).
From the results of both methods, it is evident that DRIS tended towards LF diagnostics (evident for Cu, Fe and Mn), while CND tended towards LE diagnostics (particularly for N, Ca, Mg and B).Such tendencies were not observed in the studies performed by Urano et al. (2006) and Serra et al. (2010b).The concordance between diagnostics observed for DRIS and CND methods varied among nutrients.The average concordance for all nutrients was 86.6 %, with a lower value for Mg (70.3 %), and a higher one for P (94.6 %).For macronutrients, the concordance were: N 81.1 %; K 90.5 %; Ca 86.5 %; and S 91.9 %; for micronutrients, B 82.4 %; Cu 86.5 %; Fe 87.8 %; Mn 90.5; and Zn 90.5 %.A similar result was observed by Serra et al. (2010a) for cotton crops, with an average of 87.4 %.For N, P, Ca, Mn, B and Cu diagnosis in rice crops, the use of only two fertilization response potential classes increased the concordance between the diagnostics to levels over 75 % on average.However, for Fe and Zn, even with the use of only two fertilization response potential classes, the concordance remained low (Tomio et al., 2015).
In bean plants, there was a good similarity in the diagnostics observed between CND and DRIS methods, with agreement of over 90 % for P, Ca, S, B, Cu, Fe and Zn.For the other nutrients, the agreement level was always over 79 % (Partelli et al., 2014).
The fit of linear models between DRIS indices or multinutrient variable indices and nutritional content was highly significant (p<0.01) for all nutrients (Table 5).The coefficients of determination (R 2 ) of the linear models varied between 0.61 and 0.92 for DRIS, and 0.52 and 0.93 for CND.For both methods, lower R² corresponded to Mg and higher to Mn.It was noted that even for those nutrients where the R 2 of the fitted models was low the linear fit was highly significant (p<0.01).Low R 2 were also observed by Kurihara et al. (2013).Similarly, all authors fitted highly significant models for their relationships, and were successful in determining the sufficiency ranges.
Comparing CND and DRIS methods to evaluate the nutritional status of banana from East Africa, Wairegi and Asten (2011) observed similar tendency among norms, with coefficients of determination varying from 0.96 to 0.99 for all nutritional indices.All sufficiency ranges created by the CND method provided narrower intervals than the DRIS method, despite being very close.For both, the ranges were different from those presented by Trani and Raij (1997), mainly as regards the adequate intervals (Table 5).The narrower ranges generated by the CND method can be explained by smaller concentration and dilution effects of nutrients in dry matter, caused by the multivariate functions of the calculation method (Wairegi and Asten, 2011;Wadt et al., 2013;Partelli et al., 2014).
The range similarity of the DRIS and CND methods and reduction in the amplitude of adequate contents in relation to official norms were also observed by Camacho et al. (2012), Kurihara et al. (2013), Santos et al. (2013a) and Partelli et al. (2014).Lower amplitude in the adequate ranges is highly desirable, because it increases distinction between balanced and unbalanced crops.For rice crops the adequate range delimited by the CND method, besides presenting lower amplitude, presented lower limit of the sufficiency range outside the confidence interval of the mean for Ca, Mg, S, Fe, Mn, Zn and Mo, while the upper limit was inside the confidence interval of the respective leaf content averages (Wadt et al., 2013).
Rev Bras Cienc Solo 2016; v40:e0140771 Results such as those cited above served as inspiration, whereby obtaining sufficiency ranges with small amplitude became the main objective of using DRIS and CND methods.However, it is questionable whether this is the best way to interpret nutritional content with these two methods.
The use of such methods to consider the sufficiency ranges removes the bivariate (DRIS) and multivariate (CND) character of nutrient content interpretation.The benefits such as nutrient ranking as regards the order of nutritional limitation, the formation of a medium nutritional balance index (IBN m ), the consideration of interactions between nutrients and minimization of dilution or accumulation effects in dry matter (Baldock and Schulte, 1996) are also lost.Literature ---30.0-100.0 (1) Yield higher than 18,500 kg ha -1 in 25 fields; (2) Trani and Raij (1997).DRIS: Diagnosis and Recommendation Integrated System; CND: Compositional Nutrient Diagnosis.
Rev Bras Cienc Solo 2016; v40:e0140771 Thus, this potential limitation motivated comparison of the diagnostics produced by DRIS and CND methods through nutritional balance and sufficiency range criteria.To this end, it was assumed that imbalances diagnosed in relation to excess or deficiency of nutrients implied changes in garlic yield.Therefore, the average yield of each class produced by DRIS and CND methods was studied through the evaluation criteria of nutritional status and sufficiency range.
Consequently, the greater distinction capacity of CND and DRIS methods using the nutritional status criterion, and not the sufficiency range, was evidenced.By nutritional status, the DRIS and CND distinguished in yield nine and ten classes, respectively, while the sufficiency ranges, this distinction has decreased to six and five classes, respectively (Table 6).This result may reflect the possible dilution and accumulation effects of nutrients that were not minimized by sufficiency ranges (Partelli et al., 2014).Possibly, the areas that produced higher or lower dry matter in the index leaf, were diagnosed mistakenly as deficient or excessive, damaging distinctions.First, all four groups of methods and criteria presented reductions in the average yield of the fields classified as LF for Cu (Table 6).This result, along with Cu as the element with a high number of deficient fields (Table 4), suggests the nutrient is the most limiting factor in terms of adequate nutrition of garlic in the fields diagnosed.
For N, the use of the nutritional status criterion, for both DRIS and CND methods, indicated reductions in yield caused by excess of the element (LE).None of the elaborated sufficiency ranges distinguished classes regarding the yield for the element (Table 6).Similar behavior was observed for Ca, where with the use of only the nutritional status criterion distinctions were observed among yield classes (Table 6).
However, the fields classified as LF had higher yield than those classified as NL.This result may be an indication of the negative effects of high lime doses employed for garlic cultivation in the Alto Paranaíba region.The liming criterion adopted in most of the evaluated fields aimed to increase base saturation to 80 %.
For Mg, only the CND method and the nutritional status criterion distinguished the yield between the different classes (Table 6).This diagnostic indicated similar behavior to that observed for Ca, whereby the LF class showed higher yield than the NL class.Again, the result may be an indication of the high lime doses employed in the region.
Both methods tended to show higher productivities in fields classified as LF for K (Table 6).
It is important to remember that both DRIS and CND diagnosed K as the element with the most fields classified as LF (Table 4).It is evident, therefore, that K is the nutrient applied more excessively in the garlic crop in the Alto Paranaíba region.High doses of this element can unbalance the crop nutrition and reduce yield.Finally, as occurred for Mg, the CND with the nutritional status criterion had higher distinction power, also demonstrating reductions in the average yield of the fields classified as LE (Table 6).
The P results were different from those previously found.The use of both methods using the nutritional status criterion did not distinguish yield in any of the classes (Table 6).However, the two sufficiency ranges demonstrated that contents below their lower limits (4.8 for DRIS and 5.2 for CND) promoted reductions in crop yield.Thus, it can be assumed that P foliar content lower than 5.2 g kg -1 implies limitations for the yield of garlic crops.A distinction was found among classes of nutritional limitation for Fe only through sufficiency ranges use as criteria of nutritional evaluation by DRIS and CND (Table 6).Reductions in average yield of the LF classes were observed by both DRIS and CND methods.
All methods and criteria identified increases in yield of the LE class in relation to NL class for S (Table 6).This result indicates the need to increase S supply to crops in the region.However, the DRIS method and the nutritional status criterion also distinguished the LF class as more efficient than NL.In addition, it was observed that even when no statistical differences were identified by any methods and criteria, there was a general tendency of higher productivities in the LF class in relation to the NL class.This behavior indicated a contradiction in diagnostic methods, once for both crops with high or low accumulation of S expressed high yield.Thus, either the group of methods and criteria were not efficient at diagnosing plants as regards S, or the nutrient was not limiting for garlic yield.
The diagnostics obtained for Mn classes were similar to those observed for S, except for the nutritional status criterion when there was a distinction in yield for the different classes (Table 6).In such cases, both LF and LE groups showed productivities higher than the NL class.In the same way of S, the Mn diagnostic does not allow us to form conclusions about the nutritional status regarding the element or the efficiency of the methods.
The B diagnostic differed among all of the methods and criteria.First, the DRIS method and nutritional status indicated a reduction in yield of the LE class.Moreover, the DRIS application through the sufficiency ranges showed that contents lower than 33.8 mg kg -1 reduce garlic yield.It was observed that the LF class had higher yield than the NL class with the CND method when the nutritional status criterion was used.Finally, the CND interpreted by sufficiency ranges did not distinguish any of the classes.Thus, specific research is needed to study the results obtained by these methods and criteria, in order to indicate which one produces the most adequate diagnostic (Table 6).Prod LF (kg ha -1 ) 17,192 16,202 21,066* 18,335* 18,552* 17,309 18,212* 15,385* 16,521 17,286* 16,974 Prod NL (kg ha -1 ) 17, 249 16,649 16,919 16,164 16,755 16,061 16,781 17,201 16,553 15,854 16,532 Prod LE (kg ha -1 ) 15,953* 17,488 16,017* 16,853 16,172 18,366* 15,957 18,276 17,773  Zinc was the one element in which all of the four methods/criteria distinguished yield as regards the nutritional status of classes (Table 6).Thus, the nutrient is not limiting of garlic yield in the Alto Paranaíba region.
Finally, the tests used to compare the methods and criteria of garlic foliar diagnosis do not present definitive results about the adequacy of each method.This was not necessarily because nutritional imbalances imply yield reductions.However, even with these methodological restrictions, some tendencies such as the greater distinction capacity of the nutritional status criterion for CND methods were identified.
Thus, it is suggested that long-term research should be carried out specifically to compare the adequacy and effectiveness of these methods and criteria in order to provide more accurate information as regards the best way to interpret leaf diagnoses.

CONCLUSIONS
Cu is the most limiting nutrient by lack, and K is the most limiting by excess in the diagnosed fields.
The DRIS method has greater tendency for limiting diagnostic by lack, while the CND method has a greater tendency for limiting diagnostic by excess.
CND method generated more accurate diagnoses with narrower sufficient ranges than those produced by the DRIS method.
The CND method and the nutritional status criterion presented greater capacity to distinguish classes diagnosed as regards yield.

Table 1 .
Nutritional status and fertilization response potential based on DRIS index (I A ), Multinutrient Variable Index (Ivi) and Medium Nutritional Balance Index (IEN m )Criterion of interpretation for the I A and IviNutritional statusIAor Ivi < 0 and |IA or Ivi| > IEN m LF |IA or Ivi| ≤ IEN m NL IA or Ivi > 0 and |IA or Ivi| > IEN m LE LF: limiting by lack; NL: non-limiting; LE: limiting by excess, according to Silva et al. (2004).Rev Bras Cienc Solo 2016; v40:e0140771

Table 2 .
Average and standard deviation (s) of the dual relationships (ln-transformed) between nutrient contents of garlic index leaf for the high yield subpopulation(1) ContinueRev Bras Cienc Solo 2016; v40:e0140771

Table 3 .
Average and standard deviation (s) of the multinutrient variables (Vn) and geometric average of the dry matter constituents (G) in garlic index leaf for the high yield subpopulation(1)by CND method

Table 4 .
Field frequency in each class according to nutritional status in the low yield subpopulation(1), obtained by DRIS and CND methods

Table 5 .
Statistical model and sufficiency range obtained by DRIS and CND methods through the correlation between nutrient content and DRIS and CND indices for the high yield subpopulation(1)