Coffee-Tree Floral Analysis as a Mean of Nutritional Diagnosis

JOURNAL OF PLANT NUTRITION Vol. 26, No. 7, pp. 1467 1482, 2003 Coffee-Tree Floral Analysis as a Mean of Nutritional Diagnosis Herminia E. P. Martinez, 1, * Ronessa B. Souza, 2 Javier Abadía Bayona, 3 Víctor Hugo Alvarez Venegas, 1 and Manuel Sanz 3 1 Fitotecnia and Solos Departments, Universidade Federal de Viçosa, Viçosa, MG, Brazil 2 Centro Nacional de Pesquisa de Hortaliças, EMBRAPA, Brasilia, DF, Brazil 3 Nutrición Vegetal Department, Estación Experimental de Aula Dei, CSIC, Zaragoza, Spain ABSTRACT Plant part analysis for evaluating the nutritional state of the crops is a practice commonly used. The analysis of flowers can allow an earlier diagnosis of nutritional deficiencies, excesses or unbalances, which facilitates its correction before the occurrence of irreversible losses in productivity and quality. The objective of this study were to determine the coffee tree (Coffea arabica L.) flower nutrients sufficiency ranges, to compare and correlate concentrations of nutrients observed in flowers and *Correspondence: Herminia E. P. Martinez, Fitotecnia Department, Universidade Federal de Viçosa, Av. P.H. Rolfs s=n Viçosa 36571-000, MG, Brazil; E-mail: herminia@ufv.br. 1467 DOI: 10.1081=PLN-120021055 Copyright # 2003 by Marcel Dekker, Inc. 0190-4167 (Print); 1532-4087 (Online) www.dekker.com

1468 Martinez et al. leaves collected 90 days after bloom, and to correlate the concentrations of nutrients in flowers and leaves with fruit yield. Samples of 26 experimental plots were collected. The plots were set up in nine different orchards five to nine years old and with 3000 5000 plants=ha, in the region of Viçosa, Minas Gerais State, Brazil. Eleven experimental plots were selected with mean yield greater than 7.0 kg=plant of coffee berry for the calculation of the nutrients sufficiency ranges. The concentrations of nitrogen (N), potassium (K), boron (B), iron (Fe), and zinc (Zn) were similar in flowers and leaves, whereas those of phosphorus (P), calcium (Ca), magnesium (Mg), sulfur (S), copper (Cu), and manganese (Mn) differed among the parts. There was correlation among the contents of N, Mg, Fe, Mn, Zn, and Cu in flowers and in leaves. For flowers a model of six variables and for leaves a model of eight variables explained 80% of the variation in the mean yield of the coffee tree plants. It is concluded that, flowers permit earlier diagnosis and greater precision in the diagnosis of the nutritional state of the coffee tree. Key Words: Macronutrients; Micronutrients; Coffea arabica L. INTRODUCTION Coffee is one of the main agricultural products in the composition of Brazilian exports, responsible for about 10% of the country s foreign exchange. Brazil is the world s largest coffee producer, supplying 25% of the international market. Since 1983, Minas Gerais has had the largest production among Brazilian states, with a volume of 11.6 million 60 kg bags of green coffee in 1999=2000. [1] There are three major producer areas in the state: the South of Minas, the Zona-da-Mata and the Cerrado areas. The latter, with acid soils and soils of low natural fertility, is responsible for 40% of the coffee production of Minas Gerais State. Despite the fact that the most advanced coffee tree cultivation techniques are adopted in the Cerrado area, where the soils present low natural fertility, the data of nutrient accumulation for the coffee tree reveal that this is a demanding crop, whose export of nutrients to the fruits is very high. [2] Therefore, orchards of high productivity are dependent on the use of considerable amounts of corrective amendments and fertilizers. Low correlation among soil nutrient concentrations and productivity were found in a soil fertility evaluation of five coffee tree areas of Minas Gerais State, [3] indicating that the foliar diagnosis is essential to evaluate the nutritional status of the crop, to detect deficiencies or excesses and to implement adjustments in fertilization programs. There were observed problems due to nutritional unbalances, more frequently related to micronutrients

Coffee-Tree Floral Analysis 1469 in high (>30 60-kg bags=ha of green coffee in an average of two consecutive years), medium (15 30 60-kg bags=ha of green coffee in an average of two consecutive years), and low (<15 60-kg bags=ha of green coffee in an average of two consecutive years) productivity orchards of the main areas of Minas Gerais. [4] It must be noted, however, that the results of foliar analyses of micronutrients such as iron (Fe), zinc (Zn), and manganese (Mn), with the objective of nutritional diagnosis, are imprecise, since its total concentration frequently does not reflect the magnitude of the physiologically active fraction. For coffee tree nutritional status evaluation, it is recommended that the picking of leaf samples take place after flowering, preceding the phase of rapid fruit expansion, [5] which generally takes place between the months of December and January. An earlier evaluation by means of floral analysis would be of great interest because it would facilitate the adjustment of the fertilization program at the beginning of the growth season, before the occurrence of irreversible losses in productivity and quality. Moreover, since the flowers are organs of short duration, in which metabolic reactions are not as complex as in leaves, the latter would not present substantial differences between the total concentration of the nutrient and the physiologically active fraction. [6,7] The prognosis and correction of nutritional disturbances by means of flower analysis have been developed at the Experimental Station of Aula Dei, in Zaragoza, Spain, in order to obtain earlier diagnosis of Fe-related nutritional disturbances in peach and apple trees. [6 11] Reference values were established for the interpretation of flower analysis of peach trees and correlations among concentrations of nutrients found in flowers and leaves picked 60 and 120 days after flowering. [8] These authors concluded that it is possible to use floral analysis for nutritional diagnosis of that crop. According to Sanz et al. [9] there was good correlation between the concentration of Fe in peach tree flowers and chlorophyll content in leaves that developed later on and presented Fechlorosis. The earlier diagnosis permitted correction of the disturbance, increasing fruit size and enabling earlier harvest. It was observed that, in populations of apple trees affected by Fe-chlorosis, flower analysis allowed the relation of the concentration of Fe in the flowers to the chlorophyll content of the leaves 60 and 120 days after full bloom. [10] This, in turn, permitted earlier prognosis of the onset and intensity of Fe-chlorosis. Correlation between the content of Fe in apple flowers and bitter-pit incidence in fruits was obtained by Sanz and Machín. [7] With reference to the considerations cited above, the objectives of the present study are: to determine the nutrient sufficiency range in coffee tree flowers; to establish correlations among nutrient concentrations observed in flowers and leaves; to compare the concentrations obtained from floral analysis to those obtained from leaf analysis; and to correlate nutrient content in coffee tree flowers and leaves with fruit yield.

1470 Martinez et al. MATERIALS AND METHODS Sampling and Sample Preparation Samples were taken of flowers and leaves of 130 plants from 26 experimental plots (five plants per plot). The plots were established in nine coffee tree orchards of the region of Viçosa, MG, Brazil. The plots were selected in orchards five to nine years old with high, medium, and low productivities. Homogeneous plots of 0.5 1.0 ha were established in orchards with densities of 3000 5000 plants per hectare. The bloom occurred on 13=09=2000, and flowers were collected in a fourday period after that. Complete flowers were taken from the medium portion of productive branches, from the medium third of the cup and from all cardinal points of exposure. The sampling of leaves was made 90 95 days after the flowering, according to the procedure described by Martinez et al. [5] The flowers and leaves were washed with distilled water and oven dried with air circulation at 70 C until attaining constant weight. After drying, the samples were ground in a Wiley-type mill, with a 20-mesh sieve and submitted to chemical analysis. Chemical Analysis The ground-milled material underwent wet acid digestion with nitric and perchloric acids. In the extract obtained, phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), sulfur (S), copper (Cu), iron (Fe), manganese (Mn), and zinc (Zn) were determined. Phosphorus was determined colorimetrically after reduction of the phospho molibidic complex for the C vitamin; [12] S was determined by the turbidimetric method described by Blanchar et al., [13] K was determined by flame emission photometry; and Ca, Mg, Cu, Fe, Mn, and Zn were determined by atomic absorption spectrophotometry. Boron (B) was determined after dry calcination by the azometine- H method. [14] Ammonium nitrogen was determined colorimetrically after wet sulfuric digestion [15] and nitrate nitrogen after extraction with hot water. [16] Total N was obtained by the sum of ammonium-n and nitrate-n. Harvest The plots were harvested during the months of April and May of 2001. Of the 26 experimental plots previously established, 19 were picked (95 plants), since seven of them had, inadvertently, been picked previously by the

Coffee-Tree Floral Analysis 1471 producers. The obtained yield (kg=plant of coffee berry) was corrected according to the stand, using a variable correction factor of 1.0 (3000 plants per hectare) 1.15 (5000 plants per hectare), since in denser plantations the yield per plant is smaller. In addition to yield evaluation, information was obtained about treatments and fertilizations made during this period. Statistical Procedure Correlation coefficients were calculated between concentrations of nutrients obtained in leaves and in flowers for the 26 plots sampled (mean of five plants=plot). When the correlations obtained were significant, regressions were determined that express the relationship. In addition, correlation coefficients between nutrient concentrations were calculated, both in flowers and in leaves. The data were grouped in frequency distribution, presented in the form of a histogram. The experimental plots with mean yields higher than 7.0 kg=plant of coffee berry (11 plots, 55 plants) were selected for the calculation of the reference range (RR) for the interpretation of results of coffee tree floral analyses. The concentrations and RR were calculated from the concentrations of nutrients in leaves and flowers of those plants, taking the average X, plus or minus the standard deviation of the average [S( X )], multiplied by a constant (k):, in that RR ¼ X S( X ) k, where k ¼ 1 for the nutrients whose variation coefficient VC < 20%; k ¼ 0.8 for the nutrients in that 20 VC < 40% and k ¼ 0.6 for the nutrients whose VC 40%. The reference concentrations (RC) in flowers and leaves were compared by means of variance analysis and F test. Correlation coefficients were calculated between the content of macroand micro-nutrients and the relationships N=P, N=K, K þ Ca þ Mg, P=Mn, S=Cu, and S=Zn with fruit yield (19 plots, 95 plants), as well as multiple regressions relating mean plot yield and nutrient contents in flowers and leaves. RESULTS AND DISCUSSION Although some authors [7,8] affirm that in pear, peach, and apple trees the results of flower analyses of individualized plants present better correlation with productivity and other intrinsic plant characteristics, in the present study higher correlations were obtained, and with greater degree of significance, when the mean results of each experimental plot, composed by five plants, were considered. Such behavior is justified by the fact that Coffea arabica is

1472 Martinez et al. spread by seeds and presents 5 15% of crossed fertilization, which provides larger variability among plants than that observed for fruit-cloned trees. Flower Analyses Reference Ranges Considering that a productive plant necessarily has good nutrient balance, of the 26 sampled plots, 11 with mean productivity higher than 7.0 kg=plant of coffee berry were selected to obtain concentrations and RRs for coffee tree nutritional diagnosis by means of floral analysis. That yield per plant would correspond, in uniform conditions and a stand of 3000 plants=ha, to a productivity of approximately 76 60-kg-bags=ha of green coffee. Concentrations and ranges of reference were also calculated for the nutrient contents of leaves sampled 90 days after bloom, to assure that the plants chosen as references possessed foliar concentrations of nutrients considered adequate for this culture. The results obtained are presented in Table 1. Table 1. Reference concentrations, RRs, and variation coefficients of the concentrations of macro and micronutrients in coffee tree flowers and leaves. Viçosa, MG, Brazil. (Mean data of 11 plots, 5 plants=plot). Reference concentration (g=kg) Reference range (g=kg) Variation coefficient (%) Nutrient Flowers Leaves Flowers Leaves Flowers Leaves N 24.4a 22.9a 22.9 25.9 20.3 25.5 5.96 11.54 P 2.5a 1.5b 2.4 2.6 1.3 1.7 4.61 13.43 K 22.1a 21.5a 17.9 26.3 18.2 24.8 23.9 15.42 Ca 1.7b 8.9a 1.2 2.2 8.1 9.7 49.17 8.77 Mg 1.8b 3.4a 1.6 2.0 2.8 4.0 11.01 16.75 S 1.9b 2.2a 1.7 2.1 2.0 2.4 11.04 11.09 mg=kg B 20a 22a 17 23 16 28 16.56 46.46 Cu 11b 20a 8 14 14 26 42.25 50.21 Fe 74a 72a 59 89 66 78 19.75 8.37 Mn 72b 218a 44 100 119 317 64.08 75.65 Zn 9a 10a 7 10 8 12 20.43 22.62 Note: For each line, the mean values of a same lower case do not differ from each other at the level of 5% of probability by mean of F test.

Coffee-Tree Floral Analysis 1473 It was observed that the concentrations of N, K, B, Fe, and Zn in flowers and leaves were not significantly different, while those of P, Ca, Mg, S, Cu, and Mn differed among these organs. The concentration of P was larger in the flowers, while Mg, S, and Cu were larger in the leaves. The concentrations of Ca and Mn in leaves were five and three times higher, respectively, than in flowers (Table 1). Similar concentrations of N, K, B, Fe, and Zn in leaves and flowers indicate that those organs have similar demand for these nutrients. Phosphorous concentrations in flowers were higher because P is a phloem mobile element, which accumulates in reproductive organs, where it carries out important functions in the polinic tube growth, pollen grain maturation and in the intense metabolism of the first phase of fruit formation. In peach tree flowers there were found P concentrations 1.5 1.9 times higher than those observed in leaves collected 60 and 120 days after bloom. [8] Higher Mg concentration in leaves is probably due to its role as component of the chlorophyll molecule, and that of Ca to the fact that this is an element whose transport depends on the transpiratory flow, that is higher in the leaves. The results of peach tree leave and flower analysis show the same pattern of behavior. [8] With regard to higher Mn concentration in leaves, it is necessary to consider that the soils of Center South of Brazil are generally acidic, and that the concentration observed in leaves reflects the availability of the element in the soil, and did not mean a high plant physiologic demand. The much higher Mn concentration in leaves than in flowers leads us to believe that the element is retained in mature leaves, in detriment of its transport to the areas of active growth. In calcareous soils, such sharp differences in Mn concentrations in flowers and leaves were not found. [8] In flowers as well as in leaves, the coefficients of variation obtained for macronutrients were low, in the range of 5 15%, except for VCs observed for the flower K and Ca contents, which were 23.90 and 49.17%, respectively. For Ca, high variability could result from the low concentrations observed, of the order of 1.7 g=kg. The variation coefficients observed for the micronutrients B, Cu, and Mn concentrations leaves were quite high (46.46, 50.21, and 75.65%), while in flowers only Cu and Mn presented sharp variability (42.25 and 64.08%). With the exception of Fe, variability in micronutrient concentrations was always smaller in the flowers. These results are in agreement with those cited in the literature by Sanz and Machin [7] and Sanz and Montañés, [8] who encountered larger variability in nutrient content in leaves than in flowers in peach and apple trees. These authors also found larger variability in the content of micronutrients, above all for Cu in apple tree flowers and Mn and Zn in peach tree leaves. For the macronutrients they observed larger variability for Ca, both in flowers and in leaves.

1474 Martinez et al. Except for concentrations of N and Ca, which were slightly lower than coffee tree RRs obtained for the Viçosa area, [4] nutrient content of leaves of the 11 plot plants taken as pattern are appropriate. It is worth pointing out that, for foliar concentrations of micronutrients, there was a notable coincidence between the RR obtained in the present study and that obtained by the authors cited above, which indicates that the pattern plants were appropriately selected. Correlations Between Flower and Leaf Concentrations of Macro- and Micronutrients Among the macronutrients, the N and Mg concentrations in flowers and leaves presented highly significant correlation coefficients of 0.62 and 0.71. Such correlation coefficients can be considered high for field conditions. [17] For P, K, Ca, and S there was no correlation among nutrient concentrations observed in both parts (Fig. 1). The frequency distribution of P, K, Ca, and S concentrations were quite different in flowers and leaves, while for N and Mg such distribution was similar (Fig. 2). Concentrations of Fe and Mn in coffee-tree flowers, and leaves presented highly significant correlation coefficients (0.87 and 0.85), while for Cu and Zn the correlation coefficients were smaller (0.62 and 0.45) and significant only at 5% of probability. For the Cu the relationship was quadratic, while for the other elements it was linear. For the Cu it was found that in small concentrations there are linear relationship between flower and leaf contents. Leaf concentration values higher than the maximum of the calculated reference range did not show the same relationship (Table 1, Fig. 1), indicating that, in plants with Cu concentrations above an optimum limit, there are restrictions in its transport to the flowers. Concentrations of B in flowers and leaves collected 90 days after blooming did not present significant correlation, as can be seen in the frequency distribution histograms (Fig. 3). Results obtained from 100 peach tree plants have shown correlations among the concentrations of N, P, K, Mn, and Ca in flowers and leaves picked 60 days after bloom. At 120 days after bloom, only concentrations of N, Mg, and Mn presented significant correlation coefficients. [8] Similar results for 50 peach tree plants showed significant correlations between K and Mn concentrations in flowers and leaves collected 120 days after bloom in 1994. The following year they observed significant correlations between these organ concentrations of P, Fe, and Mn. In the two years cited, the correlation coefficients were, in general, lower than those determined in the present study. In flowers, there were positive and highly significant correlations ( p ¼ 0.01) between the following nutrients: N and P, N and S, K and Ca, K

Coffee-Tree Floral Analysis 1475 Figure 1. Relationship between macro and micronutrient concentrations in coffee tree flowers and leaves collected 90 days after bloom (average of 26 plots, five plants=plot). and Mn, Ca and Fe, Mg and S, S and Cu, S and Zn, and Zn and B. Correlations highly significant and negative occurred between K and Zn and between Zn and Mn. Significant positive correlations ( p ¼ 0.05) were found between N and Mg, P and K, P and S, P and Cu, P and Mn, Ca and Cu, Ca and Mn, and Mg and Cu. Significant negative correlations occurred between K

1476 Martinez et al. Figure 2. Nitrogen, P, K, Ca, Mg, and S concentration (g=kg) frequencies distribution histograms observed for coffee tree leaves and flowers (average of 26 plots, five plants=plot). and B and Mn and B. For leaves there were positive and highly significant correlations ( p ¼ 0.01) between Ca and Mg, Mg and S, S and B, Cu and B, Zn and B, Cu and Zn. Highly significant and negative correlations occurred between K and Ca, K and Mg, and Mg and Mn. Positive and significant correlations ( p ¼ 0.05) were verified between P and K, Ca and S, Ca and Zn, Mg and Zn, S and Cu, and S and Zn. Significant and negative correlation occurred between K and S (Table 2). Many of those correlations express well-known interactions between cationic and anionic nutrients, such as the relationships between N and P, N

Coffee-Tree Floral Analysis 1477 Figure 3. Iron, Cu, Zn, Mn, and B concentration (mg=kg) frequencies distribution histograms observed for coffee tree leaves and flowers (average of 26 plots, five plants=plot). and S, P and K, K and Ca, K and Mg, K and Mn, Ca and Mn, P and Mn. It is worth noting the positive correlations between S and Cu, S and Zn, and B and Zn and the negative correlations between K and Zn and Mn and Zn. The relationships between S and Cu and S and Zn were not found in the literature as important relationships for the coffee tree, but in the present study they appear in leaves as well as in flowers. The negative correlations between K and Zn and Zn and Mn are of interest, since the lack of Zn is one of the common nutritional problems of coffee trees in the Viçosa area. [4] The high K

1478 Martinez et al. Table 2. Correlation coefficients between macro- and micronutrient concentrations in coffee tree flowers and leaves collected 90 days after flowering. Viçosa, MG, Brazil. (Averages of 26 plots, 5 plants=plot). Nutrients N P K Ca Mg S Fe Cu Zn Mn Flowers P 0.61** K 0.34 0.39* Ca 0.08 0.18 0.60** Mg 0.49* 0.03 0.13 0.21 S 0.52** 0.49* 0.23 0.05 0.51** Fe 0.29 0.33 0.06 0.52** 0.00 0.37 Cu 0.38 0.39* 0.31 0.44* 0.43* 0.50** 0.04 Zn 0.05 0.07 0.79** 0.38 0.23 0.61** 0.25 0.03 Mn 0.17 0.41* 0.82** 0.48* 0.19 0.05 0.23 0.34 0.60** B 0.05 0.17 0.39* 0.03 0.11 0.15 0.23 0.17 0.42** 0.49* Leaves P 0.37 K 0.05 0.40* Ca 0.16 0.03 0.52** Mg 0.16 0.24 0.66** 0.83** S 0.36 0.09 0.39* 0.48* 0.55** Fe 0.05 0.09 0.42 0.26 0.12 0.12 Cu 0.25 0.22 0.15 0.14 0.18 0.49* 0.12 Zn 0.24 0.25 0.08 0.41* 0.42* 0.49* 0.16 0.84** Mn 0.18 0.13 0.30 0.30 0.68** 0.24 0.05 0.02 0.23 B 0.11 0.19 0.08 0.12 0.21 0.59** 0.19 0.87** 0.70** 0.12 Note: *, **, Significant at levels of 5 and 1% of probability.

Coffee-Tree Floral Analysis 1479 demand of the coffee tree is well known, and fertilization doses that reach 450 kg=ha year of K 2 O are recommended. [18] It is also known that with conditions of moderate soil acidity, that are well tolerated by the coffee tree, it is common to find substantial Mn availability in soil solution, which can worsen and=or hinder correction of Zn deficiency. The flowers presented a larger number of significant correlations between nutrients than did leaves, which suggests that in coffee trees this organ is more sensitive than leaves in expressing nutrient interactions. Correlations Between Yield and Concentrations of Macro- and Micronutrients in Coffee Tree Flowers and Leaves The yield of coffee berry did not present good correlation with concentrations of macro and micronutrients in leaves, nor with N=P, N=K, P=Mn, K þ Ca þ Mg ratios. There was, however, negative correlation between yield and the concentrations of P, Ca, K, Cu, S=Zn, K þ Ca þ Mg, and positive correlation with N=P, N=K, P=Mn, and S=Cu ratios in flowers (Table 3). Although dry matter weight of the flowers was not evaluated, it was observed that larger flowers were collected in the most productive orchards. It is possible that a relationship exists between flower size and the development of fruits, and that those larger flowers presented smaller concentration of some nutrients due to dilution effect. That would explain the observed negative correlations. In Table 3. Correlation coefficients between flower and leaf nutrient concentrations, relationships among nutrients and mean coffee plant yield. Viçosa, MG, Brazil. (Averages of 19 plots, 5 plants=plot). Nutrients Leaves Flowers Relationships Leaves Flowers N 0.141 0.330 N=P 0.138 0.559* P 0.374 0.647** N=K 0.211 0.494* K 0.337 0.513* K þ Ca þ Mg 0.232 0.546* Ca 0.164 0.582* P=Fe 0.144 0.057 Mg 0.412 0.047 P=Mn 0.437 0.493* S 0.251 0.239 S=Cu 0.570** 0.522** B 0.193 0.006 S=Zn 0.499** 0.517* Cu 0.245 0.486* Fe 0.201 0.294 Mn 0.301 0.407 Zn 0.338 0.334 Note: *, **, Significant at levels of 5 and 1% of probability.

1480 Martinez et al. future studies it will be necessary to consider not only the concentration, but also the total content of nutrients in the flowers. To obtain one gram of dry matter, approximately 80 coffee tree flowers are needed. In general, it was observed that flower analysis was better correlated with yield than was leaf analysis. Although the concentrations of nutrients taken individually did not present good correlations with the yield, highly significant models of multiple regression with high determination coefficients were adjusted to nutrient concentration data and coffee yield, both in leaves and in flowers. For the latter, a model with six variables explains 80% of the variation in the mean yield of the plants, and for leaves a model with eight variables explains 80% of that variation (Table 4). The sign and the magnitude of the obtained coefficients (Table 4) reflect the nutritional conditions of the sampled plants. It is worth noting the negative coefficients obtained for P and Fe concentrations in the adjusted model with concentrations of leaves, as well as in the fitting with concentrations of flowers. A large part of the sampled plants presented concentrations of P and Fe higher than those of the RR (Figs. 2 and 3). The same result was found with K concentration in flowers (Figs. 2 and 3). In flowers, a model of two variables, with P and Fe, explains 69% of the yield variation. In coffee tree orchards of Minas Gerais, there were observed accumulation of P in 5 cm soil surface layer, which may be due to the continuous use of the formula 20-5-20 in fertilization, and that, as shown for the data presented here, may be resulting in plant excess of P and reduction of productivity. [19] In summary, the analysis of coffee tree flowers seems to be a sensitive and reliable technique for earlier nutritional diagnosis of coffee trees. This does not mean that floral diagnosis can substitute foliar diagnosis, because in soils with low CEC, as is the case in most Brazilian coffee tree areas, the yield seems to Table 4. Regression equations for plant production ( ^Y, kg) as a function of concentrations of macro- and micronutrients in coffee tree flowers and leaves. Regression equations R 2 Flowers ^Y ¼ 23.52 þ 0.66** N 5.37* P 0.35** K þ 2.00* Ca 6.60*S 0.80 0.02*** Fe ^Y ¼ 23.95 6.10*** P 0.01*** Fe 0.69 Leaves ^Y ¼ 6.77 12.08*** P þ 2.68*** Ca 4.05** Mg þ 7.61**S 0.04* Fe þ 0.23** Zn 0.01* Mn 0.20*** B 0.80 Note: *, **, ***, Significant at levels of 10, 5, and 1% of probability.

Coffee-Tree Floral Analysis 1481 depend predominantly on fertilization implemented during the growth season, which starts with the flowering at the beginning of the rainy season. It is believed that both diagnostic processes are important, and that under conditions of pronounced deficiencies or excesses, flower analysis permits corrections at the beginning of the growth season of the culture, thereby minimizing losses both in productivity and in quality. CONCLUSIONS In addition to allowing earlier diagnosis, flowers permit greater precision than leave in coffee tree nutritional diagnosis. The RRs for the interpretation of floral analysis are: 2.4 2.6 g=kg of P; 17.9 26.3 g=kg of K; 1.6 2.0 g=kg of Mg; 13 23 mg=kg of B; 8 14 mg=kg of Cu; 59 89 mg=kg of Fe; 44 100 mg=kg of Mn, and 7 10 mg=kg of Zn. The ranges of 22.9 25.9 g=kg of N, 1.2 2.2 g=kg of Ca, and 1.7 2.1 g=kg of S can be used as a first approximation, requiring, however, subsequent adjustments. REFERENCES 1. Saes, M.S.M.; Farina, E.M.M.Q. O Agribusiness do Café no Brasil; Pensa=Editora Milkbizz Ltda.: Sao Paulo, 1999; 230 pp. 2. Malavolta, E. Nutrição Adubação e Calagem do Cafeeiro; Fertilizantes Copas, São Paulo, w=d; 43 pp. 3. Souza, R.B.; Martinez, H.E.P.; Alvarez V., V.H.; Oliveira, J.A.; Guimarães, P.T.G.; Oliveira, M.H. Produtividade do cafeeiro em função de características químicas de solos de diferentes regiões de cultivo em Minas Gerais. In Resumos Expandidos. I. Simpósio de Pesquisa dos Cafes do Brazil; Poços de Caldas, 26 29 Setembro; Embrapa Café e Minasplan: Brasília, 2000; 1299 1303. 4. Martinez, H.E.P.; Souza, R.B.; Alvarez V., V.H.; Menezes, J.F.S.; Guimarães, P.T.G.; Oliveira, M.H. Faixas Críticas de Macro e Micronutrientes Para o Cafeeiro em Diferentes Regiões do Estado de Minas Gerais. In I Simpósio de Pesquisa dos Cafés do Brasil, I., Poços de Caldas, MG., Resumos Expandidos. v.2. Embrapa Café e Minasplan: Brasília, 2000; 1308 1310. 5. Martinez, H.E.P.; Carvalho, J.G.; Souza, R.B. Diagnose foliar. In Recomendações Para o Uuso de Corretivos e Fertilizantes em Minas Gerais,5 a Aproximação; Ribeiro, A.C., Guimaraes, P.T.G., Alvarez, V.V.H., Eds.; CFSEMF: Viçosa, Brazil, 1999; 143 168. 6. Montañés, M.L.; Val, J.; Bétran, J.; Monge, E.; Moreno, M.A.; Montañéz, L. Floral analysis: fresh and dry weight of flowers from different fruit tree species. Acta Hortic. 1997, 448, 233 239.

1482 Martinez et al. 7. Sanz, M.; Machín, J. Aplicación del análisis floral al prognóstico y diagnóstico del bitter-pit. Información Técnica Económica Agraria. 1999, 95 (2), 118 124. 8. Sanz, M.; Montañés, L. Flowers analysis as a new approach to diagnosing the nutritional status of peach tree. J. Plant Nutr. 1995, 18 (8), 1667 1675. 9. Sanz, M.; Pascual, J.; Machín, J. Prognosis and correction of iron chlorosis in peach tree: influence on fruit quality. J. Plant Nutr. 1997, 20 (11), 1567 1572. 10. Sanz, M.; Perez, J.; Pascual, J.; Machín, J. Prognosis of iron chlorosis in apple trees by floral analysis. J. Plant Nutr. 1998, 21 (8), 1697 1703. 11. Igartua, E.; Grasa, R.; Sanz, M.; Abadía, A.; Abadía, J. Prognosis of iron chlorosis from the mineral composition of flowers in peach. J. Hort. Sci. Biotechnol. 2000, 75 (1), 111 118. 12. Braga, J.M.; Defelipo, B. Determinação espectrofotométrica de fósforo em extratos de solos e plantas. Revista Ceres 1974, 21 (113), 73 85. 13. Blanchar, R.W.; Rehm, G.; Caldwell, A.C. Sulfur in plant material by digestion with nitric and perchloric acid. Proc. Soil Sci. Soc. Am. 1965, 29 (1), 71 72. 14. Malavolta, E.; Vitti, G.C.; Oliveira, S.A. Avaliação do Estado Nutricional das Plantas: Princípios e Aplicações; ABPPF: Sao Paulo, 1997; 173 pp. 15. Jackson, M.L. Nitrogen determinations for soil and plant tissues. In Soil Chemical Analysis; Jackson, M.L., Ed.; Prentice-Hall: New York, 1985; 183 204. 16. Cataldo, D.A.; Harón, M.; Scharder, M. Rapid colorimetric determination of nitrate in plant tissue by nitritation of salicylic acid. Commun. Soil Sci. Plant Anal. 1985, 6 (1), 71 81. 17. Belkhodja, R.; Morales, F.; Sanz, M.; Abadía, A.; Abadía, J. Iron deficiency in peach trees: effects on leaf chlorophyll and nutrient concentrations in flowers and leaves. Plant Soil 1998, 203, 257 268. 18. Guimarães, P.T.G.; Garcia, A.W.R.; Alvarez V., V.H.; Prezotti, L.C.; Viana, A.S.; Miguel, A.E.; Malavolta, E.; Corrêa, J.B.; Lopes, A.S.; Nogueira, F.D.; Monteiro, A.V.C. Cafeeiro. In Recomendações Para o Uso de Corretivos e Fertilizantes em Minas Gerais, 5 a Aproximação; Ribeiro, A.C., Guimaraes, P.T.G., Alvarez, V.V.H., Eds.; CFSEMF: Viçosa, Brazil, 1999; 289 302. 19. Martinez, H.E.P.; Souza, R.B.; Guimarães, P.T.G.; Alvarez V., V.H.; Oliveira, J.A.; Silva, J.A. Avaliação da fertilidade do solo de cafezais do Estado de Minas Gerais. In I Simpósio de Pesquisa dos Cafés do Brasil, I. Poços de Caldas, MG. Resumos Expandidos; Embrapa Café e Minasplan: Brasilia, Brazil, 2000; Vol. 2, 1295 1298.