FIELD EVALUATION OF WATER OR CITRATE SOLUBLE PHOSPHORUS IN MODIFIED PHOSPHATE ROCKS FOR SOYBEAN

Ten P fertilizers were collected (commercial fertilizers) or synthesized (experimental sources) in order to obtain single superphosphates varying in water and citrate solubility. A standard source of P was also produced by crystallization of the water-soluble fraction of a triple superphosphate. Eleven P sources were band applied to a medium textured Xanthic Hapludox, in Bahia, Brazil (low content of resin-extractable P) at a rate of 80 kg ha of NAC + H 2 O (neutral ammonium citrate plus water) soluble P 2 O 5 , with soybean as the crop which was grown to maturity. A check plot (control) was included in the study. Three of the P sources [single superphosphate produced from Araxa phosphate rock (PR), low-grade single superphosphate produced from Lagamar PR and the standard source of P] were also applied at rates to provide 40 and 120 kg ha of NAC + H 2 O soluble P 2 O 5 . Yield of soybean was evaluated by analysis of variance with mean comparison performed utilizing LSD lines, considering the P sources applied at a rate of 80 kg ha of P 2 O 5 + control. Regression procedures were used to study the relation between yield of soybean and rates of P 2 O 5 . The fertilizers tested performed equally well as a source of P for soybean. The level of water-soluble P did not influence fertilizer performance.


INTRODUCTION
When phosphate rocks (PR) are acidulated to form more soluble P fertilizers, P compounds are formed other than the desired NH 4 and Ca phosphates.Exhaustion of high-grade phosphate rock throughout the world will certainly increase the levels of impurity compounds, generally water-insoluble, in the final acidulated P fertilizers (Lehr, 1984).In Brazil, acidulated P fertilizers have been produced from low-grade PR due to high amounts of impurities, especially iron-aluminum oxides, present in the ore.
The presence of Fe-Al impurities in the final acidulated P fertilizer has raised the concern that the insoluble P compounds would decrease the agronomic effectiveness of the fertilizers due to a decrease in watersolubility of such fertilizers (Sikora & Giordano, 1995).Based on this concept the European Economic Community set a level of 93% of the ammonium citrate-soluble P as water-soluble P in fully acidulated P fertilizers market in the European Community (Council, 1976) but Johnston (1999), in a literature review, commented that there is no scientific basis for requiring such a high level of watersoluble P.
In the USA some studies have shown that the levels of impurity compounds currently in triple superphosphates and mono-ammonium phosphates produced in that country did not have a detrimental effect Scientia Agricola, v.58, n.1, p.165-170, jan./mar.2001 on P availability to potatoes (Mullins & Evans, 1990) and pearl millet (Mullins & Sikora, 1990) in field trials and to maize (Mullins, 1988) and sorghum sudangrass (Sikora et al., 1989) in greenhouse studies.Mullins & Evans (1990) evaluated four commercial triple superphosphates sources containing various levels of water-soluble P (81 to 94% of AOAC available P as water-soluble P) and concluded that yield of potatoes was not affected by the source of P and also that the fertilizer performance was not influenced by the level of water-soluble P. Similar results were obtained by Mullins & Sikora (1990) in ten sources of monoammonium phosphates containing from 81 to 100% water-soluble P (in the AOAC available P) to pearmillet.Prochnow et al. (1998) compared the efficiency of four experimental P sources, produced from Brazilian apatite concentrate varying in their content of iron and aluminum oxides, in its original and water-insoluble form.The authors concluded that the P availability of the water-insoluble fraction was generally lower than the original P source and that higher amounts of iron and aluminum oxides lead to lower performance of the water-insoluble fraction.
In Brazil, single superphosphates (SSP) have to contain 18% and 16% (tolerance of 10% minus), respectively, of P 2 O 5 soluble in neutral ammomium citrate + water (NAC + H 2 O) and water in order to be commercialized (Brasil, 1982).These standards discard some PRs or apatite concentrates as potential sources to produce acidulated P fertilizers.Furthermore, some materials or industrial processes utilized in order to produce SSP with better physical properties, or containing micronutrients, interfere with the water solubility of the final product and these processes should also be evaluated concerning the performance of the modified products in terms of P availability to plants.
The objective of this study was to evaluate the effectiveness of various commercial and experimental acidulated P fertilizers, varying in water and citrate solubility, and produced following the procedure to obtain single superphosphates, in order to provide P to soybean plants under field condition.

Phosphorus sources
Ten P sources 1 were produced utilizing the process to obtain SSP (TABLES 1 and 2).It can be noticed that three of the products were in the powder form (FAR, FFM and FC), six granulated and screened to pellets particle size of 1.7 -3.35 mm (FL, DUR, MR, FI, FS and EK) and one granulated to pellets particle size of 0.5 -1.41 mm (FMG).In some of the P sources materials were added, such as MgO, Concinal (obtained from the algae Lhithothamium) and micronutrients, or processes modified, in order to evaluate the improvement in the physical properties of the fertilizers (FFM, FC, FMG, DUR and MR).A standard source of P was produced by crystallization of the water-soluble fraction of a triple superphosphate -FMC.

Field Evaluation of the P sources
Field evaluation was conducted in Barreiras, Bahia, Brazil, in a medium-textured, Xanthic Hapludox containing 59%, 5% and 36% of sand, silt and clay, respectively.Resinextractable P (Raij & Quaggio, 1983) was 9 mg dm -3 and Mehlich-1 P (Mehlich, 1953) was 5 mg dm -3 which both corresponds to a low soil test level.The amounts of Ca, Mg, K, H+Al and Al were respectively 2.3, 0.9, 0.1, 1.8 and 0.0 cmol c dm -3 .Cation-exchange capacity was 5.1 cmol c dm -3 and pH in 0.01 mol L -1 CaCl 2 (2.5 solution:soil ratio) was 5.2.Potassium was added as basal dressing at a rate of 150 kg ha -1 K as KCl.Calcium sulfate dihydrate (gypsum) and a product containing micronutrients (5% Zn, 3% B, 6% Cu, 5% Fe and 17.5% Mn) were also added at rates of 400 and 40 kg ha -1 , respectively.These fertilizers were broadcast and incorporated to a soil depth of 8 cm.The amounts of nutrients applied as basal application were adequate to discard any potential side-effect when comparing the P sources.
Phosphorus sources and the standard FMC were localized in the line (3 cm below and 2 cm besides the seeds) at a rate to supply 80 kg ha -1 P 2 O 5 of NAC + H 2 O soluble P 2 O 5 .The rate of P 2 O 5 was chosen based on the P status of the soil in order to provide sufficient phosphorus for high yield of soybean and was based on the literature (EMBRAPA, 1998).The P sources were applied based on the NAC + H 2 O available P 2 O 5 instead of total P 2 O 5 since phosphate fertilizers in Brazil are marketed on the basis of their NAC + H 2 O available P 2 O 5 content.By applying 80 kg ha -1 NAC + H 2 O soluble P 2 O 5 , total P 2 O 5 and watersoluble P 2 O 5 were applied, respectively, in the range of 81.4 -101.6 and 18.7 -72.32 kg ha -1 .
In order to test some of the products in a range of rates (response curve) the P sources FL, FS and the standard source of P were also applied in rates of 40 and 120 kg ha -1 P 2 O 5 .A check plot control (no P applied) was also included.The P sources and rates were arranged in a randomized complete-block design with three replications.Field plots consisted of four rows, 4 m long and spaced at 0.5 m.
Seeds of soybean (Glycine max, cultivar FT 103) were inoculated with Rhizobium japonicum and treated with 17 g of Mo and 2 g Co per 50 kg of seeds and sown in november 24, 1998, at the rate of 15 seeds per meter.
Soybean grain was manually harvested in April 08, 1999, from the 2 meters of the two central rows and the yield calculated.

1
The term P source was preferred since many of the fertilizers tested do not follow the standards to be classified as single superphosphates according to the current Brazilian legislation (Brasil, 1982),.

Data Analysis
Yield of soybean was analyzed considering: (a) analysis of variance for the control (no P applied) + P sources treatments applied at only one rate (80 kg ha -1 P 2 O 5 ) with means comparison performed utilizing LSD (least significant difference) lines; (b) analysis of variance considering only the treatments where P was applied (control was excluded); (c) regression procedures for the factorial design between three P sources (FMC, FS and FL) and four rates of P 2 O 5 (0, 40, 80 and 120 kg ha -1 ).All the data analysis was performed using SAS software (SAS, 1985).
For the relation between yield of soybean and rates of P 2 O 5 a dummy variable multiple regression analysis was performed.This resulted in a common intercept and a single value of MSE and R 2 for the three regression equations (one for each P source).Three models (linear, semi-log and square root) were tested to describe the relationship between the parameters studied, and the one presenting the higher R 2 chosen.
The relative agronomic effectiveness (RAE) was calculated for each P source.RAE was definied as the ratio of the two slopes: where βi is the slope of the response function of the P sources tested and βFMC is the slope of the response function of the standard -FMC.This expression ranks the P sources with respect to FMC according to their agronomic potential to produce a yield response (Chien et al., 1990).
In order to evaluate if there was statistical significant difference between the three phosphorus sources in the range of rates applied a F value (= t 2 ) was calculated according to the formula:   (1) FMC: standard source of P; FAR: SSP produced from Lagamar PR; FFM: addition of 1.1% MgO to FAR; FC: Addition of 30% concinal to FAR; FL: Granulated FAR; FMG: Microgranulated FAR; DUR: Over dried FAR; MR: friable FAR; FI: Addition of micronutrients to FAR; FS: SSP produced from Araxá PR; EK: SSP produced from Togo PR.where β ia is the slope of the response function for the first P source tested, β ib is the slope for the second P source tested, SE(β ia) is the standard error for β ia and SE(β ib ) is the standard error for β ib.

Characteristics of the P sources
The P sources, excluding the standard source, presented high variability for NAC+H 2 O soluble P 2 O 5 (13.3 -19.2%), water-soluble P 2 O 5 (3.1% -16.0%) and percentage of water-soluble P 2 O 5 in the NAC+H 2 O fraction (23.3 -90.4%;TABLE 2), showing that the processes and materials, including the PR, used in the production of the P sources interfere in the solubility of the fertilizers.Actually, based on current Brazilian legislation only two of the ten P sources (EK and FS) meet the requirement to be commercialized as single superphosphates (at least 14.4% of water-soluble P 2 O 5 ).FC was the P source with the lowest water-soluble P 2 O 5 due probably to the presence of dicalcium phosphate.The percentage of water-soluble P 2 O 5 in the NAC + H 2 O fraction in the standard source of P (86.3%) was low considering that in the solution leached the only P form expected was the monocalcium phosphate monohydrate.Two possibilities exist to explain this result: (i) water-insoluble forms of Fe-Al-P still remaining among the crystals of monocalcium phosphate monohydrate and (ii) the formation of dicalcium phosphate (monetite), which is not water-soluble, during the process of crystallization.The presence of 1.6% of Fe 2 O 3 and 1.7% of Al 2 O 3 in the product obtained after crystallization reinforces the statement (i) as the possible cause for the presence of forms of water-insoluble P compounds.

Field evaluation
Yield of soybean was significantly affected when considering the control + eleven P sources applied (80 kg ha -1 ) as the independent variables (p ≤ 0.05) but the comparison of means by LSD lines showed that the only Figure 1 -Yield of soybean as affected by the amount of total P 2 O 5 (A) and water-soluble P 2 O 5 (B), considering the eleven P sources.P source (1) Yield of soybean LSD lines 1 (2, 4) LSD lines 2 (3, 4)   kg ha (2) LSD lines 1: mean comparison in the column considering all P sources + check control.
(3) LSD lines 2: mean comparison in the column considering P sources (check control excluded). ( 4 Values followed by the same letter in the column are not statistically different (p ≤ 0.05).
significant difference was met between the control with all the P sources: control < all P sources (TABLE 3).When the control was excluded and just the eleven P sources considered in the analysis of variance still no effect of P source was detected.No significant relation was found between the amounts of total P 2 O 5 or watersoluble P 2 O 5 applied and yield of soybean (Figure 1).  (1FMC: standard source of P; FS: SSP produced from Araxá PR; FL: Granulated low-grade SSP produced from Lagamar PR.

A B
(2) fi= percentage of water-soluble P 2 O 5 in the NAC+H 2 O-soluble P 2 O 5 .n.d.:not determined; similar contents of Al 2 O 3 and Fe 2 O 3 should be expected as those presented by the FAR or FL.

TABLE 1 -
Phosphate rocks utilized, pellet particle size (PPS) and details on the production of the P sources.

TABLE 2 -
Chemical analysis of the P sources.
Figure 2 -Yield of soybean as affected by the rate of P 2 O 5 .Models for the three P sources tested were not statistically different (p ≤ 0.05).

TABLE 4 -
Regression estimates for the semi-logarithmic model adjusted describing the relation between yield of soybean as affected by source and rate of P 2 O 5 .

TABLE 5 -
Relative Agronomic Effectiveness (RAE) of each source relative to the standard FMC for yield of soybean.