The importance of proton supply in phosphate rock dissolution: Comparative study of three phosphate rocks from sub-Saharan Africa

The dissolution in hydrochloric acid of three phosphate rocks (PRs) from sub-Saharan Africa was studied in the laboratory and their agronomic effectiveness compared using soils from Tanzania, Ivory Coast, Kenya and Netherlands. The objective was to investigate the potential of the PRs in supplying phosphorus under conditions of limited and unlimited supply of protons (H+). The test crop was maize (Zea mais var. LG11). Results show that calcium carbonate content is an important factor determining differences between the PRs with respect to their effectiveness. They also show that (i) differences in the dissolution of the PRs manifested more at low than at high acid concentration, (ii) availability of enough protons is an important factor in PR dissolution, (iii) low rates of PRs should be more desirable than high rates and (iv) soil and plant factors have an important role in the effectiveness of a PR.


INTRODUCTION
Phosphorus (P) is one of the nutrients often limiting crop production in sub-Saharan Africa where continuous cropping without use of fertilizers is common among smallholder farmers.The possibilities to use industrial P fertilizers are small because of their high costs.However, phosphate rock (PR) deposits abound in several countries in the region (Mew, 2000;van Straaten, 2002) and should be exploited as a source of low cost P.The direct application of such locally available P source is seen by many workers as a viable alternative (Msolla et al., 2005;Arcand and Schneider, 2006;Biswas and Narayanasamy, 2006;van Straaten, 2006).Their effectiveness upon application to soil relies on the presence of protons (H + ) and sinks of phosphorus and calcium (Savini et al., 2006).PRs differ greatly in their physico-chemical properties and hence in their effectiveness as sources of P when applied directly to soils.Coupled with this, different plants and soil conditions have varying influence on PRs.Interactions of these factors play an important role in determining the fertilizer value of a given PR.
Many PRs contain large amounts of calcium carbonate (CaCO 3 ) and have a high pH (Bolan and Duraisamy, 2003).When applied to soils, large amounts of protons are needed to make these PRs effective.This also means that the amounts of PR applied must be balanced in relation to the amounts of available protons, as high rates will work to the disadvantage of the effectiveness of the PR itself.
In this paper, the dissolution of three PRs from sub-Saharan Africa was studied in the laboratory and their agronomic effectiveness compared using soils from Tanzania (Rhodic Ferralsol), Ivory Coast (Plinthic Ferralsol), Kenya (Luvic Phaeozem) and Netherlands (Cumulic Anthrosol).The solubility of Minjingu PR from Tanzania, Khouribga PR from Morocco and Tilemsi PR from Mali under varying concentrations of hydrochloric acid (HCl) and varying shaking times was compared in the laboratory (Experiment 1).In Experiment 2, the three PRs were evaluated with respect to their agronomic effectiveness.The objectives were (i) to assess the amounts of extractable P under conditions of unlimited and limited supply of protons and how this is affected by varying shaking times and (ii) to evaluate the response of maize to the PRs.Increased concentration of the extracting acid is expected to increase the amount of P extracted.Meanwhile, at low proton concentration, the reactivity of the PRs will play an important role in the amount of P extracted.

Experiment 1: Extraction of phosphate rocks (PRs) with hydrochloric acid
The solubility of Minjingu PR in hydrochloric acid was compared with that of Khouribga and Tilemsi PRs in a laboratory experiment at Wageningen University, Netherlands.First, the PRs were analyzed for total P, P-Bray-I and P-Olsen.On the basis of the proportion of P-Bray-1 and P-Olsen (1), the PRs were graded as soft (Tilemsi PR) and hard (Khouribga and Minjingu PRs).The PRs were extracted with different concentrations of HCl for different shaking times (Houba et al., 1995).The HCl concentrations were 0.001M, 0.01M, 0.1M, and 1M.Five grams (5 g) of ground PR (100 mesh) was weighed in 50 ml shaking bottles and 35 ml of HCl of the required concentration was added (1:7 solid to acid ratio).Shaking was done for 1 min, 30 min, 6 h and 24 h in a reciprocating shaker at 150 oscillations per minute.The solution was filtered immediately after the required shaking time was over, diluted where necessary, and P was determined spectrophotometrically at 880 nm.

Experiment 2: Response of maize to the three phosphate rocks (PRs) on a Rhodic Ferralsol
The response of maize (Zea mais var.LG11) to the three PRs in a Rhodic Ferralsol from Tanzania was studied in a greenhouse experiment at Wageningen University using the double pot technique (Janssen, 1990).This technique is based on the fact that plants can simultaneously take up nutrients from soils and from a nutrient solution.When the nutrient under study is omitted in the nutrient solution the plant can only take it from the soils and the seed itself.Growth is measured by the relative increase in plant size (analogous to relative growth rate) per unit of time (t) denoted by R s The experiment was conducted for 26 days.A short duration was considered because effects of P application are more pronounced in young plants.Triple superphosphate (TSP) was included as a reference P fertilizer.Two rates of P for each of the P sources (TSP and the PRs) and two controls (with no P applied in the soil); one on complete nutrient solution [Control (+P)] and another on minus P nutrient solution [minus (-P)] made a total of 10 treatments.For the control on -P nutrient solution eight extra pots were included to make a total of 12 pots.In addition, for each of the levels of fertilizer P there was one pot on +P nutrient solution.A randomized complete block design with four blocks was used.
The levels of P were based on the P requirement for maize grown on a complete nutrient solution (+P) up to the eighth leaf stage (about 26 days old).At this stage the P mass fraction of maize is about 0.3% (Janssen, personal communication).Assuming an average dry-matter weight of about 4 g, the P content is 12 mg.The amount of the different P sources required was based on their recovery fraction and P content.In pot experiments, the recovery fraction of P from TSP may be as high as 30% (Janssen, personal communication).Given a P content of 20%, the amount of TSP required for an uptake of 12 mg P is 200 mg.This was considered the high rate, while half that amount was the low rate.Given the P content of the PRs (Table 2) and considering a recovery fraction relative to TSP (Wolf et al., 1987) of 20% for Tilemsi and 10% for Khouribga and Minjingu PRs, the low and high rates of these fertilizer materials are as shown in Table 1.
All fertilizers including TSP were ground and sieved through 100 mesh.Appropriate amount of the fertilizers was thoroughly mixed with 200 g air-dry soil and two thirds of the soils' field capacity water requirement1 added.The uniformly wetted soil was added into pots of 270 cm 3 and four maize seeds planted.The remaining one third of the soils' field capacity water requirement was added.The weight of the pots and contents were recorded.Subsequent watering was done to this weight to re-establish the field capacity of the soils.Each pot was placed on top of a larger pot of 1100 cm 3 containing nutrient solution and separated with wire gauze.Pots were covered by a polyethylene sheet to minimize evaporation loss prior to crop emergency and to create warmer conditions for germination.Ten days after planting the plants were thinned to two per pot; soil covered with sterilized gravel to minimize evaporation loss and new weight recorded for watering purposes.At later growth stages the weight of the plants were taken into account in watering2 .To avoid positional effect the pots were rotated daily.Greenhouse temperatures were maintained at 20°C during day and night using HPI lamps of 400 watts placed 1.2 m above the pots at one lamp per square meter.
The nutrient solution was changed every 5 days.Plant size was determined by measuring the length of the leaves from the base to the apex (blade and sheath) when plants on the +P solution were at the 4 th and 8 th leaf stages (Janssen, 1990).At harvest the above ground plant parts (shoots), roots in soil and roots in solution were separately weighed before and after drying at 70°C for 48 h for the shoots and 24 h for the roots.All shoot samples as well as two root samples per treatment were analyzed for P contents.The root samples were selected on the basis of dry-matter yield; for each treatment, pots with the lowest and highest dry-matter yield were considered for root analysis.This was done in order to reduce costs of analysis.

Laboratory analytical methods
Unless where stated otherwise, laboratory analytical methods used for the various determinations were: pH potentiometrically in an 1:2.5 soil to extract ratio with either H 2 O or 1 mol.KCl; organic carbon by wet oxidation (K 2 Cr 2 O 7 + H 2 SO 4 ); total N and P by continous flow analyzer after digestion with H 2 SO 4 /salicyclic acid/H 2 O 2 /selenium; P -Bray and P -Olsen spectro-photometrically at 880 nm, CEC and exchangeable bases by percolation (BaCl 2 unbuffered), Ca and K by flame photometer, and Mg by atomic absorption spectrophotometer (Houba et al., 1995).

Data analysis
For Experiment 1, no formal statistical test was performed on the data except for averages.Data of Experiment 2 were subjected to analysis of variance (ANOVA) using the statistical package STATGRAPHIC Plus version 7.0 (Manugistics Inc. 1995) to determine the effect of the different sources of P on R s , dry-matter yield and P content (Experiment 2).

Experiment 1: Extraction of phosphate rocks (PRs) with hydrochloric acid
Some of the properties of the PRs used in the study are shown in Table 2. Percent total P is highest in Khouribga PR and lowest in Minjingu PR.Meanwhile, the CaCO 3 equivalent and pH in water for Minjingu PR are the highest of the three PRs.
Table 3 shows the amount of extractable P from the PRs at different concentration of HCl averaged over the four different shaking times.There was no substantial effect on the amount of P extracted at the different shaking times.Overall, more P was extracted from Tilemsi than from Khouribga and Minjingu PRs.At high HCl concentration (1 M), extractable P was high from all the PRs.However, when based on percent of total P content, extractable P from Khouribga PR was about 10% less than from either Tilemsi or Minjingu PRs.Extractable P decreased sharply with decrease in HCl concentration.The difference between Tilemsi PR and the others became wider.The greatest difference occurred at 0.1 M HCl, where extractable P from Tilemsi PR was 14 and 24 times higher than from Khouribga and Minjingu PRs, respectively.Corresponding values at 0.01 and 0.001 M HCl concentration were 7.3 and 8.0, with respect to Minjingu, and 2.0 and 1.4 with respect to Khouribga.

Experiment 2: Response of maize to the phosphate rocks (PRs) on a Rhodic Ferralsol
Table 4 shows some of the chemical properties of the soil used in this study.The soil is slightly acidic (hence low protons content) with very low contents of available
phosphorus.Total nitrogen, organic carbon and cation exchange capacity are all medium.Table 5 shows the relative increase in plant size (R s ), dry-matter yield and P content in maize treated with different sources of P.
Severe P deficiency symptoms (purple coloration of the leaves and stems) were observed in the control pots on -P nutrient solution.Maize on the low rate of TSP and both rates of the PRs also showed P deficiency symptoms.The plants were tall and slender except for the pots treated with Tilemsi PR where they were shorter.All pots on +P nutrient solution had healthy green plants.
Highest relative increase in plant size were obtained from the Control (+P) treatment and lowest from the Control (-P) treatments.At the high and low rates, TSP gave relatively high R s compared to the PRs but apart from the high rate of Tilemsi PR the difference was not significant (P < 0.05).Among the PRs, R s did not differ significantly although there is an indication that the low rates were performing better than the high rates.
Shoot dry matter (DM) yield was highest from the Control (+P) treatment (Table 5).However, it was not significantly different (P < 0.05) from that at the high rate of TSP.The high rate of TSP differed significantly from the PRs.Tilemsi PR behaved like TSP in that yield increased with increasing P rate.There is an indication, though not significant, that the low rates of Minjingu and Khouribga were performing better than the high rates.However, when total DM yield was considered this trend was only true for Minjingu PR. Figure 1 shows the relation between DM yield (y) and total P content (x).The fitted lines for the various P sources are: TSP: y = 2.04x 0.5 -0.19x -3.25  (1) Tilemsi PR: y = 0.08x + 0.45 (2) Minjingu PR: y = 0.08x + 0.44 (3) Khouribga PR: y = 0.09x + 0.31 (4) Essentially, there were no differences between the PR sources with respect to P utilization; all showing practically the same linear relationship.Most points in Figure 1 are closer to the line representing maximum dilution of P (YPD), an indication of limited supply of P.
Only one point (from the high rate of TSP) is close to the line of maximum accumulation of P (indication of excess availability of P) and that explains why the relation for TSP is not linear (Equation 1).Highest shoot P content was from the Control (+P) treatment (Table 5).P content was significantly higher (P < 0.05) at the high rate of TSP compared to the low rate of TSP and the PR treatments.The relationship between total P content and P application is shown in Figure 2. P content increased steadily with P rate when P source was TSP or Tilemsi PR, but higher rates of Minjingu and  Khouribga slightly depressed P content.From Figure 2, the substitution value (SV) of the PRs were obtained by dividing the amount of TSP-P (standard P fertilizer) to P required to give the same P content as 200 mg PR-P.
There was no difference in SV between the PRs and on average the value was 0.1 for all the PRs.Recovery of P for TSP was 7.25% at the low rate and 8.73% at the high rate (Table 5).It was much lower for the PRs where it decreased with increasing P rates.However, with Tilemsi PR the difference between the two rates was slight compared to Minjingu and Khouribga where the difference between the low and the high rates was 2.4 and 2.8 times, respectively.Expressing P recovery from the PRs as percent of recovery from TSP, the values at the low and high rates were respectively 7.45 and 5.84 for Tilemsi PR, 8.0 and 2.75 for Minjingu PR and 8.69 and 3.03 for Khouribga PR.All these values are much lower than the 30% recovery assumed for TSP, and the 20 and 10% recovery fractions, relative to TSP, for Tilemsi PR and Minjingu and Khouribga PRs, respectively.

Extraction of phosphate rocks (PRs) with hydrochloric acid
PRs may contain CaCO 3 and other proton consuming compounds besides apatite.The CaCO 3 present in PRs as an accessory compound is readily soluble in dilute HCl and will dissolve before the inner shells of the PR particle, containing the calcium phosphates, are accessed.Together with the dissolution of CaCO 3 some P will go into solution.Under conditions of limited proton supply, most of them will be used to solve CaCO 3 , thereby increasing the concentration of Ca and hampering the dissolution of apatite.As proton supply increases the inner shells of the PR particle are accessed and proportionately more P can be solved than at low proton supply.
Considering Minjingu PR extracted at 1 M HCl, the amount of H + used for the dissolution of CaCO 3 and apatite can be found as follows: In the extraction flask, a solid: solution ratio of 1:7 was used, so per liter of 1 M HCl, 1/7 kg of the PR was added.Minjingu PR contains about 7.1 % CaCO 3 (Table 2).In 1 liter of the extraction solution there was 1/7 * 0.071 * 1000 = 10.14 g CaCO 3 , or 0.1014 mol CaCO 3 .From the reaction: Ca 5 (PO 4 ) 3 F + 7H + 5Ca 2+ + 3H 2 PO 4 -+ HF indicating that 1 mol of apatite containing 3 mol PO 4 or 93 g P requires 7 mol H + .Hence, the remaining quantity of H + after dissolution of CaCO 3 would be sufficient to solve: 0.7972/7 * 93 = 10.6 g of P from 1/7 kg of PR.
For 1 kg PR, this will be 10.6 * 7 = 74.4g P, a value close to the 73.7 g P found in the experiment (Table 3).As Khouribga and Tilemsi PRs have lower CaCO 3 contents than Minjingu (Table 2) more H + is left for the dissolution of phosphates, explaining the higher amounts of dissolved P found for these PRs (Table 3).Based on citrate soluble P, McClellan and Notholt (1986) ranked Tilemsi PR as medium and Minjingu PR as medium to high in reactivity, but our results suggest that Minjingu and Khouribga PRs have a lower reactivity than Tilemsi PR which should be due in part to the lower CaCO 3 contents of Tilemsi PR.
At low HCl concentrations, (< 0.1 M) the amount of H + that was available was not sufficient to solve all CaCO 3 , and because CaCO 3 dissolves more easily than apatite, the portion left for dissolution of phosphates was less than proportionate (Table 3).On the basis of the measured amounts of dissolved P, the amounts of dissolved CaCO 3 were calculated for the various HCl concentrations.Lines relating solved P or CaCO 3 to HCl concentration were drawn (Figure 3).The graph shows that relatively more CaCO 3 was dissolved than P, as assumed above.The trend in the effect of shaking time was not systematic; there was some indication of P reprecipitation at longer shaking time (24 h) when HCl concentration was 0.001 M. According to Deeley et al. (1987) more P should be expected with increasing shaking time.However, the removal of the products of dissolution (Ca and P) from the zone of dissolution is necessary to ensure continuation of the dissolution process according to the law of mass action, otherwise re-precipitation of P by Ca will take place (Kirk and Nye, 1986).

Response of maize to various PRs on the Rhodic Ferralsol
Dry-matter yield and P content were lower, though not significantly, at 400 mg than at 200 mg P per pot for Minjingu and Khouribga (Table 5 and Figure 1).This may be a consequence of the higher Ca 2+ concentration, following the dissolution of CaCO 3 in the soil solution, and hence the lower P concentration at the higher P rates, as explained above.Such reactions were not found for TSP and Tilemsi PR because they contain no or less CaCO 3 and, moreover, were not applied at rates as high as 400 mg P.Although in the extraction experiment Tilemsi PR appeared to be more reactive compared to Khouribga and Minjingu PRs, this was not reflected agronomically (Figure 2) given the limited supply of protons in the Rhodic ferralsol.
One of the reasons why the recovery of P was lower than assumed could be the short duration of growth.By the time of harvest a large proportion of P was from the seed and soil.Another reason is that uptake did not increase with increasing rate of application as assumed at the derivation of the rates of PR.The SV was the same for all the PRs indicating that they did not differ agronomically.Under field conditions a relative agronomic effectiveness (RAE) value for Tilemsi PR of between 85 to 90% was reported in West Africa (Henao and Baanante, 1997), while in Western Kenya values of between 70 to 75% have been reported for Minjingu PR (Lijzenga, 1998).Comparison of such results is however, difficult since the PRs were not tested under similar conditions.

Conclusion
It can be concluded from this study that CaCO 3 content of the PRs was an important factor determining differences among them.Minjingu and Khouribga PRs having relatively high CaCO 3 content were hard compared to Tilemsi PR.Differences in dissolution manifested more at low than at high proton supply.Although Tilemsi PR showed a higher reactivity than Minjingu and Khouribga in the extraction experiment, it did not perform better agronomically under conditions of limited proton supply.All had a SV of 0.1.There was an indication of poor performance at the higher rates of the hard PRs.The results show that a low availability of PR-P cannot be compensated for by a high application rate.The implications of these results are that, high rates of PR are not desirable as they will introduce more CaCO 3 hence reducing the amounts of protons that would otherwise be available for solving P from calcium phosphates in the PR.Also, laboratory extractable P does not give a correct picture of the agronomic effectiveness of a PR and that soil and plant properties are important in determining the agronomic effectiveness of a PR.The study has provided an insight into the role of protons in PR dissolution and how one can cope with situation of low proton supply in soils.This information is required to enable extension of PR use to a wider range of soils beyond those considered to be ideal.

Figure 1 .
Figure 1.Relation between maize dry-matter (DM) yield and total P content as function of P fertilizer type.YPD and YPA refer to yield with P diluted and accumulated, respectively.

Figure 2 .
Figure 2. Relation between total P content and P rate as function of P fertilizer type.

Figure 3 .
Figure 3. Relation between the solved fractions of total P (measured) or CaCO 3 (calculated) and HCl concentration.Reaction times of Minjingu PR and HCl were one minute and 24 h.

Table 1 .
Treatment details for Experiment 2.

Table 2 .
Some properties of the phosphate rocks used in the study.

Table 3 .
Extractable P from the phosphate rocks at different HCl concentration averaged over the four different shaking times.

Table 4 .
Some chemical properties of the Rhodic Ferralsol used in Experiment 2.

Table 5 .
Relative increase in plant size (R s ), dry-matter yield and P content in maize treated with different P sources.
Means followed by the same letter are not significantly different (P < 0.05); LSD, least significant difference; SE, standard error; CV, coefficient of variability.
It follows that 1 mol CaCO 3 requires 2 mol H + , and hence the amount of CaCO 3 present requires 0.2028 mol H + .If CaCO 3 would dissolve first, 0.7972 mol HCl is left for the dissolution of the apatite.Dissolution of fluorapatite can be represented by the following equation: