Chapter 4
Evaluation of phosphate rocks for direct application

Chapter 3 described the characterization of phosphate rock (PR) materials for direct application. Depending on their origin and geological history, PRs show great variability in their inherent properties, especially in their grade, beneficiation requirements and apatite reactivity. This information provides the first opportunity to assess the suitability of PR materials for direct application. In the case of their utilization in agriculture, PRs applied to soils undergo a series of chemical and biological transformations that govern their dissolution and the availability of the dissolved phosphorus (P) to plants.

This chapter reviews the methodologies for evaluating PRs for direct application in agriculture. They include solubility tests using conventional reagents, incubation studies in soils without plants, pot experiments using a test plant in controlled conditions, and field experiments integrating environmental factors, cropping systems and management practices, as well as their interactions. This chapter illustrates these approaches using selected examples from studies with PRs from West Africa (Truong et al., 1978). The information has been obtained using Taiba PR (Senegal) and Hahotoe PR (Togo), which are exploited on a large scale for export purposes. Other PRs include Arli and Kodjari PRs (Burkina Faso), Tahoua PR (Niger), and Tilemsi PR (Mali), which are mined on a small scale for local use. Gafsa PR (Tunisia) was utilized as a reference because of its high reactivity. Interest in African PRs, in particular those from sub-Saharan Africa, stems from a number of considerations. First, there is the paradoxical situation where Africa ranks first with 28.5 percent of the world’s production of PR, yet it has the lowest phosphate consumption with 2.8 percent of the world’s consumption (FAO, 1999). Second, although the PR resources of Africa are considerable in terms of both quantity and diversity, they are not exploited greatly (McClellan and Notholt, 1986; Baudet et al., 1986). All types of PR can be found. There are igneous deposits in South Africa, Zambia and Zimbabwe that are coarsely crystalline in nature and quite unreactive and unsuitable for direct application (Khasawneh and Doll, 1978). Guano-type deposits occur in Namibia and Madagascar (Truong et al., 1982). These deposits were formed recently on coral basements and are very soft and practically equivalent to water-soluble phosphate. Finally, there are sedimentary PRs that have been deposited progressively over geological time and are loosely consolidated. They contain microcrystalline particles with large specific surface areas and vary widely in terms of chemical composition and reactivity. These deposits represent 80 percent of the total world reserves. They extend from north to west and central Africa and are among the most relevant for direct application in agriculture.

For recent detailed information, the reader may refer to the FAO/IAEA international networked research project (IAEA, 2002). This project has carried out all types of the studies mentioned above in order to evaluate the agronomic effectiveness of PR sources from several deposits worldwide under a wide range of soils, climate, crops and management conditions.

Phosphate-rock solubility tests

The solubility tests of PRs using chemical extraction methods are empirical. They offer a simple and rapid method for classifying and then selecting PRs according to their potential effectiveness. They cannot be used to evaluate the amount of plant-available P because the actual agronomic effectiveness of the PR in the field depends on a range of factors. Most conventional extraction methods are based on those used at the beginning of the phosphate-industry era (Lenglen, 1935). These include the use of organic acids found in the soil after microbial metabolism and organic matter decomposition, and in root exudates that assist the roots in absorbing phosphates (Amberger, 1978). Water would be the ideal extractant because it is a natural compound (compared with alkaline or acidic electrolytes) and less disturbing of the ionic equilibrium in solid and liquid phases (Van der Paauw, 1971). However, measuring very low P concentrations in the soil solution poses many analytical difficulties. The chemical extractants include reagents such as oxalate, citrate, sodium-ethylenediaminetetraacetic acid, alkaline ammonium citrate and a number of weak acids (including oxalic, lactic, malic, acetic, citric and formic acids), that dissolve and form complexes with the P from the PRs. Various studies have evaluated these methods (Engelstad et al., 1974; Chien and Hammond, 1978; Mackay et al., 1984; Rajan et al., 1992). The most commonly used reagents are neutral ammonium citrate (NAC), 2-percent citric acid (CA) and 2-percent formic acid (FA).

Solubility measurements using conventional chemical techniques

The total P₂O₅ and CaO contents of PRs and their P solubility are obtained using conventional chemical techniques. Table 8 shows such data for African PRs. The total P in PRs ranges from 27 to 36 percent P₂O₅ and the CaO contents are higher than the P₂O₅ values. This is because of the composition of the apatite, which contains 10 Ca atoms for 6 P atoms, and the possible presence of free carbonates. In addition to P, the calcium (Ca) released from PRs can play an important role in the soil ionic equilibrium and in the nutrition of plants grown in acid tropical soils.

The solubility data for the three conventional reagents differ depending on the strength of the extractants. However, they are closely related and rank the PRs in the same relative order: Gafsa, Tilemsi and then the others follow in the same group. Table 8 also shows a good relationship between the solubility in different extractants and the CO₃:PO₄ ratio, which indicates the degree of isomorphic substitution of PO₄ by CO₃ in the apatite structure. The greater the degree of substitution, the higher will be the solubility in standard reagents (Chapter 3; Lehr and McClellan, 1972).

TABLE 8
Chemical analysis and solubility in conventional reagents of selected African PRs

Source: Truong et al., 1978.

PR solubility expressions

The expression of P solubility can be an issue. In studies with several PR sources with large variations in total P content, Chien (1993) proposed that the solubility of the PRs is expressed more appropriately as a percentage of the rock rather than as a percentage of the total P (Chapters 3 and 11). This was because unscrupulous practices, such as mixing sand with PR, can inflate the extractability figure (Chapter 11), so leading to incorrect conclusions on the reactivity of PRs.

In general, the solubility expressed as a percentage of total P is preferred for practical reasons. These are that the percentage of the total P that is soluble in a reagent indicates the reactivity of the PR. In addition, the application rate of a P fertilizer is calculated according to the total P content. The expression of PR solubility as a percentage of total P would only be recommended where it represented an inherent property of an individual PR source and/or when comparing several PRs, all PR sources having a similar total P content. However, it is possible to overcome this problem by considering the solubility together with the total P of PRs.

Another expression proposed by Lehr and McClellan (1972) is the absolute solubility index (ASI), which is estimated as follows:

ASI = [(% soluble P)/(% P in apatite)] × 100

The theoretical percentage P content of the apatite can be calculated from the dimensions of the a and c-unit cells. In this way, it is possible to eliminate the variability of the apatite content in the ores and the total P in the apatites (Lehr and McClellan, 1972). The percent-soluble P will depend on the particular test selected. The major disadvantage of the ASI method is that the crystallographic studies are expensive, and they require skilled staff and specialized laboratories with sophisticated equipment for X-ray analysis. Therefore, they should be reserved for basic studies to characterize main phosphate deposits.

Effect of particle size on solubility measurements

As PRs are relatively insoluble minerals, their geometric surface area is an important parameter determining their rate of dissolution. In the case of sedimentary PRs, the geometric surface area is about 5 percent of the total surface area because of the porous structure of PR particles (Lehr and McClellan, 1972). In igneous PR with coarsely crystallised structures possessing no internal surfaces, the geometric and total surface areas are similar.

The PR particle size is also important: the finer the particle size, the greater the geometric surface area and degree of contact between the soil and PR particles and, thus, the greater the PR dissolution rate. The cutoff point for PR particle size appears to be -100 mesh (149 µm), as the cost of finer grinding would be prohibitive compared with any gain in effectiveness. Rajan et al. (1992) reported that the increase in chemically extractable P with grinding of PRs related positively to the reactivity of the PRs. They found that the greater the reactivity of the PR, the greater the increase in extractable P on grinding. The data presented in Table 9 show a positive relationship between the fineness of particle size of North Carolina PR and the increase in citrate-soluble P of the PR (Chien and Friesen, 1992). The message is that measurements of PR solubility should be done on samples of the same particle-size distribution resulting from the same grinding process.

TABLE 9
Citrate soluble P of North Carolina PR as affected by particle size

Source: Chien and Friesen, 1992.

Effect of associated minerals on PR solubility measurements

PRs generally contain endogangue and exogangue materials, such as calcite, dolomite, gypsum, as well as quartz, iron and aluminium oxides and clays (Lehr and McClellan, 1972). Among all of these, free carbonates would have a major influence on PR dissolution because they are more soluble than apatite. They also consume a part of the chemical reagents, especially weak extractants such as NAC, used for the solubility tests. To overcome this problem of preferential reaction in the solubility measurements, Chien and Hammond (1978) discarded the first NAC extract and measured the amount of PR dissolved in the second NAC extract. Mackay et al. (1984) and Rajan et al. (1992) found the sum of four extracts to be more representative.

Associated minerals have less influence on acidic extractants (Rajan et al., 1992; Chien, 1993). By increasing the strength of the CA from 2 to 5 percent and then to 15 percent, Mackay et al. (1984) were able to extract 21, 44 and 59 percent, respectively, of the P in the PR. The influence of free carbonates decreases with the strength of the extractants. Of the three methods that are most commonly used, the 2-percent FA extractant would be preferred for a single chemical extraction procedure.

Kinetics of long-term PR dissolution

Common standard solubility tests are only qualitative in nature. This is because they use procedures that limit the dissolution of PR either by the short reaction time or by the small PR sample/extractant solution ratio. Therefore, these conventional tests focus on the estimation of the short-term efficiency of PRs.

Successive or sequential extraction procedures may improve the predictive capacity of the tests considerably by removing a greater proportion of total P without achieving complete PR dissolution.

Truong and Fayard (1995) have proposed an additional simple procedure for measuring the kinetics of PR dissolution over time. The PR sample (1 g) is placed in a funnel or a tapered tube, blocked with glasswool, and a 2-percent FA solution is dripped continuously over the PR sample at a rate of about 1 drop/s. A preliminary study showed that after 4 h and with 300 ml of solution, most of the P in PRs was dissolved. Figure 11 presents some of the results.

The advantage of this procedure is that it removes all the dissolved P and Ca. Moreover, there is no accumulation on the surface of particles preventing further PR dissolution. The extracted solutions can be analysed at selected time intervals to draw a dissolution curve with time.

In general, the curves shown in Figure 11 present two parts. The first is a zone of fast dissolution for about 50-60 min, which could be related to the short-term efficiency. The curves for different PRs are steeper and close to each other during this part. The second is a zone of slow dissolution with a pronounced change of slope, representing the long-term effect. Some PRs are difficult to dissolve completely because they are cemented by silica or occluded in iron and aluminium oxides (Lehr and McClellan, 1972). In the study shown in Figure 11, the Tilemsi PR confirms its high reactivity in the short and long term. The above procedure is more time consuming than a single extraction. However, it is not more demanding than sequential or successive extractions and it provides more information on PR dissolution over time.

FIGURE 11
Kinetics of continuous dissolution of PRs with formic acid

PR reactivity scales and crop yield response

PR reactivity scales serve not only to compare several PR sources but also to predict their potential agronomic effectiveness. Where a reactive PR is applied to a soil, it dissolves under ideal conditions. The available P will result in a good crop response if P supply is a limiting factor. However, the relationship is not direct because many factors and their interactions determine the ultimate agronomic effectiveness of PRs (Mackay et al., 1984; Rajan et al., 1996). Nevertheless, good correlations have been found for a specific set of conditions. For example, the NAC solubility of PRs correlated well with grain yield of flooded rice in Thailand (Engelstad et al., 1974), and dry-matter yield of Guinea grass on an Oxisol from Colombia (Chien and Van Kauwenbergh, 1992). However, several studies have found FA to be the best indicator of crop response (Chien and Hammond, 1978; Mackay et al., 1984; Rajan et al., 1992).

The study by Chien and Hammond (1978) illustrates the value of the FA test. They measured the solubility indices of seven PRs by various laboratory methods and the crop yield response to these PRs on two Colombian soils (Oxisol and Andosol) in greenhouse and field experiments. They found that 2-percent FA gave the highest and most significant correlation coefficient with crop yield response, followed by 2-percent CA and NAC (Table 10). The effectiveness of PRs varied widely with the P-application rate and the duration of the experiment. At low P rates, reactive PRs released P quickly and maintained their dominant position, as less reactive PRs released a small quantity of P slowly. At high rates of P application, reactive PRs released a higher proportion of P at the initial stage and the non-utilized portion was converted into less-available forms in the soil. In contrast, less reactive PRs released a greater P quantity with time, thereby providing an adequate supply to the crop. This improved their effectiveness with time.

Standardization and index of reactivity PRs can serve as raw materials for industrial processing in the manufacture of P fertilizers and for direct application in agriculture. The quality criteria for these two uses may be different as the modern fertilizer industry requires strict quality standards for the raw materials that it uses in specific manufacturing processes. Thus, not only the grade (P₂O₅ content) of the PR is important but some other threshold limits should also be considered. For example, the CaO: P₂O₅ ratio is of major importance because of its significance for acid consumption when dissolving the PR. Where the ratio exceeds 1.6, then the wet process is uneconomical. The impurities of aluminium (Al) and iron (Fe) are particularly troublesome in the wet process. The ratio (Al₂O₃ + Fe₂O₃):P₂O₅ > 0.10 is considered critical. Moreover, a ratio of MgO:P₂O₅ > 0.022 is undesirable owing to a detrimental effect on phosphoric-acid production.

TABLE 10
Correlation coefficients between reactivity scales of seven PRs and crop yield

* Significant at 5% level.
** Significant at 1% level.
Source: Chien and Hammond, 1978.

The quality factors for direct application are different. Indeed, PR sources suitable for direct application are considered to be ‘problem ores’ because of their low grade and the presence of accessory minerals and impurities (Hammond et al., 1986b). The presence of carbonates (Ca and magnesium (Mg)) as accessory minerals could be useful for plant nutrition and soil amendment. The Al or Fe contents are usually of no major consequence. Besides suitability for beneficiation, the most important factors in the assessment for direct application are grade (P₂O₅ content) and the reactivity of the apatite (solubility).

TABLE 11
Spatial variability of PR samples within the same deposit

Sources: Truong and Cisse, 1985; Truong and Fayard, 1988.

In some geological deposits, the PR minerals vary within the same deposit. Table 11 presents examples from several deposits. Some layers are more reactive than others. An adequate exploitation of the deposits should consider their quality in relation to whether the PR is to be used for industrial processing or direct application.

The reactivity-index guidelines vary for different countries. The standards of the European Community for the direct application of PRs are strict considering that in Europe most soils are not acidic, annual rainfall is moderate and the cropping is seasonal. To ensure appropriate PR agronomic effectiveness, three kinds of PRs can be sold as fertilizers. These are:

• fine PR: 25 percent total P₂O₅, 55 percent soluble in 2-percent FA, 90 percent < 63 µm, 99 percent < 125 µm;

• semi-fine PR: 25 percent total P₂O₅, 45 percent soluble in 2-percent FA, 90 percent < 160 µm;

• hard PR: 25 percent total P₂O₅, 10 percent soluble in 2-percent FA, 90 percent < 160 µm.

Regarding the United States of America, Chapter 3 presents a classification proposed by Diamond (1979) of PRs for direct application based upon their solubility and expected initial crop yield responses.

In Australia and New Zealand, most PRs are used in permanent pastures and the index of reactivity is set at 30-percent soluble in 2-percent CA (Sale et al., 1997a).

These examples suggest that the reactivity index should be adapted to the local conditions. For example, in the tropics of Latin America and Africa with very acidic soils and heavy rainfall or in Southeast Asia with plantation-state crops, the threshold values of the index could be set lower than those above.

Measurement of exchangeable P from PR by radioisotopic techniques

Isotopic techniques utilizing the beta-emitting radioisotopes 32P (half-life = 14.3 d) or 33P (half-life = 25.3 d) provide a further way of measuring the exchangeable P released from PRs while keeping the ionic equilibrium between the liquid and solid phases unchanged. Although these techniques provide precise, quantitative information, their utilization requires trained staff with adequate skills and expertise, and functional laboratory facilities that comply with radiation protection and safety regulations (Zapata and Axmann, 1995).

Applications of radioisotopic techniques include the short-term isotopic-exchange kinetics technique to measure soil P dynamics in the laboratory, in particular exchangeable P or E values at selected times (Fardeau, 1981). There are also indirect measures utilizing plants such as labile P or L values (Larsen, 1952) or from isotope dilution, the P derived from PR that is available to the plant (Zapata and Axmann, 1995).

In the African PR study, the exchangeable P values (E values) were determined first. This involved injecting a quantity of carrier-free 32P phosphate ions (R) into a phosphate-water suspension (ratio: 1 g:100 ml). At selected times, 10 ml of the suspension was withdrawn and filtered using millipore filters, and then the concentration of phosphorus (C_p) and the radioactivity (r) of the solution were measured. The isotopically exchangeable P (E value) was calculated using the general isotopic dilution formula: E = C_p × R/r. Table 12 presents the results.

TABLE 12
PR kinetics of isotopic exchange and solubility in water

Source: Truong et al., 1978.

The isotopic-exchange technique can discriminate between PRs in a very short time. After 1 min, two groups can be separated clearly: more than 99 percent of added 32P exchanged with P from Gafsa and Tilemsi PR, whereas only 75 percent exchanged with P from the Kodjari and Hahotoe PRs. Although isotopic exchange is usually a rapid process, the reactions will continue towards equilibrium. After 4 h, the difference between the two groups was confirmed from their E values. The exchangeable P is an indicator of their respective reactivities. However, the differentiation of PRs within the same group (e.g. Gafsa and Tilemsi) requires a longer time of contact of about 24 h (Fardeau, 1993) or several weeks.

The solubility of PR-phosphorus in water was very low for all the phosphates, and the values were not useful in selecting PRs (Table 12).

Reactions between phosphate rocks and soil

Soil incubation

Incubating soils amended with PRs provides the opportunity to measure the PR dissolution in selected soils with different properties. In addition, some PRs have significant quantities of free carbonates and other minerals and have the potential to modify the characteristics of the soils when the PRs dissolve. Closed-incubation studies to determine PR dissolution rates in soils have limitations because the reaction products are not removed and, therefore, the results could be experimental ‘artefacts’. Open-incubation studies, where the soil treated with PR is placed in an open container and water is added at a rate simulating local rainfall conditions, are preferable. The leachate is then collected and analysed for P, Ca and other elements. At selected intervals, soil samples are analysed for dissolved PR.

Closed-incubation studies by Mackay and Syers (1986) showed that the dissolution reaction reached equilibrium at about 50 d. Presumably, at this point, in the absence of any loss mechanisms such as plant uptake or leaching, the P concentration in the soil solution in this study had increased to the point where any further PR dissolution was stopped.

In another incubation study, Jadin and Truong (1987) compared Gafsa, a high-reactivity PR, and Hahotoe, a low-reactivity PR, with a water-soluble fertilizer, triple superphosphate (TSP), including a control without phosphate. The soil used was an Oxisol from Njole, Gabon, with a pH in water of 4.3 and a P sorption capacity of 234 ppm. The P application rate was 100 mg/kg. The available P was measured by the Olsen test as modified by Dabin (1967) to estimate the P dissolved from PRs. The data in Table 13 reflect the net result of opposite reactions occurring upon PR dissolution. These include the release of dissolved P into the soil solution, the sorption of dissolved P by the soil colloids and its conversion to forms non-available to plants. The dissolution rates based on the available P (Olsen-Dabin method) data were 84 percent for TSP, 58 percent for Gafsa PR and 35 percent for Hahotoe PR. The difference between PR sources was consistent with previous evaluation techniques, in particular for the PR reactivity.

TABLE 13
Effects of incubating PRs on the characteristics of an Oxisol from Njole, Gabon

Source: Jadin and Truong, 1987.

PR application prior to planting may be an advantage in soils of low P buffering capacity but not necessarily in soils of higher P buffering capacity (Chien et al., 1990b). The availability of the adsorbed P was measured as a function of the adsorption capacity in order to check the validity of this assumption. In the Njole study, 63 percent of the adsorbed P remained exchangeable during the desorption phase using the 32P isotopic-dilution technique (Jadin and Truong, 1987). Therefore, it is necessary to evaluate the overall P sorption capacity of the soil when considering prior PR applications to the soil.

Liming effect of PRs

Theoretical calculations can estimate the calcium carbonate equivalent (CCE) of PR as the sum of gangue mineral and the carbonate in the apatite. This can also be measured by the Association of Official Analytical Chemists method where 1 g of PR is added to 50 ml of 0.5 NHCl and the remaining acidity is measured by back-titration (Sikora, 2002). Normally, the CCE represents about 50 percent of the PR. These estimates (potential liming effect) assume full PR dissolution.

A greenhouse study assessed the potential Ca value of some PRs from South America and West Africa applied to an Ultisol soil. It found that medium and high reactivity PRs can supply Ca to plants growing in acid soils with low exchangeable Ca (Hellums et al., 1989). In another experiment with an Oxisol from Gabon, the application of Gafsa PR increased soil pH by 0.5 of a unit (Table 13). Higher PR rates can yield larger increases in pH. The effects of the Ca release from PRs were also significant owing to the Ca supply for plant nutrition, its contribution to the base saturation and the decrease in Al toxicity. This result is reflected not only by a decrease in the extractable Al but also by an increase in the level of exchangeable Ca (Kamprath, 1970).

Although the potential increases in soil pH resulting from PR dissolution are small (Sinclair et al., 1993b), they can have a significant effect on Al saturation levels in tropical soils. Work on Ultisols and Oxisols in Puerto Rico showed that Al saturation decreased from 60 percent at pH 4.2 to 35 percent at pH 4.5 and 20 percent at pH 4.8 (Pearson, 1975). These effects can improve soil chemical properties significantly, especially in very acid, degraded and problem soils such as acid-sulphate soils in Southeast Asia (Truong and Montange, 1998).

In conclusion, the liming effect of PR exists but it is of small magnitude. Realistic PR application rates (100-200 kg/ha) with an effective neutralizing value of about 50 percent are equivalent to 50-100 kg of lime per hectare. However, in spite of this small amount, medium and highly reactive PRs can have beneficial effects on the chemical properties of highly weathered tropical soils.

Greenhouse tests

The next step in the evaluation involves pot experiments using a test plant. They have the advantage of being relatively inexpensive and enabling several factors such as soil type, PR material and plant species (which influence the agronomic effectiveness of PRs), to be examined. This approach can generally control other factors that influence plant growth, such as light, moisture, temperature and disease. This is often difficult in the field. However, growth conditions and plant development in the greenhouse are usually very different from field conditions. For example, the volume, depth and stratification of the natural soil horizons in the field influence the development of the root system in different locations within the profile and regulate the storage and movement of soil water and nutrient ions. These are difficult to reproduce in a pot (Rajan et al., 1996). Considering the slow-release characteristics of PRs, a long-term field evaluation requires variations in soil and climate conditions and in the management practices of the cropping systems.

There are a number of practical considerations that are relevant to conducting pot experiments. The size of the pot depends on the test plant and the expected duration of the experiment. Pots containing 2-5 kg of soil are normally used for small-grain cereals and grain legumes to obtain grain yield. The duration of the experiment depends on the objectives of the study. Some studies recommend harvesting the plants at the period of maximum P demand by the plants, such as at the heading stage for cereals and at flowering or podding for legumes. Others focus on the effect of the treatments on the grain yield and, therefore, will grow crops to maturity. To compare a large number of PRs in a wide range of soils, small grasses are recommended such as Agrostis sp. in 100-200 g of soil per pot, or ryegrass in 500-1 000 g of soil per pot. Rapid growth allows several cuts at intervals of 2-4 weeks. Agrostis is particularly interesting because of its small-sized seeds (100 seeds/pot weigh 12-15 mg and contain 50-60 µg P) and its responsiveness to the soil P status (Truong and Pichot, 1976). The large number of seeds per pot is important for ensuring homogeneity of germination, tillering and biomass production. The P application rate should vary between 25 and 400 mg per kilogram of soil. Low rates (25 or 50 mg/kg) are often insufficient in soils with low and high P sorption capacities and the effects are not detected. High rates (400 mg/kg or more) can induce disturbances in soil ionic equilibrium (Zapata and Axmann, 1995).

The African PR study by Truong et al. (1978) compared seven PRs with different reactivities with a soluble fertilizer (TSP) and a control without P. The P sources were applied at the rate of 100 mg of P per kilogram of soil to pots containing 100 g of soil. The three soils used were an Alfisol from Niger (pH water 6.5; Langmuir’s P sorption capacity 16 ppm), a Vertisol from Senegal (pH 5.9, P sorption 1 067 ppm) and an Andosol from Madagascar (pH 4.3, P sorption, 3 818 ppm). They were all deficient in available P. Fertilizers and soils were mixed thoroughly with adequate levels of other nutrients in order to ensure that P was the only factor influencing the results. 32P carrier-free was also added to determine the isotopically exchangeable P (L values) according to Larsen (1952). The Agrostis plants were cut every month for a total of four cuts and the dry matter and P uptake were determined. The P-fertilized treatments (TSP and PRs) produced significantly greater dry-matter yields and higher P uptake and L values than the control with TSP giving the highest yields (Table 14). The actual data for the PR treatments varied from soil to soil but the relative ranking of PRs remained almost the same, i.e. Gafsa was the most reactive, followed by Tilemsi, and sometimes Tahoua was better than the others.

TABLE 14
Results of a pot experiment with selected African PRs

Values followed by the same letter are not statistically different (P = 0.05).
Source: Truong et al., 1978.

P uptake seems a more sensitive parameter than dry matter for discriminating between PRs. In general, the results confirmed that the inherent characteristics of PRs remain the most important factor determining their potential agronomic value.

In conclusion, pot experiments serve mainly to obtain preliminary information on the potential agronomic effectiveness of PR sources. They are particularly useful for integrating the effects and interactions of PRs, soils and plants under controlled conditions. Nevertheless, the results obtained in these experiments require careful interpretation.

Comparison of PRs with standard fertilizers

The agronomic performance of PRs relative to water-soluble P fertilizers is normally expressed in two ways. The first is the substitution value (SV) or horizontal approach. This is the ratio of the P rate of the standard fertilizer to that of the test fertilizer that is required to produce the same plant yield. The second is the relative response (RR) or vertical approach. This is the ratio of the response to the test fertilizer to that of the standard fertilizer when both are applied at the same rate of P. Both approaches have their advantages and their disadvantages.

The SV is useful for making an economic assessment of the PR fertilizer relative to the reference fertilizer (Chien et al., 1990a) at the desired level of productivity that a farmer may wish to achieve. The SV system has a constant value independent of P rate if the reference and test fertilizers have the same maximum yield, which is rarely the case with PRs (Rajan et al., 1991a, Ratkowsky et al., 1997). For this reason, SV values are calculated at the desired yield level (Rajan, 2002) or as a continuous function of yield (Singaram et al., 1995).

As the yield response curves with PRs and water-soluble P invariably do not share a common limiting yield, it is not possible to calculate a single RR value for each source. The recommendation is to make the evaluation over a full range of P application rates. However, this approach implies a large number of pots or experimental units, and it requires complicated supervision and logistics. A compromise involves working at moderate rates, in the region of the linear part of the response curves, where the RR would be simply the ratio of the slopes of the linear portions of the regression lines (Khasawneh and Doll, 1978). With moderate rates of 50-200 mg of P per kilogram of soil, commonly used in greenhouse experiments depending on soil texture, P status and P-fixing capacity, it has been found that the comparison of various P sources was independent of the rate of P application (Morel and Fardeau, 1989).

Another index is called the relative agronomic effectiveness (RAE) of a given P test fertilizer. This is determined by expressing as a percentage the ratio of the response to the test fertilizer (treatment - control) to the response to the standard fertilizer, when both are applied at the same rate:

RAE = [(test P fertilizer - control)/(standard P fertilizer - control)] × 100

For the RAE values to be meaningful, the difference between standard P fertilizer and the control should be statistically significant. The PR under study is the P test fertilizer while the standard P fertilizer is a water-soluble P fertilizer such as single or triple superphosphate. Based on the RAE equation, different coefficients can be calculated for crop yield or dry-matter production, P uptake, chemical extraction or L values. Table 15 shows the data obtained for the African PR study utilizing TSP as a reference fertilizer. It shows that the relative ranking of PRs is almost the same with different coefficients, but the actual values vary largely with the soils. Considering the RAEs based on P uptake, Gafsa PR was equivalent (98 percent) to TSP in the Andosol but its relative effectiveness was 54 percent in the Vertisol and 42 percent in the Alfisol. Conversely, the least reactive PR (Arli) was 42 percent equivalent to TSP in the Andosol. Thus, the ranking of PRs for direct application remains relatively constant, but the effectiveness of the PRs should be considered in relation to the soil properties.

TABLE 15
Estimated RAE coefficients based on L values and P uptake from the pot experiment

Source: Truong et al., 1978.

Effect of time

PRs are slow-release fertilizers. They require time and water surrounding the particles in order to enable the dissolution products to diffuse away from the PR particles into the soil volume. The greenhouse evaluation of the African PRs was undertaken to observe the changes occurring with time. The RAE coefficients based on L values and P uptake changed considerably between 1 and 4 months for most PRs in an Andosol (Table 16). Being very reactive, the Gafsa and Tilemsi PRs dissolved rapidly and their effectiveness remained unchanged or increased slightly after 4 months, while less-reactive PRs required time to express their potential effectiveness. The improvement in the relative effectiveness of PRs over time has been attributed to the continuation of the PR dissolution process while a low P concentration is maintained in soil solution. The improvement may also result from the depletion of P from the soluble P fertilizer as a result of P uptake by plants and the conversion of soluble P to less-available P forms. In general, less-reactive PRs need to be ground more finely in order to ensure a greater and longer time of contact between PR and the soil.

TABLE 16
Changes with time of the estimated RAE coefficients for PRs applied to a Madagascar Andosol

Source: Truong et al., 1978.

Another important factor affecting PR dissolution is the level of soil water content. Table 17 shows the influence of soil water content on P uptake resulting from the application of several PRs in an Oxisol. The effects are more pronounced for less reactive PRs.

TABLE 17
P uptake from PRs in a Madagascar Oxisol as influenced by soil water content

Relationship between PR solubility and crop P uptake

In general, increased PR dissolution is expected to increase available P, and this will result in improved P uptake and crop yield. However, the relationship is not direct because of the many factors and their interactions that affect the agronomic effectiveness of PRs (Mackay et al., 1984, Rajan et al., 1996). Normally, the predicted reactivity of PRs, assessed from solubility tests, should be validated in greenhouse and field experiments. Table 18 presents the correlation coefficients calculated utilizing data from Tables 8 (solubility) and 14 (P uptake). The coefficients are high for both the Andosol and Vertisol but low for the Alfisol. In the Alfisol, the soil pH is almost neutral and the chemical properties are not adequate for PR dissolution.

TABLE 18
Correlation coefficients between P uptake and solubility tests

Source: Truong et al., 1978.

Field evaluation

The last step of PR evaluation is to conduct field experiments in representative locations of the region or country under study. Field experiments are essential in order to provide a realistic assessment of the performance of PR in practical farming conditions. A countrywide field evaluation is valuable as the agronomic effectiveness of the PR will be affected by inherent PR properties, soil and climate conditions, cropping systems and farmer practices. However, such a programme requires a large, skilled team and a budget that few countries can afford. The National Reactive Phosphate Rock Project from Australia provides a valid example of such a programme (Sale et al., 1997a).

Several field-evaluation programmes including on-station and on-farm experiments involving multilocation trials have assessed the direct and residual effects of PRs. National programmes have been carried out in Chile and Venezuela (Besoain et al., 1999; Casanova, 1992 and 1995; Zapata et al., 1994), New Zealand (Sinclair et al., 1993c), Brazil (Lopes, 1998), Burkina Faso (Lompo et al., 1995; FAO, 2001b), Mali (Bagayoko and Coulibaly, 1995; FAO, 2001b), Togo (Truong, 1986; FAO, 2001b) and Senegal (Truong and Cisse, 1985; FAO, 2001b). For information on an international programme, the reader may refer to the recent FAO/IAEA networked research project (IAEA, 2002).

The following example illustrates the case of Burkina Faso, where the indigenous Kodjari PR had been studied extensively because of its low reactivity and its difficulty in dissolving with mineral acids (Truong and Fayard, 1987; Frederick et al., 1992). Based on geographical location, soil types, climate conditions and cropping systems, the potential areas for evaluating the direct application of this PR were grouped into three zones with the following main characteristics:

Low fertilizer rates were recommended in consideration of economic reasons and climate risks (drought). Research agronomists and extension staff estimated the fertilization rates according to the predicted response to P application at each site. Utilizing background information from past experiments and considering the available budget and technical capacities of the local staff, the limited number of treatments included: (i) a control NK without P; (ii) NPK with Kodjari PR; (iii) NPK with Kodjari PAPR (50 percent acidulated with sulphuric acid); and (iv) NPK with TSP.

A limited number of on-station-type field experiments were set up in each zone. These experiments involved a statistical design, adequate replications and technical supervision in order to obtain accurate information to serve as a benchmark for the zone. In addition, a large number of farmers were selected to participate in the evaluation programme with the same treatments, but without replications. These on-farm experiments were carried out in order to integrate the variability of farmers’ practices. Johnstone and Sinclair (1991) estimated that 40 replicates would be required in order to ensure a 90-percent probability of detecting a difference between two fertilizers that differ in P availability by 10 percent.

Table 19 presents the yield data from the two types of experiment. The average RAE values for the partially acidulated phosphate rock (PAPR) were higher than those for the PR in both types of experiments for all crops. The flooded rice cultivated in basins gave higher results, confirming that water is essential for PR efficiency in the dry savannahs of sub-Saharan Africa. A series of additional experiments evaluated the performance of Kodjari PR. Its solubility was 37 percent in FA (Table 8). The RAE values were 29 percent based on L values and 32 percent based on P uptake (Table 16). The average RAE value based on yield data was 48 percent measured in the field for the whole country (Table 19), ranging from 36 percent in the north to 60 percent in the south, depending on rainfall.

TABLE 19
Crop yield and estimated RAEs from field experiments in Burkina Faso

Sources: Frederick et al., 1992; Lompo et al., 1994.