Key roles played by phosphorus compounds in the transformation of solar to chemical energy during photosynthesis and as provider of chemical energy for biosynthesis in plants make P a singularly important nutrient element. Its low concentration and low solubility in soils (Table 1), make it commonly the key growth limiting nutrient in soils and waters (Sharpley et al., Ch. 11).

 

Table 1. Common range of amounts and concentrations of macronutrients in
A horizons of temperate region soils

	USA1	New Zealand2	Soil Solution4
	----------- g kg-1 -----------		mmol l-1
Nitrogen (NH4 NO3) Phosphorus (P) Potassium (K) Calcium (Ca) Magnesium (Mg) Sulphur (S)	0.2 - 5 0.1 - 2 1.7 - 33 0.7 - 36 1.2 - 15 0.1 - 2	-10 0.1 - 2.3 1 - 20 0.2 - 25 0.1 - 25 0.1 -2.63	0 - 7.8 < 0.0003 0.03 - 2.4 0.06 - 30 0.04 - 1 0 - 1.34

1Buckman and Brady (1972); 2New Zealand Soil Bureau (1968); 3Organic soil;
4Edmeades et al. (1985), New Zealand pastures

 

The soil parent material is the sole source of P for plant growth unless fertilizers or manures are applied. Few unfertilised soils release plant available P at rates sufficient to meet P requirements for continuous crop production, and P is commonly deficient (Sanchez and Uehara, 1980). Exceptions are younger soils formed from alluvium, glacial till and basic volcanic rocks of high P content (Wild, 1988). Even these soils may require periods of fallowing or flooding and puddling (lowland rice) to increase P availability to crops. The use of N fertilizers to boost yields has meant that even on these very fertile soils yields are often P limited (De Datta et al., 1990). Adequate P fertilisation requires large amounts of P because more than 80% of the P fertilizer may be strongly absorbed or precipitated by the soil and not be immediately available to the crop (Sample et al., 1980; Sanyal and De Datta, 1991).

In intensive agricultural systems, particularly those of western Europe, a long history of phosphatic fertilizer use has built up soil P such that crops are less responsive to current fertilizer applications. Subsidies for fertilizer costs or farm products have encouraged farmers to use heavy rates of fertilizer to remove the risk of P-limited yields. Continual research and development of sophisticated methods for soil testing, fertilizer recommendation and application have given farmers in developed countries the means to determine soil nutrient status prior to planting and amend fertilizer form and application rates for maximum crop demand. Fertilizer placement in the rooting zone of crops and use of slow-release P fertilizers have reduced the proportion of fertilizer P fixed by the soil and improved the efficiency of P use by crops. Such developments have allowed larger scale farmers in developed countries to maximise their use of other plant nutrients, particularly N.

Less expensive components of these sophisticated packages have been adopted in some developing countries to keep food production apace with population increases (FAO, 1987). Economic constraints, however, have prevented the adoption of expensive methodology for assessing and satisfying the full nutrient requirements of a crop (Runge-Metzger, Ch. 3). In general N fertilizer application, not balanced with other nutrient applications, has been adopted. Unbalanced nutrient applications are estimated to be responsible for a 20 to 50% reduction in the current efficiency of fertilizer use in Indian agriculture (FAO, 1987) and probably throughout developing countries.

The key points in any discussion of global P fertility management for the future are:

 

• Soil P deficiency has been identified as a major constraint preventing upland soils being used to produce food for the growing populations of developing countries in tropical and subtropical regions (IRRI, 1980; FAO, 1987; Stangel and Von Uexkull, 1990). Overcoming this deficiency will involve strategies to (a) maximise yields on current arable land and (b) bring additional areas of adverse P deficient soils into production (Stangel and Von Uexküll, 1990; Von Uexküll and Mutert, Ch. 9) and stop urban sprawl over agricultural land.

• Many small scale farmers in developing countries cannot afford to pay for expensive manufactured fertilizers and technology developed for their efficient use. Fertilizers must either be supplied as aid or low cost alternatives need to be developed and promoted (Runge-Metzger, Ch. 3).

• The increased urbanisation of populations creates greater dependence upon mineral fertilizers as a greater proportion of wastes from food distribution and consumption is generated remote from farms.

• Deterioration of water quality, caused by poor management of intensive agriculture in developed countries, forces reductions in fertilizer and manure use, leading to opportunities to develop low input systems of agriculture with crops designed to exploit fertilizer residues in soils. Changing fertilizer form, waste recycling and conservation tillage practices will influence methods of assessing and maintaining soil fertility.

• The decrease in the availability of high grade phosphate rocks (PR) increases the cost of manufactured soluble P fertilizer. This is increasing interest in the use of PR for direct application on suitable acid soils.

 

To design appropriate strategies to address these issues requires quantitative information on soil P status, crop responsiveness to P amendments, and agricultural practices that alter soil P status.

 

CURRENT P STATUS OF AGRICULTURAL SOILS

In uncultivated soils, the availability of P to plants is a function of the amount and form of soil P present and the rate at which it can be mobilised and transported to plant roots. The nature and stability of native P is related to the soil parent material and the extent of pedogenesis (Walker and Syers, 1976; Smeck, 1985). The rate of transport to plant roots is controlled mostly by the soil moisture regime and the soil P sorption power (or capacity) (Nye and Tinker, 1977; Frossard et al., Ch. 7).

 

The legacy of pedogenesis

Soils form by physical, chemically and biologically weathering of parent rock (Nortcliff, 1988). The native fertility of the soil is determined by dissolution of minerals and removal of nutrients through leaching or plant uptake. These processes are more rapid in wetter climates that support forest vegetation and particularly in humid tropical and sub-tropical regions. In dryer areas that support grasslands, less leaching of P and other nutrients has occurred (Stewart and Cole, 1989). As a legacy of pedogenesis, the more strongly weathered Oxisols, Ultisols and other acidic soils of the humid tropics are low in total P.

The P concentration of the parent material influences the rate of biological weathering, the rate of accession of N (Figure 1) by legumes and the rate and extent of formation of organo-mineral complexes, which are essential for fertile and productive soils (Williams and Walker, 1969; Cole and Heil, 1981). During pedogenesis, primary mineral P (e.g. apatite) is progressively mobilised and transformed into organic P (Po) and secondary mineral P (Smeck, 1973 and 1985), which is lost from the soil profile by leaching and erosion (Walker and Syers, 1976). As the rate of input of plant available P from mineral weathering decreases, to become less than the rate of P loss from the ecosystem, Po and organic matter begin to decline (Walker and Syers, 1976; Tiessen et al., 1984). Weathering changes the soil clay fraction from basic primary minerals to Al- and Fe-dominated oxides with high P sorption. Inorganic phosphate (Pi) in strongly weathered soils is associated almost entirely with these oxides (Norrish and Rosser, 1983; Frossard et al.., Ch. 7).

When soils of different parent materials with extremes of weathering are fractionated (Tiessen et al., 1984; Sharpley et al., 1987) distribution of P among fractions can be observed to follow a weathering sequence (Table 2). Total P and acid soluble (Ca-associated) Pi decrease with weathering, whereas secondary, alkali soluble Pi (Sharpley et al., 1987) increase as do stable Pi and Po. Soils of the Indian Karnataka region (Doddamani and Sesharigiri, 1988), where parent materials are similar show this sequence of P transformation (Table 3). As parent materials change, however, there can be large differences in soil P content and forms (e.g. New Zealand Ustocrepts, Table 3). Basalt provides P-rich Inceptisols in Nigeria, whereas basalt weathered to an Ultisol has lost much of the P and what is left is present in stable forms of low plant availability.

 

Figure 1. The influence of soil parent material on the relationship between the nitrogen and P content of some New Zealand topsoils (0-17.5 cm). Adapted from Walker and Adams (1958)

 

Most native forests occur on more weathered soils than native grasslands. Organic acids produced in forest litter accelerate the removal of basic elements from the root zone. Organic P derived from plants or decomposition of the litter layer are also leached (Frossard et al., 1989). In forests on highly weathered Oxisols, most available plant nutrients including P are present in the living forest and its litter layer, and the forest biomass is supported by the efficient recycling of nutrients from the litter layer atop a deeply weathered soil (Tiessen et al., 1994). Clearing and burning the forest and litter layer is the traditional method of accessing these nutrients for crop production (Palm et al., 1990). This results in accelerated nutrient loss and cropping cannot be sustained without careful soil and fertilizer management. Quantities of nutrients remaining after cropping may be insufficient to regenerate the forest, rendering soils susceptible to erosion (Lal, 1984; von Uexküll and Mutert, Ch. 9).

 

Table 2. Mean percentage of total P in soil organic (Po) and inorganic (Pi) fractions extracted with various solutions from A1 horizons of virgin calcareous, slightly weathered and highly weathered soils Sharpley et al., (1987)

Extractant	Weakly weathered calcareous (n = 41)	Slightly weathered (n = 40)	Highly weathered (n = 39)
	----------------------- % --------------------
Resin -Pi NaHCO3 -Pi NaOH -Pi H2SO4 -Pi	2 3 4 55	4 3 10 17	5 4 20 4
NaHCO3 -Po NaOH -Po   Residual Pi and Po	2 5   29	4 12   50	8 18   41
	--------------------- mg kg-1 ------------------
TOTAL P	521	512	438

 

Andisols and strongly weathered soils have high P sorption capacity and may require over 200 kg P ha-1 to raise the soil solution concentration of the plough layer to 0.2 µg ml-1, at which level P limitation of crop yield is alleviated (Sanchez and Uehara, 1980). High P sorption is a key constraint to the economic development of these soils. As soil clay content increases, P sorption increases (Figure 2), particularly if the Fe and Al oxides (or aluminosilicates) in the clay are amorphous (e.g. allophane, a dominant mineral in Andisols or Andepts). Phosphorus sorption (Frossard et al., Ch. 7) is commonly correlated to soil clay or silt content and measures of exchangeable, amorphous (oxalate extractable) and reductant soluble Fe and associated Al (reviews by Juo and Fox, 1977; Sanchez and Uehara 1980; Sharpley et al., 1984; Singh and Gilkes, 1991). A strong correlation between the various forms of extractable Al and the pH increase in a NaF-soil suspension (Perrott et al., 1976a, 1976b) has led Sanchez and Uehara (1980) and Singh and Gilkes (1991) to suggest that the NaF test be used as a P sorption index to modify estimates for crop P requirements.

If the critical soil solution P concentration required to produce an appropriate crop yield in a soil is known, P sorption isotherms can be used to estimate P requirements (Fox and Kamprath 1970; Sanchez and Uehara, 1980). Fox et al. (1980) thus provided approximate estimates of the initial P requirements of world soils (Table 4). As soil weathering increases, the P requirement increases (Table 4). Andosols are a special case with large amounts of P-sorbing amorphous aluminosilicates developing early in the weathering sequence.

Soils with the highest P requirements Andosols, Ultisols and Oxisols make up 43% of the land area of the tropics (Sanchez and Salinas, 1981) and represent areas that probably have to be brought into production to meet the future food requirements of developing countries. With low P reserves and inherent P deficiency they will require careful management, but have the potential to produce high yields if the main chemical constraints to plant growth are alleviated (Sanchez and Salinas, 1981).

Table 3. Examples of P Fractions (Chang and Jackson, 1958) in soils (mg kg-1,
% in brackets)

Soil type	pH	Al- and Fe-P	Ca-P	Occluded or residual1			Organic		Total
India (Karnataka), Doddamani and Seshagiri (1988)
Inceptisols Vertisols Alfisols Oxisols	7-8 7-8 4-7 4-7	54 (9) 55 (12) 86 (26) 144 (28)	66 (12) 89 (20) 13 (4) 6 (1)		60 (10) 55 (12) 41 (13) 70 (16)		411 (69) 257 (56) 185 (57) 267 (54)		593 457 326 498
Nigeria, Enwezor (1977c)
Tropaquepts on shale Ustropepts on basalt	4-6 4-6	49 (19) 501 (21)	13 (5) 95 (4)		80 (31) 1194 (50)		118 (46) 621 (26)		257 2387
Ustox on sandstone Udox on sandstone	4-6   4-5	40 (22) 150 (33)	9 (5)   23 (5)		53 (29) 141 (31)		81 (44)   141 (31)		184   455
New Zealand2, Williams and Walker (1969) Haynes and Williams (1992)
Ustochrept on basalt, lava Hapludand on basaltic tuff Palehumult on basalt Ustochrept on greywacke	5.4   5.5   5.1   4.9	320 (20) 190 (22) 50 (11) 212 (23)	480 (30)   0   0   25 (3)		330 (21) 190 (22) 200 (45) 119 (13)		470 (29) 500 (57) 190 (43) 435 (47)		1600   880   440   925

1Includes reductant soluble and occluded P
3Interpreted using Soil Survey Staff (1990), Dr A S Palmer, Massey University

 

INFLUENCE OF AGRICULTURAL SYSTEMS

Agricultural systems influence soil P status by the nature, intensity and frequency of: 1) cultivation and tillage, 2) crop and product removals, 3) erosion and leaching, 4) manure and fertilizer application. Arable agriculture depletes soil P (Table 5) through removal of P in the crop and soil erosion (e.g. Tiessen et al., 1983; O'Halloran et al., 1987) and smaller leaching losses (Sharpley et al., Ch. 11). Crop removals account for 1 to 40 kg P ha-1 crop-1 (FAO, 1987).

 

Figure 2. Examples of P sorption isotherms determined by the method of Fox and Kamprath (1970). Source: Sanchez and Uehara (1980).

 

Table 4. Estimated initial phosphate requirements of the major agricultural soils of the world

Soils dominated by	Area		Mean P requirement to attain solution P levels of				Fertilizer P1 required for crops with					Percentage of World’s P requirement for crops with
USDA			0.02	0.2			Low		High			Low			High
(FAO/UNESCO)			mg l-1	mg l-1			P demand
	M ha	%	- mg kg -1-				--kg P ha-1--					--- % ---
Histosols
	240	1.8	1	3			<1		<1			0.0			0.0
Aridisols (Xerosols)
	896	7.0	2	15			<1		3			0.0			0.6
Halaquepts (Solonchaks & Solonetz)
	268	2.0	7	20			1		7			0.9			0.5
Mollisols (Chernozem)
	408	3.1	7	31			1		12			0.7			1.1
Albragualfs (Planosols)
	120	0.9	16	82			<1		16			0.1			0.4
Aquepts (Gleysols)
	623	4.7	27	48			3		18			4.2			2.5
Alifisols (Luvisols)
	922	7.0	18	72			3		27			6.2			5.6
Spodosols (Podzols)
	478	3.6	13	80			2		30			2.0			3.2
Fluvents (Fluvisols)
	316	2.4	25	102			3		19			1.6			1.4
Inceptisols (Cambisols)
	925	7.0	20	132			2		50			3.9			10.3
Alfisol Ochrept (Podzoluvisol)
	264	2.0	13	220			<1		84			0.0			4.9
Vertisols
	311	2.4	30	82			6		31			4.2			2.2
Ultisols (Acrisols and Nitosols)
	1049	8.0	36	128			6		49			13.2			12.1
Oxisols (Feralsols)
	1068	8.1	161	535			23		202			54.5			48.2
Andosols Andepts
	101	0.8	486	1625			38		309			8.4			6.9
TOTAL	7678	58.4

1 Based on phosphate sorption curves from numerous sources, but mostly from unpublished data of R.L. Fox (Source: Roche, 1983)

 

As most soil P is associated with fine and light soil fractions, accelerated soil erosion leads to accelerated P loss (Tiessen et al., 1983; O'Halloran et al., 1987; Sharpley et al., 1993, Ch. 11). Cultivation normally results in the mineralisation of soil organic matter and associated Po (Table 5) (Frossard et al., Ch. 7; Barrow, 1961a; Dalal, 1977; Harrison, 1985). In general, Po mineralisation rates are more rapid in tropical soils (Table 6) where Po is an important source of available P (Adeptu and Corey, 1976 and 1977; Morris et al., 1992). In tropical soils initial net Po mineralisation rates may range from 27 to 50 kg P ha-1 y-1 for the first year of cultivation after scrub or grass fallow (Table 6), which is sufficient to provide P for two crops per year. In cooler climates where Po mineralisation rates are slower, not enough P may be mineralised during one cultivation and growing season, and a cultivated fallow may be used to provide enough mineral P, N and S for the crop and to conserve moisture. Early growing season net mineralisation rates may be 10-fold higher (Sharpley, 1985c) than those quoted for Canadian and USA soils in Table 6. However, this period of net mineralisation is followed by a net immobilisation phase as roots and crop residues with high C:P ratios decompose.

Short-term accumulation of Po probably results from immobilisation of crop residue P and any available Pi into microbial biomass and Po (Sharpley, 1985c; McLachlan et al., 1988). In this situation biomass P can be an important source of available P for the next crop. If restorative crops are used in conjunction with N inputs, either fertilizer or legumes (meadow fallow in rotation of soil 7, Table 5), then soil Po content may increase. In hotter climates where Po is mineralised rapidly, green manure crops grown to conserve N can act as effective P sources for the next crop and stimulate more efficient use of soil Pi (Tandon, 1987).

Phosphorus sorption of 8 USA agricultural soils decreased an average of -33% (range +14% to -60%) with either cultivation or cultivation plus fertilizer addition (Sharpley and Smith, 1983), presumably because Pi from fertilizer or mineralised Po decreased P sorption capacity. The exact relationship is unclear since Pi and Po may already compete for sorption sites (Morris et al., 1992). Erosive loss of fine soil particles can decrease the specific surface available for P sorption and also may result in decreased P sorption. However, if erosion is not selective of particle sizes and removes whole topsoil, ploughing will gradually mix remaining topsoil and subsoil. The inclusion of subsoil may increase P sorption (Sharpley and Smith, 1983).

Unfertilised, old arable soils (after several hundred years of cultivation in temperate climates) are characterised by low Po contents that are relatively resistant to continued cultivation while Pi declines. At Rothamsted (UK) Po mineralisation was 0.5-1.4 kg P ha-1 y-1 in unfertilised old arable soils and up to 8.5 kg P ha-1 y-1 in the 10 years following large additions of animal manure (Chater and Mattingly, 1980). More than 35 t ha-1 y-1 animal manures (FYM) were required to stabilise or increase P (Figure 3) under continuous cropping (Johnston, 1989). The largest sink for manure P was soil Pi, which was depleted rapidly after manuring ceased in 1901. Only 85 mg P kg-1 soil of the total FYM additions was Po, 75% being Pi. In 1903 only 25 mg kg-1 remained as soil Po. A Norfolk rotation containing 1 year of clover instead of a fallow increased mineralisable Po reserves from 140 to 260 mg kg-1. Thus, the inclusion of a legume increased soil C, N and Po levels (Chater and Mattingly, 1980).

 

Table 5. Cultivation and cropping effects on P and C contents of previously virgin or fallowed soils

Region		Org.	Total	Pi				Po
Soil Practice		C	P	Avail	Alkali	Acid	Total
Cultivation without P fertilizer
USA 1,2 Aeric Ochraqualf 15 cm, 60 y cotton	mg kg-1 % Change		393 -25	69 *   -50			170   18	223   -58
CANADA 3 Cryolborol 10-15cm 60 y wheat-fallow	mg kg-1   % Change	50   -32	836   -10	26**   -19	45   -4	174   -15	319   6	480   -18
NIGERIA 4 Oxic Paleudalf 15 y fallow, 2y maize	mg ka-1 % Change		466   -1		167   27	11   3	242   19	213   -35
KENYA 5 Humic Nitosol Bush to 4 y maize	mg kg-1   % Change	19   -19	505					285
NIGERIA6 Oxisol 15 cm 3 y Elephant grass 2 1/2 y mixed crop	mg kg-1   % Change	50 -12	644 -7
Cultivation with P fertilizer
USA 1 Alfic Haplorthod 50 y Potato-fallow-clover +H 2432 kg P	mg kg-1   % Change		953 102	24 78			646 170	523 -60
USA 1 Aquultic Hapludalf 31 y Corn-wheat-grass, + 1240 kg P,	mg kg-1   % Change		313   81	13   98			47   38	216 56
USA 1 8 important agricultural soils cropped 15 to 60 y	mg kg-1 % Change		574 25	168 84			240 118	328 -43
NIGERIA 4 Oxic Paleudalf 15 y fallow- 2 y maize	mg kg-1   % Change		493 1		170   29		245   22	237 -23

1 Sharpley and Smith, 1983; 2 Sharpley et al., 1989; 3 Tiessen et al., 1982,1983; 4 Adepetu and Corey, 1977; 5 Ngami, 1991; 6 Jones, 1968;
* Bray P ** Resin P

 

Table 6. Organic P mineralisation rates during the cultivation of native, fallowed and old arable soils

Site	annual mean   temp. pptn.			time of cult.	org. P	K	net org. P mineralised per season
	°C	mm	y	mg kg-1		mg kg-1	kg ha-1
CANADA 1 Blaine Lake Typic Cryoborroll	1.4	37	0 60 90	645 528 418	0.003 0.005 0.004	2.6 2.1 1.7	4.1 3.3 2.7
Sutherland Vertic cryothent Bradwell Typic cryoboroll			0 70 0 65	492 407 445 315	0.003 0.005	1.5 1.2 2.2 1.6	2.2 1.8 3.5 2.6
UK2 Old arable soil ‘Sawyers’	9.6	760	0 6 15	230 213 206	0.013 0.007 0.010	3.0 2.8 1.4	4.5 4.2 2.1
Old grassland ploughed to arable Greatfield			0 5 11 20	321 306 289 266	0.009 0.009 0.009	2.9 2.8 2.6 2.4	4.4 4.2 3.9 3.6
USA3 Mississippi Aeric Ochraqualf	16.5	1240	0 60	223 94	0.014	3.1 1.3	4.6 2.0
NIGERIA4 Oxic Paleudalf ‘2 cultivations’	26	1240	0 2	213 158	0.149	32 24	48 36
KENYA5 humic nitosol	15.9	1000	0 4	285 196	0.094	27 18	40 27

1 Tiessen et al (1983); 2 Chater and Mattingly (1980);
3 Sharpley and Smith (1983); 4 Adeptu and Corey (1977); 5 Nyamai (1991)

Wagar et al. (1980), using a sequential P fractionation (Hedley et al., 1982), found that after 1 y, most fertilizer P added to calcareous Canadian soils entered the easily extracted Pi pools. In subsequent years (2-8) these pools were depleted by crop P uptake and transformation to more stable Pi. The rate of depletion of each P pool was negatively related to the degree of chemical stability. The eventual (8 y after application) sink for P depended upon the rate of fertilizer application and nature of the soil but 30-50% remained available. On alluvial soils in India, Aulakh and Pasricha (1991) found that the largest sink for P applied annually to peanut-wheat rotations was the more strongly held NaOH-extractable Pi and Po. Increasing P application rate enhanced the formation of more stable P forms. This was seen as an inefficient use of fertilizer P in the short term.

 

 

Figure 3. Changes in the amounts of organic P (Po) and inorganic P (Pi) in an old arable soil (UK) fertilised with N P K (from 1856 to 1903) and farmyard manure (F Y M, from 1876 to 1903). (Source Chater and Mattingly, 1980)

 

Rates of net Po mineralisation in arable soils are sufficiently rapid that the product of most P added in fertilizer, manures and crop residues is Pi (Tandon, 1987). Consequently, soil tests reflecting only Pi have been successful in predicting the P responsiveness of arable cropping systems. Only in recently broken or poorly fertilised arable soils (Sharpley, 1983) does the level of Po (Table 5) greatly influence the crop response to applied P fertilizer (Jones et al., 1991; Morris et al., 1992). In general, the P fertilizer requirements of arable soils can be determined from the crop P export and the rate of loss of fertilizer P to poorly available P forms in the soil (as described by soil P sorption indices e.g. Sharpley et al., 1984).

Organic P can accumulate rapidly in P-fertilised soils growing fine-rooted grass species particularly in the presence of N fertilizer or legumes (Perrott et al., 1989; Haynes and Williams, 1992; review by Dalal, 1977). Grazing animals partition all P into dung (Barrow and Lambourne, 1962), which can have a slower decomposition rate than plant litter (Barrow, 1961b). Accumulation or depletion of Pi and Po in grazed pastures has high spatial variability depending upon animal grazing and camping behaviour (Gillingham, 1980; Saggar et al., 1990; Williams and Haynes, 1992). These animal transfers of P cause a bigger loss of P (up to 1.1 kg P per stock unit) from productive pasture areas than product losses ( 0.1 - 0.2 kg P per stock unit). Although large spatial transfers occur within the system annual total losses of P from grazed pastures are small, < 1 - 5.8 kg P ha-1 y-1, compared to those of arable crops, 2 - 40 kg P ha-1 y-1, (FAO, 1987). Mown and carted hay or silage, however, leads to losses as large as crop losses.

When calculating the P fertilizer requirements of pasture systems the product loss, soil P sorption loss, Po accumulation loss and animal transfer loss all need to be considered (Cornforth and Sinclair, 1984; Perrott et al., 1989). In unfertilised pastures, or if fertilizer is withdrawn, both the forms and amounts of soil Pi and Po need to be considered as contributing to the residual P reserves.

Plant and soil analysis are used extensively to diagnose the P status of farming systems. Adequate P nutrition at the seedling stage is important for plant development. Insufficiency of P at this stage cannot be remedied by side-dressed P because of the lack of mobility of P in soils. Therefore, pre-plant soil tests offer a better method of predicting P requirements for establishing crops. Plant analysis is more suited to permanent crops or monitoring the effectiveness of a fertilizer programme.

 

PLANT TESTS

Plant analysis provides an effective method for determining the nutrient uptake pattern of crops and confirming nutrient deficiencies, once relationships between plant part, physiological maturity and nutrient concentration are established (Moller Nielson, 1980). Phosphorus concentrations in successive crops can be logged to establish whether a fertilizer programme is having the desired effect. The DRIS (Diagnosis and Recommendation Integrated System) designed by Beaufils (1961, cited by Sumner, 1977) is probably the most suitable method for monitoring fertilizer programmes (Munson and Nelson, 1990, Westerman, 1990). Plant analysis is useful to the researcher to identify plant cultivars that are P efficient (Caradus, 1990; Hedley et al., 1993). At a given yield, the reciprocal of P concentration can be used as an index of the efficiency of dry matter production per unit P uptake. Alternatively the apparent % of fertilizer P recovered by the plant can be used to identify efficient fertilizer placement or an effective fertilizer form.

 

SOIL TESTS

In arable soils, the P nutrition of plants is limited by the transport (mostly diffusion) of Pi to plant roots (Frossard et al., Ch. 7). The rate of diffusion is determined by the soil solution P concentration (intensity factor) and the P buffer power (sorption) of the soil (Nye and Tinker, 1977). Fertilizer raises the P concentration in soil solution, which increases the amounts of Pi sorbed or precipitated on soil surfaces (Table 5). Both effects increase the rate of P diffusion to plant roots.

Indices of the abilities of soils to supply P to plants can therefore be determined (a) by extractive tests that measure the concentration of P in solution and the amount of P in a labile form or (b) from the soils phosphate (buffering) sorption characteristics. In both cases P response trials are required to establish the relationship between crop yield and the P supply index. These relationships vary with the crop, climate and soil type. Thus, large numbers of trials are required before soil tests can be used to estimate P fertilizer requirements with any accuracy (Kamprath and Watson, 1980; Tandon 1987; Kamprath, 1991).

 

Various tests have been developed in different countries to suit the forms of P present in their agricultural soils (Table 7)(Anon. 1982). The form of soil P extracted by each test is determined by its solution pH and the reaction of the ions present in the extractant with sorbed or mineral P. For instance, the HCO3- and OH- in the bicarbonate extract promote the desorption of P from CaCO3 and Fe and Al hydrous oxide surfaces. In the strongly weathered Nigerian soils, dominated by Fe-P, Al-P and Po with little Ca-P, bicarbonate is an effective extractant of P but the amount is correlated poorly with identifiable soil P fractions. Bray 1- and Truog-extractable P are highly correlated to Al- and Fe-P in such soils.

Ae et al. (1990) and Hedley et al. (1993) have shown that pigeon pea and upland rice, respectively can mobilise Fe-P in their rhizospheres. In such situations, Bray-1-, Truog- or resin P could be expected to be effective indices of P availability. Where data on soil P depletion by plants are not available, Sharpley et al., (1984) have used resin extraction results from calcareous, weakly weathered and strongly weathered soils to rank the suitability of different soil P tests. Kamprath (1991) summarises their results as follows: The Olsen extractant is suitable for calcareous and weakly weathered acid soils and less suitable on strongly weathered soils where Bray 1 and Mehlich tests are more appropriate.

Water extraction techniques (Sissingh, 1971) are mostly suited to well fertilised or calcareous soils because unfertilised acidic soils are likely to yield P concentrations in the extract below detection limits (Frossard et al., Ch. 7). Roche et al. (1980) used anion exchange resin as the preferred method to classify tropical soils into groups of P sufficiency or deficiency. Phosphorus extracted by the anion resin method of Sharpley et al. (1984) and Sharpley (1991) is strongly correlated to P extracted by Fe and Al oxide impregnated filter paper, and compares favourably as a soil P availability index (Sibbesen, 1983).

 

Soil test (reference)	pH	Soil: sol’n ratio		Extractants	Time
Bray 1 (Bray and Kurtz, 1945) Bray 2 (Bray and Kurtz, 1945)	3   1	1:7   1:7	0.03 M NH4F + 0.025 M HCl 0.03 M NH4F + 0.1 M HCl		1 min   40 sec
Colwell (Colwell, 1963)	8.5	1:20	0.5 M NaHCO3		16 hr
Egner-Riehm (Grigg, 1965)		1:20	0.1M NH4-lactate + 0.4 M Acetic acid		4 hr
Fe paper (Menon et al., 1991)	nat	1:40	Fe hydroxide impregn. filter paper		16 hr
Lactate (Holford et al., 1985)	3.7	1:50	0.2 M Ca-lactate + 0.01M HCl		1.5 hr
Mehlich (Holford et al., 1985)	1.0	1:4	0.025 M H2SO4 + 0.05 M HCl		5 min
Truog (Truog, 1930)	3.0	1:200	0.002 M H2SO4 + 0.3% (NH4)SO4		30 min
Olsen (Olsen et al, 1954)	8.5	1:20	0.5 M NaHCO3		30 min
Pw test (Sissingh, 1971)	nat	1:2 then 1:60	Distilled water standing then shaking		22 + 1 hr
Resin s (Amer et al., 1955) (Sibbesen, 1977) (Van Raij, 1986)   (Saggar et al, 1990)	nat     nat     nat	1:100	Anion exchange resin beads Anion + cation exchange resin beads Anion + cation exchange membrane		variab.     16 hr

nat. = pH of soil

 

Correlation and calibration of soil tests against crop yield pose a number of problems (reviews of Dahnke and Olsen, 1990; James and Wells, 1990; Fixen and Grove, 1990). The largest errors in soil test values are associated with soil variability and sampling technique. Designing appropriate soil sampling techniques requires a detailed knowledge of the effects of soil parent materials, pedogenesis, landscape and farming system on the distribution of labile soil P with respect to the crop rooting zone, timing of the previous fertilizer application and placement of the future application (James and Wells, 1990; Saggar et al., 1990). A way of grouping similar land areas and/or coping with soil heterogeneity is required to reduce (or measure) the spatial error involved in estimating a representative soil test value (James and Wells, 1990).

The ideal nature of crop yield response to increasing soil P test value is curvilinear, rising to a maximum yield plateau where P no longer limits plant growth. Differences in climate, other nutrient availability and soil characteristics cause variations in the maximum yield between years and sites. Yield data at each site can be transformed into percentages of maximum yield at each site to allow comparison of data between sites. Curvilinear or linear response-plateau models must be fitted to the relative yield data to test the predictive power of the soil test. The choice of model influences the description of the data and the soil test value that is required to alleviate P deficiency (Anderson and Nelson, 1975). Common models used are : Mitscherlich Y = A (1 - e-cx); Quadratic Y = A + Bx + Cx2; Logarithmic Y = a + b log x; Sigmoidal Y = a/(1 + bcx) where (0 < c < 1); and the Linear-plateau model Y = A + Bx, where x = 1, if x is above critical level and where x = < 1 below critical level.

An appropriate model will give the highest coefficient of determination. and can be used to partition soil test values into groups where low, medium or high response to fertilizer P is expected. If fertilising for maximum yield, a critical soil test value above which 90% of maximum yield is expected can be chosen. This is commonly called the 'external P requirement' of the crop. The fertilizer requirement to raise a soil test within a certain group to the external P requirement of a crop can be calculated from the trial data. The soil test can then be used as a diagnostic tool for fertilizer requirements, and is portable between similar soils, climates and crops. For arable crops this is the most common method of making fertilizer recommendations.

Unless field trials are conducted under uniform soil and climatic conditions coefficients of determination may be low for even the best model (Figure 4). The scatter of data usually results from soil heterogeneity (James and Wells, 1990) and yield variance caused by P nutrition interactions with plant water use (e.g. Tandon, 1987; Halford et al., 1988), soil P buffer power (e.g. Halford et al., 1985; Fageria et al., 1991), the nature of residual forms of P (Vijaj Kumar et al., 1991) and other factors affecting plant growth and the measurement of yield (Dahnke and Olsen, 1990). Variance associated with permanent pasture yield is higher at lower soil test values, partly because variations in plant water use and soil P buffer power have a greater impact on the P supply to plants when labile soil P is low (Saunders et al., 1987). There is considerable risk associated with making fertilizer recommendations from such groupings of data. In New Zealand, soil groups are separated by their ability to sorb phosphate in a laboratory test (Saunders, 1965) or by a field assay based on NaF reaction (Perrott et al., 1976a, b). Higher Olsen P tests are required to maximise yield on high-P sorption soils (Cornforth and Sinclair, 1984).

 

 

Figure 4. The relationship between P uptake by spring wheat in five different years and Olsen P test. Note the significantly lower critical Olsen P (10) selected by the Cate-Nelson approach than that (16) selected by 90% of maximum yield described by Mitscherlich equation (source of data Halvorson and Black, 1985).

 

In tropical soils (Sanchez and Salinas, 1981), the Cate-Nelson (1971) and Nelson and Anderson (1977), linear-plateau model for interpreting soil test-crop yield data is preferred because it tends to give lower critical soil test values and more economical fertilizer recommendations than curvilinear models. To produce practical soil test recommendations from highly variable data Saunders et al. (1987) proposed a method of fitting models to the probable minimum relative yield (PMY). This is a chosen plant yield which will only be less than the mean relative yield produced by a particular soil test value on a chosen percentage (% n) of times. Percent n may vary according to the risk the agronomist and farmer are prepared to take. Developing such techniques for variable data strengthens the argument for more fundamental research into physical, biological and chemical factors controlling the growth of plants in soils.

 

In P-fertilised soils, Fox (1981) argues that variations in the relative yield of crops can best be explained by the concentration of Pi in the soil solution rather than the quantity of labile P (extractive method) on the soil surface. Two factors control relative yield: (a) the initial concentration of P in soil solution that will support maximum plant yield, i.e. the critical solution P concentration or external P requirement, and (b) the ability of the soil to buffer the solution Pi during plant growth. By conducting plant growth experiments and P sorption studies with a range of fertilizer levels it is possible to describe the relationships between added fertilizer P, soil solution Pi concentration and relative yield, and thus estimate fertilizer. Different plant species have different critical solution P concentrations. Fox (1981) and Fox et al. (1990) show how P sorption curves constructed for different soils can be used to assess the preliminary fertilizer requirements which, for particular crops, are relatively unaffected by soil texture and mineralogy. While the method is useful for research purposes, the labour and time involved in constructing P sorption curves make it too expensive for routine soil testing laboratories. A modification of the method for determining P sorption curves (Dear et al., 1991) can be helpful in this regard.

 

Factors affecting the accuracy of extractive and intensity/buffer capacity methods

Both methods for assessing P fertilizer requirement of crops rely on the development of field calibration curves. Their effective range of use is constrained to situations where the variance in yield caused by interactive effects of other nutrients, management techniques and climate on the relationship between plant yield and soil P status is similar to that accommodated in the original calibration curve.

External factors affecting plant P demand and supply, change the shape of calibration curves by influencing plant root growth independently of soil P status. They include variations in climate, and nutrient interactions with soil amendments, such as lime, and methods of cultivation and fertilizer application. Soil moisture stress inhibits plant root growth and, in addition, decreases the diffusion of Pi to root surfaces and the mineralisation of Po. In drying soils, P limitation occurs earlier than N or S limitation (Kilmer et al., 1960) particularly on high P-sorbing soils because decreases in soil water content may increase soil solution NO3 and SO4 concentration, but P concentration may change little or even decrease (Nye and Tinker, 1977). As soil water content increases, P diffusion rates to roots increase at least cubically (Nye and Tinker, 1977), which may allow the yield plateau to be reached with lower solution and labile P concentrations. Thus, soil moisture stress commonly decreases the critical concentration of an extraction test required for 90% of maximum yield. Critical P concentrations for the Olsen test may vary three fold from season to season even when the same cultivar, soil and cropping practice are used (Figure 5).

Banding or seed placement of P fertilizer can increase the relative P supplying power. A larger amount of soluble P applied to a smaller soil volume will increase solution P concentration and decrease the soil's P buffer power (Barber, 1984). Banding and placement of P also lead to greater variability and errors in soil testing (James and Wells, 1990). Separate calibration curves for soil tests should be used for various P placement methods, irrigation or reduced tillage (which concentrate P in the surface soil). Further research is required to identify appropriate environmental, plant and soil measurements that can be used by modellers to parameterise computer simulations of plant nutrition and growth in soils. In this way soil test indices could be made more portable.

 

 

Figure 5. Relationships between P uptake by spring wheat and Olsen P test for five years with different amounts of available water. Note differences in critical Olsen P (----) test level for each year (source of data Figure 4).

 

Perrott et al. (1993) found general agreement in the literature that soil tests extracting P by desorption, near the low ionic strength found near growing plant roots and at native soil pH, prove to be the most robust indices of plant-available P where soil and fertilizer form and soil pH change. Such tests currently being evaluated are the combination of cation and anion exchange resin (Van Raij, et al. 1986; Saggar et al., 1992; Qian et al., 1992) and the iron-hydroxide impregnated filter paper test developed by Van der Zee, et al. (1987) and modified for use with soils by Menon et al., 1991 (which is highly correlated with simple resin extractable P in most soils). Van Raij and Quaggio (1990) found that liming increased leaf P concentration and P uptake for some crops grown on a Brazilian Oxisol, and that double resin-extractable P was the only soil test closely related to changes in leaf P. The greatest obstacle to accurate soil testing, however, remains in the field sampling and in the lack of understanding by farmers and advisers of the importance of field variability.

 

Static equations are often used to convert soil test and crop yield information into fertilizer requirements. The extractive and intensity/sorption methods of estimating soil P status have been used for estimating the P fertilizer requirements of crops (e.g. Greenwood et al., 1980; Holford et al., 1985) and pastures (Dear et al., 1991) using pre-determined relationship between P fertilizer application rate and increasing indices of soil P status (Sharpley et al., 1984; Sissingh, 1971; Dahnke and Olsen, 1990). Alternatively, the relationships between initial soil P test values and the effect of fertilizer addition on yield is used (Dahnke and Olsen, 1990), which gives a family of response curves similar to those showing the effect of soil moisture in Figure 5. At a single site, the relationship between a pre-crop soil test value and final crop yield integrates the process involved in P supply during the growing season. When such families of curves are produced on different sites, very variable data are obtained that are difficult to interpret (Saunders et al., 1987). The relationships also lose accuracy over time as soil P sorption and Po mineralisation characteristics change with tillage, erosion and successive removal of P by crops (Sharpley and Smith, 1983; Sharpley, 1983). The usual way to accommodate such changes, and to improve the predictive power of a soil test, is to determine the way in which the coefficients of these models change with soil characteristics, climate (Figure 5) and previous fertilizer history (Helyar and Godden, 1977a, Bennett and Bowden, 1976).

A different approach to determining fertilizer requirements is to assess and model the loss of plant available P occurring at different levels of productivity of a certain farming practice. This "steady state" approach is used for determining the fertilizer requirements of grazed pastures in New Zealand (Cornforth and Sinclair, 1982) and involves a sound knowledge of mechanisms that net immobilise or cause losses of plant available P. These mechanisms vary with each soil type, landscape unit, climate and animal production enterprise (Cornforth and Sinclair, 1982; Sinclair and Cornforth, 1984). The model calculates the rate of annual P loss, which becomes the annual fertilizer P 'maintenance requirement'. For a given farming operation and location, P loss (equivalent to maintenance fertilizer P) is related to animal productivity from which pasture dry matter production can be calculated. In a P-limited situation, relative pasture production is related to Olsen soil test values. Therefore, the Olsen P required to shift to a different steady state can be calculated. A fertilizer recommendation can be made knowing the farmer's production goals and the risk associated with a fertilizer application giving the required growth response at a certain Olsen level (Saunders et al., 1987). Over time, farmer records are established that give more accurate relationships between ‘maintenance P application’ and ‘change in soil test'. Current research is directed at making the model dynamic so that periods of fertilizer withdrawal and the effect of increased P application can be predicted. (A Metherell, pers. comm.).

Dynamic modeling of P in soil plant systems has to date only been used by Jones et al. (1991) and in the preliminary work of van Noordwijk et al. (1990). Dynamic models, however, must be developed to design P fertilizer strategies that meet environmental requirements, and avoid P loss to surface and drainage waters (Sharpley et al., Ch. 11, Van Noordwijk et al., 1990). The Jones et al. (1984) model has been incorporated into an environmental impact model (Erosion-Productivity Impact Calculator, EPIC, Williams et al., 1984). In this combination it simulated accurately the changes in soil P form and P fertilizer requirements of various cereal-fallow rotations in contrasting soils of the USA Great Plains (Jones et al., 1991). One particular advantage of this dynamic model is its power to predict how changes in cropping practice will influence future fertilizer requirements or conservation practices. For example, pasture rotations tend to transform Pi from residual fertilizer into Po. A dynamic model could predict the period in pasture required to provide sufficient mineralisable Po to supply adequate P to a subsequent annual arable crop. The need for soil testing is reduced and measures of crop yield and performance in relation to readily recorded climatic factors can be used to pin-point years when models may need adjusting. In addition, farmers can examine cropping options that maximise the use of residual fertilizer P in the soil. A disadvantage of more complex dynamic models is the investment required to measure the many soil characteristics necessary to set initial model parameters. Sharpley et al. (1984) have described various techniques for reducing this investment.

Perhaps many traditional farmers, through generations of trial and error, passed on an understanding of nutrient dynamics, as expressed for example in the Norfolk rotation. Extensive use of manufactured fertilizer has increased yields but has probably dulled this skill. Future generations of farmers will again require such knowledge to manage limited soil and fertilizer resources and adapt to a changing climate. As the mechanisms determining successful crop rotations become mathematically describable, computing techniques will allow farmers and scientists to reduce the risk in choosing practices for farming diverse sets of soils and climatic conditions.

 

 

Having determined the fertilizer requirement, an appropriate fertilizer form can be chosen. Most fertilizer requirement assessments have involved water-soluble P fertilizers in their calibration. If less soluble P fertilizers are to be used, their agronomic effectiveness relative to standard soluble fertilizers must be known. Methods of assessing agronomic performance of fertilizers have been reviewed by Barrow and Bolland (1990), Bolan et al. (1990) and Chien et al. (1990).

Modern P fertilizers are almost entirely derived from geological deposits of apatite-containing phosphate rock (PR), with small deposits from recent guano being exploited locally (Cook et al., 1990). The majority of P fertilizer sold is manufactured by completely acidulating PR with various mineral acids to produce more soluble phosphates (Table 9). Fertilizer quality has been assessed using the total P content and its solubility. Different extractive tests are used to determine solubility. The two most common tests are 1 M neutral ammonium citrate (NAC) and water.

Where the cost of 'fast-release' P fertilizer (high NAC solubility) represents a high proportion of the crop or animal product value, cheaper, less soluble P fertilizers such as directly applied PR, partially acidulated PR and thermally treated PR are alternatives. For some soil/crop situations less soluble P fertilizers can be cost effective, depending on its agronomic effectiveness relative to (RAE) a standard soluble fertilizer, its cost of purchase and application and the increased income projected from higher yields. Transportation costs can be a large part of the equation. As distance from factory increases high analysis manufactured fertilizers may become more cost effective than low analysis fertilizers. In remote areas a local PR source with low agronomic effectiveness may be more cost effective than an imported soluble fertilizer. Direct-applied PR is most effective on acid soils (Khasawneh and Doll 1978; Rajan and Upsdell 1981). Since P fertilizers exhibit a lengthy residual influence on crop yields (Halvorson and Black, 1985), long-term field trials are required to establish the true cost effectiveness of a fertilizer (Sinclair et al., 1990).

 

Table 9. Some characteristics of soluble P fertilizers

 

Fertilizer and acronym	Chemical formulae Primary P compounds	Nominal Grade (P2O5 basis)
Single super phosphate (SSP)	Ca(H2PO4)2.H2O	0-20-0
Triple super phosphate (TSP)	Ca(H2PO4)2.H2O	0-45-0
Monoammonium phosphate (MAP)	NH4H2PO4	11-48-0
Diammonium phosphate (DAP)	(NH4)2HPO4	18-46-0
Nitric phosphate (NP)1	CaHPO4, NH4H2PO4 Ca(NO3)2	16-20-0
Urea-ammonium phosphate (UAP)	CO(NH2)2, (NH4)2HPO4 NH4H2PO4	28-28-0
Ammonium polyphosphate2	(NH4)3HP2O7.H2O	10-34-0

1 Water solubility varies but generally is about 40% of total P
2 Fluid fertilizer, all others are solids

 

Much research has been conducted on the relationship between the level of water-soluble P in fertilizers and the immediate P availability. The EC requires P fertilizers to contain at least 93% of the ammonium citrate-soluble P as water-soluble P (Anon, 1976). However, greenhouse results indicated similar P availability in products that vary in their water-soluble P content from 80 to 97% (Mullins and Sikora, 1992), and 90% of the maximum crop yield could be attained with products containing 63% water-soluble P. Field research is needed to determine the level of water-soluble P required for maximum effectiveness under varying crop-soil-climate conditions.

Fluid fertilizers, such as ammonium polyphosphates, are completely water soluble and are widely used as starter fertilizers in the USA. These products are easily placed near the seed in row crops but also may be spayed or dribbled on the soil surface and subsequently incorporated into the soil. Micronutrients also may be easily incorporated into such fertilizers (Achorn and Faulkner, 1984). Availability of P applied as polyphosphates is similar to that of orthophosphates (Sample and Akin, 1984) because about half of the P in polyphosphate fertilizers is in the orthophosphate form (Terman and Engelstad, 1966), and polyphosphates hydrolyse rapidly in most soils.

Water solubility facilitates diffusion of P, which increases the probability of plant root interception of soil affected by fertilizer P. Early crop response usually is greater from water-soluble P fertilizers when root distribution is limited. This is important with short-season crops such as vegetables, especially in cool, wet soils (Engelstad and Terman, 1966).

A maximum of only 25-30% of the P applied to soils is recovered by the first crop, even under ideal conditions. The remainder is retained in the soil in forms which are slowly available to succeeding crops. Water-soluble phosphates react rapidly with soil constituents to form less-soluble Al and Fe phosphates in acid soils, or Ca phosphates in neutral to calcareous soils (Sample et al., 1980).

 

Acidic soils are suited to direct application of ground, unacidulated PR. Appropriate use requires an understanding of the properties of PRs and soils (Anon., 1978; Bolan et al., 1990a).

Phosphate rocks differ widely in their mineral constituents - apatites, crandallites, millisites, silica and calcite (Doak et al., 1965; Khasawneh and Doll, 1978). The most important fertilizer mineral is apatite (Lehr, 1980; Lehr and McClellan, 1972; Frazier and Kim, 1989). The greater the degree of carbonate substitution, the smaller the crystal size and the more water soluble the apatite (McClellan and Gremillion, 1980).

The chemical 'reactivity' of PRs can be assessed by their solubility in selected chemical extractants. Two percent citric acid is used in New Zealand, 1M ammonium citrate at pH 7 (NAC) in Australia and the USA and 2% formic acid in the EC. Sechura (SPR), North Carolina (NCPR), Gafsa, and Arad PR are among the most reactive rocks, whereas Nauru, Tennessee, Florida, Christmas Island and Duchess PR are among the least reactive (Leon et al., 1986; Quin et al., 1965, 1987)(Figure 6). Sequential extractions with citric and formic acid may be required to overcome the 'masking effect' of a high CaCO3 content (Mackay et al., 1984c) before the reactiveness of the apatite can be assessed. Whereas PR solubility in 2% formic and 2% citric acids are strongly related (Figure 7), those between formic acid and NAC are less so, particularly in the less reactive region (Figure 7). Standardisation of the tests used for quality control of RPRs worldwide is needed.

Rock phosphate should be ground to fine powders (< 150 µm diameter) to be agronomically effective (Khasawneh and Doll, 1978; Kirk and Nye, 1986a). Although fine grinding produces a faster acting fertiliser, finely ground material is difficult to spread and needs to be granulated or pelletised (Stephens and Lipsett, 1975; Mackay et al., 1980) to produce an acceptable fertilizer.

The phosphate component of a PR dissolves congruently in moist soils according to the reaction shown in Figure 8. The forward reaction requires an adequate supply of moisture and acid (H+) (Kanabo and Gilkes, 1987a) and the removal of Ca2+, H2PO4- and F- from the reaction site through diffusion, surface exchange and adsorption (Wilson and Ellis, 1984; Bolan and Hedley, 1989). Kirk and Nye (1986a) used a mechanistic model based on the premise that diffusion of ions away from the particle surface, but not the surface reaction itself, was the rate-limiting step in PR dissolution (Kirk and Nye, 1986b). The poor performance of PR in many field experiments may be explained in terms of inadequate sink strengths (Bolan et al., 1990a). Apart from moisture (Kanabo and Gilkes, 1988), factors most likely to limit PR dissolution are insufficient pH buffering capacity and an increase in Ca2+ activity in the soil solution around the RPR particles due to a lack of Ca-exchange capacity (Figure 8).

A high P sorption capacity of soil surfaces reduces P concentration around a dissolving RPR particle and enhances dissolution (Kirk and Nye, 1986b; Symth and Sanchez, 1982; Kanabo and Gilkes, 1987b). If the sorbed P is not readily available to plants, the agronomic effectiveness of the RPR may be no higher than that of TSP or SSP (Hammond et al., 1986; Bolan and Hedley, 1990).

 

 

Figure 6. Relationship between % of P extracted from a number of phosphate rocks in 2% formic acid, and their relative agronomic effectiveness RAE (Rajan et al., 1992). PR Key: Sechura, S; North Carolina, C; Gafsa, G; Youssaffia, Y; Arad, A; Khouribga, K; Jordan, J; Zin, Z; Mexico, M; Nauru, N; Florida, F. Upper case - unground, lower case - ground.

 

 

Figure 7. Relationship between percentages of PR-P extracted in 2% Formic acid and 2% Citric acid, or neutral ammonium citrate (underlined) (NAC) (Rajan et al., 1992) key as Figure 6.

 

Reactive PRs (RPR) have generally proved to be more effective in supplying P to perennial crops and permanent pastures in New Zealand than for cereal crops and pastures based on annual legumes in Australia (review: Bolan et al., 1990a). Rock phosphates are common fertilizers for plantation crops (Ling et al., 1990; Pushparajah, et al., 1990). High effectiveness with these crops partly reflects the acidic nature of the tropical soils on which they are grown. A number of studies have shown improved residual value of PR's over TSP in the second or third cropping season in Indonesia (Tambunan, 1992; Harris et al., 1985), Philippines (Briones and Vincentre, 1985), Brazil (Fageria et al., 1991) and Kenya (Bromfield, 1981). Unreactive PR's are as effective as TSP in very acid soils (Hakim and Moersidi, 1982; Fageria et al., 1991). With perennial crops and permanent pastures, slow release P can have considerable economic advantages in the long term (Sinclair et al., 1990).

High root density has a significant effect on RPR dissolution, particularly when roots are acidifying the rhizosphere (Kirk and Nye, 1986a; Nye and Kirk, 1987; Bolan et al., 1990a). Legumes can acidify their rhizospheres when they are actively fixing N2. The effect of legume rhizosphere acidification may explain why Sechura and Chatham Rise RPRs stimulate white clover growth as much as, or more than SSP at comparable rates (Gregg et al., 1981; Mackay et al., 1984).

In general, plants taking up NH4-N (Haynes, 1990), or those with alkaline uptake patterns (an excess of cations over anions absorbed) such as buckwheat and brassicas, and N2-fixing legumes should use RPR more effectively than others (Aguilar and van Diest, 1981; Bekele et al., 1983).

 

 

Figure 8. Schematic diagram showing the rate-limiting factors and variables determining phosphate rock dissolution

 

Tambunan (1992) accurately modelled the dissolution of North Carolina and Moroccon PRs over periods of 180 to 540 days. Using this model the important effect of soil water content on RPR dissolution can be determined. As surface soil dries out rapidly after rain it is important to incorporate RPR (and other fertilizers) into the plant root zone that has the highest soil water content for the longest period. Low RAEs of 10-20% obtained for RPRs on light textured soils in south-western Australia (Bolland et al., 1988) have been attributed to the short winter ‘wet season' and the drying of the surface soil between rainfall events (Bolland et al., 1986). Climatic changes that reduce the yield potential but not PR dissolution effectively make PR-based fertilizers more effective compared with soluble P fertilizers. For example, PR was an effective source of P for flooded rice in the wet season (low photosynthetic potential) but was less effective in the dry season (De Datta et al., 1980). Tambunan's modification of the Kirk and Nye model allows soils, crops and climates suited to RPR use to be selected on the basis of soil chemistry, crop type, root density, and local climatic information. It could be used to select low-cost P fertilizers for certain regions and farming practices.

 

Phosphate rocks containing a high percentage of Fe and Al associated with the apatite or containing Fe and Al minerals may prove difficult to acidulate. Low temperature (450-600oC) kiln processes have been developed to improve solubility and agronomic value of such materials (Lehr, 1980). High temperature (1000oC to 1450oC) fusion converts apatitic ores mixed with Na, K or Mg silicates, halides or carbonates into more soluble Na, K and Mg phosphates. The fused products are much less soluble than acidulated apatites. Because of high energy cost, fusion generally does not represent a less expensive technology for processing raw materials into fertilizers, and limits its use in developing countries. Basic slag (4-6% P) a by-product of the steel industry has similar chemical properties. It has been used extensively in countries with steel industries or transported back to countries where the ore originated.

 

Fertilizers containing both fast-release (water or NAC soluble) and ‘slow release' (PRs or thermal phosphates) P are ideal when there is a need for immediately available ‘starter' P and where acidic soil conditions allow continued slow release of ‘maintenance' P. Mixtures of fast and slow release P, or partially acidulated phosphate rocks (PAPR) (review by Bolan et al., 1990b) can be produced by incomplete acidulation of PRs with limited acid addition (Braithwaite, 1986) or by mixing fully acidulated fertilizers and unacidulated PRs.

In the short term, the agronomic value of directly acidulated PAPRs and SSP-RPR mixtures is influenced by their water-soluble P content, but in the long term the chemical nature of the non-water soluble P residue, which includes secondary P products such as dicalcium phosphate (DCP) and unacidulated PR is important. It is usual to measure their solubility in water or NAC and the solubility of the insoluble residue in either 2% formic or citric acids. Bolan et al., (1990b) have reviewed manufacturing variables that influence solubility.

The partial acidulation of PR offers a cost saving by reducing the quantity of acid used, especially for H3PO4 which accounts for up to 83% of the raw material costs in TSP manufacture (Braithwaite, 1986). The most important products of manufacture in terms of agronomic value are CaSO4, Ca(H2PO4)2 (MCP) and residual PR, if H2SO4 or mixed acid is used. Phosphoric acid acidulated PAPRs contain only minimal amounts (<2%) of S, but more total and soluble P. Elemental S can be incorporated at the mixing stage into H3PO4 acidulated PAPRs. In this way, high P analysis and cost effectiveness can be maintained.

Partial acidulation is a possible route for use of low-cost local PRs, which are not suitable for direct application because of their low solubilities (Hammond et al., 1986). Reviews by Hagin (1985) and Hammond et al. (1986) indicate that PAPRs made from less reactive PRs generally are less effective and effectiveness varies with soil type. H2SO4 acidulation is often the most economic and provides a valuable S supply (Friesen et al., 1987 a, b) both PRs of low and high reactivity have been acidulated. Partially acidulated PR made from relatively unreactive PR was as or more cost-effective than SSP on West African Ultisols and Alfisols (IFDC, 1985), in different climatic zones and soils in Colombia (Hammond et al., 1986; Anon, 1985)., and on both acid (De Datta et al., 1990) and calcareous (Dash et al., 1981; Anon, 1985) soils in India. In Europe, Israel and New Zealand (review by Bolan et al., 1990b), PAPRs made from H3PO4 (and/or H2SO4) acidulation of very finely ground (ca. 80% < 72µm) RPRs are tested, because they have higher P analysis than SSP and reduce transport and spreading cost.

The objective for PAPRs made from RPR has been to provide P fertilizers that are suitable replacements for basic slag and SSP or TSP (reviews: Bolan et al., 1990b; Davis, 1984). On permanent pastures or forage crops, on acidic soils, acidulated PAPRs were at least as effective as TSP, particularly in the second year (Rajan, 1987a). In long-term P fertilisation (5-13 y) for wheat, barley, maize, oats and sugar and fodder beet on a wide range of soil types, only sites with pHs d 6 showed yield responses greater than 10%.

Mixtures of TSP with PR or RPR (Chien et al., 1987) are unlikely to become popular because PR addition lowers the total P, and these mixtures contain <2% of S. In SSP-RPR mixtures, RPR addition increases P analysis and provide adequate amounts of S. SSP-RPR mixtures can contain 50% of the P as slow release P of higher residual value. The addition of small amounts of water soluble P to low reactive PRs helps to establish crops in severely P-deficient soils, thereby achieving more efficient utilisation of the PR (reviews Hammond et al., 1986, and Bolan et al., 1990b). In Western Australia, a mixture of SSP, NCPR and elemental S was very effective in reducing the leaching losses of P when compared with soluble fertilizers in sandy soils (Bolland, 1985). Animal transfer losses of P from grazed pasture could also be reduced (Mackay et al., 1987). In developing countries where attempts have been made to use indigenous PR sources, addition of SSP is the least expensive way of providing water-soluble P for crop establishment. The substitution of PR for SSP can improve the cost benefit ratio of P fertilisation, because the unit cost of P in the indigenous PR can be less than half that in SSP. On unit P basis, an SSP-NCPR mixture is currently the cheapest P fertilizer to manufacture in New Zealand.

 

MANAGEMENT STRATEGIES TO USE P EFFICIENTLY

 

The relief of a nutrient deficiency by applying fertilizer will increase plant growth and alter the requirement for other nutrients (Table 10). The yield increases gained from use of N fertilizers (Tandon, 1987; De Datta et al., 1990) have led to inefficient N use because of induced P deficiency. Similarly, increased soil P status decreases the plant uptake of other nutrients, which then become key factors in limiting yield and efficiency of P fertilizer use (Table 10). Interactions between soil acidity, lime and phosphate availability have been reviewed by Edmeades et al (1990). Use of RPR on acidic grassland soils can reduce the development of soil acidity (Sinclair et al., 1993).

 

Table 10. Examples of nutrient interactions with efficiency of P fertilizer use and situations leading to their occurrence

Nature of interaction		Mostly likely occurrence	Reference
Positive
P x N	Non-legume crops on old arable soils, particularly crops that do not immediately follow legume or pasture rotations		Metherell et al. (1980) Tandon (1987) Dibb et al, (1990) De Datta et al. (1990)
P x K	All crops, particularly root crops grown on soils containing levels of K. Strongly weathered soils. Peat soils.		Mengel & Kirkby (1982) Tandon (1987) Dibb et al. (1990) De Datta et al. (1990)
P x S	Soils low in organic matter. Soils with low anion retention. Peats. High rainfall areas. Short season shallow rooted crops not able to use subsoil S. or with high S demand e.g. legumes and brassicas		Maynard (1993) Tandon (1987) Dibb et al (1990) De Datta et al. (1990)
Negative
P x Zn	High rates of P fertilizer on neutral to alkaline soils or strongly weathered soils. Organic soils.		Mengel & Kirkby (1982) Tandon (1987) Dibb et al. (1990) De Datta et al. (1990)
P x Fe	High pH soils low in available Fe		Maynard (1993) Tandon (1987) Dibb et al (1990) De Datta et al. (1990)
P x Cu	High pH soils, Organic soils Interactions in animal nutrition in soils where molybedenum and sulphur -intake into the animal are increased by P fertilizer application		Dibb et al (1990) Grace (1983)

 

Soil organic matter management (conservation tillage, use of mulches, manures and crop residues and rotations with restorative crops such as legumes and green manures) plays a key role in efficient utilisation of soil and fertilizer P, especially on acid, P deficient soils of the tropics. Soil organic matter content interacts with most soil fertility parameters through improvement of soil structure, provision of N, S and P, increased cation exchange capacity, increased soil water holding capacity and alleviation of Al toxicity (von Uexkull, 1987). All these factors impact on plant root growth. Increased root length or mycorrhizal extension has the largest effect on plant P uptake efficiency (Hedley et al., 1994).

 

Research into methods of fertilizer placement (Tandon, 1986; Dibb et al., 1990) has four main objectives: (i) to increase the efficiency of fertilizer use by plants; (ii) to increase crop yields; (iii) to reduce the cost of fertilizer application, and (iv) to prevent plant injury by fertilizers

For small grains, application of starter P with the seed is now a standard practice which reduces the quantity of P required and thus the cost. Deep-banding of starter P has been successful in soils prone to drought (Tandon, 1987; Murphy and Beaton, 1988; Dibb et al., 1990). In general, banding and drilling is more effective than broadcasting on soils with low P availability. Results are often variable, however, depending on the amount of available soil P, soil P sorption capacity, water availability and crop type. The optimum volume of the fertilised zone appears to range from 1 - 24% of the total cultivated depth and should be in close proximity below and to the side of seedlings and seeds to maximise exposure to roots and minimise soil contact (Barber, 1984; Dibb et al., 1990). Tandon (1987) reviewed a number of examples where deep placement of P fertilizer improved crop water use efficiencies. Such results are soil type-, climate-, and crop-specific and depend upon good subsoil properties and water retention. The latter can be a function of compaction caused by previous tillage operations.

Dipping seedling roots in solutions or slurries of P fertilizer has proved an effective method of improving crop establishment and yields (Tandon, 1987). Similarly, seed coats containing neutral P fertilizers such as reverted superphosphate and PAPR are effective in improving crop establishment (Scott et al., 1985).

Whereas band placement is recommended for soluble fertilizers, broadcasting and soil incorporation is recommended for slightly soluble fertilizers to increase the contact time and soil volume in contact with fertilizer in order to increase dissolution (Kirk and Nye, 1986a, b, c). Exceptions are the establishment of plantation crop seedlings on acid tropical soils (Ling et al., 1990; Pushparajah et al., 1990). In soils with high P-sorption capacity, long contact time can reduce the availability of PR-P, but the reduction in agronomic effectiveness with time is usually less than with soluble P fertilizers. Application of P by drip irrigation is not generally recommended because strong adsorption by soils restricts the movement of P away from the point of application.

Progress with fertilizer application techniques has been made in three areas: (i) application of liquid fertilizers (Beaton and Murphy, 1988), (ii) application of animal slurries and slaughterhouse effluents (Kofoed et al., 1986), and (iii) application under reduced tillage conditions (Anon, 1984). A newly developed point injection technique allows fertilizer to be placed close to the seed or growing plant with minimal damage to the roots, and to be injected through the crop residues in reduced tillage systems (Anon., 1984). To some extent this overcomes the common problem with reduced tillage of nutrient accumulation in shallow surface soil layers (Sharpley et al. Ch. 11), which encourages shallow rooting and increases drought susceptibility. In the case of animal manure (Kofoed et al., 1986), injection overcomes the problems of noxious odours and surface applied slurries reducing the palatability of the plant produce. Shallow injection of slurries reduces the loss of NH3 by volatilisation while conserving nutrients in the root zone.

 

Fertilizer amendments designed to increase PR dissolution

The agronomic performance of a PR depends on the ability of soils and plants to supply acid. Most soils are only mildly acidic, so PR-P release is slow. Thus PR is poorly suited for cropping systems in near neutral soils that require a short term, high soil-P status. One way of overcoming slow P release is the use of elemental S and NH4+ forms of N to increase soil acidity (Hedley et al., 1990).

Sulphur or pyrites are added to low grade, unreactive PRs to improve their agronomic value and fulfill the S requirement of pastures and crops (Rajan and Gillingham, 1986; Tandon, 1987). Inoculation of Togo PR - S mixtures with S-oxidising microbes ('biosuper', Swaby, 1983) achieved yields of ground nut that were greater than those produced by superphosphate (Bromfield, 1975). Most biosuper products developed in Australia and Fiji using elemental S and ground (<150mm) PR, release P too slowly to meet the requirements of fast growing annuals and are more suited to pastoral fertilisation. A swelling clay can be added to the mixture to ensure rapid granule breakdown (Owers, 1988) or the PR can be sparingly acidulated with H2SO4. Using this approach, a low cost high analysis S:P fertilizer can be produced from RPR or PR, which may not have been suitable for full acidulation.

Through the process of nitrification, NH4- based fertilizers (or urea) rapidly acidify soils with active nitrifier populations. In addition to providing N, the dissolution of RPR and plant uptake of P are increased when such N forms are combined with RPRs by banding (Apthorp et al., 1987) granulating (Hedley et al., 1989) or compacting (Chien et al., 1987b). The initial high pH conditions resulting from urea hydrolysis may also enhance RPR dissolution through the chelating action of solubilised soil organic matter (Chien et al., 1987b).

The composting of animal manures and organic waste materials with PR (phospho-composts) has been used for increasing the agronomic value of PR (reviews: Kothandaraman et al., 1986; Tandon, 1987). The largest potential sources of organic wastes for composts are cereal straws, particularly rice straw and rice husks, which presently cause disposal problems in many countries. It is appropriate to develop composting systems (Gasser, 1985) that are capable of turning both straw and local PR materials into valuable fertilizers. Unfortunately these lignocellulosic wastes are not rapidly decomposed (L'Honeux et al., 1988; Lynch and Wood, 1985) and in an unaltered state do not provide effective substrates for acid producing bacteria or fungi such as Clostridium sp. or Aspergillus sp.. Chemical pretreatment of the straw (L'Honeux et al., 1988) or novel mixed inoculants of fungi, with lignocellulase activity, and anaerobic N-fixing bacteria (Lynch and Wood, 1985), could both improve the digestibility of straws and increase their potential for acidulating PRs.

Mahimairaja et al. (1993) have developed a method by which the NH3 that normally is volatilised from stored or composted animal manures can be trapped in organic wastes such as bark (or coconut coir ), nitrified and the protons used to dissolve PR. The method has the potential to produce low cost soluble P. More research work is required in similar areas that can provide cheap recycled N and a low cost P fertilizer for developing countries that have local PR deposits.

 

Nations whose economies are based on manufacturing industries rather than primary production are most likely to have sufficient nutrients in wastes (sewage, industrial and intensive livestock farming) to meet all their fertilizer needs. In general, industrialised nations are net nutrient importers (see Table 11 for the contrast between the UK and New Zealand) even when fertilizer is not considered in the equation (Beaton et al., Ch. 2). Animals and humans are inefficient assimilators of nutrients from their food. More than 70% of most nutrients in food are excreted. Phosphate is almost entirely excreted in feces. Few population centres have waste recycling or sewage systems that can return valuable nutrients to the primary producer in an environmentally safe form. There is an urgent need to reduce this loss of nutrients in order to slow the consumption of finite PR deposits.

 

Table 11. Estimated P balance sheet for the UK1 and New Zealand

	UK1	New Zealand2
	------------ Mg ha-1 y-1 --------
Net Losses from Farms3 Crops Livestock Total Loss	38740 19535 58275	581 7860 8441
Amount recoverable from population Domestic refuse Sewage Total recoverable   Balance   Per capita balance (person-1)	58,850 30,250 89,100   + 30,825   + 0.56	3531 1815 5346   - 3095   -0.94

1From Widdowson, 1987; population 55 million in 1984

2Population 3.3 million; using Widdowson's (1987) estimates of per capita recovery of nutrients from domestic refuse and sewage

3Does not consider fertilizer input

 

Many agricultural wastes are utilised as fertilizer because of their proximity to the farm. Much improvement, however, can be made in the way that municipal and industrial wastes are utilised. Most wastes have inherent nutrient value and if processed correctly can be used to increase crop yields. Sewage sludge is a major P source (0.06% to 0.48%P) but unfortunately frequently is contaminated with heavy metals from industrial sewage (Kofoed et al., 1986). Strict regulations are required to ensure its safe use (Kirkham, 1983; Hall and Williams, 1983). Large volume industrial wastes important for their P content and generally free from heavy metal contamination are wastes from the sugar, palm oil, rubber (Hedley et al., 1990) and milk (Gregg and Currie, 1992) processing industries. These are important P sources in tropical and subtropical P-deficient soils. Sugarcane production in the Asian and Pacific region is increasing (Anon, 1987a) and filter cake, the residue produced when cane juice is clarified, is an effective P fertilizer for sugarcane and paddy rice (Prasad, 1976; Chanchareonsook et al. 1988). The filter cake is variable in its moisture and nutrient content but can be expected to contain between 1-2% N, 0.5-2.4% P, 0.4-0.5% K and 2-3% Ca and has a pH of 7.5-9.0. Filter cake and a similar product from the sugarbeet industry have been used as soil amendments for many years. Maximum fertilizer value of these materials can be gained from heavy dressings (up to 20 Mg ha-1) applied to acid soils with high P sorption capacities. Waste from palm oil mills includes empty bunches, fibre, shell, steriliser condensate, hydrocyclone waste and centrifugal/separator waste. While the empty bunches, fibre and shell are recycled as fuel in the mills or used as mulches, the remaining three constitute the bulk of the palm oil mill effluent (POME) (Yeow, 1983). Similar effluents originate from rubber factories (John, 1981). Depending upon the type of treatment, the P content of the effluents from palm oil mills or used as mulches varies from 12 mg kg-1 in the supernatant to 1180 mg kg-1 in the bottom slurry and the effluent from rubber processing varies from 48 to 81 mg kg-1 (Chan et al., 1983b). The application of POME and rubber effluents gave positive yield responses with oil palm, cacao, grasses, maize and vegetables (Chan et al., 1980). It is estimated that in the Asian and Pacific region alone there are approximately 73,000 Mg of P are available from the sugar, rubber and oil palm industries which, according to FAO (Anon., 1987b) statistics, is equivalent to 50% of the fertilizer P imported into Asia.

Amounts of P in industrial and urban wastes can be equivalent to a large percentage of the amount of P used as fertilizer. If efficient nutrient recovery from wastes could be achieved, then countries with large urban centres, that are net food importers, have no need to import P fertilizer materials. Finite fertiliser resources could be saved for countries that do not have positive nutrient balances. Unless the cost of transporting bulky sewage and animal wastes is transferred to the cost of food, nutrient recovery can not compete with the relatively low cost of nutrient in fertilizers. Implementation of nutrient recovery schemes must therefore be political decisions based on environmental concerns rather than economic ones. Costs of nutrient recovery could be reduced if it was considered socially acceptable to ship environmentally safe wastes to coastal agricultural regions. Many of the worlds large urban centres are ports or are coastal plus much intensive agriculture is riparian, estuarine or coastal. The lower cost of marine transport compared with land transport could reduce the cost of waste nutrient recycling.

 

Selection of P-efficient plant species

Recently more emphasis has been placed on choosing plants suited to P-deficient soil conditions by selecting alternative plant species or genetically improving existing species. Plants vary widely in their abilities to survive and grow in soils of low P status (Barber, 1980). Two types of mechanism may account for these differences in yield between cultivars at different P levels (Lajtha and Harrison, Ch. 8):

 

• Cultivars having different internal P requirements, which are able to produce different yields with the same amount of P in the above ground dry matter.

• Differences in the plants' ability to extract P from the soil. due to differences in root total length and density, root hair length and density, root diameter, etc., or the extent or rate of infection with P sequestering mycorrhizal fungae; or due to root-induced changes in soil chemical conditions.

 

The amounts and length of root hairs and the rate of development of the root system in soil zones containing plant available P are important characteristics. This is particularly so in drought prone environments where root- and root hair growth must be rapid at the onset of rains so that subsequent dry periods can be survived. In dry soil, the crosssectional area for solute diffusion is reduced and the impedance of the diffusion pathway increased such that P transport to root surfaces is negligible. Rapid early root growth independent of soil P supply may be favoured by large seed-P reserves (Marschner, 1991; Hedley et al., 1993). In addition to cultivar differences in this characteristic, the agronomic management of the seed crop may have a large influence (Bolland et al., 1991).

Varietal differences in P uptake have been observed for many plant species, such as, rice (Gopalakrishna Pillai et al., 1984), wheat (Saggar et al., 1974), barley (Nielsen and Schjorring, 1983), maize (Nielsen and Barber, 1978) and white clover (Caradus, 1980). Species differences in P uptake characteristics can be used to introduce plants with greater P uptake efficiencies into soils of low P status to be fertilised with PR (Bekele et al., 1983). Varietal differences, induced by genetic manipulation, can be used in a similar manner.

 

ACKNOWLEDGMENTS.
The senior author is grateful to Drs A. Braithwaite, R.Harrison, A. Hussin and G.D.J. Kirk and R.E. White, for their contributions to earlier papers from which some discussion is drawn, Drs. R.J. Haynes, P.H. Williams and S.S.S. Rajan for supplying data for figures, Drs A. Fink, E. Sibbesen, G. Varallyay and H Von Uexküll for their workshop session on the paper and to Ms. N. Collins for typing the manuscript, and to Dr. Holm Tiessen for his patience and editing skills.

Phosphorus in the Global Environment. Edited by H. Tiessen © 1995 SCOPE. Published in 1995 by John Wiley & Sons Ltd.

---

Back to table of contents

 

Last updated: 12.07.2001