Soils contain 100 to 3000 mg of phosphorus
kg-1 almost entirely as orthophosphate (PO4). The proportion
of soil phosphate in organic compounds ranges from 29 to 65%
of the total P (Harrison, 1987). Processes involved in P cycling
can be inorganic or biological. Inorganic processes include physico-chemical
reactions, such as precipitation/dissolution and sorption/desorption.
The biological processes are initiated primarily by the uptake
by higher plants and microorganisms of P released during the
weathering of primary and secondary minerals, and include active
solubilisation of soil P minerals. The inorganic P (Pi) generated
is then recycled via complex food chains through a series of
mineralisation and immobilisation reactions.
DISSOLUTION and PRECIPITATION
Dissolution
Apatite [Ca10X2(PO4)6, where X = OH- or
F-, Ca may also be substituted with Na, Mg and PO4 with CO3]
is the most common primary P mineral. It is the main phosphate
mineral in the earth's crust, is very stable in calcareous environments,
can be sand or silt-sized (and therefore easily identified) and
can be occluded in other minerals such as quartz (Syers et
al., 1967; Lindsay et al., 1989). The first step of
P cycling is the dissolution of apatite.
The dissolution of apatite has been extensively
studied with phosphate rock (Olson, 1975) but more seldom in
natural systems where it is the first step of the P cycle (Walker
and Syers, 1976). Apatite dissolution requires a source of H+
which can originate from the soil itself or from roots or microbes
, and sinks for
Ca and P (Simllie et al., 1987;
Mackay et al., 1986). Therefore phosphate rocks have lower
agronomic value in soils above pH 6.2 (Fardeau et al.,
1988a). Gupta et al. (1990) attributed the release of
P from a sodium dominated soil to the dissolution of calcium
phosphates at high alkalinity and pH. The rate of
|
Phosphorus in the Global Environment.
Edited by H. Tiessen
© 1995 SCOPE. Published in 1995 by John Wiley &
Sons Ltd. |
|
dissolution further depends on the accessibility
of the apatite particle, its morphology (Kirk and Nye, 1986a)
and the rate of substitution of PO4 by CO3 within the crystal
lattice (McClellan and Gremilion, 1980).
Methodologies for studying precipitation/dissolution
(Table 1) have been reviewed by Pierzynski (1991). Solubility
equilibrium experiments may have limited validity because of
our inability to describe properly the chemical state of the
solid phase of the soil (Pierzynski, 1991; Yong et al.,
1992). The use of direct methods such as X-ray energy dispersive
analysis in conjunction with Scanning Electron Microscopy for
high resolution solid state NMR (Hinedi et al., 1989)
look more promising since they allow direct observations of P
minerals. New NMR spectrometers and 31P solid state probes allow
the acquisition of spectra on materials with total P contents
as low as 1 g kg-1 (Frossard et al. 1994a). Further methodological
research on the acquisition and interpretation of 31P NMR spectra
on soil samples is required.
Precipitation in Calcium systems
As early as the mid-nineteenth century
researchers related the retention of P to the presence of Ca
carbonates and Al and Fe hydrous oxides in soils (Wild, 1950).
Calcium phosphates can form by precipitation following an initial
P adsorption on to calcite (Cole et al., 1953; Freeman
and Rowell, 1981; Syers and Curtin, 1989). After an initial adsorption
of P on the surface of a pure calcite, monocalcium phosphate
[MCP, Ca(H2PO4)2] precipitates, it then transforms to dicalcium
phosphate dihydrate (DCPD, CaHPO4.2H2O), to octocalcium [OCP,
Ca8H2(PO4)6.5H2O] and finally to hydroxyapatite [HAP,Ca10(PO4)6(OH)2]
(Lindsay et al., 1989). This mineral having the lowest
solubility should ultimately control the concentration of P in
the soil solution while the intermediate precipitates are unlikely
to persist in soils.
Various Ca phosphates have been identified
in calcareous soils following P fertilisation (Lindsay et
al., 1989). Apatite is seldom observed although it is the
thermodynamically most stable P mineral (Bell and Black, 1970;
Harrison and Adams, 1987; Pierzynski et al., 1990a). This
is because of the presence of impurities such as organic acids
or Mg which inhibit the crystal growth of apatite through adsorption
on the seed crystal surface (Brown, 1981; Amjad et al.,
1984; Amoros et al., 1986; Inskeep and Silvertooth, 1988).
Thus organic acids adsorbed on DCPD crystal surfaces can act
as new nuclei for DCPD crystals hampering any further transformation
to OCP or HAP (Grossl and Inskeep, 1991). This could explain
the increased available P contents of calcareous soils amended
with an organic P sources (farmyard manure, sewage sludge) compared
to those receiving only mineral fertilizer (O'Connor et al.,
1986).
Precipitation in Aluminium and iron systems
Few well crystallised Al and Fe phosphates
have been observed in soils. Many researchers have assumed that
amorphous Fe-P or Al-P compounds would transform respectively
to strengite [FePO4.2H2O] and variscite [AlPO4.2H2O] after prolonged
aging (Lindsay et al., 1989). However the crystallisation
of strengite and variscite occurs only under very restricted
conditions (high P and cation concentrations), which seldom occur
in soils (Hsu, 1982a and b). Instead, the reaction of P with
Al oxides can result either in the formation of amorphous aluminum
phosphate (Nanzyo, 1988) or in the formation of organised phases
such as sterretite [(Al(OH)2)3HPO4H2PO4, Van Riemsdjik et
al., 1975)]. Similarly the reaction of P with an Fe oxides
such as goethite can result in the precipitation of tinticite
[Fe6(PO4)4(OH)6.7H2O, Jonasson et al., 1988)] or griphite
[Fe3Mn2(PO4)2.5(OH)2, Martin et al., 1988)] depending
on the solution and surface conditions.
Table 1. Summary of the methods used for
studying the precipitation/dissolution reactions of P in soils
and with soil minerals
|
DIRECT METHODS, IN SITU OBSERVATIONS |
|
IDENTIFICATION OF THE P SOLID PHASE:
petrographic observations
X-ray Diffraction
Differential Scanning Calorimetry
High resolution solid state 31P Nuclear Magnetic Resonance
Limits: high P concentration required
Paramagnetic impurities (Fe, Mn) preclude the observation P by
NMR.
MORPHOLOGY AND COMPOSITION OF P-RICH PARTICLES:
Electron microscope (SEM, STEM) coupled with energy dispersive
X-ray analysis (EDS) (Electron microprobe)
Limits: Local information at the micrometric scale
only (representativeness)
INFORMATION ON THE SURFACE STRUCTURE OF MINERALS:
Scanning auger microscopy/spectroscopy
Infra red spectroscopy, diffuse reflectance IR spectroscopy
Limits: These techniques require "clean minerals" |
|
INDIRECT METHODS
SEQUENTIAL EXTRACTION:
quantification of Ca-P and Fe/Al-P
Limits: solubilisation and re-precipitation of P forms
during the extraction
SOLUBILITY EQUILIBRIUM EXPERIMENTS:
identification of the P solid phases controlling the concentration
of P in the solution
Limits: solid P phases in soils not in the standard
state;
product of solubility of solid P phases in soils underestimated |
Various P minerals can be observed under
natural conditions in P rich environments (phosphatic parent
material, heavy fertilisation). Minerals of the crandallite group
[crandallite CaAl3(PO4)2(OH)5.H2O; plumbogummite PbAl3(PO4)2(OH)5.H2O;
florencite CeAl3(PO4)2(OH)5.H2O; gorceixite BaAl3(PO4)2(OH)5.H2O];
wavellite [Al3(PO4)2(OH)3.5H2O]; and barrandite [(Al,Fe)PO4.2H2O]
have been identified in soils (Norrish, 1968; Adams et al.,
1973; Kumar et al., 1991; Karathanasis, 1991; Wang et.
al., 1989). Vivianite [Fe3(PO4)2.8H2O] has been reported
in anaerobic conditions in soils rich in organic matter either
under natural conditions [peats, buried alluvial soils, (Lindsay
et al., 1989)] or following heavy additions of dairy effluent
(Guichet, 1986). Indirect evidence based on solubility equilibrium
experiments, indicate that MnPO4.1.5H2O could control the concentration
of solution P in soils (Boyle and Lindsay, 1986; Schwab, 1989).
However, such crystallised minerals represent only a small proportion
of the total soil P, most of it being precipitated as amorphous
compounds (Wang et al., 1991a). Furthermore (Wang et
al., 1991b) show that crystalline Fe can not be found in
a phosphorus rich environment all the iron being precipitated
as amorphous Fe-P compounds. In heavily fertilised soils amorphous
mixed Al-Si-P compounds prevail (Pierzynsky et al., 1990b).
SORPTION/DESORPTION
Although precipitation-dissolution reactions
are of interest, sorption-desorption reactions usually give a
better description of the uptake and release of P by soils (Syers
and Curtin, 1989). We use the term "sorption" instead
of "adsorption" because it covers the fast surface
reaction of adsorption and the long term reaction (Barrow, 1985).
There is a theoretical difference between the precipitation/dissolution
and the sorption/desorption processes: the concentration of P
in solution controls the amounts of sorbed P, whereas the solubility
product of the least soluble P compound in the solid phase controls
the dissolution and therefore the concentration of P in solution
(Syers and Curtin, 1989). In reality, the abiotic retention of
P on soils particles must be seen as a continuum between surface
reactions and precipitation.
Sorption
Phosphorus sorption on soil minerals has
been extensively studied on clean minerals (Wild, 1950; Parfitt,
1978; Sample et al., 1980; White, 1982; Barrow, 1985;
Lindsay et al., 1989; Syers and Curtin, 1989; Barrow,
1990; Sollins, 1991; Pierynski, 1991; Fardeau and Frossard, 1992).
Sorption of P on soil minerals and organic
compounds initially proceeds by a rapid reaction. When water
soluble orthophosphate ions are added to metal oxi-hydroxide
a very rapid exothermic ligand exchange reaction takes place
with the reactive surface groups. An OH- or an H2O molecule is
released from the surface and a phosphated surface complex is
formed (Parfitt et al., 1976, 1977; Goldberg and Sposito,
1985; Torrent et al., 1990, 1992). This surface complex
has been directly confirmed by X-ray photoelectron spectroscopy
(Martin and Smart, 1987).
The greater sorption of P on goethite than
on hematite (Figure 1) can be explained by a greater accessibility
to phosphates of singly coordinated OH surface groups on {110}
faces (Torrent et al., 1990). The sorption of P can be
accounted for by a total occupation of these sites by bidendate
complexes. The average maximum quantity of sorbed phosphates
reaches 2.5 mMol
P m-2 on goethite but only 0.97 mMol m-2 on hematites because reactive surface sites
are present only on a small proportion of the total surface (Barrón
et al., 1988). Similarly, the polymerisation and crystallisation
of aluminum oxide decrease the number of surface sites available
for P sorption (Sims and Ellis, 1983).
Monoesters such as glucose-1-P, glucose-6-P
and myoinositol hexaphosphate, or phosphonates are adsorbed on
the same sites and in similar ways as orthophosphate ions on
ordered Al and Fe hydroxides (Shang et al., 1990; Shang
et al., 1992; Ognalaga et al. 1994). The sorption
of P monoesters is a function of the charge density of the PO4
group (Frossard et al., 1989) and of the spatial conformation
of the molecule. Thus, the sorption of DNA on soil is governed
both by the amount of sorbing clay and by the molecular weight
of the DNA (Ogram et al., 1988). The phosphonate moiety
of the herbicide glyphosate [isopropylamine salt of N-(phosphonomethyl)glycine]
can form binuclear complexes with surface -Fe(H2O)+ groups on
goethite by ligand displacement of H2O (McBride and Kung, 1989).
The sorption of di- and tri-esters has not been studied in detail
probably because of their complexity, but deserve more attention.
The initial rapid sorption of P on soils
is followed by slow reactions which have been ascribed to either
solid (Barrow, 1985) or liquid state diffusion (Sollins, 1991).
Barrow (1985) shows that this slow reaction is endothermic and
can occur by ion exchange within the crystal lattice or by a
vacancy mechanism in which an atom moves in one direction and
the vacancy in the other direction. The latter would be more
likely to occur in imperfect crystals such as those found in
soils. The slow reaction may also result from the lower accessibility
of surface sorption sites located within the aggregates of poorly
crystallised oxides such as ferrihydrite. This porosity then
controls the reaction rate (Madrid and De Arambarri, 1985; Willett
et al., 1988; Torrent et al., 1990 and 1992).
The sorption of P on oxides is a function
of the electrostatic potential in the plane of adsorption (Bowden
et al., 1977). Therefore, P sorption by pure oxides decreases
with increasing pH, due to a decrease in the electrostatic potential.
The effect of ionic composition of the solution on anion sorption
varies with pH. Above a certain pH, sorption increases with increasing
ionic strength and below this pH the reverse occurs. The pH at
which there is no effect of the electrolyte concentration is
called Point of Zero Salt Effect (PZSE). For uniform surfaces
such as those of synthetic oxides, the PZSE coincides with the
Point of Zero Net Charge (PZNC). The effects of pH and electrolyte
on the P sorption on goethite have been described by Bowden et
al. (1980) and Barrow et al. (1980) using the mechanistic
model developed by (Bowden et al., 1977). This model also
describes the effect of time on P sorption by Fe and Al oxides
(Bolan et al., 1985).
Organic ligands can affect P sorption.
Organic anions compete with orthophosphate for similar sites
on the surfaces of oxides (Nagarajah et al., 1970; Sibanda
and Young, 1986; Violante et al., 1991; Fontes et al.,
1992) and soils (Yuan, 1980; Frossard et al., 1986; Amann
and Amberger, 1988; Hue, 1991). The destruction of organic matter
by H2O2 can increase the accessibility of sorbing sites to PO4
(Frossard et al., 1992b). Several aliphatic acids (such
as oxalic, citric, isocitric, malic, and malonic) are very effective
in decreasing P sorption, followed by aromatic acids and phenolics,
while sugars, or amino acids are not effective in reducing P
sorption (Nagarajah et al., 1970; Amann and Amberger,
1988). The effect is greater when the organic compound is added
prior to the P addition (Hue, 1991). Organic compounds can chelate
metals and prevent the reaction between metals and phosphates
(Earl et al., 1979). An increase in P sorption can occur
when the addition of organic compounds to amorphous oxides in
soils hampers their crystallisation and increases their specific
surface thus increasing P sorption (Huang and Violante, 1986;
Borggaard et al., 1990). Citrate, tartrate or formate
also can extract metallic ions from mineral surfaces creating
new P sorption sites (Traina et al., 1986a).
Complexes of organic matter and Al are
an important P sink in allophanic soils (Borie and Zunino, 1983),
but additions of various organic compounds did not alter the
P sorption in a soil derived from volcanic ash (Appelt et
al., 1975). Humic acid-goethite complexes extracted from
oxisols could sorb up to 100 mMol P g-1 despite their low specific surface (1
m2g-1) (Fontes et al., 1992).
An indirect effect of organic matter on
P sorption can result from the anaerobic decomposition of organic
matter, which causes Fe compounds to be dissolved and re-precipitated
as amorphous minerals with strong P sorption abilities (Sah and
Mikkelsen 1986, 1989; Sah et al., 1989a). Sing and Jones
(1976) and Bumaya and Naylor (1988) showed that P sorption could
either increase or decrease following the addition of organic
residues in highly P-sorbing soils. The P sorbing capacity increased
with the duration of incubation and decreased with an increasing
P content in the residue.
Complexes of humic compounds with Fe, Al
and, to a lesser extent Ca, are able to sorb orthophosphate (Levesque
and Schnitzer, 1969; Cegarra et al., 1978; Bloom, 1981;
Gerke and Hermann, 1992) possibly by the formation of ternary
compounds (HA-Metal-PO4) (Cegarra et al., 1978). Humic
acid from a peat sorbed little P until it was conditioned with
a concentrated solution of AlCl3 (White, 1982). The quantity
of sorbed P was a function of the degree of hydrolysis of Al
on the organic compound. The sorbed P:Al ratio reaches a maximum
for the maximum charge of adsorbed Al, i.e. at OH:Al ratios between
1.5 and 2.5 (Appelt et al., 1975). For iron freshly complexed
to humic substances, Gerke and Hermann (1992) found a maximum
sorption at a molar ratio of sorbed P:Fe of 1. This ratio was
10 fold lower for an amorphous iron oxide. P sorption was influenced
by pH and the presence of Ca through changes in the accessibility
of sorption sites and electrostatic interactions.
Clay minerals sorb less PO4 than oxides
(Figure 1). Two sorption surfaces can be distinguished on a clay:
the edge of the Al layer and the negatively charged surfaces.
The broken edges of aluminum layers are variable-charge surfaces
containing -Al(OH) groups with P sorption properties similar
to those of Al hydroxide surfaces Parfitt, (1978). On 1:1 clays
such as kaolinite, P sorption is a function of the specific surface
(Schwertmann and Herbillon, 1992).
The quantity of P sorbed on 2:1 clay can
greatly exceed the available edge sites and is strongly affected
both by the valency and hydration energy of cations or Fe- and
Al hydroxides on clay surfaces, and by the ionic strength of
the solution (Blanchet, 1960; Coleman et al., 1960; White
1982; Traina et al., 1986b).
Allophane-like materials sorb large amounts
of orthophosphate rapidly (Clark and McBride, 1984; Parfitt,
1989) (Figure 1) due to their very high specific surface area
(Parfitt, 1980). In addition, the slow reaction is more important
in allophane than in goethite, hematite or ferrihydrite (Parfitt,
1989). The slow reaction can result from the formation of alumino-phosphate
precipitates disrupting the allophane structure and creating
new defect sites (Parfitt, 1989).
The sorption of organic P (Po) compounds
on montmorillonite is related to its active Al content (Anderson
and Alridge, 1962) and increases below pH 5 (Greaves and Wilson,
1969). Above pH 5, sorption reaches a minimum and is confined
to the external surfaces of the clay. Interaction of a phosphonate
oxygen from dimethyl methyl phosphonate with interlamellar Ca
in montmorillonite was shown by Bowen et al. (1988). Such
sorbed species may have different properties than free ones (Mingelgrin
and Tsvetkov, 1985).
The quantity of P sorbed on calcite depends
largely on its specific surface (White, 1982; Borrero et al.,
1988). Pure calcites have a low specific surface (1-2 m2 g-1)
while soil calcites can have specific surfaces in the range of
16-500 m2
g-1 (Holford and Mattingly, 1975a) due to intermittent dissolution,
reprecipitation and incorporation of impurities. Although pure
calcite is slightly negatively charged (White, 1982), orthophosphate
can be sorbed at the surface. This sorption occurs only on 5%
of the surface of pure calcite (Cole et al., 1953), but
these sites then act as nuclei for the 'clustering' of Ca and
PO4 ions with the ultimate formation of a calcium phosphate phase.
The induction period prior to nucleation and crystal growth is
very long at low P concentration.
For whole soils, there is much evidence
that Al and Fe oxides are the most important components determining
the soil P sorbing capacity even in calcareous soils (Holford
and Mattingly, 1975a; Parfitt, 1978; Ryan et al., 1984;
Hamad et al., 1992). However the long term sorption of
P in calcareous soils is governed by calcium carbonate (Solis
and Torrent, 1989). The quantity of Fe oxides in a soil determines
the P sorption capacity but the type of oxide and their organisation
are important as well. Soils with large amounts of hematite sorb
less P than soils rich in goethite due to the lower accessibility
of sorption sites on hematite (Colombo et al., 1991).
The dispersion of an oxisol with a Na+ resin greatly increases
the accessibility of sorbed 32PO4 (Frossard et al., 1992b).
Liming can either increase, decrease or
have no effect on the P sorption capacity (Haynes, 1982). Liming
acid soils can either (i) increase the negative charge of the
soils, thus decreasing P sorption, or (ii) cause the precipitation
of exchangeable Al as hydroxy Al polymers resulting in the formation
of newly active sorbing surfaces (Haynes and Swift, 1985). The
net liming effect depends on the magnitude of these two processes.
At high P concentrations, P sorption can
be modified by the slow dissolution of Al (Stumm et al.,
1983) or Fe (Lin and Benjamin, 1990) oxides, and particularly
of amorphous compounds. Large quantities of myoinositol hexaphosphate
added to acidic soils extracted Al and Fe (Anderson et al.,
1974). Sufficiently high P concentrations (1 M) are reached in
soils near dissolving fertilizer particles, where dissolution
of minerals has been observed (Lindsay and Stephenson, 1959).
The ligand-promoted dissolution may proceed through polarisation,
weakening and finally breaking of the metal-oxygen bonds, causing
the release of metal to solution (Stumm et al., 1983)
(Figure 2). Not all oxides are similarly prone to dissolution;
the polarizing effect of a ligand or a surficial complex is not
sufficient to release the metal from goethite (Zinder et al.,
1986).
Figure 2. Dissolution of hydrous
oxides caused either by protonation of Me-OH surface groups or
by sorption of ligand on Me-OH groups (Stumm et al., 1983)
Myoinositol hexaphosphate is more strongly
sorbed than orthophosphate on Fe and Al oxides of acidic soils
(Anderson et al., 1974). However, after reaching a maximum,
sorption declines due to the dissolution of oxide surfaces. No
such surface dissolution is observed in neutral or basic soils
(McKercher and Anderson, 1989). The sorption appears to be governed
by clay and organic matter content and by the number of PO4 groups
in the monoester (Figure 3).
All these studies show that P sorption
capacity is closely related to soil properties and that, aside
from the short-lived effects around fertilizer granules, it is
little affected by fertilizer inputs at levels compatible with
financially sound farming practices (Frossard et al.,
1992a). Variations in P sorbing capacity of soils in relation
to soil properties (type) have been mapped in France on sedimentary
parent materials (Frossard et al., 1992a) and in Albania
(Sinaj et al., 1992) on sedimentary and plutonic parent
materials.
Figure 3. Sorption of organic
phosphates on a clayey soil (McKercher and Anderson, 1989)
Barrow (1983b) developed a mechanistic
model describing the sorption of P by soils. This model has three
assumptions (i) the reaction occurs between divalent phosphate
ions and variable charge surfaces; (ii) there is a range of values
of surface properties which are distributed normally and (iii)
the initial adsorption induces a gradient towards the interior
of the particle which drives a solid-state diffusion processes.
This model describes the effects of phosphate concentration,
pH, temperature and time contact. Furthermore the model suggests,
the phosphate that has reacted with a soil for a long enough
period is not "fixed" but that it can be recovered
slowly if a low enough surface activity is induced.
Sorption curves
The sorption of P can be characterised
with parameters calculated from the Langmuir equation:
S = (1)
where S is the quantity of sorbed P, c
the P concentration in the soil solution, Smax the maximum quantity
of P that can be adsorbed on the soil and k an "affinity"
constant. However the theoretical assumptions of the equation
are not realistic since sorbed phosphates carry charges which
decrease surface charge and potential of the sorbing surface
(Bowden et al., 1977). This leads to large errors in the
estimates of maximum P sorption capacity (Kuo, 1988). The Freundlich
equation often fits data better (Barrow, 1978; Ratowsky, 1986),
but cannot predict sorption maxima. Barrow (1991) modified the
Freundlich equation to account for increases in P sorption with
sorption time:
St = ca(kt)b (2)
where St is the quantity of sorbed P at
a specific time, c the concentration of P in the solution, k
a function of the temperature, and a and b are constants specific
to the soil. Fardeau and Frossard (1992) related sorption to
the amounts of P added:
St = Q - V ct (3)
where Q is the quantity of P added in a
volume V and ct the concentration of P in the solution at a time
t, and the relation ct = f(t) is described by:
ct = c1 [t + (c1/c0)1/m ] -m (4)
At t = 0 the concentration of P in the
solution c0 is Q/V, and after an infinite time ct reaches 0;
c1, is the concentration of P in the soil solution at t = 1 unit.
Desorption
Most of the P taken up by plants is derived
from the solid phase of the soil. Therefore, P desorption is
useful for describing available P. Four approaches are currently
used to estimate the quantity of P that can be desorbed from
the soils.
(1) A large number of chemical extractants
have been used to desorb phosphate from soils through changes
in pH or ionic composition of the soil solution (Roche, 1983;
Houmane et al., 1986). However, chemical reagents do not
extract homogeneous pools of phosphate but ions presenting a
wide range of mobility (Fardeau et al., 1988b) and they
ignore the kinetics of P release.
(2) Desorption in P free water or dilute
electrolyte is frequently used in sorption/desorption experiments
where P was previously sorbed. Often, only small proportions
of sorbed P can be desorbed, and this water extraction can rarely
be used in unfertilised soil because of the very low concentration
of P in solution. The amount of desorbable P increases with increasing
soil:solution ratio and increasing desorption time. This has
been modeled by Sharpley et al. (1981). Desorption decreases
with increasing sorption time prior to the desorption (Barrow,
1979). Using a mechanistic model to describe P sorption, Barrow
(1983a) quantified the effects of incubation time, temperature
and soil/solution ratio on P desorption.
(3) Desorption can be effected through
lowering the solution P concentration by introducing a P sink
such as an anionic resin or iron oxide coated filter paper into
the soil suspension (Amer et al., 1955; Van der Zee et
al., 1987). This causes the P from the solid phase to diffuse
towards the solution and to be trapped by the resin or oxide
(Abrams and Jarrell, 1992). Results obtained are strongly correlated
to the amount of P taken up by plants (Roche et al., 1980;
Lin et al., 1991; Tran et al., 1992; Menon et
al., 1991; Hedley et al., Ch. 5) and algae (Sharpley
et al., 1992). The kinetics of P release indicated that
P desorption is controlled by diffusion (Pavlatou and Polyzopoulos,
1988; Abrams and Jarrell, 1992).
(4) The amount of P present on the solid
phase and able to diffuse to the soil solution can be indirectly
assessed by the amount and rate of isotopic exchange of 32PO4
(Larsen, 1967; Le Mare and Leon, 1989; He et al., 1991).
This technique is based on the homoionic exchange between the
radioactive 32PO4 introduced in the solution and the 31PO4 located
on the solid phase of the soil. When 32PO4 ions are introduced
carrier free into a soil/solution suspension at a steady state,
the specific activity of phosphate in solution [32PO4/31PO4]
is equal, at any given exchange time, to the specific activity
of the exchangeable phosphate located on the solid phase.
= (5)
where R0 is the total radioactivity introduced
(MBq), Rt the radioactivity remaining in the solution after t
minutes of exchange (MBq); cp the concentration of P in solution
(mg l-1), EP the isotopically exchangeable P (mg kg-1). The factor
10 represents the soil:solution ratio of 1 g in 10 ml of water
so that cp*10 is equivalent to the water soluble P content of
the soil in mg kg-1. The main problem with this technique is
the measurement of low concentrations of water soluble P (Fardeau
and Jappé, 1988, Salcedo et al., 1991).
Fardeau et al., (1985) give a theoretical
formula describing the decrease of radioactivity with time (t)
of exchange:
= [t + (R1/R0)(1/n)]-n + (6)
where R0 and Rt are the same as in Eq.
(7), and R1 and R• are the quantity of radioactivity remaining in
solution after 1 minute and after an infinite isotopic exchange
period; n is an experimental factor between 0 and 0.5. Therefore:
EP = (7)
Thus, EP is a function of R1/R0, cP, R•/R0 and of exchange
time. All isotopically exchangeable P present on the solid phase
can come into solution and be available to plants. This demonstrates
that soil P can not be divided into pools containing available
P and unavailable forms of P. Instead, available P is a function
of time and most Pi must be seen as been more or less rapidly
exchangeable with the P in solution (Barrow, 1983b).
The EP values obtained with Eq. 7 can be
extrapolated to at least 3 months. Frossard et al. (1994b)
have shown that the L value of 10 soils (fertilised or not with
100 mg P kg-1 soil) experimentally measured after 13 weeks of
growth of Agrostis communis was not statistically different
from their EP value calculated for an isotopic exchange time
of 23 weeks (Figure 4). Validation of the Equation for longer
time is difficult due to the 14 day half life of 32P.
Figure 4. Comparison between
the E value calculated for 3 months according to Fardeau et al.
(1985) and L values determined experimentally by the method of
Truong and Pichot (1976) for 9 soils from widely different origins
(from Frossard et al. 1994b).
Tran et al., (1988) and Salcedo
et. al., (1991) using a double isotopic dilution have
shown the existence of an homogeneous pool of free phosphate
ions. This pool contains ions in the solution plus ions located
on the solid phase with the same mobility as the ions in the
solution. This pool is generally not different from the quantity
of P isotopically exchangeable within one minute. For t = 1 minute,
the Equation 7 can be approximated as follows:
EP1 = cP . 10 . (R0/R1) (8)
The R0/R1 ratio is a measure of the P buffering
capacity of soils (Tran et al., 1988; Salcedo et al.,
1991; Frossard et al., 1992a; Frossard et al.,
1993). Buffering capacity is considered to be very high when
R0/R1 is higher than 10; high when R0/R1 is in the range of 5
to 10; medium when R0/R1 is in the range of 2.5 to 5 and low
when R0/R1 is lower than 2.5.
The results obtained from isotopic dilution
kinetics can be interpreted by two complementary analyses: (1)
stochastic and (2) compartmental.
(1) In the stochastic analysis, Eq. 6 can
be considered as the Laplace transformation of a g function. This is
a probability function describing the distribution of the individual
rate of exchange, ki, between the ions present in the solution
and exchangeable ions located on the solid phase, k = 0 and k
= •
being excluded (Fardeau et al., 1991). There is a continuum
between 0 and the infinite for all ki values. This means that
all phosphate ions in the soil are able to move more or less
rapidly to the solution. It is possible to calculate a mean exchange
rate, km (min-1), of phosphate ions between the soil solution
and the solid phase.
km = (9)
The mean residence time of phosphate ions
in the solution, Tm (min), is:
Tm = 1/Km (10)
And the mean flux of exchange, Fm (mg kg-1
min-1) between the solid phase and the solution is:
Fm = 10.cP.Km (11)
(2) In the compartmental analysis, a compartment
is defined as an homogeneous unit in which all the phosphate
ions have the same kinetic properties and exchange at the same
rate with phosphate ions present in other compartments.
When 32PO4 ions are added to a soil suspension
at a steady state, they are introduced in the pool of free phosphate
ions and can be exchanged with the 31PO4 ions located on the
solid phase (Fardeau, 1993). This system is therefore both multi-compartmental
and mamellary as defined by Sheppard (1962) and Shipley and Clark
(1972), and can be described by the simplified model proposed
by Fardeau (1993) (Figure 5).
All compartments are connected to the pool
of free phosphate ions which is homogeneous and has a size close
to that of EP1min. These ions are almost instantaneously available
to plants. Phosphate in this compartment can be exchanged with
ions located in:
The quantity of P ions present in each
of these compartments is calculated from Eq. 7. R•/R0 is the maximum
possible dilution of the added 32PO4 and is approximated by the
water soluble P divided by the total mineral P of the soil. It
excludes Po whose release is due to hydrolysis and not exchange.
Figure 5. Schematic representation
of the multi-compartmental mamellary model of soil exchangeable
P (Fardeau, 1993).
Table 2. Total P, organic P,
P buffering capacity and water soluble P in 7 soils from various
origins
|
Soil |
R0/R1 |
cp |
n |
Total P |
Org. P |
|
|
|
mg l-1 |
|
---- mg kg-1 ---- |
|
Sandy Feralsol, Burkina Faso |
2.44 |
0.036 |
0.24 |
89 |
55 |
|
Clayey Vertisol, Sénégal |
33.3 |
0.008 |
0.27 |
265 |
71 |
|
Clayey Cambisol, France |
7.58 |
0.10 |
0.31 |
927 |
251 |
|
Silty Luvisol, France |
7.40 |
0.16 |
0.36 |
1058 |
368 |
|
Calcareous Rendzina, Morocco |
6.67 |
0.15 |
0.25 |
629 |
266 |
|
Calcareous Rendzina, France |
3.33 |
0.12 |
0.19 |
1919 |
348 |
|
Andosol, Dominica |
142 |
nd |
0.42 |
2200 |
909 |
Data from Frossard et
al. (1994b); nd: not detected.
The isotopic exchange kinetics for 7 soils
are given in Table 2. The sandy soil from Burkina Faso has a
very low P buffering capacity while the Vertisol has a very high
P buffering capacity. The clayey and silty soils from France
and the calcareous soil from Morocco have a high P buffering
capacity. The French soil developed on decalcified clay (a "terra
fusca" developed on Bajocian limestone) has a medium P buffering
capacity. The Andosol has the highest P buffering capacity. Water
soluble P ranges from a very low value in the African Vertisol
to a high value in the French Rendzina. Due to a high Si concentration
in the solution of the Andosol, the concentration of water soluble
P could not be accurately measured. Silica is the major obstacle
to a precise measurement of water soluble P (Fardeau and Jappé,
1988; Medeiros and Salcedo, Ch. 20).
The results of the compartmental analysis
are presented in the Table 3. The sandy soil from Burkina Faso
has a very low available P content. Less than 1 mg kg-1 is exchangeable
within one minute and only 13 mg kg-1 can be exchanged within
a year. A soil can be regarded as P limited when EP1min is lower
than 5 (Tran et al., 1988). The Vertisol is also very
low in available P. In this soil, low in oxides, the slowly exchangeable
phosphate must be sorbed either on the clay edges, or on the
surface of clay sheets covered by polycationic species. The heavily
fertilised clayey and silty French soils have a high available
P content. As in the Vertisol, the slowly exchangeable P of the
clayey soil probably is sorbed on montmorillonite. In the silty
soil, the slowly exchangeable P may be sorbed to iron oxides.
The two rendzinas are not P limited. The high proportion (70%)
of the Pi which can not be exchanged within a year in the rendzina
developed on the terra fusca may be related to its high iron
oxide content (Cabrera et al., 1981; Frossard et al.,
1992a). The high proportion (57%) of slowly exchangeable P (EP>1year)
found in the calcareous Moroccan rendzina, might be precipitated
with Ca. Due to the small surface area, and to the long reaction
times necessary for calcium phosphate precipitation, a significant
proportion of the P in this soils can still be exchanged within
a year.
Table 3. Calculated P contents
in the various exchangeable pools according to the multi-compartmental
analysis of Fardeau (1993).
|
Soil |
EP
1 min |
EP
1min-24h |
EP
24h-3 m |
EP
3m-1 y |
EP
>1year |
|
|
--------------------------------
mg kg-1
----------------------------------- |
|
Ferralsol |
0.9 |
3.5 |
5.9 |
2.6 |
21 |
|
Vertisol |
2.7 |
15 |
31 |
15 |
130 |
|
Cambisol |
7.8 |
59 |
140 |
67 |
400 |
|
Luvisol |
8.6 |
92 |
220 |
86 |
290 |
|
Rendzina (M) |
10 |
43 |
72 |
30 |
210 |
|
Rendzina (F) |
37 |
99 |
150 |
68 |
1200 |
The results obtained with the stochastic
analysis show that all the Pi could arrive in the solution with
a mean exchange rate varying from 9.9 min-1 in the sandy soil
to 118,000 min-1 in the Vertisol i.e. the P in the solution will
be completely renewed 10 times per minute in the sandy soil and
118,000 times in a minute in the Vertisol (Table 4). The mean
flux of exchange varies from 3.55 mg kg-1 min-1 in the sandy
soil to 9,000 in the Vertisol. Fardeau et al. (1991) suggest
that when EP1min is greater than 5 mg kg-1 and when Fm is higher
than 80 mg kg-1 min-1, P is not a limiting factor for the growth
of plants.
Table 4. Mean constants calculated
from the stochastic model of Fardeau et al. (1991).
|
Soil |
Km
min-1 |
Tm
min |
Fm
mg min-1 kg-1 |
|
Ferralsol |
9.85 |
0.10 |
3.55 |
|
Vertisol |
118,000 |
8.5.10-6 |
9,430 |
|
Cambisol |
213 |
4.7.10-3 |
219 |
|
Luvisol |
94.5 |
10.6.10-3 |
109 |
|
Rendzina (M) |
494 |
2.0.10-3 |
741 |
|
Rendzina (F) |
107 |
9.3.10-3 |
1202 |
|
Andosol |
56,700 |
17.6.10-6 |
nd |
In natural soils, a number of factors affect
P desorption. Increasing amounts of oxides decrease P desorption
(Sayin et al., 1990; Colombo et al., 1991; Tiessen
et al., 1991). The amount of desorbable P is inversely
correlated to the P sorbing capacity (Kuo et al., 1988),
and to the unoccupied portion of the sorbing capacity (Kuo, 1988).
Increasing sodicity can increase the water soluble P (Curtin
et al., 1992). Organic compounds such as oxalate can desorb
P from acid soils by complexing Al and Fe (López Hernández
et al., 1979; Fox et al., 1990; Fox and Comerford,
1992a). The soil water regime can also affect desorption. Bhadoria
et al. (1991b) showed that the effective diffusion coefficient
of P (De) increases strongly when the soil water content increases.
Reducing conditions lead to P release followed by a decrease
in solution P concentration (Ponnamperuma, 1972). The release
has been attributed to the reduction of Fe (III) leading to a
dissolution of Fe oxide and to the partial elimination of P sorbing
sites. The subsequent decrease of P in solution may be due to
the sorption of P on re-precipitated amorphous oxides (Sah et
al., 1989b). On the other hand, P sorption may protect Fe
oxides against reduction (Willett, 1986), and a rise of solution
P in reduced sediments has been observed long before the Eh conditions
were low enough to reduce Fe oxide (Nriagu and Dell, 1974). The
release of P from waterlogged soils must then be accounted for
not only by oxidation-reduction processes but also probably by
some biological mechanisms such as the bacterial reduction of
iron. Aerobic conditions prevailing in the rice root rhizosphere
induce very low pH values due to Fe oxidation, and therefore
can lead to the release of soil P to roots (Kirk, 1993). This
release may be short-lived, since the succession of flooding
and draining periods leads to an overall decrease of P desorption
(Sah et al., 1989b).
Plant Uptake and Phosphorus Supply
Root absorption
Only orthophosphate (H2PO4- and HPO4=)
appears to be taken up directly by the membrane transport systems
of plant roots (Anderson, 1980) in an active, energy dependent
process (Haynes, 1990) which occurs preferentially in young unsuberised
roots, but also along most suberised root surfaces. The process
is capable of concentrating phosphates to 10-3 M within
plant root cells and xylem tissue from concentrations ranging
between 10-5 and 10-8 M in the soil solution. Uptake takes
place against a steep electrochemical gradient and is mediated
by a H+ co-transport. The absorption of phosphate from solutions
of low concentration can be described by Michaelis-Menten kinetics.
The lowest concentration at which roots can absorb P from soil
solution (threshold concentration; Barber, 1984) varies with
plant species (reviews by Nye and Tinker, 1977; Mengel and Kirkby,
1982; Barber, 1984) and with genotype (Krannitz et al.,
1991). The solution pH and concentrations of mobile cations (e.g.
K+ and NH4+) also affect the rate of uptake (Barber and Chen,
1990).
Physico-chemical supply
The root absorbing power creates a demand
for phosphate at the rhizoplane which is greater than the rate
of phosphate transport to the root surface by mass flow of soil
water meeting an evapotranspiration demand (Nye and Tinker, 1977;
Barber, 1984). The concentration of phosphate in the rhizoplane
solution is rapidly depleted through active uptake but, because
in most soils the equilibrium condition between solution phosphate
(PL, µmol cm-3) and exchangeable soil surface phosphate
(PS, µmol cm-3 soil) is often 1000 fold in favour of the
soil surface, the diffusion of phosphate to the depleted zone
around the root is slow. As mass flow of P is negligible, in
all but heavily fertilised soils, the effective diffusion coefficient
of phosphate through soil (De, cm2s-1) becomes a major factor
limiting the rate of plant P uptake. De can be calculated from
the self diffusion coefficient of phosphate in water, (DL, cm-2s-1),
by constraining diffusion to the fractional volume of water in
a soil, q,
and recognising that the diffusion path has to circumnavigate
soil particles and air-filled pores, which effectively reduces
the diffusion coefficient by an impedance factor, f. In addition,
the power of the soil surface to sorb (PS/PL) reduces the time
allowed for effective diffusion. De is thus given by:
De = DL q f (12)
Consideration of root extension rates and
the calculated flux of phosphate to the root surface (Jr, µMol
cm-2 s-1), mostly by diffusion plus a small amount carried in
mass water flow (Vo) has proved an effective method of simulating
P uptake by plants growing in cultivated soils for short growing
seasons (Nye and Tinker, 1977; Barber, 1984), using the equation:
Jr = De+ V0 PL (13)
where PS/r is the concentration gradient
of soil phosphate at radial distances (r) away from the root
surface.
The most important soil factors influencing
P uptake are those that influence De described in Equation [12]
(i.e. soil moisture content influences both q and f and soil texture and bulk density influence
f), and the soil P sorption characteristics. Within one soil
type (PS/PL) is curvilinear with respect to PS and is represented
by the differential of the Freundlich isotherm (Kirk and Nye,
1986a). PS and the nature of the sorption isotherm used to calculate
PL and De are sensitive parameters in calculating P uptake (Silberbush
and Barber, 1983). However the sorption isotherms may not be
sufficient to predict the uptake of P. Shaviv et al. (1992)
showed that the predicted uptake of P was strongly reduced when
P sorption kinetics were also taken into account. Bhadoria et
al. (1991a) stressed the importance of using desorption rather
than adsorption isotherms to describe P uptake in soils of low
P status.
Although only limited validation of these
physico-chemical models of P supply to plants has been carried
out, they are believed to provide a mechanistically correct description
of short-term P uptake processes. Deviations of predicted from
observed P uptake are likely to be due to factors that influence
the most sensitive parameters of a model, for example: factors
influencing root extension, such as soil acidity and Al toxicity,
lack of other growth limiting nutrients, and adverse soil physical
properties; and factors influencing De, such as, heterogeneous
soil structure, soil pH (Barber and Chen, 1990) and a PS pool
which may sorb P during the period of plant growth. A low P supply
induces an increase of the root surface mostly through increased
production of root hairs (Anghinoni and Barber, 1980; Dinkelaker
et al., 1989; Anuradha and Narayanan, 1991; Föhse
et al., 1991) and/or enhanced myccorizhal association
(Plenchette, 1991).
Solubilisation of inorganic P by acidification,
chelation and reduction
The modeling approach described above should
be restricted to those plants which use only isotopically exchangeable
P such as rye grass (Fardeau and Jappé, 1976) or Agrostis
(Figure 4). Other plants (e.g. Lupinus spp., Brassica
spp, Beta spp) and associated microorganisms modify the
chemistry and biochemistry of rhizosphere soil in a number of
ways which can influence the pool of PS through the solubilisation
of soil Pi and the mineralisation of Po. Claassen (1992) recently
reported significant deviations between the uptake of P as predicted
from the model and the actual uptake by sugar beet (Beta vulgaris).
This difference can be attributed to biological processes causing
the dissolution of insoluble forms of Pi and/or Po in the sugar
beet rhizosphere (Heck and Saive, 1983). The dissolution of the
mineral P in the rhizosphere are effected by the excretion of
H+ and organic acids by roots, fungi and bacteria.
The excretion of H+ is related to the cation-anion
balance of root uptake and exudation. A plant supplied with NH4+
as the sole source of N excretes H+, while OH- is excreted with
NO3- uptake (Gahoonia and Nielsen, 1992). A net H+ efflux also
occurs when NO3- fed plants are grown in a P-deficient solution
(Le Bot et al., 1990), due to a greater uptake of cations
than anions (Moorby et al., 1988). This process was invoked
by Hedley et al. (1982) to explain the solubilisation
of calcium phosphate by Brassica napus. Bekele et al.
(1983) reported higher yields and P uptake of plants grown on
rock phosphate with NH4+ than with NO3-. Legumes acidify their
rhizospheres (Haynes, 1983) through surplus cation uptake as
a result of N2 fixing activity, and can be efficient users of
acid-soluble P (Bekele et al., 1983, De Swart and Van
Diest, 1987). Kirk and Nye (1986b) have incorporated H+ excretion
by roots and subsequent P solubilisation into a computer simulation
of plant uptake of P from soil fertilised with calcium phosphates.
Hoffland et al. (1989a) showed that
while P deficient plants continue to show a net surplus of anion
uptake under NO3 nutrition, areas behind the root tip still excreted
H+. In young rapeseed plants this was associated with accumulation
and excretion of malic and citric acids (Hoffland et al.,
1989b) in quantities sufficient to solubilise more P from a phosphate
rock than was actually taken up by the plant (Hoffland 1992).
Similar organic acid excretion has been noted for P deficient,
proteid roots of lupins (Gardner et al., 1983) and of
Banksia integrifolia (Protaceae) (Grierson, 1992). White
lupine (lupinus albus) can excrete up to 23% of its total
weight as citric acid when grown on a calcareous P deficient
soil (Dinkelaker et al., 1989). The excretion of citric
acid not only allowed for the dissolution of calcium phosphate
but also for the precipitation of Ca-citrate which acts as a
sink for Ca ions. Ae et al. (1990) showed that pigeon
pea can solubilise Fe phosphate with chelating organic acids.
Strains of carrot root cells which produce large amounts of citric
acid are able to use AlPO4 as the sole source of P (Koyama et
al., 1990), and have maximum P uptake rates three fold higher
than non adapted cells (Koyama et al., 1992).
Further research is required to determine
how the physical stage of plant growth and the nutrient status
of plant and soil influence acid excretion. Plants that excrete
acid as undissociated organic acids have advantages because the
organic anion concentrated in the rhizosphere may have chelation
and reduction abilities (Gardner et al., 1983). Plants
that are net anion absorbers are expected to raise the pH of
their rhizospheres. Increased rhizosphere pH has been shown to
convey Al-tolerance in cereals (Haynes, 1990) which presumably
can improve root extension and sustain P uptake in acid soils.
The rhizoplane and rhizosphere host a large
number and variety of free-living microflora and microfauna,
mostly living within 20 µm of the root surface (Foster,
1986). The microflora metabolises rhizo-deposited carbon, which
in young plants can account for 5-30% of the photosynthetically
fixed carbon (Lynch and Whipp, 1990). In this process such microflora
may generate low molecular weight organic compounds which have
the power to release soil P (reviews Tinker, 1983; Richards,
1987) by acidification (e.g. 2-ketogluconate), chelation (e.g.
citrate, oxalate) and reduction (e.g. sugars, citrate, hydroxamates).
Whether these compounds are produced in sufficient quantity to
influence plant nutrition remains a matter for debate (Tinker,
1984). The activity of iron-reducing bacteria can release P sorbed
on iron oxides. In soils, the solubilisation of P minerals may
result from the synergistic action of a group of microflora rather
than just one microorganism. Leyval and Berthelin (1987) showed
that the solubilisation of Ca3(PO4)2 was 2-5 times higher when
a bacterium (Agrobacterium radioabacter) was introduced
with the ectomycorhizal fungi Laccaria laccata or Paxillus
involutus than when the bacterium was alone (Figure 6). On
the other hand A. radiobacter inhibited the growth of
L. laccata in the presence of FePO4. The effects of rhizosphere
microflora are further complicated by other trophic interactions
of microflora that result in root microflora being both sinks
and sources of plant available P (Baas, 1990).
Figure 6. Growth (____), pH decrease (......),
phosphorus solubilisation (_ _ _) and immobilisation (----) kinetics
of an ectomycorrhizal fungus (P. involutus) inoculated
after 25 days with A. radiobacter (Leyval and Berthelin,
1987)
Enzymatic hydrolysis
Research on cropped soils (Sharpley, 1985)
supports the concept that Po mobilisation occurs when Pi supply
limits plant growth and phosphatase enzymes of plant and microbial
origin are induced. The turnover (Helal and Sauerbeck, 1984)
and depletion (Tarafdar and Jungk, 1987) of soil Po are greatest
in rhizosphere soil. Phosphatase activity is greater in the rhizosphere
of young and perennial plants than in bulk soil (Burns, 1985;
Burns et al., 1989; Häussling and Marschner, 1989;
Fox and Comerford, 1992b). This has permitted the observation
in vivo of acid phosphatase activity in the rhizosphere
of soil-grown plants (Dinkelaker and Marschner, 1992).
Roots under sterile conditions produce
phosphatases able to hydrolyze a wide range of Po and polyphosphates
(Dick and Tabatabai, 1986; Tarafdar and Claassen, 1988; Juma
and Tabatabai, 1988). Phosphatase activity varies with plant
species and variety (Helal, 1990) and along roots (Adams and
Pate, 1992). Kroehler and Linkins (1988) showed that Pi produced
by the hydrolysis of Po at the root surface could provide up
to 65% of the annual need of P of Eriophorum vaginatum,
while the amount of Po mineralised through enzymatic activity
by oats, barley, clover or wheat surpassed by a factor of 20
the plant uptake of P under both sterile and non sterile conditions
(Tarafdar and Claassen, 1988; Tarafdar et al., 1992).
Microbial phosphatases may also be of importance
for plant nutrition (Tarafdar and Claassen, 1988). The uptake
of P by Pinus caribea from phytic acid was strongly enhanced
when the plant was inoculated with both phosphate solubilising
bacteria and the ectomycorrhizal fungus Pisolithus tinctorius.
(Chakly and Berthelin, 1983). Macrofauna, such as earthworms
can affect the production of microbial phosphatase (Satchell
et al., 1984), possibly through increasing microbial activity
(Lavelle et al., 1992).
The production of phosphatase either by
roots or associated micro-organisms is an efficient strategy
for the acquisition of P by plants. The factor limiting plant
utilisation of Po is the availability of these Po sources rather
than the production of enzymes (Tarafdar and Claassen, 1988).
It is not clear to what extent roots and micro-organisms compete
for P sources.
Enzymatic activity is a function of soil
properties (pH, water content, surface charge, redox condition
etc.) and soil fabric (microsites) (Nannipieri, 1984). The sorption
of enzymes on clay, oxides or humic substances can change enzyme
conformation and reduce activity (Dick and Tabatabai, 1987; Quiquampoix,
1987, Nannipieri et al., 1988).
Mycorrhizal association
Mycorrhizal fungi increase the uptake of
less mobile nutrient ions such as, P and Zn from soils (Bolan,
1991; Plenchette, 1991, Lajtha and Harrison, Ch. 8). More than
80% of the higher plants can enter into mycorrhizal associations
(Gianinazzi, 1991). These associations have occurred at least
since the Devonian, allowing substantial co-evolution (Gianinazzi,
1991) and genetic adaptation (Lu and Koide, 1991). Root exudates
stimulate P uptake and increase growth of hyphae (Johnson et
al., 1991; Thomson et al., 1991; Lei et al.,
1991). The uptake and translocation of P from the soil to the
host plant is ultimately regulated by the transfer of carbohydrate
from the host to the fungus (Koide and Schreiner, 1992; Schwab
et al., 1991). High levels of infection together with
high P availabilities have been associated with a depression
of host plant growth due to the high carbohydrate demand of the
fungus (see references in Koide, 1991).
Mycorrhizae increase the physical exploration
of the soil, effectively increasing the rhizosphere volume (Leyval
and Berthelin, 1991; Jakobsen et al., 1992a; Li et
al., 1991a). The dependence of plants on mycorrhizal infection
for nutrient acquisition may be related to P requirement (Plenchette,
1991; Koide, 1991) and is inversely correlated to the fineness
of the root system (Baylis, 1975; Hetrick et al., 1988).
Mycorrhizal infection also enhances the transfer of P between
living roots and from dying roots to living roots. Inter root
transfer of P overcomes the loss of P likely to occur through
soil adsorption, leaching and immobilisation by microorganisms
(Newman, 1988; Newman and Eason, 1989).
Numerous studies have shown that mycorrhizae
dissolve insoluble P minerals (Leyval and Berthelin, 1986) through
the excretion of H+ and organic acids (Li et al., 1991b;
Lapeyrie et al., 1990; 1991), and hydrolyse Po through
phosphatases (Fabig et al., 1989; Antibus and Linkins,
1992). The presence of a varied microflora may enhance or inhibit
the use of these insoluble P compounds by plants (Chakly and
Berthelin, 1983; Leyval and Berthelin, 1987). The extent to which
insoluble P can be used by plants in field soils remains to be
demonstrated (Gianinnazi-Pearson et al., 1981).
The rate of inflow of P into mycorrhizal
roots cultivated under controlled conditions is much greater
than that of nonmycorrhizal roots (Sanders and Tinker, 1973;
Jakobsen et al., 1992a). However the relationship between
arbuscular mycorrhizae (AM) root infection and P inflow is not
clear under field conditions (Sanders and Fitter, 1992). The
capacity of transport of P from the hyphae to the root varies
greatly with the fungus species (Jakobsen et al., 1992b).
The variability in the responsiveness of plants to mycorrhizal
infection is related to both the plant demand and the soil P
supply (Koide, 1991) as well as environmental factors such as
the soil water status, or to faunal interactions (Fitter, 1985;
Coleman et al., 1988).
SYNTHESIS AND DECOMPOSITION OF ORGANIC
P
Organic P constitutes 29 to 65% of topsoil
P (Harrison, 1987), most in low molecular weight compounds (Fares
et al., 1974; Condron and Goh, 1989). Inositol phosphates
can make up to 50% of the identifiable Po, other sugar phosphates,
phospholipids, nucleic acids, phosphonates are also present but
in minor quantities (Stewart and Tiessen, 1987). A large proportion
of Po remains unidentified because of the stable nature of Po-soil
complexes, which prevent extraction of these compounds. The Po
composition of the soil is quantitatively different from that
of living organisms which indicates that specific processes lead
to the selective stabilisation of various Po compounds in soils
(Stewart and Tiessen, 1987).
Organic P is synthesised in plants and
may constitutes approximately 20-80% of P in plant vegetative
tissue, depending upon plant part, age and P supply, and 70-83%
of P in seeds. The more important Po compounds synthesised are
phospholipids (membrane structure), nucleic acids (genetic code)
and inositol phosphates (P storage). Their respective tissue
concentrations in leaves are in the range 0.01%, 0.01% and <0.0001%
and in seeds 0.002%, 0.02% and 0.05%. In general, P concentration
is higher in meristematic tissue. It is common for P in older
plant parts to be remobilised and transported to meristematic
tissue or to seeds (Koide, 1991).
MICROORGANISMS
Brookes et al. (1984) estimated
that microbial phosphate made up 3 to 24% of the Po in temperate
arable or pasture soils. Lee et al. (1990) observed that
glucose additions to a highly weathered soil increased the quantity
of microbial P, but Martin and Correll (1989) detected no Po
synthesis by biomass in wheat rhizosphere soil. A possible explanation
may be that approximately 5-30% of photosynthetically fixed carbon
is rhizodeposited (Lynch and Whipps, 1990) and probably consumed
by the population of rhizosphere microorganisms within 20 µm
of the rhizoplane (Foster, 1986), so that rhizosphere synthesised-P
and root P may not be separable. Rhizosphere organisms commonly
are high in P and often contain inositol polyphosphate granules.
Away from roots, most bacterial activity is isolated in discrete
microsites around organic residues, and the bacteria are usually
smaller and lack polyphosphate granules (Foster, 1986). Most
forms of Po are common to both plants and microbes. Scyllo-isomers
of inositol phosphates are of microbial origin (Cosgrove, 1966),
and phosphonates are produced by the soil microflora or fauna
(Newman and Tate, 1980; Nakashita and Seto, 1991).
Only a limited proportion of P is incorporated
into high molecular weight organic compounds during humification
(Fares et al., 1974; Condron and Goh, 1989). Brannon and
Sommers (1985) have shown that P incorporated into synthetic
humic compounds could resist both chemical and enzymatic hydrolysis.
This stability can be reinforced further by a physical stabilisation
or organic colloids with clay particles (Skjemstad et al.,
1986).
DECOMPOSITION OF ROOTS, LITTER AND EXCRETA
As ecosystems mature, the proportion of
P in organic forms increases and P nutrition becomes more dependent
upon the recycling of P by Po decomposition (Walker and Syers,
1976; Harrison, 1985; McLachlan et al., 1988).
Decomposition processes are the result
of activities of microbes and small animals (Richards, 1987;
Toutain, 1987a; Ponge, 1991) and involve physical breakdown,
transport, catabolism and leaching (Swift et al., 1979;
Toutain, 1987b). Microorganisms have specific degradative functions.
Few bacteria degrade pectin, cellulose or starch while most of
the fungi will. Fungi are commonly the initial agents of catabolism
(Reisinger et al., 1978), in conjunction with invertebrate
animals, protozoa and bacteria (Richards, 1987). Trophic interactions,
for example predation of microflora by amœbae and nematodes
(Cole et al., 1978; Clarholm, 1981; Anderson et al.,
1982; Coleman et al., 1988; Elliot et al., 1988)
or interactions between microflora and fauna (Lavelle et al.,
1992) increase the turn-over of the microbial populations. This
leads to increases in the diversity and activity of enzymes contributing
to the degradation of stable organic compounds (Bottner and Billès,
1987). Communition and mixing by earthworm or termite activity
allows for a greater surface area of substrate to come into contact
with appropriate organisms and enzymes (Kretzschmar, 1987; Garnier-Sillam
et al., 1987). As an example of the role of the macrofauna
on P cycling, Fardeau and Frossard (1992) showed that the quantity
of isotopically exchangeable P at 1 min increased from 2.2 mg
kg-1 in the bulk soil to 87.0 mg kg-1 in the center of a termite
(Trinervitermes geminatus) nest.
The microstructure of soil fabric offers
physical protection to the microflora (Van Veen et al.,
1984; Merckx and Martin, 1987). Chotte et al. (1994) showed
that aggregates >2 cm (isolated by wet sieving without agitation)
represented the most important habitat in a Vertisol; 50% of
the soil weight containing 30% of the total biomass was associated
with fresh or partly decomposed organic skeletons and clay plasma.
Access to the substrate by appropriate organisms and their enzymes
can be limited, either by aggregation of soil particles and microbial
metabolites entombing the residue and the initial decomposer
organism (Ladd et al., 1990), or by allelopathy of the
incumbent organisms. Aggregation may also cause decomposition
to decrease because small, water-filled pores limit oxygen diffusion
to intra-aggregate decomposers or limit access to predators (Bamforth,
1988). The size of pore openings depends on soil particle size
and orientation, while the strength of the aggregate bonding
reflects mineralogy and the nature of the bonding agent (roots,
hyphae or microbial metabolite) (Harris et al., 1966;
Tisdale and Oades, 1982; Payne, 1988). For example, amorphous
alumino-silicates form particularly strong aggregates with small
micropores which are considered to be responsible for the fine
structure and high carbon content (10-12%) of some New Zealand
Andisols under pastures (Jackman, 1964a).
Factors affecting the rate of decomposition
and fate of organic P
The rate of decomposition of Po in litter
and excreta, depends initially on the P solubility in the residue
and subsequently on the rate of decomposer growth. The rate of
decomposer growth depends on substrate accessibility to the organism
and its enzymes, the chemical nature and nutrient content of
the substrate (extent of lignification, C:N:P ratio, Table 5)
and the surrounding soil water (pH, 02 partial pressure, N, S
and P concentrations) (Vaughan and Ord 1985; Sparling, 1985).
Table 5. Phosphorus, carbon,
nitrogen content and C/P, C/N ratios of some plants, animals,
and manures.
|
|
P |
C |
N |
C/P |
C/N |
References |
|
|
mg g-1 dry weight |
|
|
Bacteria |
14-32 |
500 |
150 |
16 |
3 |
1 |
|
Actinomycetes |
15 |
500 |
110 |
33 |
5 |
2 |
|
Yeasts |
6-7 |
470 |
62 |
67 |
8 |
3 |
|
Fungi |
4-6 |
440 |
34 |
73 |
13 |
3 |
|
Soil microbial biomass |
|
|
|
10- 26 |
10-11 |
4 |
|
Earthworms |
9 |
460 |
100 |
51 |
5 |
2 |
|
Maize |
2 |
440 |
14 |
220 |
31 |
5 |
|
Lucerne |
3 |
450 |
33 |
161 |
14 |
6 |
|
Farmyard manure |
5 |
370 |
28 |
69 |
13 |
2 |
|
Tropical manure |
3-15 |
|
0-26 |
|
|
7 |
|
Sewage sludge |
25 |
39 |
68 |
159 |
6 |
8 |
1 Roberts et al., (1955)
cit. in Jenkinson (1981); Brookes et al., (1982);
2 Jenkinson (1981); 3 Jenkinson
(1981), Brookes et al., (1982); 4 Singh et al., (1991), Brookes et al.,
(1984); 5 Miller (1938); 6 Bertrand (1939), cit. in Jenkinson (1981); 7 Pieri (1989);
8 Brossard et al., 1991
Fifteen to 50% of plant P may be soluble
(Pi and monoesters, Table 6). If plant residues are incorporated
into soils this P can be rapidly adsorbed by soil surfaces (Friesen
and Blair, 1988; McLachlan et al., 1988). Sorption of
simple Po compounds by reactive clay surfaces will decrease their
rate of decomposition and may lead to their stabilisation (Tate,
1985). Chopping or grinding plant residues and incorporation
into soil improves residue-soil contact and speeds this process
(Friesen and Blair, 1988; Jenkinson, 1988). In highly competitive
root zones of permanent pasture, roots or mycorrhizal hyphae
can grow into litter and utilise soluble P (Thibaud et al.,
1988; Newman and Eason, 1989) preventing sorption by soil surfaces
or immobilisation by free-living soil flora.
The non-water soluble P (mostly diesters:
nucleic acids, phospholipids, and phosphoproteins) remaining
in the residues are used by fast growing saprophytic fungi, followed
by slower growing fungi with suites of polysaccharases. If there
is adequate moisture and the solution does not become acidic,
bacteria will be able to colonise residues quickly (Richards,
1987). The rate of decomposition at this stage is controlled
by nutrient availability, particularly N (Jenkinson, 1988). Much
of the residue P will be immobilised into decomposers and their
predators. For example, clearfelling increases biomass P in the
Oh horizon as a result of debris input to the forest floor (Hughes
and Reynolds, 1991). The new microbial Po may last for one season
because the mean turnover time of microbial biomass in cultivated
temperate soils appears to be greater than 1 year (Sparling,
1985).
Table 6. Water soluble phosphorus
in Digitaria d. and maize roots (Brossard et Laurent,
unpublished)
|
|
|
Digitaria decumbens |
Maize |
|
|
|
----------- µg
g-1 root ------------- |
|
Total P |
|
175 |
644 |
|
Water soluble |
total P |
37.1 |
107.9 |
|
|
Pi |
16.9 |
56.7 |
|
|
Po |
20.2 |
51.2 |
|
|
C |
5041 |
16463 |
|
|
C/N |
26 |
17 |
|
|
C/Po |
250 |
322 |
Results of McLachlan et al. (1988)
support these ideas. Seven to ten days after applying 33P labeled,
medic residues to a wheat seed bed, 40-50% of the litter P (approximately
equivalent to the soluble P content of the litter) could be found
in soil Pi, whereas the remaining residue Po remained either
as soil Po or biomass P for the remainder of the wheat growing
season, showing that little P mineralisation occurred due to
biomass cycling. The medic residues initially contained 0.3%
P, thus after initial soluble P loss (50%), the non water-soluble
P residue probably had a C:P ratio exceeding 300. The generalised
equation (14) derived by White (1981) shows that little P mineralisation
is expected:
= (14)
where Com/Pom and Cm/Pm are the C:P ratios
of the organic matter being decomposed and the biomass respectively,
and Cy is the growth yield.
If the Cm/Pm is taken to be 15 and the
long term growth yield Cy = 0.5, then it can be calculated that
net mineralisation of P is unlikely to occur until the 4th generation
of decomposers on the medic residues. As C:P ratios of most plant
residues are 140 or greater (Brossard and Laurent, 1992; Tiessen
and Stewart, 1983), initial P mineralisation is probably due
to the release of soluble P by cell lysis (Friesen and Blair,
1988). Mineralised biomass P is more likely to be derived from
substrates with C:P ratios near 30-50. This suggests that most
P mineralised by the biomass is derived from the decomposition
of microflora and fauna which have C:P ratios ranging from 12-45
(Tate, 1985; Brookes et al., 1984). Evidence for rapid
biomass growth on residues applied to cultivated soils followed
by periods of no net P mineralisation from the fresh residue
but extensive P mineralisation from old residues is provided
by McLachlan et al. (1988). During the first 7 days after
adding 33P labeled medic residues to the soil they found rapid
incorporation of 25-30% of the label into microbial biomass and
the percentage of 33P remaining as Po remained constant for the
next 88 days.
This sequence of events is consistent with
observations that at any one time <15% of biomass in arable
soils is active (Sparling, 1985) but carries high adenylate energy
charge and can respond rapidly to inputs of easily decomposable
residue for short periods of time before returning to an inactive
state. During the active period of P cycling in McLachan et
al.'s experiment 83% of the increase in biomass P was derived
from unlabelled soil P reserves which also supplied 60-70% of
the P taken up by the wheat crop. Presumably this resulted partly
from the limited spatial distribution of the 33P labeled, medic
residue in the bulk soil and partly from the more widespread
mineralisation of previously "protected" Po residues
made available by cultivating the soil.
Decomposition in an arable soils may proceed
as follows. Cultivation and residue incorporation or root exudation
activate microflora (Billès et al., 1986) . Decomposers
grow on insoluble P residues and immobilise soil Pi and any P
released by hydrolytic enzymes. Decomposer activity then is reduced
because the carbon substrates become more complex and the initial
decomposers may lack the appropriate hydrolases to complete decomposition.
Access to the substrate by organisms and their enzymes may be
limited by environmental factors (changes in soil fabric, aggregation
and climate).
Mechanical disturbance of soils affect
these processes, reducing aggregation (Harris et al.,
1966) and promoting Po mineralisation (Hedley et al.,
Ch. 5). Cultivation, freeze/thaw or wet/dry cycles (Harris et
al., 1966) and root growth through weak aggregates (Payne,
1988) also provide such disturbance. Cultivation decreases the
Po content of unfertilised soils (Tiessen and Stewart, 1983;
Condron et al., 1990) and limits Po synthesis in favour
of Pi accumulation in fertilised soils (Johnston, 1989; Ivarsson,
1989). Cultivation causes preferential mineralisation of diester-P
while monoesters remain (Condron et al., 1990) due to
their greater stabilisation through sorption.
In undisturbed soils, like permanent pastures,
the degree of protection of Po is high, and addition of inorganic
P fertilizer results in rapid accumulation of organic P (Jackman,
1955; Walker et al., 1959; Jackman, 1964a, 1964b; Brossard
and Laurent, 1992). The rate of accumulation of Po in regularly
fertilised pastures decreases per unit increase of total soil
P (Perrott et al., 1989). The rate of Po accumulation
is a function of plant productivity and the rate of Po mineralisation.
In legume-based pastures the rate of organic matter mineralisation
is also a function of soil nitrogen status.
Modelling the P Cycle
Most models of P dynamics have concentrated
on a specific set of soil and plant processes operating over
time scales of hours to months. While annual P budgets have been
constructed for a variety of ecosystems, there have been few
attempts to develop dynamic models of P cycling at ecosystem
level over a number of years.
The P cycle in grazed temperate perennial
pasture in New South Wales, Australia was modeled by Blair et
al. (1977). The model is based on an available P pool comprising
soil solution P plus bicarbonate extractable P. Mineralisation,
fixation, fertilizer dissolution, plant uptake and litter return,
and animal transfers were included in the model (Figure 7). A
similar P model by Katznelson (1977) gives typical pool sizes
and a qualitative description of the controls on the flows of
P for pastoral ecosystems.
Figure 7. P cycling model (Blair et
al., 1977)
The simulation model of P cycling in semiarid
North American grasslands by Cole et al. (1977) considers
solution P, labile and stable Pi and Po, decomposers, litter,
standing dead, live plants partitioned into shoots, crowns and
roots, and consumers. Labile Pi was defined as P which can enter
soil solution by isoionic exchange within an appropriate time.
Processes represented include plant uptake and translocation
of P, uptake by decomposers, and mineralisation of labile and
stable Po. Soil solution P is used to control the rates of P
uptake by plants and decomposers but, because the model was designed
to run at time steps of one day or more, P flows are taken directly
from the labile Pi pool. The P uptake by decomposers was the
largest annual flow in the simulation. Long term stability of
the model was sensitive to the rate of Po mineralisation.
The decomposition of soil organic matter
was the focus of a model of C, N, P, and K dynamics described
by Smith (1979). The initial disintegration phase of the decomposition
of plant or microbial organic matter yields free Po, which can
in turn yield Pi or be sorbed onto soil colloids in competition
with Pi sorption. The sorbed Po is protected from microbial degradation
and is only slowly released to soil solution Po. The sorption
of both Pi and Po is described by a Langmuir isotherm. The dissolution
of slowly soluble mineral P or fertilizer P are described separately.
Microbial and plant uptake of P is modeled with Michaelis-Menten
kinetics with maximum uptake rate dependent on the labile P concentration
in the cell. Growth and death of the microbial population are
allowed for.
P and N cycles in Calluna heathland
ecosystems have been modeled by de Jong and Klinkhamer (1983)
using an annual time step. The model has state variables for
total soil Pi, ash from the burning of plants or litter, root
zone and sub-root zone humus, live and dead roots, green and
brown shoots, flowers, wood and litter. Plant uptake, senescence,
burning, litter decomposition, dead roots, and humus are represented.
The only P input considered is rainfall, while leaching of Pi
and downward movement of P in humus are the outputs from the
ecosystem.
The P cycle in Calluna heathland
has been modeled again by Chapman et al. (1989) using
either annual or monthly time steps. The model includes a water
soluble P pool with P sorption and the dissolution of P from
ash and mineral compartments considered. Plant production is
controlled by available P equal to the sum of the water soluble
and adsorbed pools. The P content of aboveground plant material
is a function of plant age. The model allows for two optional
approaches to mineralisation of P from organic matter. Rates
of P release can be related to organic matter decomposition,
or the composition of other relevant compartments can be made
to remain constant, with transfers to the soil solution calculated
accordingly. Both approaches give similar results. A leaching
loss of Pi is calculated from drainage and the concentration
of P in soil solution. Aeolian losses of P by burning are also
considered.
A model of C, N, P, and S cycling in perennial
pastures in New South Wales, Australia (McCaskill and Blair,
1988) accounts the soil biomass, organic and inorganic soil pools,
grass and clover, sheep and excreta. Inorganic P was split into
plant available P and two fixed pools representing native soil
P and P of fertilizer origin. Uptake of S, P and N was based
on diffusion and plant uptake kinetics (McCaskill and Blair,
1990). Relationships were simplified to reduce the computer time
required. Nutrient uptake was allowed to occur within an effective
radius calculated from the root radius plus root hair length,
the half distance between roots, and the effective diffusivity
coefficient. Mycorrhizal infection was simulated by increasing
the effective root length. Plant uptake could be limited by either
diffusion to the root, root uptake capacity or plant demand.
To account for the effects of rhizosphere acidification, variation
of the buffer capacity with distance from the root and other
factors, it was necessary to calibrate the effective rooting
radius against field data for plant tissue concentrations.
As a submodel for the Erosion-Productivity
Impact Calculator (EPIC) model Williams (1990) and Jones et
al. (1984) developed a soil and plant P model based on a
labile P pool equivalent to P extractable with anion exchange
resin. The labile pool is buffered by an active Pi pool, the
size of which is estimated from the size of the labile pool and
an availability index calculated from P sorption over 6 months.
Relationships to derive labile P and the availability index have
been established using other soil P tests and soil properties
such as the clay content (Sharpley et al., 1984; Sharpley
et al., 1989). The active pool is in slow equilibrium
with a stable mineral P pool. Plant P uptake is controlled either
by plant demand or by a function of the size of the labile P
pool, the root mass in each soil layer, and soil moisture. The
model has pools for fresh and stable organic matter. Mineralisation
from the fresh Po pool is controlled by the C:N and C:P ratios,
soil water, temperature and the stage of residue decomposition.
The stable organic matter pool is subdivided into mineralisable
and non-mineralisable fractions depending on the length of time
under cultivation. P can be lost from the ecosystem in plant
produce, eroded sediments and surface runoff. The effects of
agricultural management are a key feature of the model.
Figure 8. The pools and flows of phosphorus in
the CENTURY model
The CENTURY model (Parton et al.,
1988; Figure 8) simulates long-term C, N, P, and S dynamics in
a variety of ecosystems with an emphasis on soil organic matter.
Plant residues and animal excreta are partitioned into structural
and metabolic components, while soil organic matter is divided
into active, slow and passive pools. Each pool has a characteristic
turnover time and range of C:N, C:P and C:S ratios. Turnover
times and the efficiency of organic matter stabilisation are
functions of soil texture. The C:P ratio of organic matter entering
each of the active, slow and passive pools varies as a function
of the labile P level where labile P is defined as phosphate
which is isotopically exchangeable or extractable with anion
exchange resin. Plant P status varies with the balance between
supply from the labile pool and plant demand. Increasing N status
is assumed to increase the availability of P to plants. Given
the monthly timestep of the model, the labile pool is assumed
to be in equilibrium with a sorbed pool equivalent to 0.1M NaOH
extractable P, in a curvilinear relationship described by sorption
affinity and sorption maximum parameters. This pool is in slow
equilibrium with a strongly sorbed pool. P can enter the cycling
pool by weathering of a parent material and can be slowly fixed
in an occluded pool.
There is still considerable scope for the
development of mechanistical P cycling models which can be applied
over long time scales in order to study the transformations of
P between various inorganic and organic soil pools and to account
for losses from the ecosystem. Transport of P between ecosystem
components in the landscape with runoff, eroded sediments and
animal excreta also needs to be considered in ecosystem level
models. Model development needs to characterise processes at
an appropriate level for application to a wide variety of soils
and managed and natural ecosystems. Plant and microbial uptake
of P and the rate of organic matter decomposition can be limited
by the availability of other nutrients, particularly N, K, and
S. Hence it is desirable that models of ecosystem P dynamics
also consider these other elements.
An alternative approach has been to model
the residual value of P fertilizer in agricultural systems where
empirical relationships are derived between crop or pasture yields
and a labile P pool (Helyar and Godden, 1977b; McCall and Thorrold,
1991). The approach lends itself to economic analysis (Godden
and Helyar, 1980; Scobie and St. Pierre, 1987). P is lost from
the labile pool either by first order linear kinetics or by an
inverse function of time since fertilizer application (Probert
and Williams, 1985).
Conclusions
The most noticeable advances of the last
15 years have been the integration of abiotic processes with
biological processes into more complex models, stressing the
role of the soil fabric and of the biological interactions (between
organisms, or between organic components and minerals). Concerning
the abiotic processes significant recent advances have been:
