| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a School of Earth and Environmental Sciences, University of Adelaide, PMB 1, Glen Osmond, South Australia, Australia 5064
b South Australian Research and Development Institute, PO Box 397, Adelaide, South Australia, Australia 5003
c CSIRO Land and Water, PMB 2, Glen Osmond, South Australia, Australia 5064
* Corresponding author (warwick.dougherty{at}adelaide.edu.au)
Received for publication November 18, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONCONCEPTUAL MODEL OF PHOSPHORUS...PHOSPHORUS SOURCESPHOSPHORUS MOBILIZATIONPHOSPHORUS TRANSPORT—...SCALE AND LANDSCAPE EFFECTS...RAINFALL SIMULATIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: CSA, critical source area • VSA, variable source area
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONCONCEPTUAL MODEL OF PHOSPHORUS...PHOSPHORUS SOURCESPHOSPHORUS MOBILIZATIONPHOSPHORUS TRANSPORT—...SCALE AND LANDSCAPE EFFECTS...RAINFALL SIMULATIONCONCLUSIONSREFERENCES |
|---|
Concentrations of P in surface runoff from intensive pasture systems are typically high, being in the range of 1 to 10 mg L–1 (Greenhill et al., 1983; Nash and Murdoch, 1997; Nexhip et al., 1997; Fleming and Cox, 1998). The concentration of P in water is one of the limiting factors to productivity in waterways (Wetzel, 1983). Consequently, the transfer of P (the collective term for those processes resulting in the loss of P in runoff) from pasture systems to waterways can contribute to the development of toxic algal blooms. There is a general consensus that improved management of P in intensive pasture systems is needed so that P transfers in surface runoff and the subsequent effects on water quality are minimized.
The significance of P transfer in surface runoff from intensive pasture systems compared with other sources is difficult to assess because P sources are highly varied within most watersheds. Consequently, the extent of the required reduction in losses of P in surface runoff is difficult to define. It is also difficult to directly link surface water quality measured at the farm or paddock scale to that at the watershed scale. Regulatory authorities generally set water quality guidelines at watershed scale. For example, in a number of Australian coastal watersheds, the general objective is to reduce P losses from agricultural industries (Anonymous, 1998, 2002), whereas a more specific water quality target of P < 0.05 mg L–1 has been set for watersheds.
There has been a significant investment in research to determine the type and magnitude of processes resulting in P transfer in surface runoff and how management influences this transfer. This research has utilized a wide range of approaches. However, little consideration has been given to the effect of scale and methodology of the research on the processes of mobilization (the process whereby P is transferred from a P source to the runoff) and subsequent concentrations, and loads of P measured in surface runoff. Small plot studies allow examination of the effect of specific factors on P transfer. Broader-scale (e.g., watershed) studies allow quantification of P transfer as a result of general management and landscape factors, but are generally unable to provide detailed information about the effect on P transfer of specific management factors. A mixture of research at a range of scales offers promise of greater insight into the influence and interaction of both management and landscape factors (Haygarth and Jarvis, 1999).
Commonly used techniques for measuring runoff P concentration, such as rainfall simulation, may allow prediction of concentrations in runoff at paddock, farm, or broader scales. However, there are almost always differences in the hydrological processes, and therefore the nature of P transfer, at these different scales. For example, runoff coefficients are high and runoff residence times short for rainfall simulation, whereas under natural rainfall at paddock or watershed scale, runoff coefficients are relatively low and residence times relatively long. An understanding of these differences and the factors influencing them may allow P mobilization and transfer to be examined more effectively. Furthermore, a clearer understanding of landscape processes will contribute to efficient management of P.
This review examines those factors affecting P transfer in surface runoff from intensive pasture systems in temperate regions. A conceptual model of P transfer is used as the basis for examining the effects of scale and landscape characteristics on P transfer. Subsequently, the efficacy of predicting P transfer at various scales is also examined. Deficiencies in our understanding and conceptualization of the scale issue are highlighted and priorities for future research identified.
Phosphorus transfer in surface runoff is specifically examined in this review because (i) in a majority of intensive pasture systems it is the primary pathway for transfer of P off-site, and consequently has been the subject of a great deal of research; (ii) the findings of some of this research are of lesser value than they might have been had "scale" issues been considered more carefully in the design and interpretation of the research; and (iii) it is the primary transport pathway dealt with in management tools such as P risk indexes.
The fate of P in runoff is strongly influenced by factors such as stream bank erosion, ground water inflows, and internal cycling within the stream by sediment–water interactions and biological action (Wetzel, 1983). The discussion of P mobilization and transfer in this review is limited to terrestrial processes involved in P transfer and does not deal with processes occurring to P "in-stream." It follows that the use of the term "watershed" refers to those upper areas of watersheds (i.e., low-order components of the landscape).
|
CONCEPTUAL MODEL OF PHOSPHORUS TRANSFER |
|---|
|
TOPABSTRACTINTRODUCTIONCONCEPTUAL MODEL OF PHOSPHORUS...PHOSPHORUS SOURCESPHOSPHORUS MOBILIZATIONPHOSPHORUS TRANSPORT—...SCALE AND LANDSCAPE EFFECTS...RAINFALL SIMULATIONCONCLUSIONSREFERENCES |
|---|
|
|
PHOSPHORUS SOURCES |
|---|
|
TOPABSTRACTINTRODUCTIONCONCEPTUAL MODEL OF PHOSPHORUS...PHOSPHORUS SOURCESPHOSPHORUS MOBILIZATIONPHOSPHORUS TRANSPORT—...SCALE AND LANDSCAPE EFFECTS...RAINFALL SIMULATIONCONCLUSIONSREFERENCES |
|---|
|
A large amount of P in the soil–pasture system can be present in organic forms. Under favorable growth conditions, irrigated pastures can produce 10 to 30 Mg of dry matter (DM) ha–1 yr–1, equating to an uptake of 30 to 90 kg P ha–1 yr–1 at a typical plant P content of 0.3%. Typically, in Australian dairy pastures, aboveground biomass at any given time is in the range of 1 to 3000 kg DM ha–1. Inefficiencies in the utilization of this pasture will result in a considerable amount of P in plant tissue being deposited on the soil surface as leaf litter. Phosphorus is also found on the surface and within the internal structures of leaf tissue. Typically, water-soluble P in pasture material is 10% of total P (Sharpley, 1981; Schreiber and McDowell, 1985), although in dried pasture values up to 70% have been reported (Jones and Bromfield, 1969). Microbial processes play an important role in mobilization (mineralization) and immobilization of the organic P pool (Bromfield and Jones, 1972; Hutchinson and Roper, 1985; Perrot et al., 1992). Temperature, moisture, and plant growth influence the cycling of P through the microbial pool and the balance between mineralization and immobilization and thus soluble soil P.
Deposition of fecal P on intensively grazed pastures can be high. Deposition of fecal P is a function primarily of the grazing intensity (Nexhip et al., 1997). High stocking rates may require substantial increases in supplementary feed that not only increase the amount of fecal P deposited on the soil surface but also increase the importation of P (within the supplementary feed). Furthermore, fecal P concentration is affected by dietary composition (Wu et al., 2000; Dou et al., 2002; Ebeling et al., 2002). Increasing dietary P above basal animal requirements increases the P content of feces, 30 to 60% of which is water soluble (Wu et al., 2000; Dou et al., 2002).
|
PHOSPHORUS MOBILIZATION |
|---|
|
TOPABSTRACTINTRODUCTIONCONCEPTUAL MODEL OF PHOSPHORUS...PHOSPHORUS SOURCESPHOSPHORUS MOBILIZATIONPHOSPHORUS TRANSPORT—...SCALE AND LANDSCAPE EFFECTS...RAINFALL SIMULATIONCONCLUSIONSREFERENCES |
|---|
|
The specification of whether the process of mobilization of P is chemical or physical is not simple. The difficulty in specifying the process arises because of a lack of routine methods for accurately distinguishing ionic phosphate. Instead, operational definitions of the nature of P have generally been employed (Leinweber et al., 2002). The most popular fractionation schemes in the study of runoff have involved filtration using 0.45-µm filters to separate particulate (>0.45 µm) and dissolved (
0.45 µm) P (e.g., Fleming and Cox, 1998; Withers et al., 2001). Phosphate ions have often been assumed to be the P fraction that is both capable of passing through a 0.45-µm filter and of reacting in the molybdenum blue method (Murphy and Riley, 1962). Although this classification is still widely used, the limitations have been acknowledged (Haygarth and Jarvis, 1999). This classification suffers from the ability of some of the colloidal particulate matter (i.e., particles 1 nm to 2 µm in diameter) (van Olphen, 1977) to pass through 0.45-µm filters (Haygarth et al., 1998; McDowell and Sharpley, 2001) and the ability of the molybdenum blue method to solubilize some organic P (Tapachak, 1983; Baldwin, 1998).
Colloids capable of passing through 0.45-µm filters are common components of soil and include oxides, clay minerals, organic matter, and amorphous material (Goldberg et al., 2000). Colloidal materials are notable for their large surface area to mass ratios and the consequent reactivity. Furthermore, these colloids can provide an important transport mechanism for pollutants (Kaplan et al., 1996; Saiers and Hornberger, 1999). Colloid transport can be promoted by low ionic strengths in the interacting water. Saiers and Hornberger (1999) observed significant increases in movement of colloidal kaolinite when ionic strength was reduced. Soil mineralogy and sodicity will also influence colloidal transport.
Phosphorus in surface runoff from intensive pasture systems is dominated by soluble or fine colloidal material (Nelson et al., 1996; Nash and Murdoch, 1997; Fleming and Cox, 2001). Exceptions to this may occur when livestock remove excessive pasture, exposing the soil, and/or physically disturb the soil surface by trampling (McDowell et al., 2003a). This may happen when pasture is overgrazed, and/or excessively wet areas of paddocks are grazed. Other areas where ground cover is likely to be poor and soil trampling occurs are around gates, feed or water troughs, and on livestock tracks (although these tend not to be so prominent in intensive pasture systems). In agricultural systems prone to erosion, such as cropping systems, P in runoff is predominantly in particulate form.
Physical Processes of Mobilization
Detachment is primarily a physical process controlled by the availability of energy that can be supplied either as kinetic energy from raindrop impact (Ahuja et al., 1982; Torri and Borselli, 2000) or from flowing water (Torri and Borselli, 2000; Truman et al., 2001). In addition, electrical forces can supply energy involved in physicochemical processes such as slaking or dispersion. Only in the case of dispersion will mineral particles become spontaneously mobilized in water. Particulate matter in surface runoff can range in size from 1 nm to 2 µm (colloidal particles) through to soil aggregates (<10 mm). It is comprised of mineral, organo–mineral compounds, macro organic matter, and soil micro- and macro-fauna.
Inherent properties of the soil–plant system can influence particle detachment and can be illustrated by contrasting a regularly tilled soil with a permanent pasture soil. The tilled soil will often have little or no ground cover and poor aggregate stability because of relatively low organic matter content. In contrast, intensive pasture systems will generally have 100% ground cover that intercepts raindrops and therefore reduces the kinetic energy from this source. As noted above, an exception to this may occur when pasture cover is reduced due to livestock activity. The pasture system also provides organic matter that contributes both directly and indirectly to the stabilization of the soil surface (Loch, 2000). The mobilization of P in particulate forms can also occur as a result of the kinetic energy of surface runoff. Truman et al. (2001) observed a linear relationship between runoff rate and sediment yield on arable soils. Similarly, Gabbard et al. (1998) observed large increases in sediment transport as a result of increased runoff rates.
The transfer of P is not only a function of the quantity of soil mobilized, but also of the concentration of P in the material transported (Haygarth and Jarvis, 1999). This phenomenon is the result of preferential mobilization of smaller particles (e.g., clays) that contain a higher P content than coarse particles, the lower amounts of energy required to transport these particles (Quinton et al., 2001; Palis et al., 1997), and their slower settling velocities. The ratio of the P content of the eroded soil to that of the bulk soil is known as the enrichment ratio. Enrichment ratios typically range from 1 to 2. Enrichment ratios may also be affected by runoff rate. Quinton et al. (2001) observed a disproportionately high loss of P in sediment from small storms. They attributed this to preferential erosion of fine clay particles of high P content during small runoff events, whereas in larger storm events, the greater energy of overland flow transported more coarse material containing relatively lower concentrations of P. Particulate P can also be derived from dust accumulated on plant surfaces (Schreiber and McDowell, 1985). This is not likely to be a large component of P losses from soils under pasture. However, it is important to consider because discriminating this source from other sources of P associated with plants is difficult.
Chemical Processes of Mobilization
Inorganic Phosphorus
The equilibrium between solid-phase P (adsorbed or precipitated) and the solution or runoff phase can be represented by a generalized model (Eq. [1]) where the solid phase is in equilibrium with the solution phase:
 | [1] |
Increases in the concentration of P in the solution phase (such as may result from fertilizer addition) will also increase the concentration of P in the solid phase. The shift in equilibrium is rate limited, such that increasing the time of reaction between solid-phase P and solution will increase the amount of P in solution (Romkens and Nelson, 1974; Sharpley et al., 1981b).
Mobilization (or in some cases immobilization) of phosphorus can occur from both a stable soil surface (runoff containing little or no sediment) and sediment entrained in the runoff. The concentration of P in the topsoil is particularly important in determining runoff P concentrations because this is where runoff interacts with the soil. As previously noted, P concentrates in the top of the soil in permanent pasture systems, with the highest concentrations being closest to the surface (Haygarth et al., 1998; Sharpley, 2003). The interaction of surface runoff with soil is greatest at the surface and declines exponentially with depth (Ahuja et al., 1981). The depth of interaction varies with slope, rainfall intensity, and kinetic energy of rain (Sharpley, 1985). Investigation of the depth of interaction has mainly been undertaken in repacked soils lacking natural structure and plants (e.g., Sharpley et al., 1981a; Ahuja et al., 1982; Ahuja and Lehman, 1983). A number of authors have attempted to simulate the effect of plants (Ahuja et al., 1982; Sharpley, 1985). Plants can change the processes of soil–water interaction significantly. Plants will reduce the kinetic energy of raindrops, thereby reducing detachment and subsequent mixing of detached material with water ponded on the soil surface. Plants also increase the depth of interaction by reducing runoff velocities and by providing less dense topsoil with greater hydraulic conductivity (Ahuja and Lehman, 1983).
The existence of a linear soil P–runoff P relationship has been observed under controlled rainfall simulation conditions by various authors (Sharpley and Syers, 1976; Pote et al., 1996; Pote et al., 1999; Torbert et al., 2002). The soil P–runoff P relationship has also been described using a "bent stick" model, the soil P concentration at the bend being known as the "change-point" (McDowell and Sharpley, 2001; McDowell et al., 2003b). Soil P–runoff P relationships have been observed to vary significantly at different times and with different soil types (Pote et al., 1999). Hesketh and Brookes (2000) found the linear relationship between soil P and P in drainage waters changed over time. Similarly, Gillingham et al. (1997) observed a linear relationship for soils with an Olsen P in the 0 to 50 mg kg–1 range, but the relationship changed with season.
Determination of a soil P–runoff P relationship at broader scales has been less successful. McDowell and Trudgill (2000) found no correlation between soil P and runoff P in a 94-ha watershed of mixed land use. They attributed the lack of any correlation to unusual rainfall and runoff conditions. Soil P status, processes of mobilization, and hydrology would most likely be highly variable in such a large area. Their study highlights the effect of hydrology and the challenges that exist in using soil P as a predictor of runoff P concentrations at broad scales.
The best relationships between soil test P and runoff P have been derived in pure soil–pasture systems at small scales where uniformity is high and variability associated with plant biomass, livestock, and manure effects is eliminated. We hypothesize that there will be large differences in these relationships at different times of the year because of the biological and organic components of the P cycle. Furthermore, the effects of grazing (e.g., changes in biomass P, trampling effects, and fecal deposition by livestock) are difficult to assess in small plots.
It is widely accepted that the kinetics of P release from soils (stable surfaces such as under pasture as well as in mixed soil–water phases in erosive systems) are controlled by diffusion rather than chemical, kinetic rates. Ogdawa and Sparks (1986) examined the effect of different mixing methods on the rate of P adsorption in a number of clay and soil systems. The rate constants for batch experiments in which suspensions of soil were shaken at 180 rpm were approximately 40 times greater than those of either static or continuous flow systems. Furthermore, the activation energies were very low in the static and continuous flow systems and were consistent with diffusion processes (Sparks, 1985). In the batch experiments the activation energies were close to those expected from chemical rate–controlled reactions, although they were still within the diffusion-controlled range.
These static and continuous flow systems where solution–soil interaction is relatively small are akin to the situation that occurs under a highly stable soil surface such as in pasture systems. The shaken, batch experiments are more likely to represent mixed solution–soil systems observed under erosive systems examined by various researchers, for example, Ahuja et al. (1982), Ahuja and Lehman (1983), and Sharpley (1985). We hypothesize that there will be large differences between bare soils and stable pasture soils in the effect of parameters such as runoff rate on P concentrations in runoff. The concentration of P in runoff will be a function of the equilibrium between the solid and solution phases. The equilibrium will be determined by three factors, namely, rate of release of P from soil to solution, time of contact, and soil P concentration.
Characterizing the processes and quantifying the rates of P release from the soil surface and the changes in these rates under differing conditions may provide a basis for understanding the differences in P mobilization at different scales. Such studies may assist in explaining the differences in measured P transfer using different techniques (e.g., rainfall simulation on small plots compared with paddock scale under natural rainfall of much lower intensity).
Organic Phosphorus
The contribution to runoff P from plant and manure P has been identified as being potentially significant (Bromfield and Jones, 1972; Sharpley, 1985; Hutchinson and Roper, 1985; Nash and Halliwell, 1999). Phosphorus leached from plant residues originates from a number of pools. First, plants may have a surface coating of dust particles that is rapidly removed on rainfall. Second, plant surfaces are also coated with a layer of phosphate ions exuded by the plant. Third, there is a slow release of phosphate ions from the cellular component of the plants.
The contribution of P to runoff P from aboveground biomass has primarily been studied using materials such as wheat (Triticum aestivum L.) straw, "hayed off" pasture, or dried and ground plant material. Early research by Bromfield and Jones (1972) examined the leaching of P from "hayed off" clover (Trifolium subterraneum L.) and phalaris (Phalaris aquatica L.) mixtures. They found that a large proportion of the P present was water soluble (60–70%) and that approximately 35 and 40% of this was leached from clover and phalaris, respectively. For a given amount of rainfall, they observed that the amount of P leached decreased with increasing intensity, suggesting that P release at longer times was controlled by a combination of kinetic and diffusion factors rather than kinetics alone.
Surface dust and phosphate ions were lost when wheat straw was leached with simulated rainfall (Schreiber and McDowell, 1985). An initial rapid rise in P concentration was observed (associated with surface dust and phosphate ions) followed by a more gradual decline (associated with leaching and diffusion of P from within the wheat straw). Similar results were obtained by Schreiber (1985), who examined leaching of P from wheat straw.
Sharpley (1981) found that application of P fertilizer (50 and 100 kg P ha–1) to soil increased the concentration and mass of P leached from the plant canopy. These increases were a result of elevated P concentration in tissue as well as the increase in plant production. The capacity for leaching of P from the plant canopy was regenerated within 24 h of a leaching event at both 50 and 100 kg P ha–1. The concentration of P in runoff derived from plant P was sufficiently high that it suppressed P release from the soil. Furthermore, P in the runoff derived from plants was sorbed by the soil. The proportion of P in runoff that was leached from the plant canopy was greater at lower fertility levels. Sharpley (1981) concluded that modeling of runoff P concentrations should include allowance for contributions from the plant canopy.
Fecal P can contribute to P in runoff (Ebeling et al., 2002; Mundy et al., 2003). Fecal P deposition rates at the soil surface can be high in intensive grazing industries such as dairying, for example, 20 to 30 kg ha–1 yr–1 (Nexhip et al., 1997; Nash and Halliwell, 1999; Mundy et al., 2003). Rates of fecal P deposited on the surface are highly dependent on dietary regime (Wu et al., 2000) and grazing pressure (Nexhip et al., 1997). Increasing concentrations of P in feces as a result of a high-P diet increased the concentration of P in runoff under simulated rainfall (Ebeling et al., 2002). Mundy et al. (2003), however, found that the contribution of fecal P to runoff P was small.
The relative dearth of studies on the effect of organic pools on runoff P is notable. This is a particular issue for intensive pasture-based enterprises where organic matter cycling can be large. The reason for the lack of such studies is not clear. It is likely to be a combination of factors, partly because of the nature of the systems being studied (e.g., bare soils or lack of variability in P pools), the difficulties associated with such studies (e.g., confounding effects of plant biomass and treatments to alter biomass such as grazing), and a lack of resources required to investigate these effects.
Incidental Mobilization
Incidental mobilization is a variation of both physical and chemical processes of mobilization and results in high concentrations of P in runoff. Its potential importance has lead to it generally being identified as a distinct mobilization process (Haygarth and Jarvis, 1999). Incidental mobilization is the direct loss of P from fertilizer or manure sources soon after their application (Haygarth and Jarvis, 1999). Phosphorus concentrations in excess of 50 mg L–1 have been observed in runoff occurring within 24 h of fertilizer application (Nash et al., 2000; Greenhill et al., 1983). These high concentrations were attributed to the direct dissolution of highly soluble P from fertilizer granules, or from fertilizer P that was poorly equilibrated with the soil. Incidental losses from dung are not as large and the effect of time since grazing and dung deposition is relatively weak compared with that for fertilizer (Nash et al., 2000; Mundy et al., 2003). This is because of the relatively low amounts of dung deposited under grazing conditions, compared with fertilizer, and the relatively small surface area of dung exposed (Nash and Halliwell, 1999). In colder regions of the Northern Hemisphere where confined livestock operations during winter result in the need to apply large amounts of manure at once, incidental mobilization of manure P may be important (Preedy et al., 2001).
|
PHOSPHORUS TRANSPORT— PATHWAYS AND PROCESSES |
|---|
|
TOPABSTRACTINTRODUCTIONCONCEPTUAL MODEL OF PHOSPHORUS...PHOSPHORUS SOURCESPHOSPHORUS MOBILIZATIONPHOSPHORUS TRANSPORT—...SCALE AND LANDSCAPE EFFECTS...RAINFALL SIMULATIONCONCLUSIONSREFERENCES |
|---|
Hillslope Hydrology
The generation of surface runoff generally will be a combination of the infiltration excess and saturation excess models proposed by Horton (1937) and Hewlett (1961), respectively. The infiltration excess model of Horton (1937) proposed that runoff would occur as a result of the rainfall rate exceeding the soil infiltration rate. The saturation excess model of Hewlett (1961) proposed that rainfall infiltrates and subsequently accumulates at various positions in the landscape as a result of subsurface processes. Consequently, infiltration in these saturated areas becomes negligible and any subsequent rain falling on these areas runs off (Fig. 4)
. Both processes were observed to occur in a detailed study of hillslope hydrology and runoff generation by Srinivasan et al. (2002). The occurrence of either process in a catchment is dependent on climatic, geomorphic, and management factors (Chorley, 1978). In intensive pasture systems, infiltration rates tend to be relatively high compared with cropping systems because of the maintenance of ground cover and favorable soil structure that results from high levels of organic matter. These factors will favor the model of runoff generation proposed by Hewlett (1961). An exception to this may occur in high-traffic areas, such as near water troughs, around gates, and stock camps, or when compaction and disruption of the surface soil has occurred due to excessive livestock activity when soils are wet.
|
|
|
|
|
The identification of areas that do generate runoff may be particularly important for the location of research sites. For example, siting rainfall simulation studies in mid- or upper slope positions that infrequently contribute to runoff may lead to inaccurate quantification of P transfer when extrapolations are made to landscapes that include other landforms (e.g., crests, plains, and valley bottoms). Systematic variations in source factors within a landscape may also contribute to such errors in extrapolation. Heathwaite and Dils (2000) observed that runoff P concentrations were greater from plots located further down the slope. In attempting to model runoff P export or concentrations, it is important to understand the runoff generation processes and the spatial and temporal distribution of runoff. Small-scale studies (e.g., rainfall simulation) will not be able to provide such information as they only provide a limited spatial and temporal description of P transfer. Modeling noncontributing areas and the use of data generated under conditions vastly different from those occurring in the watershed may provide misleading models of runoff P (Gburek and Sharpley, 1998). The study of Srinivasan et al. (2002) highlighted the complexity of studying the dynamics of contributing areas.
Consideration of the contributing areas in a watershed will aid our understanding of the processes of mobilization. If for instance we assume that the whole watershed is contributing runoff, but in reality the contributing area is only 10% of the total area, the runoff rate in the contributing area is 10 times greater than if we assume that the whole catchment is contributing. The effect of location and subsequent flow characteristics needs to be considered when designing studies examining P transfer in surface runoff.
The mapping of topographic wetness index (TWI) is an approach that has potential to predict the likely extent of VSAs. The TWI is defined as ln(a/tan ß), where a is the upslope contributing area per unit contour and tan ß is the local slope angle. Areas with greater upslope contributing area and that are flatter have a higher TWI and are therefore potentially VSAs. However, the prediction of VSAs is highly sensitive to a number of parameters associated with the models used to derive the a and tan ß values (Quinn et al., 1997). Further investigation of tools for identifying VSAs is required.
An extension of the VSA concept is that of critical source area (CSA). In the CSA concept, only P sources in that part of the landscape generating runoff actually influence runoff P concentrations; other parts of the landscape have no influence, irrespective of source factors in these areas. It is these CSAs that should be managed for most efficient reductions in runoff P concentrations. For example, the adoption of practices that reduce P export by 50% on that 10% of the watershed generating runoff will lead to a 50% reduction in P export for minimal change in management. Contrast this with implementation of new management practices across the whole watershed that will still only result in a 50% reduction in P export but are far more demanding to implement and less likely to be adopted by land managers. The identification of CSAs is critical to the management of water quality while minimizing the effects on farm profitability and increasing the likely adoption of such strategies. Careful consideration of the effect of VSA and CSA concepts in designing research is essential.
|
SCALE AND LANDSCAPE EFFECTS ON PHOSPHORUS MOBILIZATION AND TRANSPORT |
|---|
|
TOPABSTRACTINTRODUCTIONCONCEPTUAL MODEL OF PHOSPHORUS...PHOSPHORUS SOURCESPHOSPHORUS MOBILIZATIONPHOSPHORUS TRANSPORT—...SCALE AND LANDSCAPE EFFECTS...RAINFALL SIMULATIONCONCLUSIONSREFERENCES |
|---|
In relation to P transfer, scale is often poorly defined and sometimes confused. It is important to differentiate issues purely associated with scale in the sense of size alone (e.g., plots 1 vs. 10 m long), and factors that may actually result, at least partially, from the measurement technique adopted or factors associated with landscape (e.g., change in hydrological processes). A particular difficulty is that a change in scale often results in a change in a range of factors such as hydrological pathways, sources, and measurement methods. These different pathways and differences in hydrology are often lumped together under scale. For example, Cornish et al. (2002) compared various measures of runoff P concentrations at different scales. This study compared measures of P concentration in runoff from small 3-m2 plots, large plots (30 m2), paddocks (4 ha), and whole farm (140 ha) under natural rainfall. They concluded that the effect of scale on concentrations of P in runoff was small. Although not specifically defined by Cornish et al. (2002), by the nature of the study, their definition of scale included changes in size, hydrology, sources, and in-stream processes. Almost certainly, all of these factors varied among the various scales they studied.
In a similar study, Dougherty (unpublished data, 2003) compared P concentrations in runoff from small plots (2 m long) nested within larger runoff plots (50 m) across a range of fertility levels. Simulated rainfall was applied at 80 mm h–1 to the small plots and at 8 mm h–1 to the large plots. Runoff P concentrations were significantly (P < 0.05) lower (approximately 50%) from the smaller plots. However, the cause of differences in P concentrations could not be attributed specifically to plot size or rainfall intensity effects because the differences in size and rainfall intensity were confounded.
Effects of Scale on Processes Determining Phosphorus Mobilization
There is a continuum of scales at which P mobilization and transport have been examined. For convenience they have often been arbitrarily divided on the basis of the most likely processes occurring at each scale. Studies can be categorized as laboratory-, profile-, plot-, field-, and watershed-scale studies. Whereas laboratory- and profile-scale studies can be clearly differentiated, plot- and field-scale studies are less easily differentiated. Not only will the scale of the study affect the result or processes observed, but changes in scale will result in different topographic features, soil morphological properties, and landscape segments being incorporated within the study. Furthermore, source factors will tend to be more heterogeneous at broader scales. Laboratory, profile, and plot studies lend themselves to testing of specific hypotheses while controlling variability in source, landscape, and broader-scale hydrological processes. Haygarth and Jarvis (1999) proposed that profile and slope/field are the most important scales in agronomic terms and for understanding mobilization processes. Broader scales, such as the watershed, are more important in terms of understanding the effect of nutrient mobilization on regional water quality.
Because the distinction between scales is not simple, nor can the advantages and disadvantages of particular scales for studying P transfer be clearly defined. Nash and Halliwell (1999) noted that the use of plots generally involves hydrological isolation at the upslope end of the plots such that plots are drier than they would otherwise be, resulting in greater infiltration at the expense of runoff. Pathway length has also been proposed as an important determinant of runoff P concentrations (Gascho et al., 1998). However, this effect is far from simple because of the complex interaction between path length, flow rate, and therefore contact times. These interactions will be discussed later.
Heathwaite and Dils (2000) observed that P concentration in runoff from small plots was greater the further down a toposequence the plots were located, although they did not attempt to provide a reason for these differences. In a similar demonstration of the effect of hillslope position, Cornish et al. (2002) observed increases in P concentrations further down the hillslope from small plots. They attributed this to downslope movement of P-rich material over extended periods of time resulting in increased soil P. Whether this apparent pattern of soil P distribution is common or not and whether there is an influence of such a pattern on runoff water quality is unknown. The areas being examined in both cases were clearly not homogenous and differed systematically with landscape position. In both of these cases, the selection of any one segment of the landscape would have provided quite different estimates of runoff P concentrations from those derived from another segment of the landscape.
Plot studies tend to have one-dimensional flow pathways, such that surface runoff flows down a uniformly sloping plot with no cross slope. Typically such plots are relatively small and range from a few m2 to <1 ha. Often plot lengths are in the range of 1 to 100 m. In contrast, watershed or field scale studies often incorporate more complex topography (e.g., Nash and Murdoch, 1997; Fleming and Cox, 2001) such that measured concentrations are a combination of mobilization, transport, and (because of their size) landscape hydrological effects.
A larger study area does not necessarily result in longer flow lengths, rather it may result in more variable path lengths. Contrasting studies in terms of size can be used to illustrate this point. Dougherty (unpublished data, 2003) studied P mobilization in large plots of length 50 m and area of 1250 m2. A contrasting study in terms of scale was that of Fleming and Cox (2001). They studied runoff from a 2.4-ha area with surface interception barriers at the lower end of the monitored area. The average path length of surface runoff (Lo) in these studies can be defined using the following equations (Horton, 1945):
 | [2] |
 | [3] |
Ls is the sum of the stream lengths for the watershed and A is the drainage area of the watershed. Using Eq. [2] and [3], average path length on the small plots examined by Dougherty (unpublished data, 2003) was 25 m. Despite the relatively large area examined by Fleming and Cox (2001), the average path length calculated using Eq. [2] and [3] was only 30 m. Furthermore, the nature of mobilization, transport, and hydrological processes would have been complicated in the study of Fleming and Cox (2001) by highly variable path lengths, complex surface morphology, heterogeneous P sources, and variable source areas. We hypothesize that research at broader scales does not necessarily provide advantages to improving the understanding of the processes resulting in P transfer. The objectives of research must be clearly linked to the scale at which the research is undertaken.
Effect of Landscape on Processes Determining Phosphorus Mobilization
The effect of incorporating landscape in studies of P transfer is to incorporate landscape-scale transport factors and landscape hydrological pathways. At this scale, processes of mobilization and transfer inevitably become more complex. This almost certainly leads to a decline in the relative amount of information obtained on specific mobilization processes. However, the combination of a clear conceptual model of the system and landscape being examined and use of statistical techniques, such as nonlinear regression, will assist in understanding the various processes resulting in P transfer (e.g., Nash et al., 2000).
At the field scale, anything other than a perfect plane surface with slope in one direction will result in some convergence or divergence of flows and so potentially alter the mobilization–transport processes. If soil morphological features such as depth to horizon differ within plots, then this may affect plot hydrology and hydrological pathways (Nash et al., 2002). Soil physical features such as the presence of macropores may change such that in various parts of the landscape, preferential flow becomes more important (Gachter et al., 1998; Cox et al., 2000; Cox and Pitman, 2001). Landscape position may also affect moisture regimes, with areas lower in the landscape being preferentially predisposed to runoff compared with areas higher in the landscape.
Effect of Changing Flow Conditions on Runoff Phosphorus
In the field there are a wide range of factors that are likely to affect the amount of P desorbed and consequently P concentration in surface runoff. These factors will influence two fundamental properties of P source–water interaction. The first is the contact time between runoff water and P source and the second is the runoff to P source ratio. In addition, factors such as ionic strength and composition will also be affected by these two factors exerting a secondary influence on P concentration. Typical runoff flow velocities are of the order of 0.3 to 15 cm s–1 (Dunne, 1978). Presumably because of high hydraulic resistance of pastures, the typical velocities for pastures are at the low end of this range. Increasing slope length will increase the average runoff–soil contact time. Slow-flowing water has a greater contact time with P source and so will be expected to have greater concentrations of P in runoff (Haygarth et al., 2000). In an examination of scale effects, Gascho et al. (1998) compared P concentrations in runoff from plots 3 and 43 m long. Concentrations of soluble P were on average approximately 50% greater from the large plots than the smaller plots. They attributed this difference to longer residence times on the larger plots.
Fleming and Cox (1998) identified an apparent dilution effect associated with high flow rates for dissolved P. This was likely to have been the result of a combination of wider runoff to soil ratios and shorter contact times. Similarly, McDowell and Trudgill (2000) observed an apparent reduction in P concentrations associated with high flows and an inability of the labile P pool to supply P rapidly to water. Sharpley (1980) attributed a significant decrease in runoff P concentration with time to the dilution of soil solution P. These data suggest that, whereas soil has a capacity to supply P to runoff to maintain runoff water P concentrations under low to moderate flow conditions, under high flow conditions, the supply of P to solution may be too slow to maintain P concentrations. Conversely, Pionke et al. (1996) showed that dissolved P concentration increased above a threshold flow rate. These results were collected from a large, mixed-land use watershed where significant amounts of sediment could be expected to be transported. It may be that the increased dissolved P concentrations are associated with increased sediment loads that will increase the P concentration in runoff.
Overland Flow Hydrological Theory
It is hypothesized that the basic principles of overland flow hydrology can provide a basis for examining the effect of runoff path length and flow rates. For a given surface runoff pathway (e.g., hillslope), the greatest changes in flow conditions will be associated with changes in volume of runoff. Higher volumes of runoff from higher-intensity rainfall will result in a greater depth of flow and higher velocities and therefore wider runoff to soil ratios and shorter soil–water contact times, respectively. For turbulent flows, depth can be estimated by combining the continuity equation:
 | [4] |
 | [5] |
 | [6] |
Therefore, depth increases to a 0.6 power of the increase in discharge (q), the rest being attributable to changes in velocity. Therefore, for a given slope and Manning resistance coefficient, doubling the unit discharge increases depth by a factor of 1.52 and velocity by a factor of 1.32. It follows that changes in discharge per unit area will increase the solution to soil ratio and decrease the contact time. The following sections will examine the effect of time of contact and solution to soil ratio.
Effect of Time of Contact and Water to Soil Ratio
The effect of time of reaction on soil P dynamics has been explored in detail primarily in relation to the adsorption of P to soil such as occurs when fertilizer is added to soil. The timeframes of interactions between soil and runoff water are significantly shorter than those typically considered when examining dynamics of added P (e.g., Burkitt et al., 2002). In surface runoff occurring at the plot to field scale, contact or runoff residence times are in the range of 1 to 60 min (see Table 1).
The kinetics of P desorption in batch experiments have been used to explain differences in concentrations of P from rainfall simulation at different intensities on bare soil surfaces (Sharpley et al., 1981a). The kinetics of P desorption can be represented by the empirical equation:
 | [7] |
(1), and ß (1) are constants for a given soil. Sharpley et al. (1981b) fit this data to P desorption data acquired using batch experiments in the laboratory for a number of soils that had been recently treated with different amounts of P. The equation tells us that as t increases, Pd increases, but at a declining rate. Similarly, as W increases, Pd also increases, but at a declining rate. The effect of these parameters is shown in Fig. 7
. Figure 7 also shows that Pd increases with the amount of P added (or the initial amount of desorbable P present, Pd). Equation [7] can be combined with the following equation:
 | [8] |
 | [9] |
|
The utility of such relationships under pasture systems is not clear. There are likely to be significant differences between the stable pasture systems and the bare soil systems examined by Sharpley et al. (1981a) and Ahuja et al. (1982) in the development and testing of these relationships. One likely difference will be the solution to soil ratios. The high solution to soil ratios in bare soil situations are unlikely to be applicable for stable systems where pasture and the organic horizon stabilize the soil surface. Sharpley et al. (1981a) and Ahuja et al. (1982) assumed that the interaction between the runoff and soil phases involved mixing of the two similar to that occurring in batch experiments. However, in pasture systems where sediment concentrations are low or negligible, the solution to soil ratios are most likely to be significantly narrower. This is because the solution–soil interaction is essentially biphasic (i.e., water is running over the surface of the soil without any substantial physical mixing). The t term in the equation is assumed to be the total time of runoff and W, the ratio of total volume of runoff to sediment concentration. It may be more appropriate to use the instantaneous W value and average residence time of water with soil as the t value. If runoff occurred for 30 min and the average depth of runoff was 2 mm and this interacted with the top 3 mm of soil, the solution to soil ratio (W) would be 0.67. The use of 30 min as the t term may be inappropriate, rather the average residence time of water in contact with the soil may be more appropriate for pastures.
Furthermore, the effects of flow rates on time of contact and solution to soil ratio, and the subsequent effects on runoff P concentrations, relate primarily to desorption processes of phosphate ions from soil. As previously discussed, P in runoff from pasture systems is derived from a complex combination of sources including adsorbed P, precipitated P, organic P, manure P, and plant P. Whereas some of these sources will conform to the empirical relationships derived by Sharpley et al. (1981b), it is possible that not all components contributing P will. An understanding of these relationships and the effect of organic soil P on them is a critical component of understanding P mobilization in pasture-covered soils and differences with scale.
|
RAINFALL SIMULATION |
|---|
|
TOPABSTRACTINTRODUCTIONCONCEPTUAL MODEL OF PHOSPHORUS...PHOSPHORUS SOURCESPHOSPHORUS MOBILIZATIONPHOSPHORUS TRANSPORT—...SCALE AND LANDSCAPE EFFECTS...RAINFALL SIMULATIONCONCLUSIONSREFERENCES |
|---|
Rainfall simulation was originally developed with a view to examining erosion. Therefore, issues such as raindrop energy and size distribution were the focus of much of the early research on simulation methods (Meyer, 1965). Minimum plot lengths may be required to simulate various processes occurring in the field. Loch and Donnollan (1982) noted that approximately 6 to 12 m were required to observe rilling under rainfall simulation. Pasture systems generally can be regarded as stable systems with little transport of sediment occurring. Because ground cover is usually close to 100% in these systems, it is reasonable to expect that all raindrops will be intercepted by plant material before they hit the soil surface or by organic material deposited on the soil surface (Torri and Borselli, 2000). Consequently, energy input to the soil surface in the form of raindrop impact is reduced to negligible levels. Ahuja et al. (1982) and Sharpley (1985) simulated this effect by using screens over soil. They showed that fine screens simulating plant cover dramatically reduced P and sediment concentrations.
Rainfall and Runoff Intensity
Rainfall simulation has been undertaken at rainfall intensities ranging from 50 mm h–1 (McDowell and Sharpley, 2001) to 150 mm h–1 (Cornish et al., 2002). Most commonly, intensities of 80 to 90 mm h–1 for 30 to 45 min (e.g., Sharpley et al., 1981a; Pote et al., 1999) have been used. These rates presumably have been for convenience. High rates allow quick measurements to be made as well as the testing of management effects under extreme conditions when sediment losses, for instance, would be expected to be high (Loch and Donnollan, 1982). It follows that at these high rainfall intensities, runoff rates are also high. This is often exacerbated by the practice of pre-wetting soils to minimize initial differences in soil moisture contents (Pote et al., 1996; Humphry et al., 2002).
Based on the previous discussions of the processes of dissolution and the effect of hydrology, it is logical to question the effect that high rainfall intensities and subsequent runoff rates have on P concentration. Use of rainfall simulation to study P transfer requires consideration of not only the issue of the effects of scale and the processes that are measured, but also the tendency for high rainfall intensities to be used. Sharpley et al. (1981a) examined the concentrations of P in runoff at two rainfall intensities. An increase in rainfall intensity from 60 to 120 mm h–1 resulted in a reduction of P concentrations by 30 to 40%. Whereas there has been some examination of the effect of hydrological parameters under rainfall simulation in bare soil (Sharpley et al., 1981a; Ahuja et al., 1982), there have been no attempts to examine the effect of rainfall simulation intensity on runoff P concentrations under pasture. The limited interaction between soil and overland flow in these highly stable systems is likely to render these situations dramatically different from bare soils. These issues are particularly important if runoff P concentration data derived from rainfall simulations are to be used to accurately model runoff P concentrations at the field or watershed scale.
Plot Size (Length)
Plot size will have two effects, one on variability (and therefore number of plots required to achieve a given level of statistical confidence) and one on plot length (which will dictate path length and therefore times of travel and duration of contact between P source and water). The effect of increasing plot length is not simple because increasing plot length increases the discharge and therefore the depth of flow, and velocity of runoff. Therefore, increasing plot length increases contact time and P source to solution ratio. However, these increases are not linearly related to plot length, as discussed previously.
Modeling the Effect of Rainfall Simulation Parameters
We can combine our knowledge of overland flow theory and the models of P desorption previously discussed to propose a model of the effect of changes of rainfall simulation intensity on runoff P concentration. Based on a simple law of conservation, runoff rates can be defined as follows:
