| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Environmental Science Department, University of Lancaster, Lancaster, LA1 4YQ, UK
b Soil Science and Environmental Quality Team, Institute of Grassland and Environmental Research, North Wyke Research Station, Okehampton, Devon, EX20 2SB, UK
c National Soil Resources Institute, Cranfield University, North Wyke Research Station, Okehampton, Devon, EX20 2SB, UK
d National Soil Resources Institute, Cranfield University, Silsoe, Bedfordshire, MK45 4DT, UK
* Corresponding author (phil.haygarth{at}bbsrc.ac.uk)
Received for publication October 27, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: CSA, critical source area • NSI, National Soil Inventory • WSRP, water-soluble reactive phosphorus • WSTP, water-soluble total phosphorus
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
One important baseline risk indicator of potential P loss is catchment soil P status (Heckrath et al., 1995; Sims et al., 2000; Heathwaite et al., 2003). Increases in soil P concentration have been observed or inferred in many catchments (e.g., Pote et al., 1996; Edwards and Withers, 1998) as a result of agronomic management (Haygarth and Jarvis, 1999; Haygarth et al., 1998, 2000), which has historically been aimed at keeping a minimum soil P status for adequate plant growth (Sims et al., 1998; Hansen et al., 2002; Indiati and Rossi, 2002). To specify an acceptable level of soil P for environmental protection (Gartley and Sims, 1994; Sims et al., 1998; Magdoff et al., 1999) for a given location, other factors that govern the magnitude of P loss for a given period must also be taken into account. Many of these processes are well understood, but may not be well known for a specific catchment of interest. For example, the connectivity of "soil units" to watercourses is a critical factor (Gburek et al., 2002; Heathwaite et al., 2003) but one that is very difficult to estimate, even in intensively researched catchments where collected data are relatively abundant and easily available. In some cases, however, soil P status (e.g., Olsen P) is used in isolation as a P loss risk index, without much consideration of other factors (e.g., soil P threshold type approaches: Sharpley et al., 1996; Daniels et al., 2001). Approaches that include interacting processes and characteristics include index type frameworks (e.g., Lemunyon and Gilbert, 1993; Sharpley et al., 2001; Mallarino et al., 2002), conceptual models [e.g., Haygarth and Jarvis, 1999; the Phosphorus Indicators Tool, PIT (Heathwaite et al., 2003)], and more process-based models [e.g., CREAMS (Cooper et al., 1992); SWAT (Arnold et al., 1994); and INCA-P (Wade et al., 2002)].
In this paper, we study the soil P status of two different first-order "headwater" catchments on contrasting soil types, located in close geographical proximity. The study objectives are to (i) assess the distribution and variability of soil P concentration, both spatially and with soil depth; and (ii) examine the usefulness of topography, measured by the topographic index (Beven and Kirkby, 1979) as an indicator of critical source areas (CSAs) in providing a guide for soil sampling and any potential relationships with soil P [CSAs are specific areas within a catchment most vulnerable to P loss in runoff; for example, see Maas et al. (1985), Gburek and Sharpley (1998), and Pionke et al. (2000)]. If topographic index proves to be useful, this may provide a simple means of helping us assess the potential for runoff spatially and thus, potential risk of P transfer.
There is also the wider objective of assessing the implications of soil P distribution and variability for soil sampling strategies aimed at environmental protection, and considering the implications for modeling and predicting in "non-research catchments" in the UK that are data poor. We define "non-research" as catchments that are not expressly monitored or gauged for research purposes but will nonetheless be subject to modeling and prediction by use of existing national databases on, for example, soil P.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
|
|
|
The Drewston catchment (Fig. 2a; UK grid ref. SX 72495 87857) is 22 ha in size and is also a first-order headwater stream characterized by a well-drained fine loamy soil (Dystrochrept [USDA]; brown podzolic soil [UK]). The catchment is managed predominately as mixed grassland with cattle and sheep throughout. In its lower reaches, the stream is surrounded by a vegetated wetland [soft rush, Juncus effusus L.; and tufted hair grass, Deschampsia cespitosa (L.) P. Beauv.]. In contrast to the Den Brook catchment, the only overland flow observed is confined to the vegetated wetland area (Fig. 2a). There are a few tile drains in the catchment, which are of unknown status, and there are a number of minor roads that cross its upper slopes.
The annual rainfall and runoff for the two catchments for the period December 2001–November 2002 was 1111/610 mm and 1311/639 mm for Den Brook and Drewston, respectively; the annual average rainfall for the area is approximately 1050 mm (40-yr average). Total P fluxes in the streams for the same period have been estimated at between 5 and 7 kg ha–1 yr–1 and 0.5 and 0.9 kg ha–1 yr–1 for Den Brook and Drewston, respectively (Haygarth et al., unpublished data, 2005). For both catchments, the grazed land is dominated by perennial ryegrass (Lolium perenne L.) swards and receives applications of inorganic fertilizer plus manure and excretal returns. Catchment sectors of differing agricultural management were created, in general, using field boundaries (see Fig. 1a and 2a; Table 1). Catchment inputs and livestock numbers for the study period are given in Table 2 and are typical for the catchments' recent agronomic history.
|
/tan ß), where is the upslope contributing area (per unit contour length) to a given point in the catchment, and ß is the local surface slope angle (see Beven, 2001). The index represents the propensity of any point to become saturated and act as a source area for surface runoff. High values of the index occur on shallow slopes and where contributing areas are high, for example in hillslope hollows; low values occur on steep slopes and where contributing areas are small. The simple steady state theory that underlies the topographic index will not always be valid, even in areas where dynamic saturated areas occur (Barling et al., 1994; Beven, 1997; Wigmosta and Lettenmaier, 1999), and factors such as field drainage, local variability in soil permeability, and bedrock topography can modify these patterns (e.g., Freer et al., 2002). However, the pattern of values at the hillslope scale still might provide a general indication of relative propensity to saturation. Figures 1b and 2b show the spatial distribution of the topographic index for both catchments [using the multidirectional flow algorithm of Quinn and Beven (1993)]. The topographic index values were calculated from 5- x 5- and 2.5- x 2.5-m digital elevation models (DEMs) of the Den Brook and Drewston catchments, respectively. The DEMs were generated using a combination of stereography from aerial photographs and GPS surveys. Areas of high topographic index are shown in dark grayscale and areas of low topographic index are in light grayscale. High topographic indices coincide with areas of low slope and/or high accumulated upslope area (i.e., hillslope hollows and/or gullies) and lower topographic indices are associated with areas of steep slopes and/or low upslope accumulated areas (e.g., topographic ridges). These features were used subsequently in the sampling strategy, described below.
Surface runoff has been observed in the hillslope hollows and along the valley axis in both the Den Brook and Drewston catchments. In the valley bottom at Drewston, there is a semi-permanent saturated area that acts as a runoff source area and which is thought to be the result of the blockage of a very old stone drain where once there may have been a small first-order stream. Flow accumulations will also be modified by the roads in the catchment, additional drains in small hollows on the slopes, and (possibly) bedrock topography, since there are some bedrock exposures on the hillslope spurs.
In Den Brook, the relatively impermeable soil means that it remains close to saturation throughout the winter and produces surface runoff frequently, with the greatest depths in the hollows, as would be indicated by the topographic index. The major flow lines in this catchment, however, are underlain by field drains which channel some of the water to the catchment outlet. There is also a drain that takes some of the water flowing from the maize field and farm hard standing. The only roads here are close to the catchment divide.
There are thus limitations to the expected utility of the topographic index as an indicator of CSAs for P, but the pattern of the index might still be broadly indicative of the likelihood of surface runoff. There is, however, an interesting question about how this might then be reflected in the pattern of soil P status: Are critical areas subjected to more frequent overland flows that lead to the accumulation or depletion of soil P over time? Soil P status will be a balance of what is added at a point, what is removed by runoff from that point, and what is brought to that point by runoff from upslope. It has been observed, for example, that runoff on the maize field in Den Brook has resulted in significant depths of soil accumulation at the lower end of the field. This will not necessarily lead to increased P concentrations, however, since it will tend to be the coarse fraction that is retained, which has less propensity to retain P than finer materials (e.g., Walling, 2005). It is therefore possible that the CSAs for P may not be the same as CSAs for runoff.
There may be some clues to the mechanisms involved in the patterns of soil P status both in plan and in the soil profile in these catchments. To address this issue with a minimum number of samples, a stratified sampling strategy was adopted, along paired transects, which followed topographic highs (ridges) and lows (hillslope hollows) within each catchment, and for each major land use (Fig. 1 and 2 and Table 1). The transects compare areas of potentially different hydrological behavior, which may also have very different connectivity to the stream.
Spatial Samples
Implementation of the sampling strategy into a practical sampling plan was relatively simple at Den Brook where ridges and hillslope hollows were well defined, continuous, and easily observable in the field. The identification of ridges and hillslope hollows at the Drewston catchment was, however, more difficult as hillslope hollows were less pronounced and more difficult to identify from both the topographic index plot (Fig. 2b) and visually in the field. The only exception was the main axis of the catchment (i.e., the channel and/or wetland area and its extension toward the top of the catchment; see Fig. 2b). The final sampling transects identified are shown in Fig. 1b and 2b. As there were no clear topographic features in the maize field at Den Brook (Sector 3), four samples were taken: two along the contours of the field and two downslope. Similarly, four samples were taken in Sector 5a of the Drewston catchment.
Soil samples were taken along transects at approximately 30-m intervals (Fig. 1b and 2b) using a 2-cm-diameter, 7.5-cm-deep, foot-driven corer with a stainless steel cutting blade. Two sets of samples were taken at depths of 0 to 2 and 0 to 7.5 cm, as the shallower samples may be more representative of forms that are available for transfer in runoff water, representing the effective depth of interaction. A previous study by Haygarth et al. (1998) found that soil Olsen P was enriched in the surface 1 to 2 cm of soil under permanent pasture that suggested there was a need for a fine vertical resolution in sampling. Each 0- to 7.5-cm-depth sample comprised 15 individual 2-cm-diameter cores, collected randomly from a 2-m square grid. The 0- to 2-cm-depth cores were sampled from the same 15 locations but two cores were taken to obtain a sufficient amount of soil for analysis. In both cases, all subsamples were bulked to form a composite, to limit the number of separate analyses required. Each individual core had the surface vegetation and any organic mat removed before being checked for length. The 0- to 2-cm-depth cores were taken using the same corer but were cut to length. For saturated areas, where the corer was unsuitable, cores were taken as a 7.5- x 7.5-cm square (7.5-cm-deep cores) or a 15-x 15-cm square (2-cm-deep cores). The vegetation was removed and the sample was extracted with a trowel. At each sample point, five subsamples were bulked to make a composite. At all locations, recent dung deposits were avoided. Spatial sampling in both catchments occurred between 2 and 11 Oct. 2002.
Depth Samples
To test for variation in soil P with depth, some soil samples deeper than 7.5 cm were also taken and these are called "depth samples." For the depth samples, a single sampling location was identified for each catchment. The sampling points were taken from gullies, at locations where the soil P status was "similar" for the two catchments (using the information gained from the spatial survey), and represented a 10-m2 area. Within these sample points, four subsampling locations were randomly chosen within the 10-m2 area and six cores were taken (within 0.5 m2) to a depth of up to 91 cm (depending on soil depth; see Tables 5 and 6). The six cores were bulked to obtain enough soil for analysis. The cores were collected using a 2.5-cm-diameter (maximum 100-cm depth) percussion corer, driven into the ground by a hand-held electric hammer (Makita Corporation, Aichi, Japan). Depth sampling in both catchments occurred between 16 and 29 June 2003.
|
|
Dry soils were also analyzed for Olsen-extractable P using a modified version of the method outlined in Olsen et al. (1954), using alkaline sodium bicarbonate as the extractant in a 20:1 ratio, filtered through a Whatman no. 2 filter. The filtrate was analyzed for P using the molybdenum blue method (Murphy and Riley, 1962), by addition of acidified ammonium molybdate reagent and ascorbic acid.
To determine the potential variability of the analytical methods, nine separate runs of Olsen P were performed using an "in house" reference soil from different analytical batches in 2003. The soil was the same soil type as that found at Den Brook. For the nine batches, a mean value of 42.2 mg kg–1 was determined, with a maximum of 45.9 mg kg –1 and a minimum of 39.2 mg kg –1 (standard deviation 2.7). It was therefore concluded that analytical uncertainty and variability were minimal.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
For most variables measured, their distribution is positively skewed (see skewness Tables 3 and 4) with a few very high P concentrations measured at "hot spots" (discussed below). In general (although see discussion below), and as would be expected given previously reported data on the decline in P concentration with depth (e.g., Haygarth et al., 1998; Kleinman et al., 2003; Owens and Deeks, 2004; see also depth profile results below), at each sampling location the 0- to 2-cm samples had a higher concentration than the 0- to 7.5-cm samples. All P determinations had a very similar spatial pattern within the catchments. Therefore, in what follows, relationships discussed with respect to Olsen P for the 0- to 7.5-cm depth are similar for all determinations unless specifically stated.
Spatial Distribution and Variability
Den Brook and Drewston had a similar soil P status (over the samples taken) with median Olsen P levels (0–7.5 cm) of 54.5 and 49.6 mg kg–1, respectively (Fig. 3)
. This is despite the fact that in the reported year (12 Dec. 2001–31 Jan. 2003; Table 2), Den Brook received considerably more P inputs than Drewston. The soil P information is probably indicative of the fact that, over the longer term than the period we monitored inputs, soil P inputs would have been closer to similar for the two catchments. Presented in Fig. 4 and 5
are the spatial distributions of Olsen P concentration (0–7.5 cm) for Den Brook and Drewston, respectively. For Den Brook it can be seen that there are a few areas with high Olsen P concentrations that appear to be potential local "hot spots." For example, in the Den Brook catchment, the area of high concentrations in Sector 1, at the top of the ridge (R14–R18; Fig. 1b), corresponds to an area where manure applications are made in the winter when it is too wet to apply to other areas of the catchment. Additionally, the single high concentration location toward the top of the hillslope hollow (MG3; Fig. 1b) in Sector 1 is associated with an area where livestock shelter under trees and the relatively high concentrations in the hillslope hollow within Sectors 6 and 7 (RG transect) may be associated with runoff from the cow shed and/or hard standing area (Sector 2).
|
|
|
The variability of soil P is high and is made particularly high by the "hot spots." The range and standard deviation for Den Brook and Drewston (Olsen P, 0- to 7.5-cm samples) are 167.2 and 42.8, and 161.1 and 36.8, respectively. This has important implications for soil sampling strategies regarding the number and location of samples needed to give a representative value for the soil unit of interest (see also Schepers et al., 2000; Needelman et al., 2001). Indeed, given that in practice only one value of soil P status may be obtained from a composite sample, we concede that the loss of information on variability and spatial distribution for a given sample area could be very important. The problem of obtaining an adequate representative value (or distribution) of soil P status is compounded if the CSA concept is embraced. In this case, only the P status of the source areas might be relevant in determining what reaches the stream, and those source areas might make up only a small fraction of the catchment except under extreme conditions. High soil P measurements do not necessarily indicate a CSA. High soil P values can result from a high source term (either locally or by delivery from upslope) or because transport is limited so that there is a buildup of P over time. For these reasons, it is important that we learn from the variability and distribution of soil P within research catchments and use this knowledge in a pragmatic way when we need to assess soil P status within a non-research catchment. A pragmatic solution is suggested below (Catchment Effective Spatial Sampling for Soil Phosphorus Status).
Depth Profiles
There was a decline of concentration with increasing depth for water-soluble total phosphorus (WSTP) and water-soluble reactive phosphorus (WSRP) as well as for Olsen P for both catchments (Tables 5 and 6 and Fig. 6a and 6b)
. Additionally, for all determinations, apart from the WSTP at Den Brook, the 0- to 1-cm sample had a lower mean value of P concentration than the subsequent one or two sampling depths. This may reflect the higher organic matter content of the 0- to 1-cm sample for which P has a lower affinity compared with the more mineral particle-dominated sublayers. It may also reflect the fact that fine particles and/or colloids, with P attached, may have been removed from the very upper soil horizons in overland flow, with the upper horizon being the "effective depth of interaction" for runoff (Ahuja et al., 1981). For WSTP, the reduction in concentration with depth is greater in the shallower layers for Den Brook compared with Drewston, although the concentrations at the deeper layers show a similar 90% reduction from the surface samples. For WSRP, the reduction of concentration with depth is less extreme for Den Brook compared with that observed for WSTP; at Drewston WSRP showed a decline consistent with that observed for WSTP. This may in part reflect the higher P inputs to Den Brook over the monitored period (Table 3). For both catchments, Olsen P concentrations declined more steeply than the water-soluble determinations. This may reflect: (i) the more mobile nature of water-soluble fractions which can move more readily in the soil column compared to the well-bound Olsen P; and (ii) that grass roots are most efficient at utilizing and thus removing Olsen P (in relation to water-soluble fractions) from lower horizons. The rapid decline in concentration with depth for all determinants reinforces the recommendation that, when sampling soil to provide assessment of P risk to water quality, samples from shallower soil layers are of greatest significance, because the effective depth of interaction with runoff can be very shallow (Ahuja et al., 1981; Haygarth et al., 1998).
|
Soil Phosphorus and Topographic index
The usefulness of the topographic index in providing a guide for soil sampling and any potential relationships with soil P is limited because the spatial patterns of soil P are dominated by the observed "hot spots." Gburek and Sharpley (1998) and Weld et al. (2001) also reached this general conclusion, themselves indicating that soil P status was generally a function of land use and field boundaries. Thus "hot spots" obscure any strong relationship that may occur between soil P status and topographic index (see Fig. 7a and 7b)
.
|
It has been noted that in these catchments, the topographic index might be limited as an indicator of runoff source areas because of the effects of field drains, the road system (Fig. 2a), and the bedrock topography. Although surface runoff has been seen during this study in the areas of high topographic index values, even if the index was a good indicator of the occurrence of surface runoff, the results show that the topographic index alone cannot help to estimate spatial patterns of soil P status, which are the result of the balance of supply to a point and removal from that point. Critical source areas for P production in a catchment will be a function of both the occurrence of surface (and subsurface) runoff and the potential mobilization of P at a point. It is interesting to reflect on the question of the balances of supply and transport within the catchment that lead to the rather uniform pattern of P status in both catchments when the "hot spots" are excluded. Without additional information, such as local land use knowledge and/or trends or further soil testing, it would not appear to be possible to locate the CSAs for P unambiguously in these catchments.
Implications for Using National Soil Phosphorus Data
The estimation of catchment soil P status and which areas of the catchment are the most important contributors remains uncertain, even with a relatively large number of samples as may be available in research catchments such as Den Brook and Drewston. For non-research catchments the uncertainty will usually be greater. In a UK context, this is because soil P data available for non-research catchments are generally of much lower spatial resolution. For example, the UK National Soil Inventory (NSI) gives total and Olsen P values at 5-km nodes across the UK. At each node, 25 cores (15 cm deep) were taken using a 4-m grid within a 20- x 20-m square. These data include the soil type and land use at the node but in many cases may not be a reasonable measure of the soil P status of a given catchment for use in P loss prediction. This adds to the uncertainty associated with any model or predictive tool that uses such data. Figures 8a and 8b compare NSI soil P data and that collected for the present survey. Note that the NSI data used are based on all similar occurrences of similar soil type and land use across the UK. For the purposes of the comparisons in Fig. 8, the Olsen P measurements have been corrected for the difference in depth between the present survey and the NSI data (i.e., 0–7.5 cm compared to 0–15 cm, respectively) using the depth profile data presented above (under the assumption that all samples have a similar depth profile). Clearly national data do not adequately reflect catchment-specific measurements, due to localized management effects and variation in soil P (Owens and Deeks, 2004).
|
Using all data on a catchment-wide basis revealed that Den Brook and Drewston had a similar soil P status (already discussed in the earlier section on spatial distribution and variability) and both have a positively skewed distribution (skewness of 1.4 and 1.7, respectively; see Fig. 3). However, use of the stratified sampling, based on the topographic index and CSAs, allows an assessment of "catchment effective soil phosphorus status" to be made. For Drewston an arithmetic average of seven 0- to 7.5-cm samples in the wetland area gives an Olsen P value of 18.8 mg kg–1. For Den Brook, using similar 0- to 7.5-cm samples in areas where overland flow has been observed, and which are directly connected to the stream, the average Olsen P concentration is 79.5 mg kg–1. Using these estimates Den Brook has an "effective" soil P status approximately four times that of Drewston, which is indeed consistent with the trend in the respective annual P losses of approximately 5 to 7 kg ha–1 yr–1 (Den Brook) and 0.5 to 0.9 kg ha–1 yr–1 (Drewston). This trend may, in part and among other things, reflect the difference in P inputs to the two catchments (Table 2). This is consistent with the CSA concept of Gburek and Sharpley (1998) and with the topographic index–weighted export coefficients of Endreny and Wood (2003). CSA and topographic index–weighted stratified sampling may hence be a more useful indicator of catchment P status than some mean and/or composite measure of a random sample.
Identification of Critical Source Areas A Priori—Without Measurement
The a priori determination of CSAs for varied physicochemical–climatic conditions is not a simple task. It inherently requires an estimate of hydrological connectivity [including macropores and artificial drainage; see Stamm et al. (1998) and McGechan (2002)] between soil units and watercourses—a very challenging problem, even within research catchments. Alternatives in tackling this problem include CSA identification using GIS overlays of soil, slope, land use, and watercourses, etc. (e.g., Sivertun et al., 1988); the probabilistic design curve method of Gburek et al. (2002), which identifies a contributing distance for riparian areas; and the topographic index–weighted method of Endreny and Wood (2003). If CSAs are to be targeted in any mitigation strategy, the chosen approach must be at a scale that can effectively identify potentially critical areas.
Additionally, we need to extend the CSA concept to cover the vertical dimension, given vertical distributions of P that have been reported for some soils (e.g., by estimating the dominant soil horizons with respect to lateral subsurface flow). Furthermore, artificial drains and naturally occurring macropores may be CSAs themselves. For instance, Kleinman et al. (2003) found evidence of higher concentrations of P in clay films from macropore internal surfaces compared to those from bulk soil samples. A priori determination of CSAs, and hence an "effective" catchment soil P status, is therefore complex. Some alternatives to soil sampling are suggested in the following section.
Alternatives to Soil Sampling
Although alternatives to soil sampling were not investigated as part of this study, this section discusses and speculates on some possibilities. Given the obvious difficulties in sampling and thus using soil samples to assess catchment soil P status and risk to water quality, there may be alternative approaches we can use that can provide surrogate estimates of soil P status and are better "integrators" of complexity and variability around a catchment. We believe that headwater stream baseflow P concentrations may provide one such opportunity. These are inherently controlled by the physicochemical characteristics of the catchment, and in particular the soil P status of important soil units (see Rawlins et al., 2003). There are, however, inherent problems in using baseflow concentrations such as the variations in baseflow concentrations associated with seasonality, biological activity, in-stream processes, agronomic perturbations and so on, and the fact that the CSAs are temporally dynamic within storm events.
Streambed sediment samples may also provide an integrated measure of P loss from previous storm events (e.g., Jansson et al., 2000; Rawlins et al., 2003). In this case, much of the sediment may originate from CSAs active during storm events and hence are potentially more useful than baseflow concentrations. These sediments also inherently include effects of other catchment characteristics and farming practices. However, if we are to utilize streambed sediments, we need to take into account the complicating factors associated with the sediment dynamics of events preceding the sample(s), equilibrium effects with antecedent flows, and the interaction with local geological and geomorphological characteristics.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
For catchments where there are sufficient resources, soil sampling strategies should be designed to suit individual catchment characteristics and should embrace the CSA concept. Where there are insufficient resources, and, as currently available soil P data rarely exists for CSAs in isolation (i.e., particularly for non-research catchments), it may be more effective to try to utilize more "integrated" measures of catchment soil P status such as a limited number of baseflow and/or streambed sediment samples, although this is speculation and was not confirmed specifically by research in this paper.
Using the CSA concept as a basis for estimating environmental effects of given soil P levels has the advantage of implicitly focusing on those areas where both overland flow and lateral subsurface flow are more likely and hence high magnitude hydrologic events when very high proportions of annual P fluxes can occur (Ryden et al., 1973; Heathwaite et al., 1989; Haygarth et al., 2004). However, CSAs vary greatly depending on catchment physicochemical and climatic characteristics and are hence difficult to identify without measurement.
There is then an inherent limitation in the estimation of CSAs, as they depend on the estimation of connectivity and delivery, which is known to be difficult (Beven et al., 2005). Similarly, there is often a disparity between lysimeter and plot studies, where strong relationships have been seen between soil P status in the former, but at the larger catchment scale, the results and relationships become less clear, because different processes emerge to dominate at different scales [see Quinton et al. (2003) and Jordan et al. (2000) for a contradictory large-scale example and Haygarth et al. (2005) for a discussion of this in relation to "decoherence" with scale]. Furthermore, our results suggest that more research is needed into how macro-scale features of catchments can be used to determine connectivity between soil units, and hence identify CSAs where environmental protection measures will be most effective. Moreover, since observed soil P distribution is variable and is also difficult to relate to nationally available soil P data, any assessment of soil P status for determining risk of P loss is uncertain and problematic.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. L. Little, S. C. Nolan, J. P. Casson, and B. M. Olson Relationships between Soil and Runoff Phosphorus in Small Alberta Watersheds J. Environ. Qual., July 17, 2007; 36(5): 1289 - 1300. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. van der Perk, P. N. Owens, L. K. Deeks, B. G. Rawlins, P. M. Haygarth, and K. J. Beven Controls on Catchment-Scale Patterns of Phosphorus in Soil, Streambed Sediment, and Stream Water J. Environ. Qual., April 5, 2007; 36(3): 694 - 708. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. J. Chardon, G. H. Aalderink, and C. van der Salm Phosphorus Leaching from Cow Manure Patches on Soil Columns J. Environ. Qual., January 9, 2007; 36(1): 17 - 22. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||
