| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Dep. of Agroecology, Danish Institute of Agricultural Sciences, P.O. Box 50, DK-8830 Tjele, Denmark
* Corresponding author (kirsten.schelde{at}agrsci.dk)
Received 29 March 2005.
 |
ABSTRACT |
|---|
 TOPEXECUTIVE SUMMARY ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: BG, background experiment, 26. Apr. 2001 • DIP, dissolved inorganic phosphorus • DOC, dissolved organic carbon • DOP, dissolved organic phosphorus • DP, dissolved phosphorus • DRP, dissolved reactive (inorganic) phosphorus • EC, electrical conductivity • H+, experiment following herbicide application, 7 Nov. 2001 • LE-WDC, water dispersible colloids determined by Low Energy input method • NTU, nephelometric turbidity • PL–, experiment preceding plowing, 20 Mar. 2002 • PL+, experiment following plowing, 4 Apr. 2002 • PP, particulate phosphorus • PIP, particulate inorganic phosphorus • POP, particulate organic phosphorus • PP, particulate phosphorus • S1+, experiment following first slurry application, 8 May 2001 • S2+, experiment following second slurry application, 19 June 2001 • TDP, total dissolved phosphorus • TDR, time domain reflectometry • TP, total phosphorus • TRP, total reactive (inorganic) phosphorus • WDC, water dispersible colloids
|
INTRODUCTION |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Many studies of colloid mobilization and transport have been made in model systems with a homogeneous or well-defined structure to investigate single processes and interactions (e.g., Wan and Wilson, 1994; Wan and Tokunaga, 1997; Roy and Dzombak, 1997; Grolimund et al., 1998; Noack et al., 2000; Flury et al., 2002; Saiers and Lenhart, 2003a, 2003b; Laegdsmand et al., 2005). However, natural soils are structurally and chemically heterogeneous, and an increasing number of studies have focused on the mobilization and transport of colloids in natural structured soils (Jacobsen et al., 1997; Ryan et al., 1998; Karathanasis, 1999; Laegdsmand et al., 1999; El-Farhan et al., 2000; Villholth et al., 2000; Schelde et al., 2002; Petersen et al., 2003; Kjaergaard et al., 2004a, 2004b, 2004c; de Jonge et al., 2004a). Preferential flow phenomena may be largely responsible for the observed leaching patterns of strongly sorbing contaminants (Flury, 1996; Gachter et al., 1998; de Jonge et al., 1998, 2004a; Heathwaite and Dils, 2000).
In situ mobilization of colloids in natural structured soils depends on soil characteristics controlling the dispersibility of colloids from soil aggregates and may be enhanced by the exchange of high-ionic strength resident water with low-ionic strength infiltration water. The subsequent translocation of colloids depends on the prevailing conditions for transport, such as colloid stability in the soil solution, soil water content, and the pore size and geometry of the water-conducting pore system. The available information about the intrinsic and dynamic properties of the soil controlling the mobilization and transport of colloids in natural soils is still insufficient to make accurate predictions of the risk of contaminant colloid leaching (Kjaergaard et al., 2004c).
In Western Europe, land use is dominated by intensive agricultural production. The greatest environmental concerns with respect to agriculture are the leaching of P, N, and pesticides (European Environment Agency, 2003a, 2003b). The Water Framework Directive (European Union, 2000) is a legislative framework to protect and improve the quality of European water resources such as lakes, rivers, groundwater, and coastal waters. The implementation of the directive implies assessing the pressure on water bodies, including the leaching patterns in individual catchments, and measures to reduce leaching. Thus, a good knowledge of the effective processes involved in the leaching, comprising also particle-facilitated transport, is required for implementation in water resource management models (Shirmohammadi et al., 1998; Heathwaite, 2003).
Phosphorus is an essential element for plant growth, but in many regions the farm level application of fertilizers and manure has built up soil P to levels that often exceed crop needs (Sibbesen and Runge-Metzger, 1995; Sharpley et al., 1996; Dalgaard et al., 2003). Surface and subsurface runoff and erosion from high-P soils may be major contributing factors to fresh water eutrophication. According to national monitoring programs, diffuse losses of P from agricultural land make up more than 50% of the P load to Danish streams, lakes, and coastal waters, thus contributing significantly to damage ecological quality in lakes and estuaries (Kronvang et al., 2005). These diffuse losses roughly correspond to 0.5 kg P ha–1 yr–1 being lost from arable land in Denmark.
An overview of the agricultural P mobility problem and the processes controlling P loss was given by McDowell et al. (2001), while Sims et al. (1998) focused specifically on subsurface P losses. The processes controlling soil P release to surface runoff and to subsurface flow are a complex interaction between the type of P input, soil type and management, and transport processes depending on hydrological conditions. The mechanisms controlling soil P release to surface runoff at point or field scale are better known than the mechanisms and quantities of P transported by subsurface pathways (McDowell et al., 2001). Agricultural soils do not have equal potentials for P leaching; dissolved P is more susceptible to leaching in sandy or organic soils with a low P retention capacity (McDowell et al., 2001). However, on loamy and structured soils with a high P binding capacity, P may be lost by leaching due to preferential and colloid-facilitated flow phenomena (Sharpley and Syers, 1979; Heckrath et al., 1995; Grant et al., 1996; Stamm et al., 1998; Sims et al., 1998; Laubel et al., 1999; Djodjic et al., 2000; de Jonge et al., 2004a).
Earlier investigations of preferential transport of solutes and soil colloidal particles at the lysimeter, plot, or field scale have focused on (i) colloid mobilization as affected by infiltration dynamics (Ryan et al., 1998; El-Farhan et al., 2000), (ii) transport of particulate P in drain water below grassland or arable crops (Turtola and Jaakkola, 1995; Grant et al., 1996; Laubel et al., 1999; Djodjic et al., 2000; Turner and Haygarth, 2000; Heathwaite and Dils, 2000; Uusitalo et al., 2001), (iii) nitrate leaching (e.g., Bronswijk et al., 1995), (iv) leaching of pesticide (Villholth et al., 2000; Zehe and Flühler, 2001; Fomsgaard et al., 2003; Petersen et al., 2003), or (v) tillage effects on leaching of nutrients or pesticides (Djodjic et al., 2002; Fortin et al., 2002).
The target of the present study was to combine several of the above foci into a series of experiments at a single location. Thus, we investigated solute and particle-facilitated transport in a structured, tile-drained soil below grassland, as affected by a full agricultural year of conventional management, which consisted of repeated slurry application and grass cutting during spring and summer, herbicide application in autumn to terminate the crop, and early spring plowing. Using this approach, effects of plowing and slurry application on particle mobilization and leaching of P can be examined under comparable conditions in terms of equal soil type and management history and rainfall events having equal rate and duration.
The objectives of the field study were to investigate
The experiments were made on macroporous soil at the Røgen field site, Denmark. A number of laboratory studies made on undisturbed soil cores have examined the dynamics and mechanisms in leaching and particle-facilitated transport for this soil (Jacobsen et al., 1997; de Jonge et al., 1998, 2000, 2004a; Laegdsmand et al., 1999; Schelde et al., 2002). An additional objective of the present study was to see how results obtained on short soil cores with respect to amounts and timing of particle transport compared with results at the field scale.
|
MATERIALS AND METHODS |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The Røgen site has a coastal temperate climate with a mean annual precipitation of 714 mm. The mean minimum and mean maximum temperatures in January are –3 and 1.6°C, respectively, and the mean minimum and maximum temperatures in July are 10.7 and 19.8°C, respectively.
The clay tile drainage network at the field was installed at a depth of 1.1 m during 1944 to 1950. In 1993, two experimental plots were established in the upper part of the slightly sloping field. A trench was dug around each of two 5 by 5 m plots to isolate them hydrologically from the surrounding soil by plastic sheeting, penetrating to a depth of 1.1 m. After isolation, the trenches were refilled and compacted, and the whole area was managed uniformly. The upstream drain flow was diverted and conducted to the flow downstream of the plots. The water from the 5-m drain line below each plot was accessible for sampling via two observation wells. Further details on the field layout and installations are given in Laubel et al. (1999) and Villholth et al. (2000).
The plots had not been plowed since 1996 when a crop of winter wheat (Triticum aestivum L.) was established (Villholth et al., 2000). Since late 1997 the plots had been covered with permanent grass–clover (Trifolium spp.) that was left unmanaged apart from irregular grass cuttings and limited fertilization.
Six irrigation experiments were conducted within 1 yr, starting in April 2001 and ending in April 2002 (Fig. 2, Table 1). The first experiment (labeled BG, on 26 Apr. 2001) aimed to establish background information on responses to irrigation by the unmanaged plots before slurry application. Slurry was applied on 4 May and 15 June 2001, and each application was followed by irrigation experiments within a few days (S1+ on 8 May and S2+ on 19 June 2001). The grass was cut and removed from the plots before the second slurry application (June 2001) and in September 2001. On 26 Sept. 2001, the plots were treated with the herbicide glyphosate [N-(phosphonomethyl)glycine] to terminate the grass crop before plowing in accordance with common management practice. A late-season irrigation experiment was performed 42 d after the herbicide treatment (H+, on 7 Nov. 2001). In spring 2002 an irrigation experiment was conducted before plowing of the plots (PL–, on 20 Mar. 2002). The final experiment followed 2 d of plowing (PL+, 4 Apr. 2002).
|
|
|
The observation well corresponding to the treated plot was monitored to note the onset of water flow in the drains. Drain water was collected manually, starting with sample sizes on the order of 100 mL and increasing in size as the experiment advanced. Drain water was collected until the outflow had peaked and leveled off to a low value or stopped completely, up to 22 h after the onset of irrigation. The outflow rate as a function of time was quantified by measuring the elapsed time between samples and the mass of the effluent that was sampled. The effluent samples were stored at 2°C until further analysis on the following day.
In the laboratory, the effluent samples were first analyzed for turbidity, pH, and EC. Turbidity was measured with a Hach 2100AN turbidimeter (Hach, Loveland, CO) equipped with an EPA filter, measuring at wavelengths 400 to 600 nm. The nephelometric turbidity (NTU) was assumed proportional to particle concentration; the Røgen-specific correlation used for calculating particle concentration (mg L–1) from turbidity (NTU) is given by Schelde et al. (2002). Further, the samples were analyzed for total P (TP), total dissolved P (TDP), total reactive P (TRP), and dissolved reactive P (DRP). Dissolved P was defined as the P present in the supernatant after centrifuging for 10 min at 5087 g (equivalent to a cut-off particle diameter of 0.24 µm). Reactive P concentrations were determined by spectrophotometry using a colorimetric technique with ascorbic acid reduction as described by Murphy and Riley (1962). Total P concentrations were determined using acid persulphate digestion in an autoclave (120°C, 200 kPa) (Koroleff, 1983) followed by the colorimetric technique described by Murphy and Riley (1962). Subsequent to analyses, the fraction of particulate inorganic P (PIP) was estimated as the difference between TRP and DRP. The fraction of dissolved inorganic P (DIP) was set equal to the DRP, while the fraction of dissolved organic P (DOP) was determined as the difference between TDP and DRP. Finally, the fraction of particulate organic P (POP) was estimated as (TP – TDP) minus (TRP – DRP). See de Jonge et al. (2004a) for a flowchart representation of the fractionation of P forms in the effluent.
Auxiliary measurements at the plot site included climate and soil water content. The climatic variables precipitation, air temperature, and soil temperature were measured continuously during the year of experiments, using a tipping spoon rain gauge (Rain-o-matic Prof, Pronamic, Silkeborg, Denmark) and PT-100 sensors. Every 10 min there was an air temperature measurement at 2 m height and a soil temperature measurement at a depth of 10 cm. Due to problems with data logging, the time series were not complete; however, data from a neighboring climate station (20 km) were obtained to assess the total natural precipitation. Soil water content was measured intermittently before and during irrigation campaigns using the time domain reflectometry (TDR) technique. The TDR measurements were performed using the Tektronix 1502C cable tester (Tektronix Inc., Beaverton, OR, USA). The Topp calibration (Topp et al., 1980) was used to convert the measured dielectric constant to volumetric soil water content. Time domain reflectometry probes were installed vertically in the plots for manual observation of soil water content in the 0- to 20-, 0- to 50-, and 0- to 100-cm depth intervals, in two replicates per interval and plot.
At a supplementary plot that underwent the same treatments as the adjacent experimental plots, soil samples were taken before the first background experiment for further analysis in the laboratory. The soil samples were analyzed for the potential release of water dispersible colloids (WDC) according to the low-energy dispersion procedure developed by Kjaergaard et al. (2004a).
|
RESULTS AND DISCUSSION |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Resulting runoff/irrigation ratios of the two plots varied between 0 and 36% and were a little lower than previously obtained runoff/irrigation ratios at the Røgen site (14–63%, Laubel et al., 1999; and 29%, Villholth et al., 2000). Unlike the findings of Laubel et al. (1999), there was no tendency for Plot 2 to yield lower ratios than Plot 1. This finding indicates varying flow dynamics with time in the embedded soil of the plots and that there was no general failure on the lining of the plots. Water that was not collected via the drains could contribute to wetting of the plot soil, but was more likely routed downward without merging with the flow to the drains. Observed changes in water content in the 0- to 100-cm depth interval during the experiments were negligible or very small (results not shown).
For the majority of experiments, when there was no drain flow at the onset of irrigation, the first flow of water in the drains was observed after 0.3 to 2.3 h (Table 1). Water flow in the drains peaked 1 to 2 h after breakthrough and gradually decreased after the irrigation was halted (Fig. 3 ). Villholth et al. (2000) and Laubel et al. (1999) observed similar flow dynamics at the Røgen plots. Laubel et al. (1999) concluded, on the basis of tracer experiments showing rapid arrival of applied chloride, that macropore flow prevailed at the site. Other laboratory investigations of undisturbed soil cores originating from the Røgen site have also demonstrated macropore flow (Jacobsen et al., 1997; de Jonge et al., 1998, 2004a; Laegdsmand et al., 1999; Schelde et al., 2002).
|
The results indicate the following mechanisms of particle mobilization. The initial flush of particles is probably leaching of colloids that have been dispersed into the soil water before the irrigation experiments. This stock of easily mobilized colloids is replenished during periods of no-flow via transport of colloids from immobile intra-aggregate water to interaggregate water (Schelde et al., 2002). The secondary peaks of particles sometimes observed in our experiments were probably not associated with particle detachment resulting from hydrodynamic sheer of the flowing water. Schelde et al. (2002) showed how particle release from comparable soil cores under equal flow (and consequently equal sheer) conditions increased depending on the length of the pause between water infiltrations. In contrast, we assume the secondary particle peaks to be caused by decreased ionic strength of the soil water in contact with the soil aggregates. Ionic strength in the interaggregate pore water decreases when high ionic strength pore water is replaced or diluted by low ionic strength infiltration water. This in turn causes more ions to diffuse from the intra-aggregate water into the interaggregate water in response to the enhanced gradient in ionic strength. Decreasing ionic strength leads to an increasing release of particles from the aggregate surfaces and an associated higher particle concentration in the effluent (Grolimund et al., 1996; Flury et al., 2003). Observed dynamics of particle concentration and electrical conductivity of the effluent for the four experiments most abundant in drain outflow are given in Fig. 3. The most notable secondary peaks in particle concentration were observed for Plot 1 and were correlated, although sometimes only weakly, with a decreasing EC. Laegdsmand et al. (1999) observed a high initial concentration and a secondary peak in particle concentration leached from Røgen soil cores (30-cm diameter, 35 cm long). Their secondary peaks were more clearly associated with decreasing ionic strength of the soil water than was observed in our study.
El-Farhan et al. (2000) noted peaks of particles both on the rising and falling section of the hydrograph. This finding, together with those of Saiers and Lenhart (2003a, 2003b), has been referred to as a general effect associated with transients in flow conditions (DeNovio et al., 2004). Our experiments did not indicate any particle pulses associated with the end of irrigation. Indeed, some particle pulses peaked around the time when irrigation was halted (Fig. 3), but we found that these peaks started to build up even before then, while there were still no changes in the infiltration rate. Our secondary pulses seemed to be correlated with high (out)flow rates and the associated decreasing EC, rather than with a phase of drainage. The papers by Grant et al. (1996), Laubel et al. (1999), and Djodjic et al. (2000) present runoff and particulate matter in runoff versus time for individual natural rainstorms and irrigation events. Neither of these studies indicates a clear increase in PM concentration beyond the first particle peak preceding peak flow.
A flow highly dominated by macropore flow implies a limited active flow volume and relatively little exchange between the low-ionic strength infiltration water and the high ionic strength resident water. Plot 1 generally experienced delayed onset of drain flow and leached more particles per volume of water compared with Plot 2 (Table 1). This is probably due to different flow patterns in the two plots, with a higher degree of preferential flow, and thus a more limited flow-active and particle-feeding soil volume, in Plot 2.
We found pH in the drain water to vary relatively little during and between events (Table 1). Mean pH ranged between 6.4 and 7.9, with a slight trend toward decreasing pH with experiment number. We did not find any significant correlations between pH and particle mobilization dynamics within events.
Plowing Effects: Hydrology and Particle Leaching
Plowing resulted in a significant increase in particle concentrations in the effluent. For both plots the average and maximum particle concentrations after plowing (PL+) increased by factors of 4 and 3, respectively, compared with concentrations before plowing (Table 1, Fig. 4
). This is most probably because plowing exposed new soil aggregates to water flow, destabilized aggregates, and thus increased the source of water-dispersible colloids to be mobilized with subsequent rainfall events (Watts et al., 1996; Petersen et al., 2004).
|
The additional particles leached after plowing evidently originated from the disturbed plow layer. The general source of particles, however, is also the topsoil. Grant et al. (1996) and Laubel et al. (1999), using 137Cs as a marker for topsoil, showed that the sediment material carried by drain flow at Røgen originated from the topsoil. Uusitalo et al. (2001) came to a similar conclusion for a clayey soil at Sjökulla, Finland, also on the basis of measurements of 137Cs activity in the surface runoff sediments and drain flow sediments.
At this point we may compare observed totals of leached particles with the laboratory measurements of WDC, using a low energy input approach (LE-WDC; Kjaergaard et al., 2004a). LE-WDC was found to be 2300 mg kg–1 for the plow layer of the Røgen soil. This entails a source of dispersible colloids for the plots of approximately 0.8 kg m–2, assuming a bulk density of 1.4 kg L–1 and a depth of the plow layer of 0.25 m. Kjaergaard et al. (2004a) were able to mobilize about 1% of LE-WDC of clay soils with differing clay contents when exposing small soil cores to gentle irrigation (1 mm h–1) for 4 to 5 d. Conducting the same laboratory experiment with small soil cores sampled at Røgen, we collected a quite similar amount of particles (22 mg, equivalent to 1% of LE-WDC of the soil core). In contrast to this, we collected a maximum of 3 g m–2 in experiment PL–, corresponding to only 0.4% of LE-WDC. All experiments collected a total of 4.8 g m–2 for Plot 1 and 3.8 g m–2 for Plot 2, equivalent to 0.6 and 0.5% of LE-WDC, respectively. Three factors help to explain why we collected fewer particles in the field. First, the laboratory experiments on short soil cores had a runoff/irrigation ratio of 1, while the tile drains collected a smaller fraction of the irrigated water. Second, filtering effects in the subsoil (Kretzschmar et al., 1999) probably reduced the number of mobilized colloids that were effectively transported to the depth of the drains. Finally, the laboratory experiments applied a much lower irrigation rate (1 vs. 18 mm h–1 at the field plots), allowing more time for diffusive exchange of colloids between the matrix and the flowing water compared with flow conditions in the field plots. The field experiments were more dominated by macropore flow due to the high irrigation rate, implying that much of the water bypassed the soil matrix and its source of dispersible colloids.
Slurry Effects: Particle Facilitated and Soluble Phosphorus Leaching to Drains
Slurry-applied P amounted to 49 g P per plot (20 kg P ha–1) at the first application in April 2001 and 33 g P per plot (13 kg P ha–1) at the second application in June 2001. At the five irrigation campaigns following slurry application, totals of P per plot collected from the drains amounted to <0.1% of applied P (Table 2). It is clear that the losses via drains were negligible from an agronomic point of view. From an environmental point of view, the observed P losses at single events were comparable with reported annual losses from Danish forested catchments (4–14 mg P m–2 y–1; Kronvang et al., 2005), but lower than the estimated mean annual loss of 50 mg P m–2 from agricultural land in Denmark. Mean and maximum concentrations of total P observed in the drains were high, on the order of 0.5 to 2 mg P L–1 and reaching a temporary maximum of 4.8 mg P L–1 for Plot 1 in experiments SL1+ and PL+ (Table 2). These concentrations were clearly higher than the reported mean concentrations of 0.042 to 0.059 mg P L–1 in runoff from "undisturbed" forested catchments with little agricultural activity (Kronvang et al., 2005). The observed mean and maximum P concentrations at Røgen also exceeded all drainage concentration observations reviewed by Sims et al. (1998; their Table 1). Heathwaite and Dils (2000) found comparable maximum concentrations of 0.7 and 0.5 mg L–1 at the 100- and 150-cm depths, respectively, when collecting water samples from piezometers installed in a grazed and slurry-treated grassland hill slope. Preedy et al. (2001) observed high maximum concentrations of total P (7 mg L–1) at 24 to 32 h after dairy slurry application to grass plots on a clayey soil. The rate of slurry P application in the study of Preedy et al. (2001) (2.9 g P m–2) was moderately higher than our application rates of 2 g P m–2 (SL1) and 1.3 g P m–2 (SL2). However, their plots had been grazed by sheep and beef cattle before the experiments, thus adding to the P load by manuring; moreover, their water samples were a composite of surface and lateral subsurface flow (0–27 cm), which is considerably closer to the P source compared with our drains at the 1.1-m depth.
|
|
The H+ experiment in November 2001 took place well beyond the time of manure applications. At this stage, P originating from the cattle manure had presumably been either sorbed to the topsoil or taken up by the grass crop during the growing season. The content of dissolved P in the effluent was still quite high at H+ (30–50% of TP) and markedly higher than the modest content of <20% observed toward the end of the winter (Table 2). The relatively high proportion of dissolved P may be attributable to the decomposition of grass residues. Nutrient mineralization had progressed since the herbicide-aided termination of the grass about 40 d before H+, and in the absence of crop uptake there was potential for nutrient leaching, including P leaching, with subsequent rainfall events (Sharpley and Smith, 1989; Havis and Alberts, 1993; Moretto et al., 2001).
The observed off-season dominance of particulate P over dissolved P leaching, in our case during winter and early spring, may be a consequence of diminishing contributions from the sources that provided a high soluble input at earlier stages (fresh manure, decomposing grass). Investigating soil cores sampled at Røgen in early spring, de Jonge et al. (2004a) also found that 75% of leached P was transported in a particle-facilitated manner. Turtola and Jaakkola (1995) compared surface and subsurface runoff losses of P for a 3-yr period with a heavy clay soil in Finland used for barley (Hordeum vulgare L.) and grass ley. The majority of drainage P was lost as particulate P, especially for the barley crop that was not as heavily fertilized as the grass ley. Heathwaite and Dils (2000) found PP to dominate in drain flow below grassland all year; however, it was most pronounced during the winter months, past the time of spring slurry application and summer dairy herd grazing.
At every experiment inorganic P forms were more prevalent (54–75% of TP) than organic P forms in the effluent (Table 2, Fig. 5). Dissolved inorganic P (orthophosphate) is generally considered relatively immobile in soil because of its absorption to soil particles via adsorption–precipitation processes (Frossard et al., 2000; Preedy et al., 2001). The substantial amounts of DIP and PIP that we observed in the effluent may be due to the prevailing macropore flow processes, causing storm water to bypass most of the soil matrix at the Røgen plots. Our observations are in accordance with Turner and Haygarth (2000), who found total P export during a drainage year from four soil types to be dominated by (dissolved) inorganic P. On the other hand, our results are in contrast to Toor et al. (2004), who found organic P forms to represent 77 to 91% of TP in leachate from fertilizer-treated lysimeters originating from grassland. Similarly, Chardon et al. (1997) found that more than 75% of TP was DOP when collecting leachate from lysimeters that had been fertilized with cattle manure or mineral N.
A large disparity between various field studies with respect to observed amounts of inorganic versus organic P, and dissolved versus particulate P, is to be expected. The leaching patterns of P depend on several factors, such as soil type, hydrology, event size, types of fertilizer, and field management (Turner and Haygarth, 2000; McDowell et al., 2001), variants that may obscure comparisons among field studies.
A further note concerns the fact that most prior P leaching studies calculated DOP as a residual between TDP and TRP determined in filtered samples using standard 0.45-µm filters. Most likely a fraction of smaller colloids remained in the filtered samples, causing the DIP fraction and the DOP residual to be altered due to interactions between PO4 and colloids during analysis (Sinaj et al., 1998; McDowell and Sharpley, 2001; Hens and Merckx, 2002). The possible interference in the quantification of P forms by the presence of colloidal particles guided our choice of centrifugation corresponding to a lower cut-off diameter of particles of 0.24 µm. It would have been highly relevant to include even lower cut-off diameters in the study (Haygarth et al., 1997; Heathwaite et al., 2005). However, as was also stated by Turner et al. (2004), such analyses complicate and lengthen the analytical work, which was incompatible with our field setup and the number of drain water samples for analysis following each experiment. Therefore it was beyond the scope of our work to pursue the importance of the different size classes of the colloidal material. A recent study (Heathwaite et al., 2005) lends us some confidence that we captured the majority of colloidal P. These authors found the larger-sized particles (>0.45 µm) to be the most important carriers of TP in runoff from slurry treated field plots, while for untreated land very small particles (<0.001 µm) also were important carriers.
In all experiments the concentrations of PIP and POP were closely correlated to particle concentration. Correlation coefficients between PIP (or POP) and particle concentration typically exceeded 0.8 (Table 2). The total P associated with particles (PP) was thus quite constant for irrigation events (Fig. 6a ). The concentration of PP was linearly related to the concentration of particles in the drain water, with regression line slopes (Fig. 6b) varying between 3.2 mg P g–1 particles (BG, Plot 2) and 5.8 mg P g–1 particles (SL1+, Plot 1). The range of PP densities across events may also be seen from the tabulated "PP per particle" in Table 2. It was clear that the amount of P associated with particles increased after slurry application and then decreased to an intermediate concentration level. Many investigators (Heathwaite and Dils, 2000; Bottcher et al., 1981; Culley et al., 1983; Heckrath et al., 1995; Grant et al., 1996; Djodjic et al., 2000) have also observed significant positive relationships between particulate matter and particulate P concentrations in effluent. Some additional results, representative of Danish conditions on an event basis, have been included in Fig. 6b. Laubel et al. (1999) observed P content of particulate matter in drain water at Røgen to be constant and on the order of 2.8 mg P g–1 particles. de Jonge et al. (2004a) performed irrigation experiments on Røgen soil cores and found cumulated PIP and cumulated POP to be closely correlated to the accumulated mass of leached particles. Their data (Fig. 6b) represents an intermediate slope of 4.1 mg P g–1 particles. Further, Kjaergaard and de Jonge (unpublished data, 2005) examined effluent water sampled from Danish drain line observation sites and also found rather constant relationships between PP and particulate matter. Data from the Estrup and Silstrup sites (structured sandy loam) have been added to Fig. 6b. We conclude, in accordance with Laubel et al. (1999) and de Jonge et al. (2004a), that the constant fraction of PP throughout experiments indicates the source of mobilized particles to be the same whenever and wherever the particles are transported during an event, during the early stage or later during the event. In addition, Fig. 6 indicates that for Danish structured soils, the PP concentrations generally range between 3 mg P g–1 particles (soils not recently fertilized) and 6 mg P g–1 particles (soils recently amended with animal manure). Grant et al. (1996) presented relationships between pooled weekly concentrations of PP and particulate matter in drainage water for four Danish catchments. Observed PP densities in particulate matter in three catchments were within the above limits (3–6 mg P g–1). Slightly higher PP densities (7 mg P g–1) were found in a catchment with significant drain water contributions from grazed and occasionally waterlogged grassland where P release to the soil water solution may be enhanced due to reducing conditions (Jensen et al., 1999; McDowell et al., 2001). Results by Rubæk et al. (1999) further support the reliability of the range in P densities found for Danish conditions. Rubæk et al. (1999) examined clay-sized separates from (long-term) unfertilized, mineral-fertilized, and animal manure–amended sandy loam soils. For the long-term unfertilized soil they found a total P content of 2.6 mg g–1 clay separates. For the regularly, but not recently, fertilized soils they found P contents of 3.9 mg total P g–1 clay separates.
|
Implications for Phosphorus Leaching
A plot or field-scale experiment is an important step to acquire knowledge that enables the prediction of particle-facilitated transport at larger scales, such as at the catchment scale. When evaluating the results of the current study we must bear in mind two aspects: (i) the plots were tile-drained, and (ii) the results were obtained by monitoring storm events.
Subsurface drainage systems are convenient for investigating the leaching from field soils. A number of papers (e.g., Skaggs et al., 1994) discuss the effect of tiles on the leaching patterns of nutrients and pesticides. Drainage systems are usually installed to avoid water logging. Therefore, artificial drainage changes the major flow patterns from surface storage and runoff to subsurface leaching to drains (Brown et al., 1995; Gachter et al., 1998; Dils and Heathwaite, 1999; Simard et al., 2000). Stamm et al. (2002) argued that effluent from tile drains may not be representative of water reaching shallow groundwater in undisturbed soils since the establishment of the tile drains creates artificial preferential flow paths. Their results were obtained in a weakly structured loamy soil. Also, Uusitalo et al. (2001) hypothesized that the drainage excavation itself was the main pathway for suspended particles to the drains. Their results were obtained on heavy clay soils whose structure is much destroyed when dug out, allowed to dry, and returned as backfill around drain pipes. At the Røgen site the soil is neither heavy clay nor unstructured loam, and we regard the effects of the drainage excavation to be negligible after 50 yr of settling. A network of natural macropores is present (Laubel et al., 1999) to distribute soil water down the profile. If artificially created pathways above the drainpipe were important for the leaching from the plots, we would have expected higher runoff/irrigation ratios than the maximum of 36% obtained. We consider the leaching patterns and timing observed at the Røgen site to be representative of leaching from relatively flat, well-drained and structured soils, soil types that are quite common for Danish agricultural land (Krogh and Greve, 1999).
In our study we focused on storm events because it has been shown that these account for the majority of annual particle and particle-facilitated losses (Grant et al., 1996; Laubel et al., 1999; Uusitalo et al., 2001). Drainage water during storm events also appears to be enriched in dissolved matter compared with base flow conditions (Grant et al., 1996; Heathwaite and Dils, 2000). The type of storm we produced by irrigation in terms of duration and intensity was one that would only rarely occur in Denmark. Thus, inducing six such storms within 1 yr is extraordinary, and we cannot use the results of total losses of particulate matter and P to predict an average annual loss at Røgen. The intense irrigation was applied to promote preferential flow and to allow identification of effects of management practices on the leaching of dissolved and particulate matter.
Preferential flow affects mobilization and transport of colloids through the soil in different ways. The chemistry of the pore water depends on the mixing of infiltration water with resident water. With an increasing degree of preferential flow, a decreasing fraction of the soil water actively participates in water movement. Reduced displacement of high-ionic strength resident water with low-ionic strength infiltration water and reduced diffusive exchange of ions will delay the dispersion of colloids from soil aggregates. Under the storm conditions that we induced, the macropores probably came to act as the most important source of mobile colloids. This led to an initial rapid depletion of the mobile colloids in the pore water followed by a slow diffusive release of colloids from aggregates near the macropore walls. Kjaergaard et al. (2004c) applied low irrigation rates to small cores of structured clay soil and obtained larger accumulated leaching of colloids in soil exhibiting matrix-dominated flow behavior than in soil with dominating preferential transport. This highlights that quantitative significant colloid mobilization in the topsoil is not restricted to macroporous soils. However, once colloids have been mobilized to the flowing water, preferential flow in the subsoil will increase the transport velocity and reduce the recapture (filtration etc., Kretzschmar et al., 1999) of colloids compared with matrix dominated flow.
|
CONCLUSIONS |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
In all cases we found rapid arrival of water to the drains and flow conditions indicating macropore flow. The turbidity of the water samples generally fluctuated in accordance with earlier laboratory findings on cores of soil originating from the same site—just after the onset of water flow particle concentration was high, peaking before flow peak. The particle concentration sometimes showed a secondary peak at maximum drain flow and then gradually decreased with flow rate after irrigation was terminated. We attribute the colloid leaching pattern to an initial depletion of a high concentration of colloids that has built up in the pore water before the onset of water flow. The eventual low particle concentration is due to diffusion-limited release of colloids from immobile intra-aggregate water to the mobile water. Secondary particle peaks are attributable to decreasing electrical conductivity in the intra-aggregate water as low-conductivity infiltration water progressively dilutes pore water. We observed no particle peaks on the falling section of the hydrograph, meaning that particle release could not be explained as a function of flow transients.
The results showed that seasonal variability and cultivation practices caused significant intra-annual variability in the leaching of particulate and dissolved P. Just after slurry application in April and June, dissolved P dominated the P loss to the drains. This was probably due to (i) significant amounts of dissolved P in the slurry matrix applied at the grassland surface and (ii) slurry application coinciding with springtime conditions with rising soil temperatures and increased microbial activity and turnover of soil organic matter. At the irrigation storms in autumn and winter, particle-facilitated transport of P came to dominate over dissolved P transport, reaching a maximum in early spring when more than 80% of P loss was in the particulate form.
Field studies have often concluded that no-tillage management systems conserve macropores in the topsoil and thus enhance leaching of contaminants compared with conventional tillage systems. Our study indicates that in the latter systems, the plowing process promotes particle mobilization and exposes new aggregates to the soil solution, increasing the risk of particle-facilitated and dissolved P leaching in storms occurring shortly on the inversion of the soil. We observed an almost fourfold increase in the leaching of TP per volume of leachate when comparing irrigation experiments just before and just after plowing.
We found the amount of P associated with particulate matter in the drains to be constant within events, but varying between storms. Comparing P densities in particulate matter from the present study with previous results for Danish structured clay soils, we found P density to range between a maximum of 6 mg g–1 for soils that have recently been fertilized with animal manure and an approximate minimum of 3 mg g–1 for soils not recently fertilized.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
