| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Alberta Agriculture and Food, Conservation and Development Branch, 206, 7000, 113 St. Edmonton, AB, T6H 5T6
b Alberta Agriculture and Food, Irrigation Branch, 100, 5401, 1st Ave S., Lethbridge, AB T1J 4V6
* Corresponding author (joanne.little{at}gov.ab.ca)
Received for publication November 16, 2006.
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
Abbreviations: DPS, degree of phosphorus saturation • DP, dissolved phosphorus • DRP, dissolved reactive phosphorus • FWMC, flow-weighted mean concentration • PP, particulate phosphorus • PSI, phosphorus sorption index • STP, soil test phosphorus • TP, total phosphorus
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
The prediction of phosphorus losses from land has been a major focus of research during the past decade. Many researchers have reported a direct linear relationship between phosphorus concentrations in soil and levels of dissolved phosphorus in runoff (Sharpley et al., 1977, 1978; Daniel et al., 1994; Pote et al., 1996; Torbert et al., 2002). However, most of these relationships have been derived from rainfall simulations at laboratory or small-plot scales and may not adequately represent relationships from natural rainfall at field, catchment, or watershed scales since variables are site and soil specific (Young and Mutchler, 1976; Mannaerts, 1992). Laboratory- and plot-derived relationships must be validated at field scales as complex and scale-dependent hydrological processes govern the amount and forms of phosphorus loss (Bloschl et al., 1995; Le Bissonais et al., 1998; Nash et al., 2002). Furthermore, while many relationships have been developed to predict losses of dissolved phosphorus, water quality guidelines are based on total phosphorus (TP) as dissolved and particulate forms contribute to eutrophication (Wetzel 2001). Conventional methods of filtering and analyzing a water sample for dissolved phosphorus forms can overestimate biologically available phosphorus (Fisher and Lean, 1992; Hudson et al., 2000); thus, TP is recommended as a more meaningful measurement of phosphorus in surface waters (Wetzel, 2001).
Although strong relationships between soil test phosphorus (STP) and runoff phosphorus have been observed at lab and plot scales, results from field-scale research have been confounded by variation in soil types, hydrology, and management. Sharpley et al. (1996) reported that the relationship between STP and dissolved reactive phosphorus (DRP) in overland flow at the field scale varied with soil type, management, and runoff episodes. Sharpley et al. (2002) found that extraction coefficients (slopes of the regression lines) increased with greater erosion or reduced soil cover due to greater interaction between soil and overland flow, and they suggested that an erosion function may also be necessary to predict soil phosphorus release. Sharpley (1995) and Sharpley et al. (1996) concluded that field-scale coefficients are too variable to allow the use of a single or average relationship for all soils under the same management due to the inherent variability among soils and to the soil-specific nature of soil phosphorus release to overland flow. However, Vadas et al. (2005a) proposed that a single extraction coefficient could be used to approximate DRP release from soil to runoff, based on lab-scale and plot-scale results from 30 soil types. Since extraction coefficients from field-scale studies are not well documented, an understanding of the relationship between STP and phosphorus in runoff in conditions representing local climate, soil type, land use, and management is needed.
While most researchers have reported poor relationships between STP and TP at the field-plot scale (Andraski and Bundy, 2003; Kleinman et al., 2004), Schroeder et al. (2004) found stronger relationships between STP and TP (r2 = 0.69) than between STP and DRP (r2 = 0.56) at the field-plot scale. Kleinman et al. (2004) found that boxes packed with disturbed field soil yielded increased concentrations of TP relative to the grassed field plots from which the soil had been taken. This was attributed to reduced infiltration, increased surface flow, bare soil conditions, and increased erosion for the lab-scale packed boxes relative to the field plots. Erosion is much more scale-dependent than dissolution and is therefore difficult to replicate at the lab scale. However, as TP is used in determining the trophic status of surface waters and in water quality guidelines, it is important to estimate TP losses from agriculture.
The degree of phosphorus saturation (DPS) may also be an important factor in determining runoff P losses. The DPS is a measure of how saturated soil sorption sites are with phosphorus, and is influenced by a number of variables, including aluminum, iron, calcium, clay, organic matter, pH, and carbon/phosphorus ratio. Sharpley (1995), Pote et al. (1996), and Hooda et al. (2000) reported that soils with similar STP levels have yielded different amounts of runoff phosphorus due to differences in phosphorus sorption capacity (PSC). Vadas et al. (2005a) found a split-line relationship where DRP rapidly increased at DPS values greater than 12.5% for noncalcareous soils.
The question of whether an environmentally oriented soil sampling method is more appropriate for understanding the relationship between STP and phosphorus in runoff has not been resolved, particularly for Alberta conditions. Kleinman et al. (2000) proposed a soil chemical approach for determination of soil P sorption thresholds (i.e., change points) related to soil P transfer to waterways. These thresholds are based on P sorption saturation levels in the soil and they delineate a critical soil P loading level above which any added P may be lost more readily via surface runoff or leaching. Split-line models have been used to determine thresholds where the relationship between STP or DPS and dissolved reactive P (DRP) in runoff or drainage are split into two sections, one with greater P loss per unit soil P than the other (Hesketh and Brookes, 2000; McDowell and Trudgill, 2000). Quantity/intensity (Q/I) relationships such as these have been used to identify change points in several recent studies (McDowell and Sharpley, 2001; Maguire and Sims, 2002a, 2002b; Indiati and Sequi, 2004; Nair et al., 2004; Casson et al., 2006).
The main objective of this study was to determine the field-scale relationship between STP and runoff TP and DRP from field-sized catchments or "microwatersheds" under spring snowmelt and rainfall conditions in Alberta. We also examined whether a variety of depths and spatial representations of STP improved the prediction of phosphorus losses, and we explored the use of the DPS as an alternate method of predicting phosphorus export at the field scale.
|
Materials and Methods |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
|
|
|
/tan ß) topographic index was also calculated, where is the accumulated upslope contributing area that drains to a given point, and ß is the local slope angle (Quinn et al., 1995).
Soil Sampling and Analysis
A minimum of six, three-point transects, plus additional points with a range of topographic index values, were sampled at each site to achieve a density of one sample per 1 to 5 ha. The exception was the 2-ha STV site where only three sampling points were selected. A satellite-based navigational system, or differential global positioning system (horizontal accuracy <1 m) was used to locate sampling points and navigate back to the same points. Fall sampling in 2002, 2003, and 2004 was conducted after all field operations were completed. A subsample of points (n = 5 to 10) identified by high wetness index values was sampled after seeding and fertilizing had been completed in the spring of 2003, 2004, and 2005.
An excavation method was used to obtain representative portions of fertilizer bands or manure and soil using a 19- by 50-cm frame placed diagonal to crop rows. The frame dimensions were changed after the fall of 2003 to 11 cm by two times the fertilizer bandwidth and placed perpendicular to the seed row. Soil samples were excavated from the 0- to 2.5-cm, 2.5- to 5-cm, and 5- to 15-cm layers. One frame per sampling point was used for non-manured fields, while two frames per sampling point were used at the remaining sites. The excavated soil layers were thoroughly mixed in the field, and a 500-g subsample was shipped in coolers to the laboratory.
Soil samples were dried and ground to pass through a 2-mm sieve and a 5-g subsample was removed for STP analysis. Samples taken in the fall of 2002 and the spring of 2003 were analyzed for STP using the modified Kelowna extraction method (0.015 M NH4F, 1.0 M HOAc, 0.5 M NH4OAc) of Ashworth and Mrazek (1995) and the remaining samples were analyzed using the modified Kelowna extraction method (0.015 M NH4F, 0.25 M HOAc, 0.25 M NH4OAc) of Qian et al. (1991). Values of STP for the 0- to 5-cm and 0- to 15-cm soil layers were calculated by proportional weighting of the measured results. A large volume (20 to 25 L) of soil from the 0- to 15-cm layer collected at each site was used as a reference sample. The reference samples were dried, ground to pass through a 2-mm sieve, well mixed using a cement mixer, then subsampled using a sample splitter. Ten subsamples per site were analyzed at each laboratory used in the study. Soil test phosphorus values were then standardized to the reference sample results to account for differences in methodology, using factors calculated from the difference between STP concentrations at one lab relative to concentrations measured at the final lab. The mean adjustment factor was 1.14 with a standard deviation of 0.19 mg kg–1.
The phosphorus sorption capacity was characterized at six transects per site for the 0- to 2.5-cm layer using samples from the fall of 2003. A subsample of six points sampled in the fall of 2002 and the fall of 2004 at each of the manured sites was also analyzed. A calcium chloride method was used to measure the phosphorus sorption index (PSI) of each soil (Casson et al., 2006). The DPS was determined as the ratio between STP to PSI plus STP (Indiati and Sequi, 2004). The PSI results from the fall of 2003 were used in the DPS calculations for the fall of 2002 and 2004 at the non-manured sites. The soils at the non-manured sites did not receive any organic amendments during the study so it was assumed that the PSI values would remain stable with time.
Soil Test Phosphorus Sampling Strategies
Five sampling strategies of STP were calculated for each soil layer:
Only the fall samples from the 3 yr were used to calculate the STP sampling strategy means, and the STV site data were not included in the sampling strategies because there were only three sampling points at the 2-ha site.
Water Measurement and Sampling
Most sites were instrumented with circular flumes (Samani et al., 1991). The PON site was initially instrumented with a 0.61-m H-flume, which was replaced with a circular flume in June 2003. The STV site was bordered on the down-slope edge with a trough, which directed runoff water into a 0.15-m trapezoidal flume. Stage was recorded at 5-min intervals using a float potentiometer. Circular flumes were calibrated using the Water Ware software program developed by Samani et al. (1991). The resulting calibrations were then plotted in TableCurve 2D, version 3 (Jandel Scientific Software, 1994) to fit an appropriate curve to the data. Once a curve was selected and applied to the stage readings, a correction factor was applied to account for any inactive head in the flume.
The sites were also equipped with staff gauges and float potentiometers to record stage, Lakewood TP10K5 thermistors (Lakewood Systems Ltd., Edmonton, AB, Canada) to measure air temperature, and Davis tipping bucket rain gauges (Davis Instruments Corp., Hayward, CA), which were replaced with Texas tipping bucket rain gauges (Texas Electronics Inc., Dallas, TX) in May 2004. Sites were powered with two 15-W solar panels, and rechargeable 12-V batteries. Each site was equipped with integrated dataloggers and cellular communications technology obtained from ROM Communications that allowed real-time monitoring of the sites (ROM Communication Inc., Kelowna, BC, Canada). When flow or precipitation was detected, data were reported on a website and field staff alerted. The STV site was equipped with a Lakewood Ultralogger (Lakewood Systems Ltd., Edmonton, AB, Canada), a float potentiometer, and a meteorological station, and a technician collected all flow data and water samples from this site. Staff gauge readings were recorded at all sites during field visits and were used as backups for the real-time flow data recorded by the ROM dataloggers.
Water samples were taken by ISCO 6700 automated water sampling devices (Teledyne Isco Inc., Lincoln, NE), equipped with 24, 1-L ProPaks and disposable polyethylene inserts. The ISCO samplers were programmed to sample 100 mL every 15 min once changes in stage were detected.
Water samples were retrieved daily during runoff events and immediately transported in coolers to the laboratory. Water samples were analyzed within 24 h for pH and electrical conductivity, and within 30 d for TP (persulfate digestion: APHA 4500E). Subsamples were filtered on arrival (0.45-µm filter) and analyzed within 48 h for DRP using the ascorbic acid method of Murphy and Riley (1962), and within 30 d for dissolved phosphorus (DP) (persulfate digestion: APHA 4500E). Selected samples were analyzed for additional parameters, including total suspended solids (TSS). Blanks filled with deionized water, as well as a prepared standard of known phosphorus concentration, were submitted to the lab with each batch of samples as part of a quality assurance/quality control program.
Data Analysis
Runoff Phosphorus Calculations
To calculate flow-weighted mean concentrations (FWMCs), water chemistry data were linearly interpolated to 1-min intervals using Proc Expand in SAS (SAS Institute Inc., 2003). The interpolated concentration data were then matched to the flow data and instantaneous loads were calculated for matching values by multiplying flow and concentration data. The area under the curve was then integrated to estimate total loads and flow volumes using a SAS area macro. Seasonal FWMCs were then calculated by dividing the total load for all events by the total flow volume. Where possible, missing flow data were supplemented by manual staff gauge readings. However, in some cases, mean concentrations had to be substituted for days with missing flow data.
Statistical Analysis
Statistical analyses of the soil and water data were completed using SAS version 9.1 (SAS Institute Inc., 2003). Differences between means were tested using the Least Squares Means test in the PROC MIXED procedure, with variance components as the variance structure, and a Fisher's protected LSD test. The REG procedure was used to relate measures of STP to seasonal FWMCs of runoff phosphorus and the Type III sums of squares in the Mixed procedure was used to determine if there were significant differences in slopes and intercepts between regression equations. A significance level of 0.05 was used throughout this study. A PROC NLIN split-line model was used to determine the environmental soil P thresholds (change points). The NLIN procedure required estimation of linear (d + ex) and quadratic (a + bx + cx2) estimation parameters, and solved for the threshold between the linear and quadratic regressions by iterative re-evaluation of the equation. The STP values that corresponded to change points were subsequently determined from the relationships between STP and runoff P.
|
Results and Discussion |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
|
Soil Test Phosphorus Sampling Strategies
There were few significant differences among the STP sampling strategies in the 0- to 2.5-cm and 0- to 15-cm soil layers in the fall of 2002 (Table 3). This was also true for the 0- to 5-cm layer (data not shown). Significant differences were observed in the 0- to 15-cm layer at the GPC site in the fall of 2002, 2003, and 2004, where the runoff contributing area STP was significantly less than the site mean and random STP values (Table 3), and in the 0- to 5-cm and 0- to 15-cm layers at the LLB site in the fall of 2004 where the runoff contributing area STP was significantly lower than the site mean and the representative random STP values (data not shown) because manure was not applied in the wet lower landform positions in the fall of 2004. The differences among the five STP sampling strategies in the 0- to 2.5-cm layer ranged from 3 to 9 mg kg–1 for the non-manured sites and from 32 to 70 mg kg–1 at the manured sites (Table 3). Soil test phosphorus in the 0- to 15-cm layer at the PON site was 1.1 times greater in the runoff contributing area samples compared with the site mean and the representative random samples, though they were not statistically different. This is lower than the fourfold increase in "effective" soil P status observed by Page et al. (2005) when they collected samples from areas with overland flow that were directly connected to the stream and compared to a site mean in a grassland catchment in the United Kingdom.
Despite reports of accumulation of phosphorus in lower landscape positions in Alberta (Penney et al., 2003), there were no significant differences by landform position at five of the seven cultivated sites and results at one of the sites was conflicting, with more phosphorus in the mid or upper landform positions (Nolan et al., 2007). As such, few differences were observed between the landform area-weighted representation and the site mean. Although these values could not be statistically compared as they were calculated from area-weighted means, the landform area-weighted values were similar to the other STP representations. This is partly due to the similarity in STP in different landforms at conventionally tilled sites and the low proportion of lower landform areal extents at reduced and no-till sites (Table 2), where greater differences in STP by landform position were observed (Nolan et al., 2007).
Page et al. (2005) noted that important information on the variability and spatial distribution of STP for a given sampling area can be lost when samples are averaged. However, Daniels et al. (2001) concluded that when sampling soil phosphorus in pastures, current sampling strategies for agronomic soil tests can adequately account for spatial variability to produce a single, appropriate estimate of STP, if the recommendations are followed with respect to the required number of samples. Similarly, Needelman et al. (2001) concluded that field mean STP in hog and poultry manure-amended fields could be used to characterize STP for applications that are not sensitive to small errors in STP estimates. In a study of manured and non-manured soils in Manitoba, Slevinsky et al. (2002) reported that there were no differences in STP levels measured in the 0- to 15-cm layer using a composite of 15 random points per field or using the average of four representative benchmark samples per field.
Degree of Soil Phosphorus Saturation
The PSI in the 0- to 2.5-cm layer was significantly higher at the GPC site than at any other site (442 mg kg–1, Fig. 2) due to the greater clay content and more recent cultivation (Table 2). The range among PSI values at the five non-manured sites (296 mg kg–1) was much larger than the range among STP values in the fall of 2003 (8 mg kg–1; data not shown). The PSI at the heavily manured PON site (49 mg kg–1) was significantly lower than at all other sites. The DPS in the 0- to 2.5-cm layer was significantly lower at the ungrazed grassland STV site (5%) than at any other site, while DPS at the GPC site (10%) was significantly lower than at the other cultivated sites (Fig. 2). The DPS was significantly higher at the heavily manured PON site (91%, Fig. 2) than at all other sites. Casson et al. (2006) showed that the PSI of medium- and coarse-textured Alberta soils decreased significantly with increasing manure rates while the DPS increased significantly to values >90% for a coarse-textured soil and >70% for a medium-textured soil after 8 yr of annual manure application. There were no temporal differences at the manured sites in DPS over the 3-yr period at any of the sites (Little et al., 2006).
|
|
|
The DRP and TP FWMCs from the STV site in 2003 were within the range of the non-manured sites despite an STP level that was about one-third of the non-manured sites (Table 5). Concentrations of DRP and TP were comparable to those reported by Timmons and Holt (1977) from native grasses in Minnesota. The relatively high values may be due to leaching of DRP from the large amounts of vegetation cover and surface thatch at this site. In addition, freezing and thawing of plant material dramatically increases the amount of nutrients that can be leached (Timmons et al., 1970; Bechmann et al., 2005). In 2004, the DRP and TP FWMCs were much lower, possibly due to the much smaller volume of runoff.
At the manured sites, levels of DRP and TP were greatest in the spring of 2003 following manure applications in the fall of 2002. Individual values at the PON site in the spring of 2003 were as high as 24 mg L–1 DRP and 108 mg L–1 TP, with FWMCs of 16.5 mg L–1 DRP and 23.5 mg L–1 TP. The manure was applied just before freeze up and poorly incorporated in late 2002. The PON site had a very high DPS (Fig. 2), suggesting that it had little capacity to bind phosphorus. In addition, TSS concentrations were elevated (data not shown) and accumulation of sediment in the flume was observed during field visits, indicating selective sampling of sediment from the H-flume. Therefore, samples with extreme TSS concentrations and TP/DP ratios greater than 10 were deemed to be outliers and removed from the dataset. Even with these extreme values removed, the spring 2003 TP FWMC was still three times greater than from any other runoff event.
In contrast, the LLB site had manure applied to the portion of the watershed nearest the flume in the spring of 2002, which allowed greater opportunity for phosphorus to be adsorbed by soil and mixed with the subsurface soil by intensive tillage following the spring manure application and fall harvest. The DRP FWMCs values at the LLB site were an order of magnitude lower than at the PON site. Previous studies have indicated that when soils have received surface applications of manure, the manure phosphorus overwhelms the soil phosphorus and becomes the major source of phosphorus to runoff instead of the soil (Pierson et al., 2001; Kleinman et al., 2002). Therefore, STP is often not an accurate representation of runoff-available phosphorus. However, the differences between amended and unamended soils are much less if the manure has been incorporated (Kleinman et al., 2002) or has had time to equilibrate with the soil (Eghball et al., 2002).
Summer runoff FWMCs at the manured sites were more variable, with individual event DRP FWMCs ranging from 0.84 to 3.01 mg L–1 at the LLB site and from 5.25 to 6.63 mg L–1 at the PON site. Summer 2003 and 2005 runoff values from the LLB site were much greater than in 2004 as the portion of the site nearest the outlet received manure in the spring of 2002 and the fall of 2004. Declines in summer runoff were likely related to decreased levels of STP due to the equilibration of the manure with the soil, dilution by tillage, and crop uptake.
Flow-weighted mean concentrations of DRP were not significantly different among runoff types; however, the REN and THC sites had significantly higher concentrations of TP and lower DP/TP ratios in rainfall runoff (Little et al., unpublished data, 2005) compared with snowmelt runoff. Concentrations and DP/TP ratios were not different between rainfall and snowmelt runoff at the manured PON and LLB sites. This may be related to greater sediment losses from the increased erosion of unfrozen soils and/or greater precipitation intensity from rainfall compared to snowmelt.
Relating Phosphorus Concentrations in Soil and Runoff
Seasonal FWMCs were used to relate phosphorus concentrations in soil and runoff. Spring snowmelt runoff results were related to the soil sampling results from the previous fall, while summer runoff events were related to the soil sampling results from the spring of the same year. Results from the spring runoff in 2003 at the PON site were excluded due to the recent application of manure that was poorly incorporated just before freeze up and the selective sampling of sediment by the ISCO due to sediment accumulation in the H-flume.
Soil Test Phosphorus Sampling Strategies Comparison
Strong linear relationships were found between all STP sampling strategies and the spring snowmelt FWMCs of DRP and TP (Table 6). Reports in the literature have suggested that soils in critical source areas can have greater influence on phosphorus loss in runoff than soils in other areas of the field (Gburek and Sharpley, 1998); however, in our study, there were no significant differences among the regression equations for all STP sampling strategies. Coefficients of determination for the site mean were similar or slightly greater using the landform area-weighted or runoff contributing area sampling strategy means. Representative random sampling and a random subset of samples also produced similar results to using the site mean. Differences among the five STP sampling strategies were minimal (Table 3), and this was reflected in the similarity among regression results (Table 6). This was partly due to the observation that few differences by landform position were detected and that variable management practices, such as the uneven distribution of manure at the LLB site or conventional tillage at the GPC and WAB sites, obscured expected differences in STP by landform position (Nolan et al., 2007). The similarity between regression equations may also be partly attributed to the observation that most of the runoff was generated from large spring snowmelt events, which generated runoff from a greater proportion of the field due to restricted infiltration on frozen soils compared with summer precipitation events.
|
Soil Test Phosphorus and Runoff Phosphorus Relationships
There were strong linear relationships between STP and DRP and TP FWMCs for the 0- to 2.5-cm and 0- to 15-cm soil layers (Fig. 3). The relationships for the 0- to 5-cm layer (not shown) were similar to the 0- to 2.5-cm layer. Due to the relatively narrow range of STP among the non-manured sites, the manured sites drive the relationships (Fig. 3). The relationships between STP in the 0- to 15-cm layer and FWMC DRP (r2 = 0.0008) and TP (r2 = 0.052) were extremely poor when the manured sites were omitted. The distribution of values improved with time since manure was not applied to the PON site after the fall of 2002 or between the fall of 2002 and the fall of 2004 at the LLB site, which resulted in lower STP values from these sites as the manure was incorporated into the soil by tillage and manure phosphorus became equilibrated with the soil. Additional data from summer runoff events that corresponded with increased STP levels at the non-manured sites helped to improve the distribution of points. However, there were no observations within the STP range of 75 to 150 mg kg–1, as even a single manure application can rapidly increase the STP levels in soil (Volf et al., 2007). Corresponding changes in STP and the DRP and TP FWMCs at the manured sites support that the relationship is linear and other studies in Alberta have also found a linear relationship within this range (Volf et al., 2007; Wright et al., 2006).
|
Many studies have reported strong linear relationships between STP and DRP in simulated runoff at lab and field scales (Wright et al., 2003; Vadas et al., 2005a). However, very few have developed relationships with TP (Schroeder et al., 2004), which combines dissolved and particulate fractions. Particulate phosphorus (PP) concentrations can be impacted by several additional factors related to erosion including tillage (Zhao et al., 2001), event size (Quinton et al., 2001), crop cover, and clay content of the soils (Calhoun et al., 2002). These factors are often difficult to evaluate under lab- or plot-scale rainfall simulations because erosion processes operate differently at larger scales. However, incorporation of an erosion factor to account for PP was not necessary in our study, since 90% of the runoff was generated by spring snowmelt from frozen soils and PP was only a minor component in summer runoff from manured sites (Little et al., 2006).
Although previous studies have found that surface runoff interacts with only a very shallow depth of soil (Sharpley et al., 1978; Sharpley, 1985), the relationships with spring and summer runoff had similar predictive power among all three depth layers. Statistical comparisons of the relationships indicated that the slopes and intercepts of the relationships for all three layers were not significantly different, although slopes and intercepts tended to increase with increasing depth. It was anticipated that STP from shallower sampling depths may have a stronger relationship with runoff phosphorus because the majority of runoff occurred during spring snowmelt when frozen soil restricts infiltration and minimizes the interaction between runoff and soils. However, given that STP results among all three layers were highly correlated in our study (r2 = 0.99, df = 26), it was not surprising they predicted runoff phosphorus equally well. Andraski and Bundy (2003) also reported increased slopes for the relationship between DRP in simulated rainfall runoff and STP in the 0- to 15-cm soil layer compared with the 0- to 2-cm soil layer, but concluded that taking account of increased STP levels in the shallow layers did not improve relationships with DRP compared to those measured in the 0- to 15-cm layer. Vadas et al. (2005b) combined data from rainfall simulator studies representing 30 soil types throughout the United States at 0 to 5 cm, 0 to 15 cm, and 0 to 20 cm and found that STP measured from shallow samples in phosphorus-stratified soils gave a similar assessment of STP available to DRP in runoff as deeper samples in well-tilled soils.
Extraction coefficients for DRP and TP at the microwatershed scale were greater than those reported from laboratory rainfall simulations in Alberta (Wright et al., 2003). The strength of the TP relationship was much greater at the microwatershed scale (Wright et al., 2003). The DRP fraction was also a very small proportion of TP (0.08) for the laboratory rainfall simulations compared to the DRP/TP ratio of 0.55 for the microwatershed results (Wright et al., 2003). The bare and re-packed soil conditions of the laboratory simulations may have contributed to the large proportion of PP. Andraski and Bundy (2003) also reported low DRP/TP ratios in field-plot-scale rainfall simulations. Conversely, Volf et al. (2007) reported average DRP/TP ratios of 0.70 from field-plot-scale rainfall simulations, with greater ratios from manured sites compared to non-manured sites and lower ratios immediately following manure application compared to 1 yr later.
Possible explanations for the higher proportion of DRP measured at the microwatershed scale relative to the laboratory and small plot scales include the longer time that runoff is in contact with soil at the field scale, which may increase concentrations of dissolved phosphorus in runoff water compared to the plot scale (Nash et al., 2002). The PP fraction of TP tends to be favored in small plot-scale and lab-scale studies, due to the comparatively high kinetic energy of overland flow that increases the detachment of soil particles (Nash et al., 2002). Variations in topography at field scales also offer greater opportunities for the deposition of PP than at plot scales. Manure application and incorporation may have increased infiltration and reduced detachment in the field-plot-scale simulations of Volf et al. (2007), resulting in higher DRP/TP ratios than other small plot-scale studies. In addition, snowmelt runoff, which accounted for the majority of runoff in the microwatershed study, was not measured at the plot or lab scale. Snowmelt tends to have higher proportions of DRP than rainfall-generated runoff since frozen soils reduce the detachment of soil particles (Hansen et al., 2000). Higher ratios of DRP to TP in snowmelt were observed at the non-manured sites, but not at the manured sites (Little et al., 2006).
Degree of Phosphorus Saturation and Runoff Phosphorus Relationships
The DPS in the 0- to 2.5-cm layer showed a split line relationship to runoff phosphorus concentrations (Fig. 4). The relationships were described by split line model equations, and explained similar amounts of variation as the STP equations (Fig. 3). Change points could not be determined between DPS and DRP and TP without inclusion of the manured sites. Change points were determined for all sites at a DPS value of 32% for DRP and 22% for TP (Fig. 4). These change point values corresponded to STP values of 37 and 34 mg kg–1, which are approximately half the agronomic threshold of 60 mg kg–1. The agronomic threshold is the level at which crops in most Alberta soils will not respond to the addition of phosphorus (Howard, 2006). There were slightly higher r2 values for DPS and runoff P relationships (Fig. 4) compared with the STP and runoff P relationships (Fig. 3a and 3b). However, the soils used for DPS relationships were a subset of the soils used for STP relationships. Andraski and Bundy (2003) reported that phosphorus saturation explained similar amounts of variability in runoff phosphorus concentrations at one site, but explained less variability than STP at two other sites. Vadas et al. (2005a) and Andraski and Bundy (2003) reported that the relationship between STP and DRP concentrations in runoff was not improved using alternative STP extraction methods compared with the agronomic sampling methods currently in use. Our results suggest that this may also be true for TP concentrations in runoff.
|
|
Conclusions |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
Although a number of different STP sampling strategies were examined, a simple average of all soil sampling points was as good a predictor of runoff phosphorus concentrations as a landform area-weighted mean representation and a subsample of points within the runoff contributing area. A random subset of samples and representative random samples also produced similar results. There were no significant differences in the slopes or intercepts in any of the relationships using different STP sampling strategies.
There were no significant differences among the relationships using different soil sampling depths of 0 to 2.5 cm, 0 to 5 cm, and 0 to 15 cm. Therefore, it is likely that an agronomic soil sampling depth of 0 to 15 cm can be used to predict phosphorus in runoff from agricultural land in Alberta.
Snowmelt runoff accounted for 90% of the runoff volume from the eight sites during the 3-yr study. Although large residual distances were observed for some summer events, the relationship between TP and STP is likely adequate for predicting phosphorus concentrations in most runoff events, given that the vast majority of runoff occurs during spring snowmelt.
Strong relationships were found between DPS and the FWMCs of DRP and TP; however, the relationships were not linear. Predictive abilities were similar to those observed for STP. Change point values corresponded to STP values that were half the agronomic threshold of 60 mg kg–1. Although the DPS holds promise for predicting runoff and leaching losses of phosphorus, modified Kelowna STP is the standard for agronomic sampling in Alberta and our results suggest that there is no strong reason to move toward another soil test.
While several studies have examined the relationship between STP and DRP, few have reported relationships between STP and TP. In comparison with other Alberta studies, extraction coefficients for DRP were greater than lab-derived values. Since our study was based on field-scale results from Alberta, these relationships should provide the basis for phosphorus modeling in Alberta.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. Salvano, D. N. Flaten, A. N. Rousseau, and R. Quilbe Are Current Phosphorus Risk Indicators Useful to Predict the Quality of Surface Waters in Southern Manitoba, Canada? J. Environ. Qual., August 24, 2009; 38(5): 2096 - 2105. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Fiener and K. Auerswald Effects of Hydrodynamically Rough Grassed Waterways on Dissolved Reactive Phosphorus Loads Coming from Agricultural Watersheds J. Environ. Qual., February 6, 2009; 38(2): 548 - 559. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Vadose Zone Journal | ||||
| Soil Science Society of America Journal | Journal of Plant Registrations | The Plant Genome | |||
