| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Center for Agroforestry, School of Natural Resources, University of Missouri, Columbia, MO 65211
b Department of Soil, Environmental and Atmospheric Sciences, School of Natural Resources, University of Missouri, Columbia, MO 65211
* Corresponding author (udawattar{at}missouri.edu).
Received for publication October 20, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: TP, total phosphorus
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The main mechanism by which P is lost from row-cropped agricultural land is by runoff carrying both soluble and particulate forms of P (Quinton et al., 2001; Sharpley et al., 1994). For example, Catt et al. (1998) showed that losses of P from experimental plots in the UK occur mainly in particulate forms and are consequently greater in surface runoff than drain flow. Particulate P is usually the largest fraction of P in runoff from row-crop production systems due to greater losses of sediment in this type of production system (Sharpley et al., 1992).
Finer soil particles can carry a higher concentration of sorbed nutrients than the bulk soil (Sharpley and Smith, 1991). Comparing P loss and suspended sediment loss, Wall et al. (1996) showed that the ratio of TP to sediment loss increases with decreasing sediment loss because low erosion or sediment transporting runoff events carry smaller clay-size sediment particles, which have more readily sorbed P. This, in turn, results in a larger mass of P per unit of total solids from small runoff events. However, the relationship between sediment loss and P loss appears to vary greatly from one area to another because geochemical properties of runoff sediment vary according to the source (Williams et al., 1980; Grobler and Silberbauer, 1985). In southern Idaho, Westermann et al. (2001) showed that TP in runoff was not statistically related to soil test P but was linearly related with sediment concentration. In the Coastal Plain of Maryland, P and sediment concentrations in runoff were correlated, but the correlation relationships differed among the 17 watersheds that were studied (Jordan et al., 1997).
Claypan soils in central and northeast Missouri have an argillic horizon with a content of 40 to 50% smectitic clays between 10 and 80 cm below the surface. These soils are characterized by very low infiltration rates (Jamison and Peters, 1967). Due to this low permeability, claypan soils perch water and create lateral flow above the claypan (Blanco-Canqui et al., 2002). In northern Missouri, the relative proportions of surface and subsurface flow in a claypan soil on row-cropped watersheds varied between management, with greater subsurface flow on a claypan soil watershed with a riparian buffer compared with a watershed with grass filter strips (Schmitt, 1999). This observation is further supported by a study in forested watersheds where increased tree cover and subsequent rooting enhanced water infiltration (DeWalle et al., 1988). Claypan soils in the study area are classified in hydrological group D and are capable of producing runoff volumes of over 75% of the precipitation, depending on antecedent soil moisture (Schwab et al., 1993; Watson, 1979). Since claypan soils generate high flow rates during runoff events, significant losses of soil P may leave agricultural watersheds in runoff. Spatial variation in depth to claypan across agricultural fields also has been observed to influence crop yields (Kitchen et al., 1999) and water and chemical movement (Blevins et al., 1996; Blanco-Canqui et al., 2002). However, little research is available examining soil P loss in claypan soils on a watershed scale.
Factors affecting the loss of P in runoff include runoff volume, sediment loss, forms and concentration of soil P, and depth of mixing of the soil and water (Cassell et al., 1998; Sharpley et al., 1994). Other parameters, such as precipitation and soil surface characteristics that vary temporally and spatially, are also important in determining P loss on a watershed scale (Gburek et al., 2002). Phosphorus exports from watersheds also vary widely over individual storm events, annual cycles, and long periods of time (Cassell et al., 1998; Quinton et al., 2001). In a 28-yr study, Edwards and Owens (1991) showed that the largest three erosional events caused more than 50% of the soil loss measured in Coshocton, Ohio. This observation has also been supported by research in the UK (Morgan et al., 1986) and Nigeria (Lal, 1976). Small rainfall events that occur more frequently and require less energy to detach soil particles can also cause significant soil losses. Quinton et al. (2001) showed that smaller events accounted for a greater proportion of P loss over a 6-yr study than infrequent larger events.
The concentration of P in runoff is largely determined by timing of precipitation and vegetative cover because precipitation provides the major source of energy for transport. In a simulated rain study, Edwards et al. (2000) showed that the magnitude of P loss was related to the proximity of preceding rainfall. Therefore, antecedent soil moisture affects P transport (McDowell and Sharpley, 2002).
Row-crop agriculture under conventional tillage as practiced in much of the U.S. Midwest increases P loss due to the presence of a bare soil surface during spring and late fall. The soil in this condition is susceptible to raindrop impact and often contains added P from fertilizer or animal waste. In northeastern Missouri, approximately 35, 28, 23, and 14% of the precipitation occurs in spring, summer, fall, and winter months, respectively. Approximately 34 to 36% of the precipitation occurs between October and March when the ground is rarely covered with crops (Owenby and Ezell, 1992).
We hypothesized that the presence of claypan soils, characterized by a high-clay subsoil that restricts drainage, would increase P losses in runoff. The objectives of this research were to (i) examine the effects of landscape and watershed characteristics on soil P loss over time by comparing three adjacent watersheds, (ii) determine the impact of small and large runoff events and runoff characteristics (duration and peak height) on TP loss, and (iii) evaluate the effects of the timing of rainfall on P loss in relation to the type and duration of vegetative cover.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Thirty-year mean annual precipitation in the region is 920 mm per year, of which more than 66% falls from April through September (Owenby and Ezell, 1992). Mean annual air temperature is approximately 11.7°C with an average monthly low of –6.6°C in February and an average monthly high of 31.4°C in July (Owenby and Ezell, 1992). Snowfall averages about 590 mm per year, and snow can stay on the ground for extended periods.
Agricultural activities in the watersheds including the crops grown, field preparation, fertilizer regime, and grain yields from 1991 to 1997 are summarized in Table 1. Before 1991, the field containing these watersheds was in a corn (Zea mays L.)–soybean [Glycine max (L.) Merr.] rotation with planting occurring in a north–south direction perpendicular to the contour, except for some areas near the waterways. Beginning in 1991, planting was in a straight row perpendicular to the slope and then beginning in 1996 was on the contour. During the study period, the land preparation method was no-till with the exception of 1992 when watersheds were field-cultivated with a John Deere (Moline, IL) E1000 field cultivator.
|
Stevens Type F water level recorders (Stevens Water Resources, Beaverton, OR) were used to measure runoff from 1991 to 1995. During that period, water level recorders remained in the field throughout the year. Isco (Lincoln, NE) bubbler flow measuring devices and Isco 3700 samplers replaced float recorders in August 1995. Isco instruments were installed each year in late February or early March to record flow rate and sampling times and collect runoff samples. These samplers were removed from the field in late December to protect them from possible damage due to freezing. Therefore, the sample collection period during 1995 and 1996 extended from March to December. In 1997, the sample collection period was from March to the end of June. Flow measuring devices engage the sampler to withdraw a 135-mL sample of runoff after each 25 m3 of flow occurs. Therefore, runoff samples were flow-weighted and collected for individual storms. During the 7-yr study, 67, 67, and 64 runoff events generated runoff samples on the east, center, and west watersheds, respectively. For some consecutive events (maximum 2), samples were not separated by event. An average of 43 runoff events across the three watersheds over the 7-yr period did not generate sufficient runoff to activate the sampler. The runoff collection device of the west watershed was damaged by lightening and, therefore, it missed three sampling events. Flow rate, water level, runoff duration, and sample intake time data were downloaded from the Isco flow meters to a laptop computer following runoff events.
Isco flow meter data were used to estimate runoff duration and maximum flow rates (Isco, 1998). Isco flow-level data show the beginning and end of a flow event with the corresponding time. The duration of a flow event was determined using these corresponding times. The runoff level data were used to record the maximum flow rate of a runoff event. Flow duration and respective maximum flow levels were difficult to determine from runoff charts that were used from 1991 to 1995 since charts were replaced once a week and the recording of more than one event often occurred during the week. Therefore, data presented for runoff duration and maximum flow are from 1995 to 1997.
Chemical and physical analyses of composite runoff water samples were performed in the Forest Hydrology Laboratory at the University of Missouri. A known volume (20–250 mL depending on the sediment concentration) of a well-mixed runoff sample was filtered through a preweighed Whatman (Maidstone, UK) 934-AH glass microfiber using a vacuum pump to estimate sediment concentration (American Public Health Association, 1992). These filters were dried at 105°C to a constant weight and their dry weights were recorded. The difference between two dry weights and the filtered sample volume was used to estimate the concentration of total suspended sediment.
Unprocessed samples were refrigerated at 4°C until analysis. From 1991 to 1994, acid digested unfiltered runoff samples were analyzed by the ascorbic acid–molybdate colorimetric procedure using a Technicon (Tarrytown, NY) Autoanalyzer (Technicon, 1978). After 1994, TP was determined using an ammonium peroxidisulfate digestion followed by an ascorbic acid–molybdate procedure on a Lachat (Milwaukee, WI) Quickchem Automated Ion Analyzer (Liao and Marten, 2000). The detection limit for the method was 0.9 µg L–1. Several samples from 1991 to 1994 were reanalyzed using the ammonium peroxidisulfate digestion and the Lachat Autoanalyzer to compare the two procedures. Results showed no significant differences in TP concentration between the two digestion and analytical procedures.
Statistical Analysis
Statistical analyses of the data were performed using Statistical Analysis Systems software (SAS Institute, 1999). Least square linear regression analysis (PROC REG) was used to describe relationships between runoff and TP loss, runoff duration and TP loss, and peak flow and TP loss. The nonlinear relationships between TP and sediment loss were determined with SAS (PROC NLIN). Differences between regression coefficients or slopes of any two regressions (p < 0.05) were determined by testing the homogeneity of regression coefficients.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
|
During the 7-yr study, the east, center, and west watersheds lost a total of 8.35, 8.93, and 11.22 kg TP ha–1, respectively. The annual TP discharges were larger for the west watershed than for the other two watersheds except for 1994 and 1997 when the east watershed had larger losses (Fig. 3). In 1994, average Bray-1 soil test P was lower on the west watershed (47 kg ha–1) compared with the other two watersheds. The east watershed (54 kg ha–1) had the highest average Bray-1 soil test P while the concentration was intermediate on the center watershed (52 kg ha–1). Despite having lower soil test P, the west watershed lost more TP than the other two watersheds over the whole 7-yr study and during five individual years. In an irrigation study, Aase et al. (2001) also observed that total P in runoff and soil P were not related.
Despite the close proximity of the three watersheds in this study, several inherent physical differences existed among the three watersheds that may have affected observed P losses possibly due to their effects on surface and subsurface flow of water across the watersheds. The area of the center and west watersheds was larger than the east watershed (Table 2). In addition, the east, center, and west watersheds had total slope lengths measuring 234, 425, and 383 m and with corresponding 2.1, 1.3, and 0.9% slopes along the entire watershed (Fig. 1, Table 2). The lowest 100 m of east, center, and west watersheds had slope segments with 3, 2, and 1.75% slopes, respectively.
Depth to claypan (Bt horizon) varied by landscape position (Table 2) and this characteristic may also influence surface and subsurface flow because soils with a shallower depth to the claypan would tend to have higher rates of lateral flow (Blanco-Canqui et al., 2002). The depth to the claypan in the upper one-third of each watershed was deeper than the lower one-third. Soils in the lower one-third of each watershed were sampled from erosional side slopes and dissected areas compared with more leveled areas in the upper one-third of each watershed. The depth to the claypan was much greater at all landscape positions in the west watershed than the other two watersheds. Waterways had the deepest soils above the claypan in each watershed. The average depth to the claypan in the waterways was 49, 62, and 57 cm for the east, center, and west watersheds, respectively. Despite greater depth to the claypan in the cropped area and grass waterways on the west watershed, more P was lost from the west watershed.
The grass waterways located at the base of each watershed may have reduced TP loss on the three watersheds. Approximately 44, 35, and 27% of the total watershed length on the east, center, and west watersheds, respectively, was occupied by grass waterways (Table 2). The lower proportion of the watershed length in grass waterway may have been one reason that the west watershed lost 1.34 and 1.25 times more TP than the east and center watersheds, respectively, during the 7-yr period. In findings reported by Dillaha et al. (1989), orchard grass filter strips of 9.1 and 4.6 m in width removed 79 and 61% of the P in runoff, respectively. Strips of varying widths have been shown to remove from 22 to 89% of TP in runoff (Sharpley et al., 1992; Patty et al., 1997; Schmitt et al., 1999).
Another reason that may be causing observed differences in TP loss among the watersheds is that ground water and surface runoff during a large precipitation event can cross topographically defined catchment boundaries, so the collection area of the runoff becomes uncertain (Garrison et al., 1987). The west and center watersheds, which have road boundaries on the west and south sides, respectively, in addition to topographic boundaries, could have experienced this additional source of surface runoff during large rainfall events.
Crop Rotation, Fallow Period, and Total Phosphorus Loss
During the 7-yr study, the watersheds were planted to corn for 3 yr and to soybean for 4 yr (Table 1, Fig. 3). Average annual watershed losses were 1.70 and 1.10 kg TP ha–1 during corn and soybean years, respectively. Higher average annual loss of TP during corn years compared with soybean years could be because P fertilizer is more often applied to corn and not to soybeans. However, in this study, P fertilizer was applied to corn only in 1993. During the corn year of 1993, watersheds also received 42% more precipitation than the long-term mean causing higher losses of TP in runoff (Fig. 3). The average annual corn and soybean TP losses over the duration of the experiment (excluding 1993) were 1.12 and 1.10 kg ha–1 yr–1, respectively.
The annual TP losses during crop and fallow periods were different among years (Fig. 3). Approximately 48 to 100% of the annual total loss occurred when the ground was free of crops. During the 7-yr study, 34.4 to 66% of the annual precipitation occurred during the fallow period with a 55% mean for the 7-yr study period (Fig. 2). The study was initiated in April 1991 and the reported entire loss for 1991 occurred after the crop was harvested. In contrast, most of the TP loss occurred within the cropping period during the high rainfall year of 1993 (Fig. 2 and 3). However, in all other years, a higher proportion of the TP losses occurred during the fallow periods (Fig. 3). For example, in 1995, when rain continued for several weeks during the corn planting period, the preplanting period loss alone accounted for 89% of the total annual TP loss.
Another possible reason for the higher TP loss during the fallow periods is that the high clay content of the subsoil in these watersheds restricts vertical soil water percolation (Jamison and Peters, 1967). Due to this low permeability, argillic horizons act as a barrier directing the vertical flow horizontally above the clay pan (Blanco-Canqui et al., 2002). At the Midwest Research Claypan Farm near Kingdom City, MO, Blanco-Canqui et al. (2002) observed that approximately 98.5% of applied water in the upper end of the monolith moved laterally through the soil layer above the restrictive argillic horizon after 48 h. These clay soils produce a large volume of surface runoff during periods of saturation in the spring and early summer, especially when the soil is free of any vegetation. A comparison of TP losses in 1993 and 1995 supports the finding that restrictive layer and the amount of precipitation together were the main contributing factors for the TP loss on these watersheds, especially during the fallow period. In a year such as 1995, when rain occurred more frequently, antecedent soil moisture conditions may have been higher than in a normal year. A laboratory study showed that antecedent soil moisture condition affects P loss (McDowell and Sharpley, 2002). Higher antecedent soil moisture conditions also may have contributed to greater TP losses when rain occurred more frequently.
Results of this study suggest that crop cover may have substantially reduced the TP loss from the three row-crop watersheds in 1993, in spite of 42% more rainfall than the long-term mean. At the Claypan Experimental Farm in Missouri, Ghidey and Alberts (1998) showed that fallow-period soil loss was five times higher than during the cropping period under continuous corn or soybean. The observed difference in TP loss during the crop and fallow periods may result from less interaction of runoff with surface soil due to the effects of greater vegetative cover during the crop period (Sharpley et al., 1996).
Runoff Characteristics and Total Phosphorus Loss
Runoff volume and TP loss were highly significant for the three watersheds (r2 = 0.64–0.73, p 
0.001; Fig. 4). Runoff in the east watershed had the best relationship with TP loss (r2 = 0.73; Fig. 4A) while the west watershed had the weakest relationship (r2 = 0.64; Fig. 4C). However, slopes between TP loss and runoff relationships among the watersheds were not significantly different (p 0.05). The slopes of the regression lines for center and west watersheds were the same (0.0006) while the east watershed had a slightly smaller slope (0.0004) (Fig. 4). Figure 4 also shows that the relationship between runoff and TP loss was largely determined by a larger number of small runoff events and only a small number of larger events that occurred during the 7-yr measurement period.
|
The largest five runoff events had an average concentration of 0.93 mg TP L–1 as compared with the mean (0.87 mg L–1) for the entire study period (Fig. 5). The mean of the largest five events and the grand mean without the largest five events (0.86 mg L–1) were not significantly different (p 0.05). The more frequently occurring small runoff events account for a larger portion (73% of the total) of the TP loss compared with the infrequent large events. Although infrequent larger events remove proportionately larger quantities of P from row-crop agriculture, more frequent small events cumulatively remove larger quantities over an extended period of time. Our study shows that the mean concentration of TP in larger runoff events was not significantly different, but due to the larger runoff volume, proportionately larger amounts of TP were removed from the three watersheds. For the purpose of water quality protection, the USEPA has recommended a maximum level of 0.1 mg TP L–1 in surface runoff water (Daniel et al., 1998). Total phosphorus concentrations in runoff from this study's watersheds were consistently greater than the critical value established by the USEPA.
|
0.001) of the variation in TP loss. Larger flow events appeared to be responsible for removing significant amounts of TP from row-crop agriculture in the study watersheds. Maximum flow height explained 57 to 75% of the variation in TP loss (Table 3). Both the east and west watersheds had similar r2 (0.75 and 0.74, p 0.001) while the r2 for the center watershed was only 0.57 (p 0.01). In the combined data set, the maximum flow height explained 66% of the variation in TP loss from the three watersheds.
|
Sediment and Total Phosphorus Loss
Total P and sediment loss relationships were significant and explained 47 to 58% of the variation in TP loss by sediment (Fig. 6). The west watershed had the weakest relationship (r2 = 0.47, p 0.001) among the three watersheds (Fig. 6C). The sediment loss on the east (r2 = 0.56, p 0.001) and center (r2 = 0.58, p 0.001) watersheds explained more than 56% of the variation in TP loss (Fig. 6A and 6B). The three watersheds had a similar pattern for TP and sediment loss. All three watersheds had a larger number of events with small amounts of TP and sediment loss and a smaller number of events with large amounts of TP and sediment losses. Therefore, we combined the data from the three watersheds to observe the relationship between TP and sediment loss. In the combined data set with 193 observation pairs, sediment loss explained 55% (p 0.001) of the variation in TP loss. It is clear that one way to reduce TP loss in runoff is to reduce the sediment loss in runoff. The control of sediment loss is more effective to reduce P in runoff as sediment loss is more related to TP loss than soil test P concentration (Daverede et al., 2003).
|
Wall et al. (1996) showed that smaller events could carry proportionately more P than larger events as small events remove more clay-size particles with higher amounts of attached P than contained by larger soil particles. Studying soil and P loss on soil trays, McDowell and Sharpley (2003) noticed that more P was lost in soil during the first 30 min of flow than the second 30 min due to selective erosion of finer particles. Our results showed that smaller events removed P at an increasing rate followed by a decreasing rate. The results show that up to a certain sediment loss, TP loss increased at an increasing rate, possibly due to the removal of P-enriched surface soils. As sediment loss continues to increase subsequent runoff events removed soils with less P. Therefore, larger events did not remove P at the same rate. Conversely, when runoff events remove larger suspended loads during high flow, the suspended soils are characterized by a wide range of particles. The phenomenon is supported by the TP concentration data as the concentrations between the largest events and the whole study were not significantly different.
The suspended sediment and TP loss ranged from 0.023 to 1171 kg ha–1 and 0 to 0.832 kg ha–1, respectively, during the 7-yr study across the three watersheds. The mean TP concentration for losses in sediment <10 kg ha–1 was 0.06 kg TP ha–1, which was significantly (p 0.01) smaller than the mean TP concentration (0.32 kg TP ha–1) of sediment >100 kg ha–1 (Table 4). During the 7-yr study, the runoff events with the largest five sediment losses accounted for 24% of the 7-yr TP loss. Figures 4 and 6 show that small events are more prevalent than larger events, so the greater frequency of smaller events leads to higher cumulative P losses than for larger events.
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Runoff volume, maximum flow rate, runoff duration, and the presence of vegetative ground cover were the main factors that affected P loss in the three watersheds from individual runoff events. Above-average precipitation, especially when it occurred within a 2- or 3-mo period, caused the greatest observed losses of total P in runoff. The highest TP losses most often occurred during fallow periods before crops were planted and after they had been harvested. Decreasing runoff volume and flow rate and extending the period of vegetative cover through changes in agricultural conservation practices may be the most effective long-term strategies to reduce P losses from agricultural watersheds in the claypan region.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Related articles in JEQ:
This article has been cited by other articles:
|
|
 |
|
  A. N. Sharpley, P. J. A. Kleinman, A. L. Heathwaite, W. J. Gburek, J. L. Weld, and G. J. Folmar Integrating Contributing Areas and Indexing Phosphorus Loss from Agricultural Watersheds J. Environ. Qual., June 23, 2008; 37(4): 1488 - 1496. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. N. Sharpley, P. J. A. Kleinman, A. L. Heathwaite, W. J. Gburek, G. J. Folmar, and J. P. Schmidt Phosphorus Loss from an Agricultural Watershed as a Function of Storm Size J. Environ. Qual., March 1, 2008; 37(2): 362 - 368. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. R. Sistani, G. E. Brink, and J. L. Oldham Managing broiler litter application rate and grazing to decrease watershed runoff losses. J. Environ. Qual., March 1, 2008; 37(2): 718 - 724. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. K. Udeigwe, J. J. Wang, and H. Zhang Predicting Runoff of Suspended Solids and Particulate Phosphorus for Selected Louisiana Soils Using Simple Soil Tests J. Environ. Qual., July 17, 2007; 36(5): 1310 - 1317. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. S. Algoazany, P. K. Kalita, G. F. Czapar, and J. K. Mitchell Phosphorus Transport through Subsurface Drainage and Surface Runoff from a Flat Watershed in East Central Illinois, USA J. Environ. Qual., April 5, 2007; 36(3): 681 - 693. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
