JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (18)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Baker, D. B.
Right arrow Articles by Richards, R. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Baker, D. B.
Right arrow Articles by Richards, R. P.
Agricola
Right arrow Articles by Baker, D. B.
Right arrow Articles by Richards, R. P.
Related Collections
Right arrow Surface Water Quality
Right arrow Watershed and Landscape Processes
Right arrow Best Management Practices
Right arrow Ecosystem Management
Right arrow Nutrient Management
Journal of Environmental Quality 31:96-108 (2002)
© 2002 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America

SPECIAL SUBMISSIONS
Findings from the USDA-sponsored Lake Erie Agricultural Systems for Environmental Quality Project

Phosphorus Budgets and Riverine Phosphorus Export in Northwestern Ohio Watersheds

David B. Baker* and R. Peter Richards

Water Quality Laboratory, Heidelberg College, Tiffin, OH 44883

* Corresponding author (dbaker{at}mail.heidelberg.edu)

Received for publication August 12, 2000.

   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 RATIONALE FOR USE OF...
 METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Phosphorus (P) budgets for large watersheds are often used to predict trends in riverine P export. To test such predictions, we calculated annual P budgets for 1975–1995 for soils of the Maumee and Sandusky watersheds of northwestern Ohio and compared them with riverine P export from these watersheds. Phosphorus inputs to the soils include fertilizers, manure, rainfall, and sludge while outputs include crop removal and nonpoint-source export via rivers. Annual P inputs decreased due to reductions in fertilizer and manure inputs. Annual outputs increased due to increasing crop yields. Net P accumulation decreased from peak values of 13.4 and 9.5 kg P ha-1 yr-1 to 3.7 and 2.6 kg P ha-1 yr-1 for the Maumee and Sandusky watersheds, respectively. Thus, P budget analysis suggests that riverine P export should have increased throughout the study period, with smaller increases during more recent years. However, detailed water quality studies show that riverine export of total phosphorus (TP) has decreased by 25 to 40% and soluble reactive phosphorus (SRP) by 60 to 89%, both due primarily to decreases from nonpoint sources. We suggest that these decreases are associated with farmers' adoption of practices that minimize transport of recently applied P fertilizer and of sediments via surface runoff, coupled with changes in winter weather conditions. In comparison with most Midwestern watersheds, rivers draining these watersheds have high unit area yields of TP, low unit area yields of SRP, and high ratios of nonpoint source– to point source–derived P.

Abbreviations: GLWQA, Great Lakes Water Quality Agreement • IJC, International Joint Commission • SRP, soluble reactive phosphorus in water • TP, total phosphorus in water • TSS, total suspended solids


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
RATIONALE FOR USE OF...
 METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
IN THE 1950S AND 1960S, water quality in Lake Erie was deteriorating rapidly, due in large measure to excessive P loading and associated eutrophication (Federal Water Pollution Control Administration, 1968; International Joint Commission, 1970). Programs to reduce P loading to Lake Erie were initiated by the International Joint Commission (IJC) under the Great Lakes Water Quality Agreement (GLWQA), which was signed by the United States and Canada in 1972 (International Joint Commission, 1974). This agreement called for initiation of a 1.0 mg P L-1 effluent limit for P at all municipal sewage treatment plants located within the watersheds of the lower Great Lakes and having discharges greater than 3770 m3 d-1 (one million gallons per day).

The 1972 GLWQA also called for studies to establish P target loads for each of the Great Lakes and to indicate the size of the load reductions needed to reach those target loads. Modeling studies of the relationships between P loading and eutrophication in the Great Lakes led to the establishment of a target load of 11000 Mg P yr-1 for Lake Erie (Thomas et al., 1980; Bierman, 1980). These models suggested that reductions in the external P loading to 11000 Mg would reduce algal growth sufficiently to prevent the development of anoxia in the Central Basin of Lake Erie. The base year P loading to Lake Erie was set at 20000 Mg, using 1976 point-source loads and, because of large annual variations in P loading from tributaries, the tributary P loading associated with average tributary flows (Thomas et al., 1980).

Although by 1982 point-source loading to Lake Erie was reduced from more than 15000 Mg P yr-1 in 1972 to less than 3000 Mg P yr-1 (Dolan, 1993), detailed tributary loading studies indicated that even when all sewage treatment plants with flows greater than 3770 m3 yr-1 reached effluent concentrations of 1.0 mg P L-1, an additional load reduction of 2000 Mg P yr-1 would be needed to reach the target load. In 1983, a Phosphorus Load Reduction Annex to the 1978 GLWQA was signed, committing the United States and Canada to efforts to achieve the additional 2000 Mg reduction, largely through reductions in agricultural nonpoint-source loading (International Joint Commission, 1988).

To achieve the final 2000 Mg yr-1 reduction, load reductions were allocated to various political jurisdictions, based on their relative P contributions to Lake Erie (International Joint Commission, 1988). A reduction of 1700 Mg P yr-1 was assigned to the United States, of which 1390 Mg P yr-1 was subsequently assigned to Ohio (USEPA, 1986). In Ohio, the P reductions were further allocated to individual counties, based on their cropland acreage in the Lake Erie basin. Since 75 to 80% of the TP loading was particulate P associated with total suspended sediment (TSS) transport, plans for reducing nonpoint sources of P focused exclusively on reductions associated with various erosion control programs. Phosphorus reduction credits were awarded to counties based on increasing acreage of various conservation measures relative to 1982 levels. Phosphorus reduction credits of 1.44 kg P ha-1 were allocated for increases in set-aside, Conservation Reserve Program, and hay acreage; 1.14 kg P ha-1 for increases in wheat (Triticum aestivum L.) and oat (Avena sativa L.); and 0.82 kg P ha-1 for conservation tillage. These reduction credits were based on the proportional reduction in erosion associated with these practices, and the P content of the suspended sediments from northwestern Ohio rivers.

In the Lake Erie phosphorus reduction programs, no reduction credits were assigned to fertilizer management. Since the P content of the soil was viewed as very large relative to annual applications of P fertilizers, it was believed that any responses to reductions in fertilizer use would be slow to develop and therefore would not address the immediate problems in Lake Erie. However, substantial reductions in P fertilizer application occurred in the early 1980s (Richards et al., 2002a). During the 1970s, many farmers had been applying P fertilizers at rates designed to increase P soil test levels. This practice was reduced substantially as farmers moved toward applications designed to replace P removal by crops.

Watershed nutrient budgets have been used for many years to assess potential nutrient buildup in watersheds (Lowrance et al., 1985; National Research Council, 1993). Their use has increased substantially in recent years, particularly in considering the effects of large livestock operations on P buildup in soils. Use of watershed-level nutrient budgets and associated models provides a way to predict trends in P soil test levels and associated changes in nutrient export to aquatic ecosystems (Aschmann et al., 1999; Sharpley et al., 2000; Beegle et al., 2000; David and Gentry, 2000). One objective of the Lake Erie Agricultural Systems for Environmental Quality program has been to evaluate changes in P budgets of large agricultural watersheds in the Lake Erie basin during the 1975–1995 period and to consider the possible effects of these changes on P loading to Lake Erie. This paper addresses that objective.

The study watersheds (Fig. 1) include those portions of the Maumee River basin and the Sandusky River basin upstream from the U.S. Geological Survey (USGS) stream gaging stations where long-term water quality monitoring studies are underway (Baker, 1985, 1988, 1993). For the Maumee River, the gaging station is located at Waterville, OH (USGS Station 04193500) and for the Sandusky River it is near Fremont, OH (USGS Station 04198000). At these points in their drainage networks, the Maumee River drains 16395 km2 and the Sandusky River drains 3240 km2. Row crop agriculture comprises more than 75% of the land use in both watersheds, with soybean [Glycine max (L.) Merr], corn (Zea mays L.) and wheat covering 71 to 72% of the land area in both watersheds in 1995 (Ohio Department of Agriculture, 1996). Population densities in the Maumee and Sandusky watersheds are about 46 and 43 persons per square kilometer, respectively, with urban land occupying 1.2 and 0.9% of the land area. Richards et al. (2002b) provide additional information regarding the study area.



View larger version (63K):
[in this window]
[in a new window]
  		Fig. 1. Locations of the study watersheds in the Lake Erie basin.

 

   RATIONALE FOR USE OF SOILS BUDGETS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 RATIONALE FOR USE OF...
METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Since numerous studies indicate the importance of P soil test levels on P export (Carpenter et al., 1998; Sharpley et al., 2000), this study has focused on developing budgets for the soils compartment of watersheds. Here, we define the soils compartment as all the soils within the watershed boundaries wherein pools of P reside. The P budget consists of a mass balance of P for the soils compartment, with the major inputs and outputs shown in Fig. 2 . P inputs to the soil include fertilizers, manure, rainfall, and sludge applications from municipal sewage treatment plants. Point-source P discharges from sewage treatment plants and industries bypass the soil and contribute directly to the total river export from the watershed. Outputs from the soil include removal by crops and loss as nonpoint-source loading to the stream system. The nonpoint-source P load is derived primarily from surface runoff. However, because of the extensive use of tile drainage in these watersheds, some P from leachate is also included in the river export. We have not included P produced through soil weathering, although weathering is included as a P input to soils in the more comprehensive watershed ecosystem P budgets developed by Cassell et al. (1998) and Aschmann et al. (1999). For ecosystem budgets, they include animal and human food imports and animal product exports, and they distinguish between grain export versus grain used as animal feed within the watershed. For a soils budget, as described above, these additional considerations are unnecessary.



View larger version (41K):
[in this window]
[in a new window]
  		Fig. 2. Phosphorus inputs and outputs from the soil compartment of agricultural watersheds.

 

   METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 RATIONALE FOR USE OF...
 METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Data for agricultural inputs and crop removal outputs were derived from county-level agricultural statistics as reported by the state departments of agriculture in Ohio, Indiana, and Michigan (Ohio Department of Agriculture, 1975–1996; Indiana Department of Agriculture, 1976–1996; Michigan Department of Agriculture, 1976–1996). The Maumee watershed drains all or portions of 22 counties, while the Sandusky watershed drains portions of 9 counties. The percentages of each county in each watershed are shown in Table 1. To convert the county data into watershed data, the county data were multiplied by the fraction of the county lying within the watershed, and then the adjusted county data were summed for all of the counties in the watershed. No attempt was made to adjust the county allocations based on land use for the portions of the counties falling within the boundaries of the study watersheds. To complete the watershed budgets, the agricultural data were coupled with information on sewage P inputs as sludge, atmospheric deposition, and watershed nonpoint-source loading. All data represent annual P inputs or outputs reported as metric tons (Mg) and as kilograms P per hectare (kg P ha-1). The unit area inputs and outputs were calculated by dividing the total watershed values by the total watershed area, not the area of cropland.


View this table:
[in this window]
[in a new window]
  		Table 1. Percentage of county areas falling within watershed boundaries.

 
Watershed Inputs
Annual P inputs as fertilizer from 1979 to 1995 are based on county fertilizer sales data (Ohio Department of Agriculture, 1975–1996). For the period from 1975–1978, fertilizer sales data were obtained directly from semiannual fertilizer sales reports published by the Ohio Department of Agriculture. Data for Indiana counties were obtained from semiannual Indiana fertilizer tonnage reports (Office of the Indiana State Chemist, 1975–1995). Fertilizer sales data from the single Michigan county were not obtained, so data from the Ohio and Indiana counties were extrapolated to the entire watershed. Fertilizer sales data includes sales of all packages greater than 4.54 kg (10 pounds). Thus it includes all cropland and most noncropland applications, such as commercial lawn applications and golf courses.

Annual P inputs as manure were based on the numbers of beef cattle, dairy cows, swine, and sheep in each county, as reported by the state departments of agriculture. The numbers of animals for each county were adjusted to the proportions of counties present in each watershed, as described above, and summed for each watershed each year. The manure P inputs for each watershed were then calculated using conversion factors in the Ohio Livestock Manure and Wastewater Management Guide (Ohio State University, 1992). These conversion factors, in kg P yr-1 animal-1, are 17.5, 12.2, 5.14, and 1.05 for beef cattle, dairy cows, hogs, and sheep, respectively. These are very similar to the conversion factors used by Goolsby et al. (1999) in their assessment of nutrient sources in the Mississippi–Atchafalaya River basin.

Phosphorus inputs through precipitation were based on atmospheric deposition studies for Lake Erie as reported by the IJC. These deposition rates are calculated from precipitation chemistry collected at stations around Lake Erie (Rathke and McRae, 1989). The average deposition rate for the period between 1982 and 1995 was 361 Mg P yr-1. This average rate was extrapolated to the Maumee and Sandusky watersheds, based on the ratio of their areas to that of Lake Erie, giving deposition rates of 231 Mg P yr-1 for the Maumee watershed and 46 Mg P yr-1 for the Sandusky watershed. Since atmospheric deposition represents less than 1% of the total P inputs into these watersheds, annual variations were ignored and the above rates were applied for the entire study period. Atmospheric deposition of P is sufficiently small relative to other inputs that it was ignored in watershed input calculations by David and Gentry (2000) and Goolsby et al. (1999). Bennett et al. (1999) did include atmospheric deposition in their P budget for the Lake Mendota watershed in Wisconsin.

Phosphorus entering municipal sewage treatment plants is either incorporated into sludge, most of which is spread onto agricultural fields, or discharged into streams as point-source inputs. Since sewage treatment plants in the Lake Erie basin remove P, sewage sludge contains much more P than is discharged directly into rivers or lakes. Treatment plants report P concentrations in their effluent so that annual P discharges into watershed streams can be tracked by the IJC. Annual data for six treatment plants in the Sandusky watershed and 60 treatment plants in the Maumee watershed for the period from 1975 to 1994 were provided by the IJC. The IJC also provided information on industrial point-source P discharges from 1988 to 1994 for 57 plants in the Maumee watershed and 20 plants in the Sandusky watershed. The 1988 and 1989 data were averaged to estimate industrial inputs for the 1975 to 1987 period.

Average municipal point-source P loading for the 1975–1977 period was used to estimate the annual P delivered as sludge to watershed soils for the entire study period. For the 1975–1977 period, point-source discharges equaled approximately 20% of the total P reaching sewage treatment plants and 80% was contained in the sludge. We assumed that P moving to the sewage treatment plants has increased at a rate of 2% per year, due to increases in population and to the extension of sewer lines into unsewered areas. The 1985 adoption of a detergent P ban in Ohio resulted in a 35% reduction in P moving into the plants in that year. Based on these assumptions, sludge P inputs changed from 964 Mg in 1975 to 895 Mg in 1995 for the Maumee watershed and from 123 to 114 Mg for the Sandusky watershed.

For 1995 the combined sludge P and municipal point-source P discharges for the Sandusky watershed totaled 126 Mg. This value compares closely with an estimated human P loading of 137 Mg, based on the sewered population of the watershed (50000) and a human loading rate of 2.74 kg P yr-1 person-1, as reported by Goolsby et al. (1999).

Although approximately 50% of the people in these watersheds are served by septic tanks, the P contributions from the unsewered populations were not included in the budget. The proportion of the septic tanks that have failed in the sense that P-containing liquid effluent is moving off-site is uncertain. Where there is off-site movement of such effluent, an unknown but probably small proportion of that effluent would probably reach either cropland soils or the stream systems. While some P-containing effluent does reach stream systems and results in elevated P concentrations during low flow conditions, the resulting concentrations are in the same range as those from nonpoint sources during high flows. Consequently, while septic tank effluent can result in ambient water quality problems during low flows, such effluents are probably insignificant relative to total riverine P export from these watersheds. In these watersheds, most of the sludge pumped from septic tanks is disposed of as input to municipal sewage treatment plants, where it forms a small proportion of the P entering municipal sewage treatment plants.

Watershed Outputs
Annual removal of P by crop harvests was based on county production figures for corn, soybean, wheat, and hay obtained from state department of agriculture reports. Phosphorus removal for each crop was calculated using the crop removal rates recommended in the Ohio Livestock Manure and Wastewater Management Guide (Ohio State University, 1992). These rates in kg P m-3 are 2.09, 4.54, and 3.62 for corn, soybean, and wheat, respectively, and 2.38 kg Mg-1 for hay. These crop removal rates are similar to those used by Goolsby et al. (1999).

Annual river exports of P from the soil compartments of the watersheds are based on the nonpoint-source components of the total annual river export. The nonpoint-source export is calculated by subtracting point-source P loading into the river from TP export from the river. In this procedure, we assume that 100% of the point-source inputs are delivered through the stream system to the monitoring station where river export is measured. This procedure probably overestimates point-source contributions to total river export. During low and medium stream flows, much of the P entering these streams from point sources moves into at least temporary storage on the stream bottom, through either biological uptake or sediment adsorption (Baker, 1980). It is unlikely that all of this point-source P is washed out during high flow events.

River export of TP and SRP is based on the monitoring programs of the Heidelberg College Water Quality Laboratory. These procedures, which include sampling storm events throughout the year, are detailed elsewhere (Baker, 1988, 1993; Richards and Baker, 2002). Export data for the Maumee watershed were not available for the 1979–1981 water years. Phosphorus exports were estimated for these three years based on seasonal flow data, and average seasonal flow-weighted mean concentrations from the preceding and following three-year periods. No data are available for SRP export for 1985 in either watershed.


   RESULTS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 RATIONALE FOR USE OF...
 METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Phosphorus Inputs to Watershed Soils
Average P inputs over the study period are shown in Table 2. More than 75% of the P inputs in both watersheds has come from fertilizer. Manure has accounted for about 20% of the P inputs, while sludge and rainfall have accounted for about 2 and 1%, respectively.


View this table:
[in this window]
[in a new window]
  		Table 2. Average annual P inputs for the Maumee watershed (1976–1995) and Sandusky watershed (1974–1995).

 
Trends in annual P inputs over the study period are shown in Fig. 3 . Fertilizer inputs have decreased substantially in both watersheds from their peak values in 1979–1981. For the Maumee watershed, the annual fertilizer P inputs for 1993–1995 were 39% lower than the annual inputs for 1979–1981. For the Sandusky watershed, the 1993–1995 inputs were 30% lower than the 1979–1981 inputs. In both watersheds, much of the decrease occurred in 1982 and 1983. Since that time, P fertilizer applications decreased less rapidly in the Maumee watershed while they have leveled off in the Sandusky watershed. Manure P inputs have decreased steadily over much of the study period. Between the 1976–1978 average and the 1993–1995 average, manure inputs decreased 13% in the Maumee watershed and 28% in the Sandusky watershed. This decrease is associated with decreasing numbers of farm animals (Richards et al., 2002a). Phosphorus from sludge and rainfall (not plotted) represent such a small portion of the P inputs that time trends in their inputs have little effect on the overall P budgets.



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 3. Annual P inputs from fertilizer, manure, and sludge for the Maumee and Sandusky watersheds, 1975–1995.

 
Phosphorus Outputs from Watershed Soils
Average annual P outputs by crop removal and nonpoint-source river export over the study period are shown in Table 3. The contributions of corn, soybean, wheat, hay and nonpoint-source river export are very similar in the two watersheds, due to their similarities in land use, agriculture, and geology (Richards and Baker, 2002; Richards et al., 2002a,b). Corn and soybean each account for about 35% of the total P export. Wheat, hay, and river export account for about 14, 5, and 10% of the P export, respectively.


View this table:
[in this window]
[in a new window]
  		Table 3. Average annual P outputs for the Maumee watershed (1976–1995) and Sandusky watershed (1974–1995).

 
Changes in annual P removal associated with crop harvests are shown in Fig. 4 . Annual variations in planting acreage and crop yields result in large variations in P removal by crops. While crop acreage has decreased slightly in both watersheds, crop yields have increased (Richards et al., 2000a). The net effect of these changes has been a gradual increase in crop removal of P during the study period in both watersheds.



View larger version (38K):
[in this window]
[in a new window]
  		Fig. 4. Annual P outputs from corn, soybean, wheat, and hay for the Maumee and Sandusky watersheds, 1975–1995. (Same legend for both graphs.)

 
The average annual river export of P, along with the average municipal and industrial point-source and nonpoint-source contributions to that export, are shown in Table 4. For the Maumee and Sandusky watersheds, point sources comprised 10.8 and 6.6% of the total river export, respectively. Nonpoint sources clearly have dominated river export of P for both watersheds.


View this table:
[in this window]
[in a new window]
  		Table 4. Average annual point-source and nonpoint-source components of P export for the Maumee watershed (1976–1995) and Sandusky watershed (1974–1995).

 
Annual exports of point-source and nonpoint-source P for the study watersheds are shown in Fig. 5 . These results illustrate the large annual variability that characterizes the nonpoint-source component of P export by river systems. During the study period, municipal point-source discharges decreased from 30.7 to 12.5 Mg P yr-1 in the Sandusky watershed and from 241 to 120 Mg P yr-1 in the Maumee watershed, based on comparisons of three-year average figures for 1975–1977 and 1992–1994.



View larger version (80K):
[in this window]
[in a new window]
  		Fig. 5. Annual point-source and nonpoint-source exports of P for the Maumee and Sandusky watersheds, 1975–1995.

 
The average contributions of SRP to the TP export over the study period are shown in Table 5. For both rivers, SRP export composed about 15% of the TP export. Changes in the SRP composition of the P export over the study period are shown in Fig. 6 . The SRP export has declined more rapidly than TP export and, consequently, SRP comprises a smaller portion of the TP in the later years of the study.


View this table:
[in this window]
[in a new window]
  		Table 5. Average annual exports of total phosphorus (TP), soluble reactive phosphorus (SRP), and particulate + soluble non-reactive phosphorus (TP - SRP) for the Maumee watershed (1976–1995) and Sandusky watershed (1974–1995).

 


View larger version (76K):
[in this window]
[in a new window]
  		Fig. 6. Annual export of total phosphorus (TP), soluble reactive phosphorus (SRP), and particulate + soluble nonreactive phosphorus (TP - SRP) for the Maumee and Sandusky watersheds, 1975–1995.

 
Net Phosphorus Accumulation
Changes in the total inputs, total outputs, and net accumulation of P for the study period are shown in Fig. 7 . For both watersheds, the P inputs have decreased over the study period, while the P outputs have increased. The linear regression lines (Fig. 7) illustrate the general directions of these changes. As the inputs and outputs converge, the net accumulation (input - output) in the watersheds decreases (Fig. 7). For the period of peak fertilizer application rates (1979–1981), net accumulation rates in the watersheds were about 13.4 kg P ha-1 yr-1 in the Maumee watershed and 9.6 kg P ha-1 yr-1 in the Sandusky watershed. These accumulation rates were equal to about 66.2% of the P fertilizer application rates in the Maumee watershed and 49.6% in the Sandusky watershed. For the 1993–1995 period, net accumulation rates dropped to 3.7 kg P ha-1 yr-1 for the Maumee and 2.6 kg P ha-1 yr-1 for the Sandusky. These accumulation rates represent 29.9 and 19.0% of the much-reduced fertilizer application rates of the 1993–1995 period. Over the entire study period the average net accumulation rates were 8.2 kg P ha-1 yr-1 in the Maumee watershed and 5.6 kg P ha-1 yr-1 in the Sandusky.



View larger version (36K):
[in this window]
[in a new window]
  		Fig. 7. Annual inputs, outputs, and net accumulation of P for the Maumee and Sandusky watersheds, 1975–1995.

 

   DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 RATIONALE FOR USE OF...
 METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Uncertainties in Phosphorus Budget Calculations
In addition to the uncertainties associated with the use of average P composition of crops and manures and average manure production per animal species, there are several uncertainties associated with the particular approach used in this study. The use of county-level fertilizer sales data for areas as small as the Sandusky watershed introduces uncertainty into the fertilizer input calculations. Transport of fertilizer across county boundaries could influence the results. Since the average unit area fertilizer inputs are very similar for the Sandusky watershed and the much larger Maumee watershed (Table 2), the problem of fertilizer transport across county boundaries does not appear large or net transport is small. If cross-county fertilizer transport is insignificant, then cross-county manure transport should also be insignificant, since manure transport would be much more costly than fertilizer transport.

Nonpoint P export by rivers from the watersheds is a small portion of the total P export from cropland soils, averaging 10% in both watersheds (Table 3). However, this flux certainly underestimates erosional movement of P on the cropland within the watersheds. Particulate P associated with the total suspended solids fraction accounts for about 80% of the TP export. Particulate P transport is therefore closely linked to sediment transport in these watersheds. For the Maumee and Sandusky watersheds, the total suspended sediment export is approximately 10% of the gross erosion that occurs in the watersheds (Baker, 1988). Much of the eroded soil simply moves downslope within fields. Also, more sediment moves from the edges of fields into streams than is exported from the watersheds by riverine transport. Large quantities of sediment are deposited in drainage ditches that must be periodically dredged to provide adequate outlets for tile drainage. Sediment deposition on floodplains and in wetlands, riparian corridors, and reservoirs represents possible sinks for P that are not included in this cropland budget. Thus, the nonpoint riverine export of P may significantly underestimate erosional removal of P from cropland soils.

Comparisons with Phosphorus Budgets in Other Watersheds
David and Gentry (2000) developed a P budget for the state of Illinois using similar techniques to those used in this study. Both the absolute P input rates and the trends in P input rates for fertilizer are very similar to those observed in this study, on an input per unit area basis. The years with the highest fertilizer application rates (1979–1981) were the same in Illinois and in our watersheds. For Illinois, the peak application rates were about 15.1 kg P ha-1 yr-1, while for the Maumee and Sandusky the rates were 20.1 and 19.3 kg P ha-1 yr-1, respectively. The higher fertilizer application rates in the Ohio watersheds are due, at least in part, to the higher percentage of cropland in these watersheds (75%) than in the state of Illinois (63%).

Manure P application rates in 1975 were about 2.3 kg P ha-1 yr-1 in Illinois and about 4.8 and 4.3 kg P ha-1 yr-1 in the Maumee and Sandusky watersheds, respectively. Manure inputs declined steadily for both areas, dropping to 1.7 kg P ha-1 yr-1 in Illinois and to 4.1 and 3.3 kg P ha-1 yr-1 in the Maumee and Sandusky watersheds.

Net P accumulations were very large in the 1970s for both Illinois and the Ohio watersheds. This resulted in a substantial buildup of P soil test values in our watersheds (Calhoun et al., 2002). Net accumulations are variable from year to year, but have dropped to zero in Illinois and to values of about 3.7 and 2.6 kg P ha-1 yr-1 in the Maumee and Sandusky watersheds. However, for Illinois, David and Gentry (2000) calculated net inputs as fertilizer inputs minus crop removal, while in this paper we include manure inputs and nonpoint river exports, along with fertilizer inputs and crop removal, as part of the net input. Adjusting for these differences in calculations of net inputs puts the Illinois and northwestern Ohio budgets closer together.

Our results differ from those of Bennett et al. (1999) in their P budget for the 686-km2 Lake Mendota watershed in Wisconsin. They calculated a P budget for 1995 and observed that P inputs of 1307 Mg (19.1 kg P ha-1) significantly exceeded P outputs of 732 Mg (10.7 kg P ha-1), giving rise to a net accumulation rate of 8.2 kg P ha-1. Phosphorus soil test data for the surrounding county (Dane County) increased by 40% from 1974–1977 to 1986–1990, indicating the effects of historical net P inputs.

Goolsby et al. (1999) provided detailed information on P inputs and outputs for 42 interior watersheds within the Mississippi River basin for 1992, based on calculations similar to those used in this study. They also provided river export information for these same watersheds. Although they did not report P budgets, we have used the information they provided to calculate budgets for the 15 watersheds with the highest unit area P exports (>0.80 kg P ha-1). Inputs included fertilizer, manure, industrial point sources, and municipal point sources while outputs included crop removal, pasture grazing, and river export. For these 15 watersheds, inputs ranged from 3.6 to 17.0 kg P ha-1 yr-1 while outputs ranged from 3.9 to 17.8 kg P ha-1 yr-1. Net accumulation ranged from -2.5 to 2.9 kg P ha-1 yr-1, with 9 of the 15 watersheds having net accumulation less than 1 kg P ha-1 yr-1. Combining data from all 15 watersheds indicated that net accumulation was 11% of total inputs. These 15 watersheds include most of the watersheds in the Mississippi Basin, where cropland is the dominant land use. For the watersheds dominated by cropland, unit area total inputs, total outputs, and net accumulation for the interior basins of the Mississippi were similar to those in this study.

Based on data from 1987, the National Research Council (1993) found that, for the Corn Belt region, net P accumulation equaled 56% of the total P input. For Illinois, Indiana, Iowa, and Ohio, which cover most of the watersheds discussed above, the National Research Council indicated that net accumulation, as a percentage of total inputs, was 54, 65, 45, and 64%, respectively. The more recent studies, including those of Goolsby et al. (1999), David and Gentry (2000), and this study suggest that net inputs have decreased substantially in recent years.

Magnitude of River Exports
With average TP export of 1.35 and 1.40 kg P ha-1 yr-1, the Maumee and Sandusky watersheds have high export rates relative to most large Midwestern watersheds. David and Gentry (2000) found average annual export rates for 1980–1997 to vary from 0.7 to 1.1 kg P ha-1 yr-1 for six major watersheds in Illinois. The area-weighted average for the state was 1.0 kg P ha-1 yr-1. Only three of the 42 interior watersheds in the Mississippi drainage studied by Goolsby et al. (1999) had higher export rates than the Maumee and Sandusky rivers. These included the Illinois River at Marseilles, with a P yield of 1.90 kg P ha-1 yr-1. The Illinois River carries the sewage effluent P from Chicago in addition to agricultural runoff. The other two rivers had low population densities and low percentages of cropland but high suspended sediment export. These were the Grand River near Sumner, MO with a P export rate of 1.83 kg P ha-1 yr-1, and the Big Black River near Bovina, MS with a P export rate of 1.45 kg P ha-1 yr-1. The export rates for the Maumee and Sandusky are also high relative to most of the agricultural watersheds included in the National Water-Quality Assessment Program (United States Geological Survey, 1999).

Point-source contributions to the Sandusky and Maumee watersheds account for a much smaller proportion of total river export than that for Mississippi watersheds, even though the Maumee and Sandusky have relatively high population densities. David and Gentry (2000) report that for Illinois rivers, including the Illinois River that receives the sewage effluents from Chicago, sewage inputs were equivalent to 47% of the total river export. Excluding the Illinois River drops the contributions of sewage inputs to 33%. For the Maumee and Sandusky rivers, sewage inputs account for 8.2 and 5.5% of the riverine export of total P (Table 4).

Using the data provided by Goolsby et al. (1999), we calculated the total municipal and industrial point-source P inputs to streams and the total riverine P export for the 15 interior watersheds with the highest riverine export rates. Industrial and municipal point-source P inputs were equivalent to 38.7% of the total riverine P export for these 15 watersheds. For two of these watersheds, the Illinois River and the Great Miami River in Ohio, the reported sewage P inputs exceeded total riverine P export. This suggests that delivery of P from point sources through rivers is less than 100% or that riverine P export is underestimated.

The smaller percentage of point-source contributions to riverine P export in the Maumee and Sandusky watersheds (Table 4) reflects the high degree of P removal required at sewage treatment plants in the Lake Erie basin coupled with the large unit area P export from nonpoint sources in these watersheds. Since Mississippi watersheds have much higher contributions from point sources than do the northwestern Ohio watersheds, the proportion of the riverine P export from nonpoint sources would be lower in the Mississippi watersheds. Agricultural activities in the Maumee and Sandusky watersheds apparently export considerably more P per unit area than most row crop watersheds in the Mississippi drainage. This is probably associated with the high clay content of soils and the corresponding high clay content of total suspended sediments exported from these northwestern Ohio watersheds (Antilla and Tobin, 1978).

Soluble reactive P accounts for a smaller proportion of the TP export for the Maumee and Sandusky watersheds than for most of the row crop watersheds of the Mississippi drainage. David and Gentry (2000) found that 38% of the TP was in the dissolved form in Illinois rivers. For the Middle Mississippi drainage, SRP comprised 28% of the TP export (Goolsby et al., 1999). Over the study period, SRP export averaged about 15% of TP export for the Maumee and Sandusky watersheds (Table 5) and had dropped still lower by the end of the study period (Fig. 6). Factors contributing to the low proportions of SRP export relative to TP export for the Maumee and Sandusky watersheds probably include the low P inputs from point sources and high particulate P export associated with the high clay content of total suspended sediments exported from these watersheds.

Trends in Riverine Phosphorus Export
The water quality data sets for the Maumee and Sandusky watersheds have been examined extensively for the purpose of detecting trends in riverine transport (Richards and Baker, 1993, 1997, 2002; Moog and Whiting, 2002) and linking those trends with changing agricultural practices and/or with weather variations or trends (Forster et al., 2000; Moog and Whiting, 2002; Richards and Baker, 2002). Because the large, weather-induced, annual variations in nonpoint loads (Fig. 5) can obscure responses to management changes, the trend analyses commonly adjust the data for variations in discharge and season. This adjustment assumes that a substantial portion of weather-induced annual variations in loading is reflected in seasonal and annual variations in discharge. Normalizing for discharge and season should reduce this aspect of weather effects.

The discharge-normalized data show 65 to 89% reductions in SRP concentrations and loads for both the Maumee and Sandusky rivers for 1975–1995. Total P reductions of 25 to 45% are also noted, along with smaller reductions in suspended solids and small, statistically insignificant increases in nitrate export. The specific size of these changes varies somewhat, depending on the statistical procedures employed. For example, the sizes of change vary depending on whether data are aggregated into daily, monthly, annual, or storm event concentrations (Richards and Baker, 1997), on whether the trends track flow-weighted or time-weighted concentrations (Richards and Baker, 1993), and on the statistical method employed (Richards and Baker, 2002).

Actual loads transported to Lake Erie have changed in the same directions as discharge-normalized loads. In Table 6, we have compared the average annual and seasonal loads for two 10-year periods, 1975–1984 and 1986–1995, for the Sandusky River. Relative to the first 10-year period the average annual loads in the second period showed a 29% reduction in TSS, a 27% reduction in TP, a 57% reduction in SRP, and a 9% increase in nitrate N. Average annual stream flow decreased by 13% and precipitation decreased by 5%. These comparisons assume that a 10-year period is sufficient to characterize "baseline" by averaging weather-related variations in pollutant loading. Changes calculated in this manner would be expected to be smaller than changes comparing the first and last years of loads over a 20-year period, where changes in various management practices occurred throughout most of the 20-year period. Changes similar to those shown in Table 6 for the Sandusky River also occurred in the Maumee River.


View this table:
[in this window]
[in a new window]
  		Table 6. Changes in average seasonal{dagger} and average annual exports between two 10-year periods in the Sandusky watershed.

 
The largest reductions in P loading occurred during winter. For the Sandusky River, winter SRP export dropped by 35 Mg and TP export by 107 Mg (Table 6). For these same periods, average municipal point-source P loading dropped by 3.7 Mg. Reductions in municipal loading account for a very small proportion of the reductions in riverine P export observed during these studies.

Possible Causes of Reductions in Nonpoint, Riverine Phosphorus Export
While there have been substantial decreases in fertilizer and manure P inputs to the watersheds, and P exports via crops have increased, net inputs are still positive, and the P content of the soils should still be increasing, albeit at a much slower rate than in the 1970s. Calhoun et al. (2002) compared P soil test levels for samples collected in 1996–1998 at 302 soil survey sites in northwestern Ohio with soil tests on archived soil samples collected between 1953 and 1988 at the same locations. The mean and median for tests from 1996–1998 were 30 and 45% higher, respectively, than mean and median of samples collected between 1953 and 1988. If P export were strictly proportional to soil P content, both SRP and TP export should now be reaching peak values. Since both SRP and TP export have decreased, other factors must be operating to reduce riverine P export. These other factors may include various erosion control best management practices (BMPs) that farmers have adopted as part of the Lake Erie Phosphorus Reduction Program, possible changes in P fertilizer management beyond simply the amounts of P applied, and changes in weather conditions.

Nutrient budget analyses at the level of large watersheds, such as in this study, do not take into account the observations that large proportions of P export can arise from relatively small proportions of the total watershed area (Pionke et al., 1997; Gburek et al., 2000). Minimal targeting of P reduction efforts to critical P source areas has occurred in the Lake Erie P reduction programs. Since critical P source areas were not identified in the study watersheds and the level of conservation treatment of these areas was not specifically tracked, it is not possible to provide estimates of the amount of P reductions to expect. However, several factors support the assumption that critical P source areas have received at least as much conservation treatment as cropland as a whole.

Most of the soils in the Maumee and Sandusky watersheds are either poorly or very poorly drained, due primarily to their high clay content (Myers et al., 2000). Thus, surface runoff, and its attendant clay and particulate P transport, probably originate over a larger fraction of the watershed cropland than in watersheds with more permeable soils. Even if adoption of conservation tillage were random across the landscape, given the large size of these watersheds and the extent of adoption of conservation tillage, a substantial portion of the critical P source areas must have been treated with conservation tillage.

Other conservation programs, such as the USDA's Conservation Reserve Program (CRP), were targeted to highly erodible lands. By 1995, 85% of the highly erodible land in the Maumee watershed was treated and 97% in the Sandusky watershed (Richards et al., 2002a). Since highly erodible land probably includes some of the critical P source areas, the high levels of treatment of highly erodible land would result in high levels of treatment of the associated critical P source areas.

The above arguments suggest that erosion control programs have reached a considerable portion of the critical P source areas in these watersheds. Farmer treatment of these areas should have resulted in decreased movement of particulate P from cropland to streams and offers an obvious explanation for the decreases in riverine export of P and of total suspended solids from the study watersheds. Myers et al. (2000) attributed reductions in TSS export from the Maumee watershed to adoption of conservation tillage.

The large decrease in SRP export was not anticipated at the time the Lake Erie P control programs were initiated. Plot studies at that time were showing increased SRP export with no-till systems, due to buildup of P levels in the surficial layer of soil. The P budget analyses presented above, and the accompanying increases in P soil test values, also would suggest an increase in SRP export.

In the early years of this study, most of the SRP export from these watersheds occurred in the winter (Table 6). Large declines in SRP export occurred at the same time the large declines in fertilizer application were occurring (Calhoun et al., 2002). A possible explanation for the large declines in SRP export is evident from a closer examination of the winter SRP data. Figure 8 illustrates the concentrations of SRP and corresponding flows during successive winter seasons for the period from 1975 to 1992. During the 1975–1982 period, many winter storm events were characterized by very high SRP concentrations. This pattern largely disappeared during the 1983–1992 period. These results suggest that when P fertilizer was being applied at buildup rates, fertilizer management practices were less important to farmers. Relatively large amounts of P fertilizer may have been applied to soil surfaces in the fall and early winter, some of which was not subsequently incorporated into the soil. This provided a ready source of P for dissolution and export during snow melt and/or rainfall-runoff events in winter. As farmers reduced their P fertilizer application rates, they may also have improved their fertilizer management techniques, shifting to more incorporation of fertilizer. Consequently, winter runoff events in recent years have lacked high SRP concentrations. Unfortunately, data are unavailable to document any changes in P fertilizer application practices during that period.



View larger version (38K):
[in this window]
[in a new window]
  		Fig. 8. Soluble reactive phosphorus (SRP) concentrations and stream flow for samples collected during successive winters (January, February, and March) in the Sandusky River, 1975–1992.

 
An additional set of factors that may have affected winter runoff of SRP and nitrate has been the changes in climate and hydrology. Moog and Whiting (2002) note that the amount of snow cover has decreased significantly during the study period. Also, minimum winter temperatures have increased. Table 6 indicates that winter stream flow decreased by 29% and precipitation by 19% in the Sandusky watershed. Yet nitrate export increased by 12% during the winter season. A possible explanation of the increased nitrate export would be increased winter tile flow associated with warmer winter temperatures. If a greater proportion of winter precipitation were moving off cropland through tile systems, an increase in nitrate export would be expected, along with a decrease in SRP export. In this region, with the large clay content of the soils, tile effluent has very low SRP concentrations. The coincidence in timing between the large reductions in winter SRP export and the large reductions in P fertilizer applications suggests a link between these effects. Although weather variations may account for portions of the reductions in riverine export of SRP and TP, the sudden disappearance of elevated SRP concentrations during winter storms is difficult to explain through weather effects. Withers et al. (2000) note the importance of reducing pollutant transport from fields for obtaining reductions in P export in the short term.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 RATIONALE FOR USE OF...
 METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
Analyses show decreasing P inputs and increasing P outputs from the soils of the Maumee and Sandusky watersheds during the 1975–1995 period. However, inputs still exceed outputs so that the soils are, on average, still accumulating P. Soils testing data generally support this assessment. Based on P budget analyses, riverine P export would still be expected to be increasing, although less rapidly during more recent years.

However, detailed water quality studies in these watersheds indicate decreasing rather than increasing riverine P export. These downward trends are evident in both actual loads delivered to Lake Erie and in loads that have been adjusted for discharge and season. These reductions in P export can be explained, in part, by farmer adoption of erosion control practices aimed at reducing particulate P delivery to streams.

At the same time farmers reduced fertilizer P applications, they may also have taken steps to minimize loss of P fertilizer via surface runoff. Although historical data on fertilizer application techniques are not available, such changes would explain the large reductions in SRP concentrations during winter runoff events that occurred at the same time fertilizer application rates dropped. Weather trends, especially reductions in winter snow cover and increases in minimum temperatures, may also be involved in the reductions in P export and increases in nitrate export that have occurred.


   ACKNOWLEDGMENTS
 
This research was funded by a grant to The Ohio State University, Case Western Reserve University, and Heidelberg College from USDA-CSREES. We thank our colleagues for stimulating and insightful discussions about the causal relationships between water quality trends and trends in agricultural practices. We also thank the IJC for providing the municipal and industrial point-source P loading data.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 RATIONALE FOR USE OF...
 METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Prepared as part of a grant from the USDA State Cooperative Research, Education and Extension Service (CSREES).


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 RATIONALE FOR USE OF...
 METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 J. Environ. Qual.Home page
L. E. Gentry, M. B. David, T. V. Royer, C. A. Mitchell, and K. M. Starks
Phosphorus Transport Pathways to Streams in Tile-Drained Agricultural Watersheds
J. Environ. Qual., January 25, 2007; 36(2): 408 - 415.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
R. P. Richards, F. G. Calhoun, and G. Matisoff
The Lake Erie Agricultural Systems for Environmental Quality Project: An Introduction
J. Environ. Qual., January 1, 2002; 31(1): 6 - 16.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
R. P. Richards, D. B. Baker, and D. J. Eckert
Trends in Agriculture in the LEASEQ Watersheds, 1975-1995
J. Environ. Qual., January 1, 2002; 31(1): 17 - 24.
[Abstract] [Full Text] [PDF]

Home page