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


     

Published online 27 October 2006
Published in J Environ Qual 35:2132-2145 (2006)
DOI: 10.2134/jeq2006.0157
© 2006 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
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 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 Google Scholar
Google Scholar
Right arrow Articles by Schilling, K. E.
Right arrow Articles by Spooner, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Schilling, K. E.
Right arrow Articles by Spooner, J.
Agricola
Right arrow Articles by Schilling, K. E.
Right arrow Articles by Spooner, J.
Related Collections
Right arrow Surface Water Quality
Right arrow Watershed and Landscape Processes
Right arrow Cover Crops

TECHNICAL REPORTS

Landscape and Watershed Processes

Effects of Watershed-Scale Land Use Change on Stream Nitrate Concentrations

Keith E. Schillinga,* and Jean Spoonerb

a Iowa Department of Natural Resources, Iowa Geological Survey, 109 Trowbridge Hall, Iowa City, IA 52242-1319
b Soil & Water Environmental Technology Center (SWETC), North Carolina State University, Box 7637, Raleigh, NC 27695-7637

* Corresponding author (kschilling{at}igsb.uiowa.edu)

Received for publication April 20, 2006.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The Walnut Creek Watershed Monitoring Project was conducted from 1995 through 2005 to evaluate the response of stream nitrate concentrations to changing land use patterns in paired 5000-ha Iowa watersheds. A large portion of the Walnut Creek watershed is being converted from row crop agriculture to native prairie and savanna by the U.S. Fish and Wildlife Service at the Neal Smith National Wildlife Refuge (NSNWR). Before restoration, land use in both Walnut Creek (treatment) and Squaw Creek (control) watersheds consisted of 70% row crops. Between 1990 and 2005, row crop area decreased 25.4% in Walnut Creek due to prairie restoration but increased 9.2% in Squaw Creek due to Conservation Reserve Program (CRP) grassland conversion back to row crop. Nitrate concentrations ranged between <0.5 to 14 mg L–1 at the Walnut Creek outlet and 2.1 to 15 mg L–1 at the downstream Squaw Creek outlet. Nitrate concentrations decreased 1.2 mg L–1 over 10 yr in the Walnut Creek watershed but increased 1.9 mg L–1 over 10 yr in Squaw Creek. Changes in nitrate were easier to detect and more pronounced in monitored subbasins, decreasing 1.2 to 3.4 mg L–1 in three Walnut Creek subbasins, but increasing up to 8.0 and 11.6 mg L–1 in 10 yr in two Squaw Creek subbasins. Converting row crop lands to grass reduced stream nitrate levels over time in Walnut Creek, but stream nitrate rapidly increased in Squaw Creek when CRP grasslands were converted back to row crop. Study results highlight the close association of stream nitrate to land use change and emphasize that grasslands or other perennial vegetation placed in agricultural settings should be part of a long-term solution to water quality problems.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
NONPOINT SOURCE (NPS) pollution is a major cause of surface water impairment in the USA (USEPA, 2003). Export of NPS pollution from the midwestern region of the USA is receiving increasing attention due to concerns regarding excessive nutrient enrichment and eutrophication in streams (Turner and Rabalais, 1994; Vitousek et al., 1997; Dodds and Welch, 2000; USEPA, 2000) and development of hypoxic conditions in the Gulf of Mexico (Rabalais et al., 1996, 2002; Goolsby et al., 1999; Burkart and James, 1999). Nitrate-nitrogen (nitrate) export from the state of Iowa, located in the middle of the U.S. Corn Belt, has been identified as a major contributor to Mississippi River pollutant loads (Goolsby et al., 1999).

The amount of agricultural land in a watershed is well understood to be a good predictor of NPS pollution in streams (Hill, 1978; Mason et al., 1990; Jorden et al., 1997; Schilling and Libra, 2000). Average annual nitrate concentrations (mg L–1) in Iowa's rivers from 1996 through 1998 was approximated by simply multiplying a watershed's row crop percentage by 0.1 (Schilling and Libra, 2000). Furthermore, agricultural land use strongly affects the hydrology of watersheds (Schilling and Wolter, 2005). The percentage of row crop land in a watershed largely governs the partitioning of total streamflow into baseflow and stormflow (runoff) components by delivering more total discharge and baseflow to streams per unit area (Schilling and Wolter, 2005). Baseflow, in particular, is significantly related to row crop intensity in Iowa (Schilling and Libra, 2003; Schilling, 2005). Nitrate is primarily delivered to Iowa streams through groundwater discharge as baseflow and tile drainage (Hallberg, 1987; Schilling, 2002a; Schilling and Zhang, 2004).

Considerable research has demonstrated that agricultural conservation practices utilizing perennial cover reduce NPS pollution in streams. Along stream corridors, perennial riparian buffers have been shown to influence the amount, timing, and pathways of water and pollutants that move through them (e.g., Peterjohn and Correll, 1983; Jorden et al., 1993; Hill, 1996; Bharati et al., 2002; Lee et al., 2003; Schultz et al., 2005). In field studies Randall et al. (1997) found that nitrate concentrations in drainage water from alfalfa and perennial grasses were 35 times lower than drainage water from corn and soybean fields. Brye et al. (2000) compared the hydrologic budgets of restored prairie and cultivated corn ecosystems and found that prairie maintained greater soil water content in the soil profile, had a larger evapotranspiration (ET), and significantly less drainage. Leaching losses of nitrogen and phosphorus were also higher from managed corn systems compared to restored tallgrass prairie (Brye et al., 2001, 2002). On a watershed scale, recent modeling studies have suggested that a conversion of substantial portions of the landscape to perennial cover offers promise for improving water quality (Coiner et al., 2001; Nassauer et al., 2002; Vache et al., 2002). Vache et al. (2002) predicted that targeted agricultural conservation practices (buffers, wetlands, grassed waterways, filter strips, and field borders) could potentially reduce nutrient loadings by 54 to 75% and sediment loadings by 37 to 67%. Dinnes et al. (2002) suggested that diversifying plant rotations in watersheds could better utilize water during vulnerable leaching periods occurring in the spring and fall.

One perennial cover option available to Iowa is the reintroduction of perennial grasses to the agricultural landscape (Schilling, 2001; Jackson and Jackson, 2002). Iowa was once part of the tallgrass prairie ecosystem that covered 67.6 million ha in the USA, of which more than 99.6 to 99.9% has been lost (Samson and Knopf, 1994). Although the plowdown of prairies occurred primarily between the 1850s and 1890s (Smith, 1992), perennial cover remained a part of the landscape through crop rotations of sod crops (oats, Avina sativa L.; hay) with annual crops (corn, Zea mays L.; soybean, Glycine max (L.) Merr.). The balance of sod vs. annual crops was about fifty-fifty through the 1950s (Jackson, 2002). However, from the mid-20th century to the present, soybean production has increased dramatically and replaced many sod-based rotations. Between 1940 and 2000, soybean production in Iowa increased from 1000000 acres to approximately 11000000 acres, so that combined with minor increases in corn production, total row crop area (corn and soybeans) increased approximately 30 to 40% during this time (Iowa Agricultural Statistics, 2001). Similarly, nitrogen fertilizer use in Iowa significantly increased from 1965 to 1981, generally averaging between 900000 to 1.0 million tons per year in the 1990s (Iowa Agricultural Statistics, 2001). In conjunction with the land use change, a two- and threefold increase in nitrate concentrations has been observed in the Cedar and Des Moines rivers in Iowa between 1940 and 2000 (Iowa Department of Natural Resources, Geological Survey Bureau, 2001; Schilling, 2005).

It is evident that (1) NPS pollution from agriculture is a major problem in Iowa and the agricultural Midwest, and (2) perennial cover in an agricultural ecosystem may reduce NPS pollution loading to streams. However, the effectiveness of the introduction of perennial cover into an agricultural landscape to reduce NPS pollution in a stream is relatively untested at a watershed scale. The Walnut Creek Watershed Monitoring Project was established in 1995 as a 10-yr NPS monitoring program in conjunction with watershed habitat restoration and agricultural management changes implemented by the U.S. Fish and Wildlife Service (USFWS) at the Neal Smith National Wildlife Refuge (Refuge) in Jasper County, Iowa (Fig. 1). A large portion of the Walnut Creek watershed is being restored from row crop agriculture to native prairie and savanna (Drobney, 1994; Schilling and Thompson, 2000). For the project, the Walnut Creek watershed (treatment watershed) was paired with a highly agricultural control watershed (Squaw Creek) to evaluate effects of large-scale prairie restoration on stream water quality. Although Squaw Creek watershed was designated a control, land use change also occurred in this watershed during the project. In the late 1990s, substantial portions of the Squaw Creek watershed were converted from Conservation Reserve Program (CRP) grasslands back to row crop. Thus, land use changes in the Walnut/Squaw creek paired watershed study involved both decreasing and increasing row crop land use over a 10-yr period.


Figure 1
View larger version (47K):
[in this window]
[in a new window]
  		Fig. 1. Location map including locations of prairie plantings, subbasins, and stream sampling locations.

 
The purpose of this study was to evaluate the response of stream nitrate concentrations to changing row crop and grassland land use patterns in two agricultural watersheds and assess the timeframe needed for observing changes in stream nitrate concentrations resulting from land use change.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Study Area
The watersheds of Walnut Creek (5218 ha) and Squaw Creek (4703 ha) are in the southern Iowa drift plain landscape region, an area characterized by steeply rolling hills and well-developed drainage (Prior, 1991). Basin characteristics in both watersheds are very similar and make them well-suited for a paired watershed design (Schilling and Thompson, 2000). Soils consist mainly of silty clay loams, silt loams, or clay loams formed in loess and pre-Illinoian till with many soils characterized by moderate to high erosion potential. The watersheds are underlain by 6 to 30 m of pre-Illinoian till overlying Pennsylvanian Cherokee Group shale, limestone, sandstone, and coal.

The study area is in a humid, continental region with average annual precipitation of around 750 mm. Highest monthly rainfall totals typically occur in May and June, although large storms occurring throughout the summer can lead to rapid rises in discharge. High stream flows are also associated with major snowmelt events that often occur in late winter and/or early spring. Three USGS stream gauges used in this study are in the upper (WNT1) and lower (WNT2) portions of the Walnut Creek watershed and at the Squaw Creek watershed outlet (SQW2) (Fig. 1). Discharge in both watersheds tends to be flashy, displaying rapid responses to precipitation. Stream discharge at WNT2 has ranged from a high of 56276 m3 h–1 to a low of 2 m3 h–1 (Schilling et al., 2006).

Monitoring Design
The project utilized a paired watershed design with Walnut Creek designated as a treatment watershed and Squaw Creek a control. Paired watershed studies offer increased statistical power to detect changes in water quality from land treatment (Spooner et al., 1987; Clausen and Spooner, 1993; Loftis et al., 2001). The approach typically involves two monitoring periods, calibration and treatment, and two watersheds, treatment and control. In typical paired watershed studies, two similar watersheds are monitored for a calibration period and then a treatment is imposed on one of the watersheds (i.e., prairie restoration in Walnut Creek). A change in the relation of a variable of interest (e.g., nitrate) between treatment and control watersheds is then considered a treatment effect (Loftis et al., 2001).

This project differed from typical paired watershed studies because a calibration period was not utilized for two principal reasons: (1) pretreatment data collection was not sufficient to derive relations between treatment and control watersheds during a calibration period before the refuge was established (Schilling and Thompson, 1999); and (2) land treatment implemented in the Walnut Creek watershed gradually occurred throughout the entire monitoring period. For these reasons, a regression approach to detect gradual change was used in the paired watershed study rather than a typical pre/post paired study.

Land Cover Tracking
Detailed land use and land cover mapping was conducted by field survey in Walnut and Squaw creek watersheds in 2005. Using a map of agricultural field boundaries in the Walnut and Squaw Creek watersheds, a tablet PC was used with a geographic information system (GIS) interface to enter land cover and conservation practices descriptions for each field boundary into the GIS database. Conservation practices mapped included tillage practices, grass waterways, terraces, CRP grasslands, and other common USDA-funded conservation practices. The results of the field mapping project were used as the final 2005 land cover for the monitoring project. Land cover in 1990 was determined by visual interpretation of 1990 photographic imagery. The 2005 field boundaries were overlain on 1990 land cover imagery and the land cover for each field boundary was entered into the GIS database. In addition, USFWS personnel documented annual prairie planting areas in the Walnut Creek watershed. The GIS coverages of prairie planting sites were made available to track annual land use changes within the refuge boundary.

Data Collection
Surface water samples were collected at upstream and downstream locations in Walnut and Squaw Creek watersheds on a weekly to bimonthly basis from 1995 through 2005. Three subbasin sites in both watersheds were also sampled from April through September each year (Fig. 1). The upstream sampling point on Walnut Creek (WNT1) was above the refuge boundaries and allowed an evaluation of upper basin effects on water quality. For the 10-yr period, approximately 205 and 144 water samples were collected at the main stem and subbasin sampling sites, respectively. Water samples were analyzed for nitrate by University of Iowa Hygienic Laboratory using USEPA Method 300.0.

Three USGS stream gauging stations were monitored continuously during the study using bubble-gauge sensors and recorded by data collection platforms at 15-min intervals (Rantz, 1982). Stream discharge was computed from the rating curve developed for each site (Kennedy, 1983). Precipitation was measured at the three USGS gauging sites using standard tipping-bucket rain gauges attached to the stream gauge building. Hydrograph separation into baseflow and runoff components was performed on streamflow data collected at the three USGS gauging sites using an automated method developed by Sloto and Crouse (1996). A local-minimum method was utilized, which essentially connects the lowest points on the hydrograph and provides estimates of daily baseflow discharge between local minimums by linear interpolation (Sloto and Crouse, 1996).

Daily nitrate loads at the three stream gauging sites were estimated using the USGS program ESTIMATOR (Cohn et al., 1989, 1992; Gilroy et al., 1990) and summarized by water year. Load data were normalized on a unit area basis by dividing the total annual load at each gauging station by the watershed area above the gauge. In the case of the Walnut Creek watershed, the load per unit area between the two gauge sites was determined by subtracting the load estimated at WNT1 from WNT2.

Statistical Methods
Statistical analyses were performed according to the guidelines of Spooner et al. (1987) and Grabow et al. (1998, 1999). To test for the gradual change in chemical concentrations over time a multiple linear regression analysis was performed. A simplified form of the equation is given by:

Formula 1[1]
where Y is either the water quality variable or log of the variable for the treatment watershed (Walnut Creek), X1 is the same water quality variable (or log) for the control watershed (Squaw Creek), X2 is elapsed time, and ß0, ß1, and ß2 are regression parameters. In this equation, the estimate of ß2 indicates the magnitude of change over time. By including covariates (e.g., variable X1), the analysis blocks out much of the hydrologic variability and the change should be isolated to the effect of treatment, which in this case is being modeled as time (X2). Multiple covariates were used to develop the regression equation, including streamflow, upstream nitrate concentrations, control nitrate concentrations, and seasonality.

Streamflow data (total discharge and baseflow) were highly skewed and log-transformed before use as covariates in the regression equations. No transformation was needed or performed on the nitrate data. The time-series data were also examined for temporal autocorrelation. Temporal autocorrelation is the correlation of values from the samples taken on 1 d with samples taken from previous sample dates. Nitrate and discharge data showed significant autoregressive, lag 1, or AR(1) time series patterns. Autocorrelation (AR(1)) ranged from 0.60 to 0.86 for nitrate, and between 0.72 and 0.84 for log discharge at WNT2 and SQW2. It should be noted that discharge data evaluated in this paper refers to discharge measured on the day of sampling, not the entire data set. Discharge measured on the day of sampling was needed to serve as an explanatory variable for evaluating anion concentrations measured on that day.

Corrections for autocorrelation were made using explanatory variables and autocorrelation time-series analysis. To determine if added correction for autocorrelation was required by the use of time-series analysis, the tests for autocorrelation were performed again on the residuals from the multivariate regression models. PROC AUTOREG in SAS 9.1 was utilized to run the time-series regressions.

Seasonal patterns were also important to make appropriate adjustments. Seasonality could be seen at higher lags in the autocorrelation function (ACF). It was evident from the monitoring program that nitrate concentration data exhibited strong seasonality, with peaks occurring in May and/or June and late fall. Thus, seasonal adjustments between each month were made to account for the seasonality in the statistical trend analysis. A cycle or sin/cosine option was not used because the cycles were not of uniform width and the bimodal peaks were not the same magnitude. Essentially, the data were ‘corrected’ for the average mean value of all the samples taken in a given month over the 10-yr monitoring period. By adding a ‘month’ grouping or class variable to the statistical models, tests could be made to adjust for changes between months but retain nearly all degrees of freedom and account for the variations due to seasonality in the statistical models. A regression model that included seasonal adjustments would take the form

Formula 2[2]
where ß3i is the parameter estimate for season i. There would be 12 values for ß3i when a season is broken down into months.

Gradual changes in nitrate loading rates were not implicitly assessed in this study because daily loads were estimated values. Statistical tests to detect gradual changes over time were believed to be more appropriate for measured concentration data than estimated loads. However, the relation of discharge to stream nitrate concentrations was indirectly addressed by considering total discharge and baseflow as potential explanatory variables in the nitrate regression equation.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Land Cover Changes
In 1990, land use in both Walnut and Squaw Creek watersheds was dominated by row crops of corn and soybeans, averaging approximately 70% in both watersheds (Table 1). Land use in subbasins ranged between 71.3 and 77.6% row crop in Walnut Creek and 34.6 to 85.6% in Squaw Creek (Table 1). From 1990 through 2005, the amount of land in row crop increased 9.2% in Squaw Creek and decreased 25.4% in Walnut Creek. In Squaw Creek, the increase in row crop area from 1990 to 2005 was due to conversion of agricultural grassland previously enrolled in the CRP back to row crop production. This primarily occurred between 1996 and 2000 following national agricultural legislation passed in 1996. The land cover change was particularly evident in subbasins SQW4 and SQW 5 where row crop percentage increased by 26 and 29%, respectively, with a corresponding decrease in grass land cover. In the upper Walnut Creek watershed (WNT1), row crop acreage increased by 8% between 1990 and 2005.


View this table:
[in this window]
[in a new window]
  		Table 1. Summary of land cover in Walnut and Squaw creek watersheds (1990 and 2005).

 
In Walnut Creek watershed, row crop land use decreased from 69.4 to 54.5% between 1990 and 2005 as a result of prairie restoration by the USFWS at the Neal Smith refuge (Table 1). From 1992 to 2005, an average of approximately 90 ha of prairie were planted each year, with areas planted in 1994 and 1995 exceeding 150 ha (Fig. 2). As of 2005, 1224 ha of land in Walnut Creek watershed were planted in native prairie, representing 23.5% of the watershed. In the subbasins, restored prairie accounted for 14.3 to 45.9% of the land area with the greatest percentage of prairie conversion occurring in subbasin WNT5 (Table 1).


Figure 2
View larger version (22K):
[in this window]
[in a new window]
  		Fig. 2. Summary of annual prairie plantings by USFWS at the Neal Smith NWR.

 
The amount of land owned by the refuge but farmed on a cash-rent basis totaled 194 ha in 2005, or 3.7% of the watershed. In these areas improved agricultural management practices are mandatory. No fall application of fertilizer is allowed and a maximum of 112 kg ha–1 of nitrogen is allowed on conventional corn. The remaining land within the refuge boundary in the watershed consists of cool season grass or woods and comprises approximately 511 ha (9.8%). As of 2005, the USFWS controlled approximately 37% (1929 ha) of the Walnut Creek watershed above the WNT2 gauging station.

Hydrology
Precipitation and streamflow were variable during water years 1996 through 2005 (Table 2). Annual precipitation varied from 380 to 1056 mm during the project whereas discharge varied from 109 to 422 mm at WNT2 and 85 to 430 mm at SQW2. Average total discharge was slightly higher in Walnut Creek than in Squaw Creek, but baseflow discharge was less (129 mm compared with 136 mm, respectively). The percentage of streamflow as baseflow was higher in Squaw Creek (62.4%) than Walnut Creek (57.3%). Differences in baseflow and baseflow percentage were noted within the Walnut Creek watershed, with greater baseflow occurring in the upper portion of the Walnut Creek watershed (WNT1) compared to the entire watershed as measured at WNT2 (Table 2).


View this table:
[in this window]
[in a new window]
  		Table 2. Summary of annual precipitation, discharge, and baseflow measured at three gauging stations.

 
Seasonally, the greatest monthly discharge occurred in May each year when more than 50 mm of streamflow occurred (Fig. 3). Discharge in June typically exceeded 25 mm, and together the months of May and June accounted for more than 35% of the annual streamflow. The months of February, March, April, and July accounted for similar percentages of the annual total streamflow (approximately 10% for each month). Baseflow fraction was lowest in February and May each year and highest in late fall through January when the baseflow percentage was near 80% (Fig. 3). Monthly baseflow fractions reflected increased runoff during snowmelt in February and spring rainfall in May.


Figure 3
View larger version (27K):
[in this window]
[in a new window]
  		Fig. 3. Box plot of total discharge and average monthly baseflow percentage (open squares) by month at WNT2 and SQW2 gauging stations. Box plots illustrate the 25th, 50th, and 75th percentiles; the whiskers indicate the 10th and 90th percentiles; and the circles represent data outliers.

 
Nitrate Concentrations
Nitrate concentrations ranged between <0.5 to 14 mg L–1 at the Walnut Creek outlet (WNT2) and 2.1 to 15 mg L–1 at the downstream Squaw Creek outlet (SQW2) (Fig. 4). Overall mean nitrate concentrations were 1.7 mg L–1 higher at SQW2 than WNT2, and highest at the upstream monitoring sites in both watersheds, averaging 11.2 mg L–1 at WNT1 and 12.4 mg L–1 at SQW1. Concentrations exceeded 10 mg L–1 at WNT2 and SQW2 32.8 and 51.5% of the time, respectively. Both Walnut and Squaw Creek watersheds showed a similar temporal pattern of detection, with higher concentrations observed in the spring and early summer months coinciding with periods of application, greater precipitation, and higher stream flow (Fig. 4).


Figure 4
View larger version (45K):
[in this window]
[in a new window]
  		Fig. 4. Time series of nitrate concentrations measured at upstream downstream sites in the Walnut and Squaw creek watersheds.

 
Annually, mean nitrate concentrations ranged between 10.0 and 12.7 mg L–1 at WNT1, 6.8 to 9.5 mg L–1 at WNT2, 10.5 to 13.8 mg L–1 at SQW1, and 8.2 to 11.5 mg L–1 at SQW2 (Fig. 5). Highest mean annual concentrations at the four main stem sites all occurred in 1998, whereas lowest annual concentration varied among water years 2000, 2002, and 2004. Greater differences among water year nitrate concentrations occurred in the subbasins (Fig. 6). In Squaw Creek subbasins SQW4 and SQW5, a large increase in annual nitrate concentrations occurred during water years 1996 through 2005. In SQW4 subbasins, mean annual nitrate concentrations increased from mean values between 2.0 and 2.9 mg L–1 in water years 1996 through 1998 to values greater than 10.2 mg L–1 in water years 2003 through 2005 (Fig. 6). During water years 1999 through 2003, mean annual nitrate concentrations in the SQW4 subbasin increased an average of 1.6 mg L–1 per year. In the SQW5 subbasin, nitrate concentrations decreased slightly from water year 1996 through 1998 to water year 2000 through 2002, then increased quite substantially in water years 2003 through 2005. Mean annual concentrations at SQW5 increased from 5.1 mg L–1 in 2000 to 15.1 mg L–1 in 2005.


Figure 5
View larger version (35K):
[in this window]
[in a new window]
  		Fig. 5. Box plot of nitrate concentrations by water year at upstream and downstream sites in the Walnut and Squaw creek watersheds.

 

Figure 6
View larger version (46K):
[in this window]
[in a new window]
  		Fig. 6. Box plot of nitrate concentrations by water year at subbasin monitoring sites in the Walnut and Squaw creek watersheds.

 
In Walnut Creek subbasins, decreasing nitrate concentrations were observed in all subbasins from water year 1998 through 2004, and then a higher mean annual value was noted in water year 2005 (Fig. 6). Mean annual concentrations decreased to a low value of 8.0, 7.7, and 3.1 mg L–1 in subbasins WNT3, WNT5, and WNT6, respectively. Concentrations in these subbasins increased to 10.1, 10.2, and 5.5 mg L–1 in water year 2005. However, water year 2005 was characterized by dry conditions in the late summer and fall that dried up tributary streams and prevented water samples from being collected in August and September. Thus, average nitrate concentration values in 2005 were probably weighted higher than normal without collection of nitrate data from the typical low nitrate months of August and September.

Monthly nitrate concentrations exhibited clear seasonality, with higher concentrations occurring during May, June, and July (Fig. 7). Nitrate concentrations were typically lowest in August, September, and October when stream flow was also at a minimum and biological uptake of nitrogen was particularly evident. Although sampling frequency was limited during November and December, concentrations generally increased in the late fall. This pattern is consistent with long-term monthly nitrate concentrations in the Raccoon River in Iowa (Zhang and Schilling, 2005).


Figure 7
View larger version (33K):
[in this window]
[in a new window]
  		Fig. 7. Box plot of nitrate concentrations by month at upstream and downstream sites in the Walnut and Squaw creek watersheds.

 
Nitrate Loads
Although this study focused on changes in measured stream nitrate concentrations, chemical load data suggested that nitrate export was less in lower Walnut Creek compared to upper Walnut Creek and Squaw Creek (Table 3). Average annual nitrate losses from upper Walnut Creek (WNT1) were similar to Squaw Creek (30.3 kg ha–1 and 26.1 kg ha–1, respectively) and these areas had considerably greater nitrate losses than lower Walnut Creek containing the restored prairie (17.8 kg ha–1). At all sites, annual losses of nitrate were greatest in 1998 when they ranged from 32.8 to 58.9 kg ha–1, and were least in 2002 when they ranged from 7.3 to17.4 kg ha–1 (Table 3). Flow-weighted concentrations followed a similar pattern exhibited by the nitrate mass losses (Table 3). Flow-weighted concentrations of nitrate were higher in Squaw Creek (8.6 mg L–1) and upper Walnut Creek (10.4 mg L–1) than lower Walnut Creek (4.9 mg L–1). Nitrate loading patterns and flow-weighted concentrations following completion of the 10-yr study were similar to those reported after the first 5 yr of the project (Schilling, 2002a). Although not evaluated herein, Schilling (2002a) reported that export of nitrates from Walnut and Squaw creek watersheds occurred primarily with baseflow.


View this table:
[in this window]
[in a new window]
  		Table 3. Annual loss of nitrate and flow-weighted concentrations from various watershed areas.

 
Trends in Nitrate Concentrations over Time
Development of regression equations to evaluate changes in nitrate concentrations over time in the treatment and control watersheds first required assessment of suitable explanatory variables for use in regression equations. Upstream nitrate concentrations, control watershed concentrations, and baseflow discharge were all found to be significant explanatory variables in assessing trends over time at WNT2 (Fig. 8). Upstream nitrate concentrations were significantly related to downstream concentrations in both the Walnut and Squaw creek watersheds, with r2 values of 0.82 and 0.78, respectively (P < 0.001). Regression relations examined between treatment and control pairs were also highly significant, with downstream relations (WNT2 vs. SQW2) having a slightly higher r2 value than the upstream pairs (WNT1 vs. SQW1) (0.82 and 0.78, respectively; P < 0.0001). Interestingly, the nitrate concentration in the upstream Walnut Creek watershed (WNT1) was very similar to the concentrations in the upstream Squaw Creek (SQW1) station (slope almost 1, 0 intercept, and r2 = 0.78). Similarly, total discharge and base flow discharge were significantly related at WNT2 and SQW2. The slope between the discharge values was near 1.0 and the intercept was not statistically different from 0.0 (r2 = 0.87 for both discharge and baseflow). This implied that the discharge values from the downstream Walnut Creek and Squaw Creek watershed at the time of sampling were very similar. Nitrate had a stronger relationship with baseflow as compared to total discharge, with baseflow correlation coefficients at WNT2 (0.73) slightly lower than at SQW2 (0.77). This was consistent with previous data (Schilling, 2002a) and other Iowa watersheds (Schilling and Lutz, 2004). Relations were positive, that is, as discharge or baseflow increased, so did nitrate concentrations. For consistency among the sites, downstream baseflow was used as the explanatory variable or covariate for detection of trends for nitrate at all monitoring sites.


Figure 8
View larger version (16K):
[in this window]
[in a new window]
  		Fig. 8. Relation of nitrate concentrations at WNT2 to regression covariates.

 
All the multiple regression models have date as a trend variable. The addition of covariates that were shown to have significant relationships with WNT2 conditions was examined. From the previous sections, it was determined that for most stations and variables the important covariates included season, discharge, upstream concentrations, and control watershed. These were variables for which data were collected as part of the experimental design of the project and have a mechanistic basis for inclusion in the models. Since the four covariates are themselves highly correlated, it is possible that the addition of all covariates was not the best model. The appropriate covariates were added to the model in the order of their individual significance and were only retained if they added significant information beyond the other covariates.

For the downstream station in Walnut Creek (WNT2) the best set of covariates or explanatory variables included month (season), WNT1 (upstream nitrate concentration) and SQW2 (downstream control watershed nitrate concentrations). Results of the trend analyses and a summary of the covariates used are shown in Table 4. Although an adjustment for season and baseflow discharge at WNT2 alone did not indicate a statistical trend, the addition or either the paired site or the upstream concentration did result in a statistically significant decrease over the 10 yr sample time period. Base flow discharge explained a significant amount of variability in nitrate at WNT2 (regression with date, month, and WNT2Qb), but baseflow discharge became nonsignificant when upstream WNT1 concentrations were added to the trend model (r2 = 0.85) because the upstream concentration was highly correlated with discharge. The addition of the paired downstream Squaw site (SQW2) was a significant covariate. Since nitrate concentrations increased in the control watershed (Squaw Creek), two trend models are provided in Table 3 (one with the control watershed concentration covariate and one without).


View this table:
[in this window]
[in a new window]
  		Table 4. Trend tests for changes in nitrate concentrations over time at monitoring sites, adjusted for appropriate covariates as indicated.

 
For the Walnut Creek outlet (WNT2), the trend analysis indicated that nitrate concentrations decreased 0.119 mg L–1 yr–1 or 1.2 mg L–1 over the 10-yr project period when the Squaw Creek control watershed was utilized as a covariate. Without adjusting for the control, the decrease was 0.7 mg L–1 over the 10-yr period. Interestingly, without the upstream covariate, there was no significant trend in nitrate at WNT2. There was an increase in upstream WNT1 nitrate concentration over time (Table 4). In Squaw Creek watershed, nitrate concentrations increased 1.9 mg L–1 over 10 yr at the downstream site SQW2 and 1.1 mg L–1 over 10 yr in the upstream Squaw stations SQW1.

The magnitude of change was also estimated for each of the subbasins (Table 4). The decrease in nitrate concentrations for each subbasin was significant and of greater magnitude compared to the downstream WNT2 station. Nitrate concentrations decreased 3.4, 1.2, and 2.7 mg L–1 at WNT3, WNT5, and WNT6, respectively. All subbasins in Squaw Creek increased in nitrate concentrations, with subbasins SQW4 and SQW5 having quite dramatic increases. Over the 10-yr monitoring program, nitrate in surface water in SQW4 and SQW5 subbasins increased 11.6 and 8.0 mg L–1, respectively. The magnitude of increase in the Squaw Creek subbasins was considerably greater than the decrease measured in Walnut Creek subbasins.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Results from surface water monitoring in Walnut and Squaw creek watersheds indicated that changes in nitrate concentrations occurred during the 10-yr monitoring program. Nitrate concentrations significantly decreased in the Walnut Creek watershed, both at the watershed outlet and in monitored subbasins. On the other hand, nitrate concentrations significantly increased in the Squaw Creek watershed, again both at the watershed outlet and more significantly in subbasins. Evidence from both watersheds points to the effects of land use change on watershed nitrate concentrations.

In Walnut Creek, prairie restoration and land management changes implemented at the Neal Smith NWR reduced stream nitrate concentrations. Nitrate reductions are thought to be the result of several factors, including reduced water and nitrate flux through the soil under perennial cover compared to row crop systems, reduced fertilizer N inputs on refuge-owned crop lands, and reduced overland flow N contributions during runoff events. Nitrate concentrations at WNT2 significantly decreased (1.2 mg L–1 yr–1 in 10 yr), whereas in subbasins where land use changes comprised a greater proportion of the watershed area, nitrate concentrations decreased up to 3.4 mg L–1 in 10 yr. The statistical model developed to estimate the nitrate concentration reductions included explanatory factors to account for seasons, variable discharge, and changes occurring in other areas. Hence, changing concentrations in Walnut Creek cannot be attributed simply to changing weather patterns since this factor was assessed by evaluating effects of seasons and discharge. In addition, the paired design of the study allowed for changes in Walnut Creek to be compared against conditions in other watershed areas to evaluate whether changes occurring in the treatment watershed were any different than changes occurring in the control watershed. Therefore, explanations for the decreasing nitrate concentrations in the Walnut Creek watershed focused on land use change implemented by the USFWS at the Neal Smith NWR.

The decrease at WNT2 occurred despite an increasing trend in nitrate concentration at upstream site WNT1. Given the significant correlation of nitrate between the upstream and downstream sites, dilution of stream water with lower nitrate concentration inputs must have occurred between the two sites to produce an otherwise decreasing trend at WNT2. However, it was also evident that contributions of nitrate from upstream areas dominated the nitrate concentrations at the watershed outlet. Evidence from chemical load data (Table 3) also suggests that headwater regions in Walnut Creek contributed a greater proportion of nitrate to the stream. Nitrate losses and flow-weighted concentrations from upper Walnut Creek were nearly double those from lower Walnut Creek containing the restored prairie. The load data were consistent with results from two synoptic surveys (Schilling, 2001, 2002b) that demonstrated that the lower portion of the watershed containing the refuge contributed less nitrate loads than headwater regions. It was estimated that in 1999, 84% of the nitrate load in Walnut Creek was coming from nonrefuge land, which at the time comprised 66% of the total land area of the watershed (Schilling, 2001). It would appear that once nitrate was delivered to the stream network from row-crop-dominated headwater regions, the downstream watershed area containing the prairie diluted the stream nitrate concentrations, but concentrations remained elevated at the watershed outlet.

In the Squaw Creek watershed, a different relation of nitrate concentrations to land use change emerged. Significant increases in nitrate concentrations were measured in two subbasins (>8 mg L–1), and nitrate at the Squaw Creek watershed outlet increased by nearly 2 mg L–1 during the 10-yr project. In the two subbasins with increasing nitrate (SQW4 and SQW5), the amount of land in row crop increased by 26 and 29%, respectively, with a corresponding decrease in CRP grassland cover. Nitrate concentrations changed quite dramatically in SQW4, increasing by over 11 mg L–1 in the span of 10 yr, with most of the change concentrated within a span of 4 yr (1999 through 2003). Even in Walnut Creek, an increasing trend in stream nitrate concentrations was evident in upstream WNT1 where row crops in the watershed area increased by nine percent. It is unknown whether the increase in nitrate concentrations in these areas is attributed to increased fertilizer inputs or mineralization of organic N, but the influence of row crop land cover on stream nitrate concentrations was plainly evident.

Water quality changes in stream nitrate concentration from land use change were more easily measured in smaller watersheds. The rate of decrease in downstream Walnut Creek nitrate was less than the decreasing slope measured in smaller subbasins, and the rate of increase in Squaw Creek was considerably greater in the subbasins than watershed outlet. The downstream watershed outlets (WNT2 and SQW2) integrated water contributions from a large landscape area and because of this did not isolate areas of change particularly well. With headwater contributions of stream nitrate playing such an important factor in downstream nitrate concentrations, changes in stream nitrate concentrations at the watershed scale were easily obscured by upstream areas. When areas of land use change were isolated at the subbasin scale, substantially greater water quality changes were observed.

The amount of change in nitrate concentrations in both watersheds (10 sites) was significantly related to the degree of change in row crop land cover that occurred from 1990 through 2005 (Fig. 9). Although the comparisons of nitrate and row crop changes across time involved somewhat different time periods, the strong link between row crop land cover and nitrate concentrations was apparent. Whereas converting row crop to native prairie at the Neal Smith NWR reduced the amount of row crop in the various watershed areas and reduced stream nitrate, converting CRP grass back to row crop in Squaw Creek increased the amount of row crop and greatly increased stream nitrate. The regression slope suggests that for every change in row crop of 10% in the Walnut and Squaw Creek watersheds, a change of 1.95 mg L–1 nitrate may be expected to occur in a 10-yr monitoring period. The slope of the relation (approximately 0.2) was higher than the relation of row crop to mean annual stream nitrate concentrations reported by Schilling and Libra (2000). They stated that mean annual nitrate concentrations in Iowa's streams could be approximated by multiplying a watershed's row crop percentage by 0.1. However, Schilling and Libra (2000) also noted that the slope of nitrate concentrations vs. row crop increased with decreasing watershed size. Thus an increased slope of nitrate concentration changes to row crop changes was consistent with the previous findings because the project watershed areas are fairly small. Moreover, this project assessed the changes over time whereas the Schilling and Libra (2000) work considered long-term average values of stream nitrate. The steeper slope of stream nitrate vs. row crop land use measured in this project suggests that changes in stream nitrate concentrations may respond quite rapidly to changes in row crop land cover in small watersheds.


Figure 9
View larger version (17K):
[in this window]
[in a new window]
  		Fig. 9. Relation of change in nitrate in stream nitrate concentrations (as determined by statistical methods) to change in percentage of land cover in row crops in watersheds and subbasins.

 
Recent modeling studies (Worrall and Burt, 1999, 2001) examined the effects of land use change on nitrate concentrations and questioned how much land use change is safe. They used export coefficients and structural models to show that plowing up permanent pastures rapidly releases soil nitrogen to streams and that this process dominates over sequestration of soil N from converting lands back to permanent or temporary grassland. In this study, the rate of nitrate increase following grassland conversion to row crop was greater than the rate of nitrate decrease following conversion of row crop land to prairie. Despite similar degrees of land use change in Walnut and Squaw creek subbasins showing the greatest nitrate concentration changes (ranging from a decrease in row crops from 27 to 31% in Walnut subbasins to an increase in row crops from 26 to 28% in Squaw subbasins), the rate of increase in nitrate concentration in Squaw Creek was more than double the rate of decrease in Walnut Creek. However, this type of comparison is complicated by several factors. Some of the difference in rate of change may be traced to more gradual implementation of prairie restoration in Walnut Creek compared to more rapid plowdown of CRP grassland in Squaw Creek. Furthermore, tile drainage contributions from areas converted back to row crop would increase the rate by which changes in water quality could be observed. In contrast, most drainage tiles located in Walnut Creek prairie restoration plots were plugged or pulled wherever encountered by refuge staff. However, both watersheds have similar hydrogeology and placements of land use changes in the watersheds were rather piecemeal based on field boundaries and property ownership. Neither watershed had land use changes located for maximum water quality effect. Thus, whereas differences in implementation may account for some of the differences in the rate of nitrate concentration change in streams, project monitoring data support modeling that suggests nitrate concentrations more rapidly increased following conversion of grassland to crops than decreased following conversion back to grassland.

Lag Times for Detecting Changes
The rate of change, or the lag time needed for observing water quality change, is also governed by the hydrogeology of the watersheds. Uplands in Walnut and Squaw creek watersheds consist of loess mantling pre-Illinoian till, whereas their floodplains are comprised of mainly silty alluvium. In the absence of tile drainage, nitrate leached from soils moves with shallow groundwater to discharge to streams. In the Walnut and Squaw creek watersheds dominated by low-permeability glacial materials and glacial-derived alluvium, groundwater flow velocities are slow. The time needed for observing changing nitrate concentrations in streams resulting from land use change would be dependent on the velocity of groundwater flow to deliver nitrate to streams.

Overall, the distance that groundwater would have flowed during the 10-yr monitoring project can be estimated by

Formula 3[3]
where V is the average linear velocity (m s–1), K is the hydraulic conductivity (m s–1), dh/dl is the hydraulic gradient (dimensionless) and n is the porosity. Assuming the K of the upland loess to be 0.2 m d–1, the gradient to be 0.04, and the porosity to be 0.3, the estimated groundwater flow velocity is 0.027 m d–1 (Weisbrod, 2005), or alternatively, 98.6 m in 10 yr. This would suggest that land use changes located at a distance beyond approximately 100 m from a stream would not be expected to have an effect on baseflow water quality during the 10 yr of monitoring. Most of the upland prairie plantings are located beyond this distance to a perennial stream. In the floodplain, a similar assessment of travel distance can be made, though with less certainty due to variable alluvial stratigraphy. Depending on whether groundwater flow was concentrated in coarser or finer alluvial sediments, groundwater may have flowed from 34 to 1157 m in 10 yr. Considering that the width of the Walnut Creek floodplain varies between 183 and 366 m, it is believed that water quality improvements from all but the most recent prairie plantings occurring on the Walnut Creek floodplain have probably arrived at the stream and are affecting watershed water quality. This would be consistent with dilution of upstream nitrate concentrations occurring as stream water moves through the lower portion of the Walnut Creek watershed.

However, travel times in agricultural watersheds, like Walnut and Squaw creek, are often confounded by inputs from subsurface tile drainage. In upland areas of Walnut Creek, prairie restoration largely occurred in areas where tile drainage is minimal, not well maintained, or has been removed. In these areas, the lag time between nitrate concentration reductions from upland prairie restoration to streams will likely be governed by groundwater velocity. In contrast, headwater regions of both Walnut and Squaw Creek watersheds are tile-drained. Most stream initiation points in both watersheds occur as tile outlets from headwater catchments, with first-order streams often beginning at road crossings with tile drainage discharging into a road culvert. In these tile-drained upland areas, land cover can have a proportionally large effect on water quality since subsurface water bypasses slow groundwater transport and is rapidly directed to streams via tiles. The effects of tile-drained headwater contributions on stream water quality were particularly evident in the Walnut Creek watershed where nitrate concentrations and loads were considerably higher in upper Walnut Creek than in lower Walnut Creek. Contributions from row-cropped headwater regions generally dominated the nitrate contributions from the lower portion of the watershed containing the restored prairie. In Squaw Creek, the rapid change in stream nitrate concentrations measured at subbasins SQW4 and SQW5 may have been facilitated by tile drainage, although the actual extent of tile drainage in these subbasins is unknown.

Overall, the mean residence time for groundwater in a groundwatershed (the average amount of time needed for groundwater to "turn over" in a groundwater catchment area) can be approximated by (Haitjema, 1995):

Formula 4[4]
where n is the aquifer porosity, H is the saturated aquifer thickness, and N is the areal recharge rate due to precipitation. For the Walnut Creek watershed n is assumed to be 0.3, H is estimated to be 6.1 m (20 feet), and N is equal to the long-term average baseflow in Walnut Creek, 129.5 mm (5.1 inches). The mean residence time for groundwater in Walnut Creek is estimated to be approximately 14 yr. Considerable uncertainty lies in the estimate of H, but it was derived as an approximation of the thickness of saturated loess and oxidized till in upland settings, and a midrange estimate of saturated alluvium in the floodplains. In actuality, H could be less than 3 m in sloping bluffs where pre-Illinoian till outcrops, or greater than 12 m in some floodplain settings. The residence time is essentially the mean of a cumulative frequency distribution of travel times in the watershed and would imply that, on average, 14 yr is needed for groundwater to drain from the watershed, with some groundwater draining faster to streams and some draining much slower. As discussed above, it is likely that the travel time for groundwater in floodplains is less than the 14-yr average, but the travel time for groundwater located in uplands is probably much greater than 14 yr. Hence, the amount of time needed to detect the water quality changes due to groundwater from all areas of the watershed is ultimately on the order of several decades.

A decadal timeframe for observing the effects of land management on groundwater nitrate concentrations was similarly noted by Tomer and Burkart (2003) at three agricultural catchments in western Iowa. In their study, groundwater nitrate concentrations in upslope landscape positions were still influenced by the legacy of past agricultural nitrogen management conducted more than 25 yr earlier. Thus, study results presented herein and those of Tomer and Burkart (2003) suggest that improvements in water quality may substantially lag behind changes in agricultural management.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
The Walnut Creek Monitoring Project began with an ambitious goal to implement a water quality program to document water quality improvements resulting from large-scale watershed restoration and management. Project results indicate that prairie restoration in an agricultural watershed can improve water quality with regards to nitrate concentrations and loads. Planting 25.4% of the Walnut Creek watershed in native prairie resulted in a reduction of nitrate of approximately 1.2 mg L–1 over 10 yr. Whereas this reduction cannot be considered overly substantial, it does shed light on the difficulty of detecting nitrate concentration changes in an agricultural setting where natural and anthropogenic nitrogen sources are ubiquitous. Upstream contributions from tile-drained, upland row crop areas had a significant effect on downstream water quality such that prairie restoration occurring in the core of the watershed primarily had the effect of diluting upstream nitrate contributions.

Nonetheless, native prairie restoration should be viewed as a viable conservation strategy for improving water quality in streams. Data from the Walnut Creek project extends the plot scale results (Randall et al., 1997; Brye et al., 2001) to a watershed scale to confirm that reintroduction of perennial grasses in the agricultural landscape can serve to reduce nitrate loss to streams. Project results highlighted the close relation of stream nitrate concentrations to land use change from row crops to grasslands. In Walnut Creek, converting row crop to grass reduced nitrate concentrations over time, but in Squaw Creek, stream nitrate concentrations rapidly increased when grasslands were converted back to row crop. This situation is analogous to historical conditions in Iowa that demonstrated how baseflow and stream nitrate concentrations increased in the 20th century as row crop acreage increased. Thus, it must be emphasized that grasslands, or other perennial vegetation placed in agricultural settings for water quality benefits should be part of a long-term solution to water quality problems if the water quality benefits are to be fully realized.

Early in the project, Schilling and Thompson (2000) wondered if the size of the Walnut Creek watershed was too large to detect water quality changes. Results suggest that water quality changes were greater and much easier to detect in small subbasins compared to the watershed outlets. Considering that Walnut Creek is a rather small 12 digit HUC in Iowa, project results should be kept in mind when expectations are raised for detecting water quality improvements from changing land use in larger watersheds. However, since all subbasins comprise part of larger and larger watershed areas, perhaps documenting improvements in stream water quality from conservation practices should be focused on small subbasins where changes can be detectable in shorter time frames. Detecting water quality improvements in larger watersheds will likely require a dedicated long-term monitoring effort on the order of several decades.

Finally, results from the Walnut Creek Monitoring Project attest to the necessity of conducting long-term monitoring to evaluate the effects of land use change and conservation practices on water quality. Lag times for observing water quality improvements are rarely less than several years long, and lag times of decades are the norm rather than the exception. In the Walnut Creek watershed, it took more than 3 yr of monitoring before the first statistically significant changes in nitrate concentrations were detected. Long term monitoring is needed to factor out the effects of climate and account for possible improvements in water quality that may take many years to show up in a stream. Many watershed conservation projects funded to improve water quality in streams often claim success but lack accountability for measuring actual water quality benefits. Without monitoring, potential water quality improvements from watershed projects may go unnoticed and unappreciated as the public debates the costs and benefits of programs designed to reduce NPS pollution in streams. Moreover, as the Walnut Creek project results demonstrated, monitoring will also establish realistic expectations for success to educate the public that solutions to NPS pollution problems are not easy and are not quick.


   ACKNOWLEDGMENTS
 
The Walnut Creek Nonpoint Source Monitoring Project was supported in part by Region VII of the U.S. Environmental Protection Agency through a 319-Nonpoint Source Program grant to the Iowa Department of Natural Resources. Additional support was provided by U.S. Department of Agriculture Cooperative State Research, Education, and Extension Service grant 2004-51130-03120. The Neal Smith National Wildlife Refuge contributed greatly to the success of the project.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES