Groundwater loading of nitrate-nitrogen and phosphorus from watershed source areas to an Iowa Great Lake

Journal of Great Lakes Research 42 (2016) 588–598

Contents lists available at ScienceDirect

Journal of Great Lakes Research journal homepage: www.elsevier.com/locate/jglr

Groundwater loading of nitrate-nitrogen and phosphorus from watershed source areas to an Iowa Great Lake Keith E. Schilling a,⁎, Matthew T. Streeter a, Deborah Quade b, Mary Skopec b a b

Iowa Geological Survey, University of Iowa, Iowa City, IA, United States Iowa Department of Natural Resources, Des Moines, IA, United States

a r t i c l e

i n f o

Article history: Received 4 December 2015 Accepted 21 March 2016 Available online 18 April 2016 Communicated by Harvey Thorleifson Index words: Lake Groundwater Nutrients Iowa Okoboji Mass balance

a b s t r a c t Groundwater discharge to a lake can be an important component to water and nutrient budgets. In this study, we evaluated groundwater loading of nitrate-nitrogen (NO3-N) and phosphorus (P) to West Lake Okoboji, Iowa, using a watershed-based approach based on groundwater recharge and land cover class. Our objectives were to assess groundwater level fluctuations and nutrient concentrations under representative land use classes and develop an allocation model for groundwater nutrient loads based on land cover class. Monitoring wells were installed at 21 locations around the lake and sampled during a three-year study period. Groundwater quality varied among the land cover types with average NO3-N concentrations the highest beneath cropped fields (8.8 mg l−1) and residential areas (2 mg l−1), and P concentrations ranging between 0.05 and 0.1 mg l−1 throughout the region. NO3-N loads were the highest under cropped fields and this source accounted for approximately 90% of the NO3-N, whereas P loads were more evenly distributed among source areas. Groundwater recharge averaged approximately 76 mm year−1 for vegetated areas and substantially less for urban areas. Based on mass balance, groundwater discharge may account for 80% of the NO3-N in the lake compared to 10% of the P. Results are instructive to more effectively target implementation of conservation practices to major nutrient loading areas for reduction of NO3-N and P delivered to the lake. © 2016 International Association for Great Lakes Research. Published by Elsevier B.V. All rights reserved.

Introduction Groundwater discharge to a lake (also referred to as “exfiltration”) is an important, if often underappreciated, component of water and nutrient budgets. Recent reviews by Rosenberry et al. (2015) and Lewandowski et al. (2015) identified 13 different hydrologic and nutrient-related reasons for why groundwater discharge to lakes tends to be neglected, among which heterogeneities in spatial and temporal discharge of groundwater and nutrient concentrations to lakes were commonly cited. Quantifying nutrient discharge to lakes is critical considering that eutrophication is a major threat to lake ecosystems (Smith, 2003; Conley et al., 2009). New approaches for quantifying groundwater contributions of nitrogen (N) and phosphorus (P) to lakes are needed (Meinikmann et al., 2013; Robinson, 2015). Methods of quantifying groundwater nutrient flux to lakes tend to rely on groundwater observation wells near lake shorelines and application of Darcy's Law (LaBaugh et al., 1997; Harvey et al., 2000), seepage meters and minipiezometers (Lee and Cherry, 1978; Shaw and Prepas, 1990), modeling (Anderson and Munter, 1981; Hunt et al., 2003) or a combination thereof (Simpkins, 2006). Unfortunately, these methods

⁎ Corresponding author. E-mail address: [email protected] (K.E. Schilling).

often produce significantly different results, due in part, to differences in measurement scales (Simpkins, 2006). In a study of Lake Arendsee in Germany, Meinikmann et al. (2013) combined well measurements with an integrative groundwater recharge calculation to estimate groundwater P loads to the lake. In their approach, the mean annual lacustrine groundwater discharge to a lake was equal to mean groundwater recharge in the watershed and direct groundwater discharge to the lake was established using lake bed temperature profiles (Meinikmann et al., 2013). Indirect groundwater discharge occurs to the Great Lakes through baseflow to tributary streams that discharge into the lakes (Kornelsen and Coulibaly, 2014; Grannemann et al., 2000). A method of estimating groundwater loads discharged to lakes based on recharge and land use holds promise for many lakes in Iowa where lake water quality is known to be influenced by agricultural activities occurring in the lake watershed (Arbuckle and Downing, 2001). West Lake Okoboji located in northwest Iowa (Fig. 1) is considered one of Iowa's premier tourist destinations, attracting more than 1,000,000 annual visitors and supporting abundant recreational boating and fishing opportunities (http://www.vacationokoboji.com/). The 15.57 km2 lake is part of a chain of lakes known as the Iowa Great Lakes, consisting of Spirit Lake, East Lake Okoboji, Upper and Low Gar and Minnewashta lakes. Over the past 15 years, concentrations of NO3-N nitrogen and total phosphorus in West Lake have ranged from b0.06 to 0.22 mg l− 1 and b0.02 to 0.08 mg L−1, respectively (IDNR,

http://dx.doi.org/10.1016/j.jglr.2016.03.015 0380-1330/© 2016 International Association for Great Lakes Research. Published by Elsevier B.V. All rights reserved.

K.E. Schilling et al. / Journal of Great Lakes Research 42 (2016) 588–598

589

Fig. 1. Location of West Lake Okoboji, Iowa, monitoring well installations and regional baseflow sites.

2015). Although not particularly elevated compared to other Iowa lakes (IDNR, 2015), it is the ratio of N and P that influences algae species composition, productivity and trophic status (Arbuckle and Downing, 2001; Downing and McCauley, 1992). Hence, it is important that sources of N and P to West Lake Okoboji be quantified so that future eutrophication threats to the lake and the resulting impacts on the regional economy can be mitigated. In the case of West Lake Okoboji, efforts directed at quantifying groundwater nutrient flux using point measurements near the lake shoreline (i.e., wells, seepage meters) would be prone to substantial error. As the deepest natural lake in Iowa, reaching a maximum depth of 41.4 m, the lake penetrates a thick sequence of heterogeneous glacial sediments in the subsurface that contribute an unknown amount of groundwater to the lake. It would be exceedingly difficult and expensive to intercept potential groundwater flow paths where they discharge (or exfiltrate) into the lake. Furthermore, groundwater flow paths in the watershed are complex and influenced by many factors such as subsurface tile drainage and ponds/wetlands, and N and P are often transformed along flow paths due to biogeochemical processes.

To address these challenges, our study focused on quantifying the contribution of nutrient concentrations and loads from various land cover source areas in the watershed. Using a watershed-based approach, we quantified the groundwater loading of NO3-N and P by land cover class to assess the potential for these land areas to contribute nutrient loads delivered to West Lake Okoboji through groundwater recharge. The objectives of this study were to: 1) assess groundwater level fluctuations and recharge in the West Lake Okoboji watershed; 2) quantify groundwater nutrient concentrations under representative land use classes; and 3) develop a load allocation model for groundwater nutrient inputs to the lake that can be used to target future nonpoint source load reduction strategies. Hydrogeological setting Iowa Great Lakes Region (IGLR) is located along the southwestern edge of the Des Moines Lobe (DML) ice sheet, the southernmost extension the Laurentide Ice Sheet that surged into Iowa approximately 15,000 years ago (Bettis et al., 1996). The initial advance of the DML

590

K.E. Schilling et al. / Journal of Great Lakes Research 42 (2016) 588–598

ice stagnated across the landscape and was followed by several readvances over the next 3000 years. The resulting landscape in the Iowa Great Lakes Region is due to younger ice advances overlapping older ice advances; thus creating the “knob and kettle” terrain associated with landscapes that form in direct contact with slowly disintegrating ice. The Iowa Great Lakes Region served as an important drainage outlet for the Des Moines Lobe (DML). The Okoboji Lake Outlet was an important drainage outlet along the northwestern edge of the DML ice sheet. This outlet now drains an interconnected system of lakes and sloughs, including West Okoboji Lake and drains into the Little Sioux River. Extensive sand and gravel deposits provide ample evidence of the huge volumes of water and sediment that were discharged by melting ice during late glacial times. All of the lakes mentioned above owe their existence to the disintegration of glacial ice during Late Wisconsin time. Upland Wisconsin-age glacial deposits are classified as the Dows Formation: two glacial till members (Morgan and Alden) are the dominant members of the formation (Fig. 2). Both members are loam in texture and considered matrix-dominated tills. However, the supraglacial Morgan Member is highly variable in texture and may contain channels of sorted sands, silts, silty clays, and gravels. Recent mapping studies on the DML have documented a landscape dominated by supraglacial landforms and sediment assemblages near the land surface (upper 3 to 6 m). The Morgan Member overlies the homogeneous and denser basal till of the Alden Member. The extensive gravels associated with the Okoboji Outlet and the Little Sioux River are classified as glacial outwash of the Noah Creek Formation while fine textured lake and paludal sediments associated with the region comprise various members of the DeForest Formation (Bettis et al., 1996; Quade et al., 2012). The West Okoboji Member of the DeForest Formation (Fig. 2) represents the organic-rich silt deposits associated with lake sedimentation. These deposits interfinger with sandy and gravelly sediments (Triboji Member) associated with ice-push ridges (ridges resulting from the pushing action of a lake's ice sheet against the shoreline). Due to the contrast in permeability between the coarser-textured Morgan and Triboji members and the underlying denser Alden Member, lateral groundwater flow in the upland deposits is primarily concentrated in the upper units. Test drilling for the project indicated that the Morgan Member is 3 m to greater than 6 m thick in the upland areas; whereas near the lakeshore, the Triboji Member is typically 2–3 m thick and often overlies the Morgan and Alden Members. At depth, well logs indicated more than 50–70 m of Dows Formation, pre-Late Wisconsin and pre-Illinoian diamicton overlying bedrock consisting of mid-Cretaceous age Dakota Formation comprised of the lower dominated sandstone Nishnabotna Member

and an upper shale and mudstone dominated Woodbury Member (Witzke et al., 2010). Groundwater flow into West Lake Okoboji occurs through direct discharge of groundwater through the Triboji and Morgan members and in baseflow discharged to small perennial streams that flow into the lake. Baseflow discharge is supplemented in many areas by discharge from drainage tiles that underlie many agricultural fields in the area. Row crop fields in the poorly drained soils of the Des Moines Lobe are often artificially drained with subsurface drainage tiles (Schilling et al., 2015). These tile systems are typically installed at a spacing of approximately 24 m and at depths of approximately 1.2 m (see Singh et al., 2006). Although precise locations of field tiles in the West Lake Okoboji watershed are unknown, larger tile drainage systems organized into managed drainage districts are located in the northern portion of the watershed (Fig. 3). Subsurface flow into smaller field tiles and larger drainage district tiles that discharge into streams or directly into the lake provides a rapid conduit for groundwater transport and discharge into the lake. Methods and materials The West Lake Okoboji watershed boundary and 2009 high resolution land cover maps were obtained from the Iowa Department of Natural Resources (IDNR) Geographic Information System (GIS) library (https://programs.iowadnr.gov/nrgislibx/). Land cover classes were grouped into major land cover classifications, namely crops (corn and soybeans), urban areas (including roads, buildings, parking lots, etc.), golf courses, residential areas (including houses and yards), and perennial vegetation (grass and forest). Another land cover type (bioswale) was added to the water quality assessment to account for these urban best management practices (BMPs) which are becoming increasingly utilized in the region to reduce impact from stormwater runoff. However, bioswales were not included in the watershed-scale groundwater load allocation model because the areas draining these BMPs were not large enough (b 10 ha) to be comparable to the other classifications. Monitoring wells were installed at 21 locations in the West Lake Okoboji watershed (Fig. 1) to assess groundwater conditions under representative land cover types in the watershed in a variety of landscape positions (upland, mid-slope, lakeshore). The water table wells screened either the Morgan Member (upland and mid-slope) or Triboji Member (lakeshore). All wells were installed using a truck-mounted, hollow-stem drilling rig. Well depths varied from approximately 3 m near the shoreline, 4.5 m in mid-slope positions and 6 m in upland landscape positions. A 1.5 m long factory-slotted PVC well screen and a solid

Fig. 2. Conceptual model of West Lake Okoboji area stratigraphy.

K.E. Schilling et al. / Journal of Great Lakes Research 42 (2016) 588–598

591

Fig. 3. Land use map of West Lake Okoboji watershed. Drainage district tile lines are larger regional systems and should not be confused with tiles in individual fields. Locations of field tiles are unknown but are suspected to be in every row crop field to some degree.

PVC riser were installed in the borehole. A silica sand filter pack was poured around the screen, bentonite chips were added to provide a seal and drill cuttings were backfilled in the rest of the borehole. Water levels in all of the wells were measured on an approximate monthly basis by a local volunteer using an electronic water level indicator. Five wells were further equipped with a pressure transducer to record hourly water level fluctuations during the study period. Four of the wells were equipped with vented miniTROLL transducers (30 psi, In-Situ, Inc) and one well (OK1) was equipped with a non-vented model (30 psi absolute pressure miniTROLL, In-Situ, Inc.). A baroTROLL (In-Situ, Inc.) was used to correct the absolute pressure to atmospheric conditions. Regional climate was monitored daily at a weather station in Spirit Lake, Iowa, located approximately 5 km north of the lake. Daily

precipitation was downloaded from the Iowa Environmental Mesonet (http://mesonet.agron.iastate.edu/). Baseflow (BF) is often considered a proxy for diffuse recharge in watersheds with gaining streams (Scanlon et al., 2002; Risser et al., 2005) and we used a simple water budget method to estimate recharge in the West Lake Okoboji watershed. The hydrological equation for a watershed can be written as P ¼ ET þ Q þ ΔS

ð1Þ

where P is the amount of precipitation, ET is the evapotranspiration, Q is the stream discharge, and ΔS is the change in storage, considering that the aquifer is unconfined and that there is no cross-basin water transfer. Over long periods of time, the change in storage is minimum, and the

592

K.E. Schilling et al. / Journal of Great Lakes Research 42 (2016) 588–598

equation can be simplified to P = ET + Q. A similar water balance equation can be written for the groundwater system in the watershed, R ¼ BF þ Q P þ ΔSG

ð2Þ

where R is the net recharge to groundwater, i.e., the recharge minus evapotranspiration from the water table, BF is the baseflow, Q P is due to well pumping, and ΔSG is the change in the groundwater volume. Assuming negligible well pumping and change in groundwater storage, R = BF. Scanlon et al. (2002) noted that pumping, ET and underflow to deeper aquifers should be considered when equating Qb to R. In the case of West Lake Okoboji watershed, there is little pumping of the water table aquifer and underflow to glacial till is negligible. Baseflow has been used to estimate the indirect groundwater contribution to the Great Lakes (Holtschlag and Nicholas, 1998; Neff et al., 2005). In our case, since the IGLR is located in the Little Sioux River watershed, we evaluated baseflow calculated at two U.S. Geological Survey gage sites to provide an estimate of net groundwater recharge in the watershed (Fig. 1). The Little Sioux near Spencer (#06604440) is located near the outlet of the IGLR but the gaging record only extends back to 2010. Moreover, a regional rural water system is located within two km of the USGS gage site at Spenser. Although this well field is tapping a deeper Quaternary aquifer that is not considered hydraulically connected to the Little Sioux alluvium (Thompson, 1986), there is potential for well field influence on baseflow levels. For this reason, we used the Little Sioux River at Correctionville (#06606600) to estimate groundwater recharge for the IGLR for a longer 25 year period from January 1990 to December 2014. The hydrogeology of the Little Sioux River watershed is consistent across the extent of the watershed, consisting of windblown silt (loess) overlying fine-textured glacial till in the uplands and coarse-textured alluvial deposits found along the modern day Little Sioux River valley (Thompson, 1986). Unconfined groundwater flow is directed toward the Little Sioux River with little interaction with deeper groundwater residing in deeper glacial till or Cretaceous bedrock located approximately 80 m below the land surface. There are no surface water diversions or reservoirs in the river. The USGS PART program (Rutledge, 1998) was utilized to generate baseflow separation values from the streamflow data. The PART program uses streamflow as an input array, searches the array for values that fit an antecedent recession equation, and sets baseflow equal to streamflow as long as the value is not followed by a daily decline of more than 0.1 log cycle (Rutledge, 1998). If the specified daily decline does occur, another formula is used to determine baseflow values. Resulting baseflow values were summarized in units of mm over the entire study area. Water samples were collected from the monitoring wells on 10 occasions during the 2012 to 2014 study period. However, due to drought in the region during the study period, some wells were periodically dry and water samples could not be collected. In two cases, the wells were dry for the entire study period. Water levels in wells were measured to the nearest 0.01 ft (3 mm) at the time of sampling. Water samples were collected using a peristaltic pump and analyzed in the field for temperature, specific conductance (SC), pH, dissolved oxygen (DO) and oxidation–reduction potential- (ORP) using a YSI Model 556 water quality meter. Measurements were made within ±0.1 C, ±0.2 pH units, ±0.1%, ±0.2 mg/l and ±20 mv for temperature, pH, SC, DO and ORP, respectively. Water samples for laboratory analysis were field filtered through a 0.45 μm glass fiber filter and transported on ice to the State Hygienic Laboratory, in Coralville, IA. Samples were analyzed for NO3-N nitrogen and total dissolved phosphorus (TDP) according to EPA Method 300.0 and 4500-P, respectively. A load allocation model was developed for West Lake Okoboji by combining the groundwater recharge estimates with the measured groundwater nutrient concentrations. The term “load allocation” is used by the U.S. EPA total maximum daily load (TMDL) program to

designate the sum of all the nonpoint source loads contributing to an impaired waterbody. Although West Lake Okoboji is not impaired for nutrients, we adopted the terminology to best describe the allocation of NO3-N and TDP loads among the land cover classifications. Mean nutrient concentrations (mg/l) were multiplied by mean annual recharge (in mm) to estimate the annual nutrient loading rate per ha (kg/ha or “yield”) for each land cover class. Multiplying the yield by the total area of the land cover class provided an estimate of the total mass (in Mg for NO3-N and kg for P) delivered to the lake from nonpoint sources. Results Watershed land cover classification Land cover in the West Lake Okoboji watershed is dominated by annual crops of corn and soybeans (2497 ha) and perennial vegetation (2046 ha), particularly in the western and northern portions of the basin (Fig. 3). The eastern side of the lake contains much of the commercial area (427 ha), whereas residential areas (764 ha) ring the lakeshore. The lake is a tourist destination and four golf courses exist in the watershed (129 ha). Hydrology Annual precipitation measured in Spirit Lake, Iowa, during the 2012–2014 study was below normal. Annual totals were lower in 2013 (492 mm) and 2012 (562 mm) compared to 2014 (620 mm), but all three years were well below the 30-year average for the region (752 mm). Lower amounts of local precipitation in 2012 mirrored regional patterns of moderate to severe drought which occurred in the latter half of 2011 and all of 2012 (Ikenberry et al., 2014). However, while much of Iowa received normal rainfall amounts in 2013, the Okoboji area remained exceedingly dry, and only a wet June in 2014 (261 mm; Fig. 4) brought the 2014 total amounts to those approaching normal conditions. Water table depths varied considerably during the monitoring period (Figs. 4 and 5). At a monthly measurement frequency, representative upland, midslope and lakeside wells showed a range of water table fluctuation that related to their landscape position (Fig. 4). All the wells showed seasonal patterns of increasing in the spring and early summer with precipitation inputs and decreasing in mid-summer through fall in

Fig. 4. Monthly precipitation (P) patterns and groundwater level fluctuations in three wells representative of upland midslope and lakeside landscape positions. Monthly baseflow measured at the Little Sioux River at Correctionville (USGS station #06606600) is provided.

K.E. Schilling et al. / Journal of Great Lakes Research 42 (2016) 588–598

593

Fig. 5. Continuous water table measurements from four wells in West Lake Okoboji region beneath a) urban pavement, b) lakeshore, c) upland perennial grass, and d) upland crop.

response to evapotranspiration. Near the lake, water table depths were also controlled by the lake level and fluctuated over a narrow range of 0.67 m. In contrast, wells located mid-slope and in upland areas fluctuated considerably more, with a maximum fluctuation of 1.96 m and 4.02 m, respectively (Fig. 4). Continuous water level measurements revealed similar seasonal patterns of fluctuation in lakeside (Fig. 5b) and upland wells (Fig. 5c, d). However, the water table beneath pavement in the urban land cover did not exhibit any seasonal trends and showed a narrower range of fluctuation (0.47 m). Much of the fluctuation in water table depths may be attributable to the flush-mount well location in an active parking area where vibrations and disturbances may have influenced the transducer readings. It was clearly evident that the water table pattern under pavement was very different than the other sites. The lack of seasonal water table fluctuation suggests little groundwater recharge occurring beneath paved areas. Groundwater recharge in the watershed was estimated using average annual baseflow in the Little Sioux River which receives drainage from the IGLR. Baseflow separation performed for two locations in the Little Sioux River watershed indicated exceptionally dry conditions during the 2012–2014 study period. At the Little Sioux River near Spencer, baseflow averaged 57.2 mm for the three years, whereas at the site further downstream at Correctionville, baseflow averaged 76.5 mm. Baseflow at both sites was highly correlated with each other (r = 0.89; p b 0.05). Because of the possible influence of well pumping near the Spencer gage and the longer record available at Correctionville, we used baseflow from Correctionville to characterize diffuse groundwater recharge in the West Lake Okoboji watershed. Monthly baseflow in the Little Sioux River ranged from b 10 mm in December 2013 to 41.1 mm in June 2013, and totaled 63.7 mm in 2012, 90.4 mm in 2013 and 75.4 mm in 2014. Baseflow exhibited strong seasonality and was significantly correlated with water table fluctuations at the three landscape positions at a monthly scale (r = 0.73 to 0.79; p b 0.05). Significant correlation of baseflow in the Little Sioux River to water table levels in the West Lake Okoboji watershed supports the use of baseflow to estimate groundwater recharge in the watershed (Schilling, 2009). For the urban pavement site, baseflow was not significantly correlated with water table fluctuations (p N 0.1). Since the water table beneath pavement only fluctuated 16% as much as upland

wells, we assumed that groundwater recharge beneath urban commercial areas was less than the watershed-wide recharge. For purposes of estimating nutrient loading rates, we assumed that groundwater recharge beneath urban pavement was 16% that of the rest of the watershed. Water quality Groundwater quality varied among the land cover types sampled in this study. NO3-N concentrations were the highest beneath cropped fields, ranging from 1 to 17 mg L−1 and averaging 8.8 mg L−1 among the wells and sampling events (Fig. 6). Average concentrations were higher in 2012 (11.2 mg L− 1) compared to 2013 (7.3 mg L− 1) and 2014 (6.3 mg L−1) as the drought persisted in the area (Table 1). NO3N concentrations in residential areas and urban pavement were similar (~2 mg L−1) whereas average concentrations beneath perennial grass (b 0.1 mg L−1) and a golf course (0.3 mg L−1) were lower. Although the average NO3-N concentrations beneath residential areas was 2 mg L−1, the median concentration among four locations was 0.8 mg L− 1 (Fig. 6). Concentrations were uniformly higher at one

Fig. 6. Box and whisker plot and statistical summary of groundwater NO3-N concentrations measured at various land cover classes during the 2012–2014 monitoring period.

594

K.E. Schilling et al. / Journal of Great Lakes Research 42 (2016) 588–598

Table 1 NO3-N and TDP concentrations and load allocation by land use category to West Lake Okoboji. NO3-N (mg L−1 l) Area (ha) Urban Golf course Perennial Residential Crop Total

427 129 2046 746 2497

Groundwater recharge (mm)

Loading rate (kg/ha)

Groundwater load (Mg)

Proportion of total GW load (%)

2012

2013

2014

2012

2013

2014

2012

2013

2014

2012

2012

2013

2014

0.52 0.10 2.11 11.23

0.25 0.33 0.05 2.56 7.31

3.15 0.17 0.06 1.66 6.28

10.2 64 64 64 64

14.4 90 90 90 90

12 75 75 75 75

0.00 0.33 0.07 1.35 7.19

0.04 0.30 0.05 2.31 6.58

0.38 0.13 0.05 1.24 4.71

0.0 0.2 0.7 5.3 93.8

0.1 0.2 0.5 9.4 89.8

1.2 0.1 0.5 5.1 9

P concentration (mg/l) Urban Golf course Perennial Residential Crop Total

0.053 0.077 0.100 0.088

0.050 0.037 0.060 0.073 0.049

0.050 0.053 0.116 0.300 0.063

0.00 0.04 0.13 1.01 17.95 19.13

2013 0.02 0.04 0.09 1.72 16.43 18.30

2014 0.16 0.02 0.09 0.93 11.75 12.95

Groundwater recharge (mm)

Loading rate (kg/ha)

Groundwater load (kg)

Proportion of total GW load (%)

10.2 64 64 64 64

0.000 0.034 0.049 0.064 0.057

0.0 4.4 100.8 47.7 141.2 294.1

0.0 1.5 34.3 16.2 48.0

14.4 90 90 90 90

12 75 75 75 75

older residential neighborhood (5.3–7.2 mg L−1), but we observed fluctuations at all residential areas (b 0.1 to 6.7 mg L−1). Average NO3-N concentrations in a bioswale were similar to residential areas. In contrast to NO3-N, TDP concentrations did not exhibit much variability (Fig. 7), with average concentrations ranging between 0.05 and 0.1 mg L−1 among all land covers. The highest TDP concentrations measured in the study were found in the bioswale well. In this well, TDP concentrations were particularly high in late 2014 (1.6 to 2.9 mg L−1), but before this time were consistently found between 0.11 and 0.55 mg L−1. TDP concentrations measured in residential wells averaged 0.1 mg L−1, but ranged up to 0.67 mg L−1in one well. Concentrations beneath cropped areas and a golf course were consistently less than 0.1 mg L−1. Other water quality parameters did not vary systematically with land cover and were grouped together (Table 2) to compare to nutrient concentrations. NO3-N concentrations were positively correlated with depth to water (r = 0.43) and dissolved oxygen concentrations (r = 0.46), and negatively correlated with specific conductance (r = −0.28). TDP concentrations were negatively correlated with ORP (r = −0.29) but not significantly correlated with any other field measurements (r b 0.12; p N 0.1). Load allocation Groundwater loads in the West Lake Okoboji watershed were apportioned by land cover classification and year (Table 1). NO3-N yields and loads were the highest under cropped fields and this land cover class accounted for approximately 90 to 94% of the NO3-N present in shallow groundwater in the watershed. However, due to the drought

0.007 0.033 0.054 0.066 0.044

0.006 0.040 0.087 0.225 0.047

3.1 4.3 110.5 49.2 109.6 276.6

2.6 5.2 178.4 167.9 117.7 471.7

1.1 1.5 39.9 17.8 39.6

0.5 1.1 37.8 35.6 25.0

conditions, NO3-N yields from cropped fields were not particularly high, ranging from 4.7 to 7.2 kg ha−1. NO3-N contributions from residential areas accounted from 5 to 9%, whereas other sources were less than 0.7% (Table 1). The potential contribution from residential areas would be lower if median concentrations (0.8 mg/L) were used instead of mean values (2.4 mg/l). In this case, the residential contribution would be approximately 2–3% and the contribution from cropped fields would range from 94 to 97%. Using average concentrations and assuming that all shallow groundwater sources discharged into the lake, groundwater sources could conceivably supply approximately 17 metric tons (Mg) of NO3-N to the lake per year. Groundwater loads of TDP were more evenly distributed among source areas (Table 1). TDP concentrations did not vary as much as NO3-N so groundwater P loads were related more to land areas of the various land covers. Distribution of TDP from crops and perennial areas accounted for approximately 37% and residential areas accounted for 23% of the shallow groundwater P loads. On the other hand, TDP loads produced from commercial areas and golf courses were less than about 1%. Assuming all shallow groundwater discharged into the lake, groundwater sources could contribute approximately 350 kg per year of TDP to the lake per year. It is important to note that our study period occurred during three years of below normal rainfall in the region. Average groundwater recharge during the study (76 mm) is among the lowest rates of groundwater recharge (baseflow) experienced in the region over the last 25 years (Fig. 8). Consequently, the groundwater loads of NO3-N and TDP were also low relative to historic conditions. Given the average groundwater concentrations associated with the land cover classes measured in this study, groundwater NO3-N and TDP loads may have reached 104 Mg and 1740 kg, respectively, during an exceptionally wet year (Fig. 8). However, in most years, NO3-N and P annual loads tend to be less than approximately 40 Mg and 700 kg, respectively. Discussion Groundwater sources of lake nutrient loads

Fig. 7. Box and whisker plot and statistical summary of groundwater dissolved phosphorus concentrations measured at various land cover classes during the 2012– 2014 monitoring period.

Study results suggest that groundwater sources of NO3-N and TDP can be important nutrient sources to West Lake Okoboji. Greater groundwater loading rates of NO3-N than TDP is consistent with the primary method of transport for each nutrient. In Iowa and throughout the glaciated Midwest, it is well understood that NO3-N is leached through soils and delivered to rivers and streams through groundwater discharge as baseflow and tile drainage from row crop fields (e.g., Schilling and Libra, 2000; Schilling and Zhang, 2004; David et al., 2010, Robertson and Saad, 2013). On the other hand, most P is delivered to rivers from agricultural runoff (Jacobson et al., 2011) and from stream bank erosion (Zaimes et al., 2008; Palmer et al., 2014). Although most P

K.E. Schilling et al. / Journal of Great Lakes Research 42 (2016) 588–598

595

Table 2 Groundwater quality conditions in the West Lake Okoboji watershed.

n Mean Median Stdev Min Max

Temperature (deg C)

pH

Specific cond. (μmhos cm−1)

Dissolved oxygen (mg L−1)

Oxidation reduction potential (mv)

NO3-N (mg/L)

Dissolved phosphorus (mg L−1)

133 13.16 13.91 5.53 −0.30 22.77

133 6.30 6.31 0.36 5.56 7.79

133 821 759 253 498 1689

133 3.25 2.42 2.30 0.72 10.44

129 63 88 88 −294 195

133 2.34 0.41 3.38 0.05 17.00

132 0.15 0.06 0.31 0.01 2.90

in runoff is sediment bound (Jacobson et al., 2011), dissolved P delivery to rivers and lakes can also be significant in some areas (Smith et al., 2015). Hence, groundwater loading patterns are consistent with these nutrient transport mechanisms indicating greater NO3-N loads in shallow groundwater compared to P. More than 90% of the groundwater NO3-N load in the watershed was associated with row crop areas. This percentage is consistent with nonpoint source contributions of NO3-N assessed in other watersheds. For example in the Raccoon River, 92% of the NO3-N load is derived from nonpoint source agriculture (Jha et al., 2010) and in the Des Moines River, the percentage was even higher (95%, Schilling and Wolter, 2009). Indeed, the relation between NO3-N concentrations in Iowa streams to row crop land use is so strong that it is possible to approximate the average annual NO3-N concentration in surface water by multiplying the watershed's row crop percentage by 0.1 (Schilling and Libra, 2000). However, the average NO3-N concentration beneath cropped areas (8.8 mg L−1) and the annual NO3-N yields (4.7 to 7.2 kg ha−1) were lower than expected. In similar tile drained areas on the Des Moines Lobe, NO3-N concentrations in tile water and streams routinely exceed 20 mg L−1 (Ikenberry et al., 2014; Tomer et al., 2003) and yields commonly exceed 20 kg ha−1 (Ikenberry et al., 2014; Jha et al., 2010; Schilling and Wolter, 2009). We suspect that the lower concentrations and yield measured in this study are due to the drought conditions that persisted during the project. In the study by Ikenberry et al. (2014) that overlapped in time with this project, they reported NO3-N yields for three central Iowa drainage districts during the 2012 drought year to be 5.5 kg ha−1, which compares favorably to yields estimated in this study for the West Lake Okoboji watershed. However, it is worth noting that in the year following the 2012 drought, water yield in the central Iowa watersheds increased from 31 to 223 mm, NO3-N concentrations averaged 32 mg L−1 and NO3-N yields increased to 40 kg ha−1 (Ikenberry et al., 2014). Hence, it should be expected that NO3-N concentrations and loads from cropped areas in the Okoboji area will similarly increase coming out of the drought. Increasing groundwater recharge combined with the release of stored N in soils may produce NO3-N loads from cropped fields that greatly exceed the estimates reported for this project period. In this case, the proportion of NO3-N

Fig. 8. Groundwater recharge measured during the 2012–2014 study period compared to long-term patterns (25 years) with implications for NO3-N and P loading rates to the lake.

from cropped fields would be expected to increase and so would the concentrations of NO3-N in the lake water. For example, increasing annual groundwater recharge to 200 mm and NO3-N concentrations to 20 mg L−1 in a wet year would result in a load to West Lake Okoboji of more than 100 Mg. Although our study focused on annual loading rates, seasonal patterns of greater recharge in the spring combined with higher NO3-N concentrations would concentrate losses of NO3-N in some months more than others. In particular, monthly NO3-N concentrations and yields from tile-drained croplands are typically highest in April to June (Ikenberry et al., 2014), and this time period coincides with rapidly increasing water table levels in the watershed (Fig. 4). Other sources of NO3-N, while less significant than crop sources, are nonetheless informative. Residential areas provided a secondary source of NO3-N to the lake (5–9%). Groundwater sources of NO3-N in residential areas includes leaky sewers, septic tanks and urban fertilizer applications, among other sources (Wakida and Lerner, 2005). Consistent with these sources, specific conductance was notably higher under urban and residential areas (~ 930 μmhos cm− 1) compared to crop areas (~ 630 μmhos cm− 1) and may indicate the presence of other subsurface contaminants in addition to NO3-N (e.g., chloride). Golf courses were not a significant source of NO3-N, either as yield kg ha−1 (b 0.3) or as a groundwater contribution to the lake (b 0.2%). This is consistent with research that suggests that concentrations of nutrients discharged from well-maintained turf are well below levels of major concern (King et al., 2007). For example, Cohen et al. (1999) conducted a review of published and unpublished studies conducted on golf courses throughout the U.S. and found that of 849 individual groundwater samples made available to the study, the average and median NO3-N concentrations were 1.6 and 0.45 mg L−1, respectively. The median concentration reported for the U.S. is similar to the concentrations measured in the golf course at West Lake Okoboji (0.3 mg L−1). Finally, perennial vegetation, including grasslands and forest, had very low groundwater NO3-N concentrations and yields consistent with groundwater studies of prairie systems (Kemp and Dodds, 2001; Schilling and Wolter, 2001). For example, Schilling and Wolter (2001) reported that concentrations of NO3-N in baseflow (i.e., groundwater seepage) were b1 mg L− 1 in subwatersheds draining prairie areas compared to concentrations N 8 mg L− 1 in subwatersheds draining predominantly row crop areas. Commercial areas were not a major source of either NO3-N or P to the lake primarily because of a lack of groundwater recharge. Because the well hydrograph beneath pavement did not fluctuate seasonally like groundwater in other wells, we estimated that urban areas under pavement received only 16% as much groundwater recharge as vegetated areas. In urban areas, infiltration is essentially sealed off due to the impervious cover (Hamel et al., 2013; Schoonover et al., 2006). Hydrologic alteration of streamflow hydrographs becomes noticeable when impervious cover exceeds 10% of the drainage basin (Booth and Jackson, 1997) and groundwater recharge is reduced (Jeong et al., 2014). Our results from West Lake Okoboji differ from those of Meinikmann et al. (2013) who reported that urban areas contributed a large portion of the lacustrine groundwater discharge (LGD) to a lake in Germany. They estimated groundwater recharge using an evapotranspiration (ET) method and treated urban areas as grasslands with

596

K.E. Schilling et al. / Journal of Great Lakes Research 42 (2016) 588–598

only 30% of the area being sealed. Although small grassy areas are present in some commercial areas of West Lake Okoboji watershed, they are primarily found in the residential areas considered in our study. Hence, the commercial (urban) areas evaluated in our study largely consisted of paved, sealed areas with little to no groundwater recharge. Groundwater P loading to West Lake Okoboji was more evenly distributed among source areas since TDP concentrations did not vary substantially beneath different land covers (Table 1). Groundwater P concentrations were generally between 0.05 and 0.1 mg L−1 throughout the region, although concentrations beneath residential areas in 2014 were higher (0.3 mg L− 1). The groundwater P concentrations measured in the Okoboji area are typical of concentrations found in Iowa groundwater (Burkart et al., 2004). Dissolved P concentrations averaged 0.074 to 0.09 mg L−1 in fractured Wisconsin till (Dows Formation) and 0.093 mg L− 1 in a northwest Iowa alluvial aquifer (Burkart et al., 2004). Groundwater concentrations at West Lake Okoboji are lower than those found in groundwater discharging to Clear Lake in north-central Iowa (Simpkins et al., 2001). Clear Lake is Iowa's third largest glacial lake but has suffered from nutrient enrichment as lake P concentrations increased from 0.06 to 0.19 mg L− 1 from the 1970's to 2000 (Burkart et al., 2004). Median groundwater TDP concentrations from outwash sands and gravels discharging into Clear Lake (0.16 mg L−1) are more than double median groundwater levels discharging into West Lake Okoboji (0.06 mg L−1; Table 2). Highest TDP concentrations detected during this study were found in groundwater present in a bioswale, where levels ranged between 0.1 to 2.9 mg L− 1 (Fig. 5). Concentrations averaged approximately 0.5 mg L− 1 from 2012 to mid-2014, then increased to 2.9 mg L− 1 in the fall of 2014. The bioswale receives runoff from a large parking lot at an amusement park and was constructed in the late-2000's as an urban BMP demonstration project. Although specific details of fill media are not known, it likely consists of a mix of sand, organic wood chips and peat and it is observed to be capped with dense grass and forb vegetation. Installation of correct fill media is needed if bioretention cells are to remove nutrients (Hunt et al., 2006). For example, in two bioretention cells in North Carolina, effluent TDP concentrations averaged 0.5–2.2 mg L−1 (Hunt et al., 2006), concentrations which are similar to values measured in this study. Hunt et al. (2006) hypothesized that TDP was leached from decomposition of the fill material and we suspect a similar mechanism for the Okoboji bioswale. An increase in TDP concentration in the bioswale in late 2014 appears to be associated with strongly reducing conditions developing in the system; ORP decreased from approximately 150 mv to −225 mv during the fall period. P will commonly be released under hypoxic or anoxic conditions as a result of microbial Fe reduction (Sharpley, 1995). Because bioswales at West Lake Okoboji receive drainage from b10 ha of urban area at this time, the lake contribution of TDP from bioswales does not appear to be significant. However, the use of bioretention practices are continuing to be installed in urban and residential areas of the watershed. Hence, we recommend that future bioretention practices consider using low P-index soils to better retain P.

NO3-N concentration of 0.095 mg L−1. A recent lake survey indicated that the NO3-N concentration in the lake was 0.12 mg L− 1 in 2014 (IDNR, 2015). Hence, if all shallow groundwater loads of NO3-N were placed in the lake, groundwater sources could account for approximately 80% of the measured NO3-N concentration in the lake. Given the range of annual groundwater source loads (Table 1) and using the same measured lake concentration, groundwater loading could conceivably account for 60 to 90% of the lake NO3-N concentration. A similar approach can be used to estimate the potential contribution of groundwater P loads to the lake. The average lake concentration of total P from 2012 to 2014 was 0.021 mg L− 1 (IDNR, 2015). Considering that the average groundwater P load (347 kg) into the lake water volume produces a lake concentration of 0.0019 mg L− 1, we estimate the groundwater contribution to lake P concentration to be approximately 9%. On an annual basis and using the same average lake concentration, groundwater loads could conceivably account for approximately 7 to 13% of the lake P concentration. An estimate of annual groundwater nutrient loads delivered to West Lake Okoboji from various land use types could be refined with additional measurements or modeling. In this study we did not consider the lag time between groundwater recharge and discharge and nutrient loading into the lake. Groundwater flow from distal watershed areas to the lake could take many years or decades to occur, providing ample opportunities for nutrient attenuation along groundwater flow paths (Pint et al., 2003; Kornelsen and Coulibaly, 2014; Robinson, 2015). However, in agricultural regions underlain by extensive tile drainage systems, groundwater travel times are significantly reduced (Schilling et al., 2015). For example, in the 7443 Bear Creek watershed in central Iowa, mean groundwater travel times were reduced from 83 years (no tile) to 1.1 years with a tile drainage density of 0.04 m−1 (Schilling et al., 2015). Hence, determining the true groundwater travel time distribution in West Lake Okoboji watershed, and the time lag between nutrient sources and the lake, will require detailed information on the location of drainage tiles and better understanding the complex distribution of groundwater flow paths in the watershed. Groundwater flow models have often been used for this purpose (e.g., Hunt et al., 2003; Rosenberry et al., 2015), but they may be premature for the West Lake Okoboji watershed without additional field assessment and data collection. Additionally, our study could be refined with better characterization of groundwater recharge by landscape position and land cover classes. Although using baseflow in this study was appropriate, it did not allow for differentiation in recharge patterns among vegetation classes. Quantification of recharge through ET-based models (e.g., Meinikmann et al., 2013) or other methods (Scanlon et al., 2002) would improve the load allocation. Further, our study was conducted during a period of drought which likely influenced nutrient loading amounts. In particular, the NO3-N loading from crop areas was likely low compared to similar crop areas in Iowa, but the patterns of NO3-N loading allocated to various land covers were probably unaffected to a significant degree. Nutrient load reductions

Potential for shallow groundwater loads to impact lake nutrient levels The potential for NO3-N and P loads in shallow groundwater to impact nutrient loading in West Lake Okoboji can be estimated if we assume that all groundwater loads discharge into the lake every year. This estimation clearly represents a worst-case scenario as groundwater travel times vary in the watershed and N and P concentrations are subject to transformation, retardation and loss along flow paths. Nonetheless, the potential groundwater contribution of NO3-N and TDP loads can be estimated using a mass-balance approach and lake concentration data. The average groundwater NO3-N load for the 2012–2014 study period was 16.8 Mg. If this mass was placed into the water volume of West Lake Okoboji (1.758 × 108 m3; unpublished data, Iowa Department of Natural Resources, Spirit Lake, IA), it would produce a lake

Given that shallow groundwater in the watershed contains abundant NO3-N and P which may discharge into the lake, what nonpoint source BMPs might be appropriate for reducing watershed-scale nutrient levels? Load reductions targeted toward reducing NO3-N lost from agricultural croplands would address the largest source of NO3-N. The Iowa Nutrient Reduction Strategy (INRC) published in 2013 is a science and technology based framework to reduce nitrogen and phosphorus delivered to Iowa rivers and the Gulf of Mexico from point and nonpoint sources (INRS, 2013). The INRS evaluated 14 different NO3-N reduction practices in three main categories: nitrogen management, land use changes, and edge-of-field treatment. Of the nitrogen management strategies, reducing nitrogen application rates and planting cover crops were considered the practices with greatest potential effect,

K.E. Schilling et al. / Journal of Great Lakes Research 42 (2016) 588–598

with NO3-N reductions of approximately 10% and 30%, respectively. Dinnes et al. (2002) also suggested that improved nitrogen management would reduce the amount of NO3-N exported with drainage water. Installing wetlands or bioreactors at the outlet of drainage tile systems has similar reductions (22% and 18%), although a substantially larger load of NO3-N can be reduced with wetlands intercepting large drainage districts tiles compared to bioreactors designed to intercept smaller tiles. The effectiveness of wetland denitrification varies greatly with hydraulic loading rate, with reductions ranging from approximately 20 to 70% (Crumpton et al., 2008). Tomer et al. (2003) suggested that a constructed wetland placed at the outlet of a tile-drained basin could reduce NO3-N concentrations by 3 mg L−1 and remove 18–34% of the NO3-N fluxes. Converting row crop land to perennial systems achieved reductions of 70–80%, whereas utilizing extended rotations of alfalfa in 4–5 year rotations with crops would achieve reductions of more than 40%. Conversion of cropland to perennial grasslands has been ongoing in the West Lake Okoboji watershed and currently the amount of land in perennial vegetation (2046 ha) is only slightly less than the amount of area in cropland (2497 ha). This practice should be continued in the watershed as it has the greatest potential to reduce groundwater NO3-N potentially delivered to the lake. Groundwater sources of P are more ubiquitous than NO3-N and reducing P loads in groundwater will be a challenge. It is quite possible that groundwater P concentrations beneath cropping systems cannot be reduced any further since they are already comparable to concentrations beneath perennial vegetation. Of the P sources assessed in this study, BMP efforts should be targeted to reduce P loading from residential areas. A more detailed source assessment is needed to quantify P contributions from urban lawns, leaky sewers or septic systems. Use of urban BMPs including bioretention cells should be encouraged, albeit with more attention given to using low P index fill media. However, because groundwater P contributions to the lake are minimal (~10%), this implies that P is being delivered to the lake from other pathways, including most prominently, as particulate P in agricultural runoff. More traditional agricultural practices targeted towards reducing runoff and sediment export, including no-till (90% reduction), sedimentation basins (85%) land use change (75%) fertilizer reductions (17%) and cover crops (29%) are appropriate for reducing P from agricultural areas (INRS, 2013). Conclusions In this study, we quantified the groundwater loading of NO3-N and P by land cover class to assess the potential for these land areas to contribute nutrient loads delivered to West Lake Okoboji through groundwater recharge. During a three-year study period characterized by below normal precipitation, groundwater recharge averaged approximately 76 mm year−1 for vegetated areas and approximately one-tenth of this amount for paved urban areas. Groundwater quality varied among the land cover types with average NO3-N concentrations highest beneath cropped fields (8.8 mg L−1) and residential areas (2 mg L−1), and P concentrations ranging between 0.05 to 0.1 mg L−1 throughout the region. NO3-N loads were the highest under cropped fields and this source accounts for approximately 90% of the groundwater NO3-N in the watershed, whereas residential areas may account for an additional 5 to 9%. Groundwater P loads were more evenly distributed among source areas with contributions from crop and perennial areas nearly equal (37%) and residential areas contributing 23% of the P load. Urban commercial and golf course areas were not major sources of NO3-N or P. Based on mass balance, groundwater discharge may account for as much as 80% of the NO3-N in the lake compared to approximately 10% of the P. Our results are informative for more effectively targeting implementation of BMPs to major nutrient loading areas for reduction of NO3-N and P delivered to the lake.

597

Acknowledgments Funding for equipment and supplies was provided by the Okoboji Homeowners Association. Drilling was provided by the Natural Resources Conservation Service. John and Julie Clark are gratefully acknowledged for their assistance in measuring monthly water levels at monitoring wells.

References Anderson, M.P., Munter, J.A., 1981. Seasonal reversals of ground water flow around lakes and the relevance to stagnation point and lake budgets. Water Resour. Res. 17, 1139–1150. Arbuckle, K.E., Downing, J.A., 2001. The influence of watershed land use on lake N:P in a predominantly agricultural landscape. Limnol. Oceanogr. 46, 970–975. Bettis III, E.A., Quade, D.J., Kemmis, T.J., 1996. Hogs, bogs, and logs: Quaternary deposits and environmental geology of the Des Moines lobe. Iowa Dep. Nat. Resour. Geol. Surv. Bur, Guidebook Series No. 18 (170 pp.). Booth, D.B., Jackson, C.R., 1997. Urbanization of aquatic system: degradation thresholds, stormwater detection and limits of mitigation. J. Am. Water Resour. Assoc. 33, 1077–1090. Burkart, M.R., Simpkins, W.W., Morrow, A.J., Gannon, J.M., 2004. Occurrence of total dissolved phosphorus in unconsolidated aquifer and aquitards in Iowa. J. Am. Water Resour. Assoc. 40, 827–834. Cohen, S., Svrjcek, A., Durborow, T., Barnes, N.L., 1999. Water quality impacts by golf courses. J. Environ. Qual. 28, 798–809. Conley, D.J., Paerl, H.W., Howarth, R.W., Boesch, D.F., Seitzinger, S.P., Havens, K.E., Lancelot, C., Liken, G.E., 2009. Controlling eutrophication: NO3-N and phosphorus. Science 323, 1014–1015. Crumpton, W.G., Kovacic, D.A., Hey, D.L., Kostel, J.A., 2008. Potential of Restored and Constructed Wetlands to Reduce Nutrient Export from Agricultural Watersheds in the Corn Belt. Final Report: Gulf Hypoxia and Local Water Quality Concerns Workshop. Am Society of Agricultural and Biological Engineers, St. Joseph, MI, pp. 29–42. David, M.B., Drinkwater, L.E., McIsaac, G.F., 2010. Sources of nitrate yields in the Mississippi River basin. J. Environ. Qual. 39, 1657–1667. Dinnes, D.L., Karlen, D.L., Jaynes, D.B., Kasper, T.C., Hatfield, J.L., Colvin, T.S., Cambardella, C.A., 2002. Nitrogen management strategies to reduce NO3-N leaching in tiledrained Midwestern soils. Agron. J. 94, 153–171. Downing, J.A., McCauley, E., 1992. The nitrogen:phosphorus relationships in lakes. Limnol. Oceanogr. 37, 237–252. Grannemann, N.G., Hunt, R.J., Nicholas, J.R., Reilly, T.E., Winter, T.C., 2000. The Importance of Ground-water in the Great Lakes Region. U.S. Geological Survey. Water-Resources Investigations Report 00–4008, Lansing, MI. Hamel, P.E., Daly, E., Fletcher, T.D., 2013. Source-control stormwater management for mitigating the impacts of urbanization on baseflow: a review. J. Hydrol. 485, 201–211. Harvey, F.E., Rudolph, D.L., Frape, S.K., 2000. Estimating ground water flux into large lakes: application in the Hamilton Harbor, western Lake Ontario. Ground Water 38, 550–565. Iowa Nutrient Reduction Strategy (INRS), 2013. http://www.nutrientstrategy.iastate.edu/ documents (accessed August 13, 2015). Holtschlag, D.J., Nicholas, J.R., 1998. Indirect ground-water discharge to the Great Lakes. U.S. Geol. Surv. Open File Rep. 98–579 Lansing, MI. Hunt, R.J., Haitjema, H.M., Krohelski, J.T., Feinstein, D.T., 2003. Simulating ground waterlake interactions: approaches and insights. Ground Water 41, 227–237. Hunt, W.F., Jarrell, A.R., Smith, J.T., Sharkey, L.J., 2006. Evaluating bioretention hydrology and nutrient removal at three field sites in North Carolina. J. Irrig. Drain. Eng. 132, 600–608. Ikenberry, C.D., Soupir, M.L., Schilling, K.E., Jones, C.S., Seeman, A., 2014. Nitrate-nitrogen export: magnitude and patterns from drainage districts to downstream river basins. J. Environ. Qual. 43, 2024–2033. Iowa Department of Natural Resources (IDNR), 2015. Iowa lakes information system. http://limnology.eeob.iastate.edu/lakereport/chemical_report.aspx (accessed August 10, 2015). Jacobson, L., David, M., Drinkwater, L., 2011. A spatial analysis of phosphorus in the Mississippi River basin. J. Environ. Qual. 40, 931–941. Jeong, J., Kannan, N., Arnold, J.G., 2014. Effects of urbanization and climate change on stream health in north-central Texas. J. Environ. Qual. 43, 100–109. Jha, M.K., Wolter, C.F., Schilling, K.E., Gassman, P.W., 2010. Assessment of TMDL implementation strategies for the Raccoon River, Iowa. J. Environ. Qual. 39, 1317–1327. Kemp, M.J., Dodds, W.K., 2001. Spatial and temporal patterns of nitrogen concentrations in pristine and agriculturally-influenced prairie streams. Biogeochemistry 53, 125–141. King, K.W., Balogh, J.C., Hughes, K.L., Harmel, R.D., 2007. Nutrient load generated by storm event runoff from a golf course watershed. J. Environ. Qual. 36, 1021–1030. Kornelsen, K.C., Coulibaly, P., 2014. Synthesis review on groundwater discharge to surface water in the Great Lake Basin. J. Great Lakes Res. 40, 247–256. LaBaugh, J.W., Winter, T.C., Rosenberry, D.O., Schuster, P.F., Reddy, M.M., Aiken, G.R., 1997. Hydrological and chemical estimates of the water balance of a closed-basin lake in north-central Minnesota. Water Resour. Res. 33, 2799–2812. Lee, D.R., Cherry, J.A., 1978. A field exercise on groundwater flow using seepage meters and min-piezometers. J. Geol. Educ. 27, 6–10.

598

K.E. Schilling et al. / Journal of Great Lakes Research 42 (2016) 588–598

Lewandowski, J., Meinikmann, K., Nutzmann, G., Rosenberry, D.O., 2015. Groundwater — the disregarded component in lake water and nutrient budgets. Part2: effects of groundwater on nutrients. Hydrol. Process. 29, 2922–2955. Meinikmann, K., Lewandowski, J., Nutzmann, G., 2013. Lacustrine groundwater discharge: combined determination of volumes and spatial patterns. J. Hydrol. 502, 202–211. Neff, B.P., Day, S.M., Piggott, A.R., Fuller, L.M., 2005. Base flow in the Great Lakes Basin. U.S. Geol. Surv. Invest. Rep. 2005–5217. Palmer, J.A., Schilling, K.E., Isenhart, T.M., Schultz, R.C., Tomer, M.D., 2014. Streambank erosion and loads with a single watershed: bridging the gap between temporal and spatial scales. Geomorphology 209, 66–78. Pint, C.D., Hunt, R.J., Anderson, M.P., 2003. Flowpath delineation and ground water age, Allequash Basin, Wisconsin. Ground Water 41, 895–902. Quade, D., Tassier-Surine, S., McKay, R., Giglierano, J., 2012. Surficial Geology of the Okoboji 7.5 Quadrangle, Dickinson County, Iowa. Iowa Geol. Survey Open File Map OFM-12-10. Risser, D.W., Gburek, W.J., Folmar, G.J., 2005. Comparison of Methods for Estimating Ground-water Recharge and Base flow at a Small Watershed Underlain by Fractured Bedrock in the Eastern United States. U.S. Geol Surv. Sci. Invest. Rep. 2005–5038. Robertson, D.M., Saad, D.A., 2013. SPARROW models used to understand nutrient sources to the Mississippi/Atchafalaya river basin. J. Environ. Qual. 42, 1422–1440. Robinson, C., 2015. Review on groundwater as a source of nutrients to the Great Lakes and their tributaries. J. Great Lakes Res. http://dx.doi.org/10.1016/j.jglr.2015.08.001. Rosenberry, D.O., Lewandowski, J., Meinikmann, K., Nutzmann, G., 2015. Groundwater — the disregarded component in lake water and nutrient budgets. Part 1: effects of groundwater on hydrology. Hydrol. Process. 29, 2895–2921. Rutledge, A.T., 1998. Computer Programs for Describing the Recession of Ground-water Discharge from Streamflow Records-Update. US Geol. Surv. Water Resour. Invest. Report 98–4148 (43 pp.). Scanlon, B.R., Healy, R.W., Cook, P.G., 2002. Choosing appropriate techniques for quantifying groundwater recharge. Hydrogeol. J. 10, 18–39. Schilling, K.E., 2009. Investigating local variation in groundwater recharge along a topographic gradient, Walnut Creek, Iowa, USA. Hydrogeol. J. 17, 397–407. Schilling, K.E., Libra, R.D., 2000. The relationship of nitrate concentrations in streams to row crop land use in Iowa. J. Environ. Qual. 29, 1846–1851. Schilling, K.E., Wolter, C.F., 2001. Contribution of baseflow to nonpoint source pollution loads in an agricultural watershed. Ground Water 39, 49–58. Schilling, K.E., Wolter, C.F., 2009. Modeling nitrate-nitrogen load reduction strategies for the Des Moines River, Iowa using SWAT. Environ. Manag. 44, 671–682.

Schilling, K.E., Wolter, C.F., Isenhart, T.M., Schultz, R.C., 2015. Tile drainage density reduces groundwater travel times and compromises riparian buffer effectiveness. J. Environ. Qual. 44, 1754–1763. Schilling, K.E., Zhang, Y.K., 2004. Contribution of baseflow to nitrate-nitrogen export in a large agricultural watershed, USA. J. Hydrol. 295, 305–316. Schilling, K.E., Wolter, C.F., Isenhart, T.M., Schultz, R.C., 2015. Tile drainage density reduces groundwater travel times and reduces riparian buffer effectiveness. J. Environ. Qual. 44, 1754–1763. Schoonover, J.E., Lockaby, B.G., Helms, B.S., 2006. Impacts of land cover on stream hydrology in the west Georgia piedmont, USA. J. Environ. Qual. 35, 2123–2131. Sharpley, A.N., 1995. Dependence of runoff phosphorus on extractable soil phosphorus. J. Environ. Qual. 24, 920–926. Shaw, R.D., Prepas, E.E., 1990. Groundwater–lake interactions: I: accuracy of seepage meter estimates of lake seepage. J. Hydrol. 119, 105–120. Simpkins, W.W., 2006. A multiscale investigation of ground water flow at Clear Lake, Iowa. Ground Water 44, 35–46. Simpkins, W.W., Drenner, K.B., Bocchi, S., 2001. An Analysis of Hydrogeology, Groundwater Discharge and Nutrient Input to Clear Lake. In: Downing, J.A., Kopaska, J., Bonneau, D. (Eds.), Clear Lake Diagnostic Report to Iowa DNR. Iowa Dep. Nat. Resourc., pp. 180–209. Singh, R., Helmers, M.J., Qi, Z., 2006. Calibration and validation of DRAINMOD to design subsurface drainage systems for Iowa's tiled landscapes. Ag. Water Manage. 85, 221–232. Smith, V.H., 2003. Eutrophication of freshwater and coastal marine ecosystems a global problem. Environ. Sci. Pollut. Res. 10, 126–139. Smith, D.R., King, K.W., Williams, M.R., 2015. What is causing the harmful algal blooms in Lake Erie? J. Soil Water Conserv. 70, 27A–29A. Thompson, C.A., 1986. Water Resources of the Ocheydan-Little Sioux alluvial aquifer. Iowa Geological Survey Bureau Open File Report, p. 86-3. Tomer, M.D., Meek, D.W., Jaynes, D.B., Hatfield, J.L., 2003. Evaluation of nitrate-nitrogen fluxes from a tile-drained watershed in central Iowa. J. Environ. Qual. 32, 642–653. Wakida, F.T., Lerner, D.N., 2005. Non-agricultural sources of groundwater nitrate: a review and case study. Water Res. 39, 3–16. Witzke, B.J., Anderson, R.R., Pope, J.P. 2010. Bedrock Geologic Map of Iowa, scale: 1: 500,000, Iowa Geol. Water Surv., Open File Digital Map OFM-10-1. Zaimes, G.N., Schultz, R.C., Isenhart, T.M., 2008. Streambank soil and phosphorus losses under different riparian land-uses in Iowa. J. Am. Water Resour. Assoc. 44, 935–947.