Livestock manure driving stream nitrate

Abstract

Growth and consolidation in the livestock industry in the past 30 years have resulted in more total farm animals being raised on fewer Iowa farms. The effects of this on stream water quality at the landscape scale have largely gone unexplored. The main objective of this work was to quantify the effects on stream nitrate levels of livestock concentration in two western Iowa watersheds relative to seven other nearby watersheds. To achieve this objective, we used data on high-frequency nitrate concentration and stream discharge, commercial nitrogen fertilizer use, and manure-generated nitrogen in each watershed. Our analysis shows much higher stream nitrate in the two watersheds where livestock concentration has been greatest, and little difference in commercial fertilizer inputs with the widespread availability of manure N. Reducing N inputs and better management of manure N, including analysis of crop N availability in soil and manure, can reduce uncertainty regarding fertilization while improving water quality.

Introduction

The state of Iowa, located in the U.S. Midwest, has long been one of the country’s leading producers of hogs, cattle, poultry, and eggs. Currently Iowa exceeds all other states in egg and pork production and is fourth in production of feeder cattle (USDA 2018). Iowa has also been a leading producer of both corn (Zea mays L.) and soybeans (Glycine max [L.] Merr.), frequently topping all other U.S. states in harvested totals of these commodities (Jones et al. 2018a). Co-locating crop production and livestock within the state has created efficiencies of production, transportation, and fertilization.

High yield agriculture, such as that conducted on nearly 70% of Iowa’s area (USDA 2018), depends on addition of nitrogen fertilizers. Various forms of fertilizer nitrogen are used throughout the state to enhance crop yields, especially those of corn. Most of this nitrogen is applied as formulations of ammonia/ammonium (NH₃/NH₄⁺) and nitrate (NO₃–N) generated from industrial processes, but also animal manures where available. The use of industrially produced nitrogen fertilizers emerged as an important component of U.S. and Iowa agriculture following World War II (Commoner 1977). Before 1945, nearly all nitrogen inputs used to fertilize Iowa corn fields came from legumes such as alfalfa and clovers, and animal waste. However, after this time, the use of inorganic nitrogen fertilizers increased 13-fold from 1945 to 1972 as they quickly became affordable and widely available (Commoner 1977). Livestock, especially cattle, consumed the alfalfa and clover, but commercial fertilizer allowed farmers to forgo hay crops and cattle. This enabled many Iowa farmers to specialize on corn and soybean production (Hendrickson and James 2005). The demand for animal protein, however, continued to increase with world population and increased income levels (Delgado et al. 2001). With fewer farmers wanting or needing livestock, those that continued with livestock production were able to greatly enlarge their operations. This is especially evident in Iowa with hog production. In 1980, 65 000 Iowa farmers raised a total of 13 million hogs; by 2002, the number of hog farmers had dwindled to 10 000, but total hog numbers increased to 14 million (Herriges et al. 2005). This dramatic shift in production resulted in many hogs being concentrated in certain areas of the state and a geographical alignment with buyers, packing houses, feed and equipment suppliers, and haulers (Honeyman and Duffy 2006). Similar scenarios have also played out with cattle and poultry. This agricultural specialization that has occurred in Iowa is consistent with changes that are still occurring worldwide (Liu et al. 2017).

This transition from diverse, multi-species farms to ones specializing in corn and soybean production with a subset of the latter raising concentrated livestock has produced both efficiencies and negative environmental consequences. It has long been known that nitrogen fertilization correlates with stream nitrate in the U.S. Cornbelt (Klepper 1974) with impacts on municipal water supply (Hatfield et al. 2009) and Gulf of Mexico Hypoxia (Rabalais et al. 2002). However, because nitrogen inputs cycle through plant biomass and into and out of soil organic matter (Jackson et al. 2000), and because of the time lag of pollutant transport to streams via groundwater pathways (Van Meter et al. 2017), it is nearly impossible to trace stream nitrate back to commercial fertilizer, animal manure, legumes, or soil organic matter. Hence, many have attempted to gain insights on nitrate sources and pathways using nitrogen budgeting (David et al. 1997; Libra et al. 2004; Jones et al. 2016).

The intensity of crop and livestock production in Iowa has made the state a major contributor to Mississippi River basin nitrate loads (David et al. 2010; Jones et al. 2018a, b). Nitrate loading from Iowa appears to be increasing (Jones et al. 2018a), especially in the Missouri River and its Iowa tributaries (Sprague et al. 2011; Li et al. 2013), and Iowa contributes up to 89% of the annual Missouri River nitrate load even though Iowa areas draining to the Missouri comprise only 3% of the total watershed area (Jones et al. 2018a). Northwest Iowa, which drains to the Missouri River, is an area where livestock production has been concentrated in recent years (Andersen and Pepple 2017). The overall objective of our research was to assess whether the manure generated from high animal densities drives stream nitrate levels in the region. Using high-frequency river monitoring data collected from nine western Iowa watersheds draining to the Missouri River, two of which have a much larger animal density than the others, and comparing the water quality data to crop area, fertilization, and livestock populations, we show that river nitrate levels are linked to agricultural and livestock management.

Materials and methods

Study area

The nine western Iowa watersheds selected for study are shown in Fig. 1 and Table 1. These nine watersheds were selected because they all drain to the Missouri River and were instrumented with real-time, continuous nitrate sensors co-located with a discharge measurement station. Areas upstream of the water quality and discharge monitoring locations constitute 23% of Iowa’s area and 74% of the state’s area that drains to the Missouri River. Agricultural land use dominates each catchment and large point source discharges are absent, with no cities greater than 10 000 population draining into any of the watersheds.

Fig. 1.

Nine western Iowa watersheds are evaluated here. The number corresponds to the watershed number shown in Table 1. The green circle designates monitoring location near the outlet

Table 1.

Watersheds in the study area along with crop area, livestock densities, commercial N fertilizer application rates, and nitrogen generated from animal manure

Figure 1 Map No.	Name	Iowa area (km2)	Area fractions						Animal Units (AU) ha−1 2017	kg ha−1
			2012			2017				Corn + soybean area			Corn area only
			Corn	Soybean	Total corn–soybean	Corn	Soybean	Total corn–Soybean		Commercial N application rate (2012)	Manure N generated	Commercial N + generated manure N	Commercial N + generated manure N
1	Rock River	1748	0.50	0.33	0.81	0.44	0.34	0.78	5.00	116	111	227	369
2	Floyd River	2295	0.48	0.34	0.82	0.47	0.35	0.82	3.53	112	117	229	379
3	Monona-Harrison Ditch	2331	0.42	0.31	0.73	0.38	0.33	0.71	1.31	122	37	159	261
4	Little Sioux R.	8350	0.45	0.34	0.79	0.42	0.35	0.77	1.04	119	40	159	262
5	Soldier River	1049	0.46	0.31	0.76	0.42	0.33	0.75	0.76	128	22	150	240
6	Boyer River	2202	0.49	0.31	0.81	0.45	0.34	0.79	1.13	138	38	176	235
7	W. Nishnabotna R.	3434	0.44	0.32	0.76	0.40	0.35	0.75	0.68	121	23	144	241
8	E. Nishnabotna R.	2862	0.39	0.31	0.69	0.36	0.33	0.69	0.53	106	32	138	233
9	W. Nodaway R.	1974	0.33	0.29	0.63	0.32	0.32	0.64	0.45	91	24	115	201

Agricultural data

County-level data for the latest available (2012) commercial nitrogen fertilizer were obtained from the US Geological Survey National Water Quality Assessment project (Gronberg and Spahr 2012). County-level manure data were obtained from Gronberg and Arnold (2017). There is reason to believe that the 2012 commercial fertilizer data are relevant in the present day because changes in crop areas from 2012 to 2017 were small in the nine watersheds, e.g., − 6.2, + 5.4, and − 1.3% for corn, soybean, and total corn plus soybean, respectively, and statewide commercial fertilization rates have not changed appreciably since 1990 (Hatfield et al. 2009).

Data for animal populations were collected from two sources. Recent (2018) data for animal numbers were obtained from the Iowa Department of Natural Resources (IDNR) Animal Feeding Operations (AFO) database (IDNR 2018a). The IDNR’s database is mostly limited to regulated facilities; therefore, numbers obtained from this source are likely to represent less than the actual number of animals raised in these areas. When calculating total animal units (AU) in a watershed, the population of a species is multiplied by the equivalence factor shown in Table 2 (IAC 2018). Historical county-level hog (1980–2012) and cattle (2002–2012) population data were obtained from USDA (2018) and adjusted to each watershed area based on the portion of the county that lies within the individual watershed. Watershed-level hog and cattle populations for 2018 were obtained from IDNR (2018a). The county-level areas planted with corn and soybeans in 2012 and 2017 were obtained from USDA (2018) and adjusted to the county’s area portion within each watershed.

Table 2.

Factors used to calculate total animal units (AU). Populations are multiplied by the factors shown to quantify total AU

Animal species	Factor
Horses	2.0
Mature dairy cattle	1.4
Slaughter or feeder cattle	1.0
Immature dairy cattle	1.0
Hogs > 25 kg	0.4
Hogs 7–25 kg	0.1
Turkeys > 3 kg	0.018
Chickens > 1.4 kg	0.01
Turkeys < 3 kg	0.0085
Chickens < 1.4 kg	0.0025
Fish	0.001

For the purposes of constructing a rough agronomic N budget for each watershed, inputs included commercial N (CN), N generated by manure (MN), fixation N (FN) from the previous year’s soybean crop while outputs included N harvested in the grain (GN) (Eq. 1).

$$\text{CN}+\text{MN}+\text{FN}-\text{GN}=\text{N surplus or deficit}.$$ 1

Biological N fixation of soybean in 2016 was calculated according to Barry et al. (1993) using county-level crop areas and soybean yields adjusted to the area portion lying within each watershed. Nitrogen harvested in corn grain was calculated using watershed crop yields and the measured average for Iowa corn reported by Blesh and Drinkwater (2013). Export of N in the harvested soybeans was calculated assuming 6.4% N in soybean seeds according to the USDA protocols using the Crop Nutrient Tool (2009).

Hydrology

Watershed precipitation totals for 2017 were estimated based on data collected at 22 stations within the individual watersheds and averaging data from each watershed location. These data were obtained from the Iowa State University Mesonet network (2017). Discharge data for all the sites were generated by the U.S. Geological Survey (USGS 2018). These 15-min interval data were aggregated into daily averages and the water yield was calculated by dividing total annual discharge by watershed area.

Water quality

High frequency (15 min) 2017 NO₃–N concentration data were obtained from the University of Iowa’s Water Quality Information System (Jones et al. 2018b). This network of real-time water quality sensors measures NO₃–N concentrations at about 65 sites throughout Iowa including those shown in Fig. 1. Data quality for the network is governed by a QA/QC plan adopted from USEPA and USGS protocols. Basic QA protocols include systematic monitoring of incoming water quality data, remote monitoring of field sensor/system health (e.g., battery voltage and signal strength), automatic data review through the use of data thresholds and limits, and use of data descriptors for denoting state of data review. Measurements are generated by the Hach Nitratax sc plus (Loveland, CO, US) nitrate sensor and accuracy is verified through regular collection of grab samples that are lab-analyzed. Extensive details about measurement and quality control protocols can be found at Jones et al. (2018b). Data from the IDNR ambient monitoring program were used (IDNR 2018b) for periods when high-frequency data were missing (i.e., equipment malfunction and Dec–Feb). Linear interpolation was used to estimate NO₃–N concentrations on days with no NO₃–N data. Daily average NO₃–N concentrations were multiplied by daily average discharges and then summed to obtain annual NO₃–N loads and yields (load per watershed area). Flow weighted average (FWA) NO₃–N concentrations were calculated by dividing total load by total discharge. Minnesota areas draining to the Rock and Little Sioux Rivers were used when calculating yields.

Results

Agricultural

Current density of animals ranged from 0.45 (West Nodaway) to 5.00 AU ha⁻¹ (Table 1). The Floyd (3.53 AU ha⁻¹) and Rock (5.00 AU ha⁻¹) watersheds had much higher animal densities than the other seven watersheds (average 0.84 AU ha⁻¹). Cattle and hogs are by far the largest contributors to AU units in all watersheds, and historical data for these species are shown in Fig. 2, illustrating how the concentration of hogs has risen since 1980 and cattle since 2002. Hog densities have increased since 1980 in the Rock, Floyd, Monona-Harrison Ditch, and Little Sioux watersheds and declined in the others, with the decline especially pronounced in the West Nodaway (− 80%) and increases largest in the Floyd (+ 126%) and Rock watersheds (+ 269%). Overall, hog populations increased 41% since 1980, but when the Floyd and Rock watersheds are excluded, the increase is only 4.2%. The Floyd and Rock watersheds also have the highest current cattle densities at 1.14 and 1.70 per ha, respectively. Since 2002, the average cattle population grew by 37%, but increased only 0.01% when the Rock and Floyd watersheds are excluded. Large declines in cattle populations occurred in the Soldier (− 46%) and West Nodaway (− 74%) watersheds.

Fig. 2.

Hog and cattle densities in the nine studied watersheds

Areas planted with corn and soybean were obtained for 2012 and 2017 (Table 1) for comparison with available fertilization and water quality data. Overall in the nine watersheds, the total corn–soybean area was 1.3% lower in 2017 compared to 2012, with 6.2% less corn area and 5.4% more soybean area. Between watersheds, the biggest increase from 2012 to 2017 was in the West Nodaway (+ 1.6%) while the largest decrease was in the Monona-Harrison Ditch watershed (− 2.7%). Total corn–soybean area declined from 2012 to 2017 in all watersheds except the Floyd and West Nodaway. The cropped portion of each watershed ranged from 0.64 (West Nodaway) to 0.82 (Floyd) with an overall average of 0.74 in 2017.

The latest available fertilization data are from 2012 and are listed in Table 1. The commercial rates plus generated manure are based on 2012 crop areas and vary from 115 (West Nodaway) to 229 kg ha⁻¹ of combined corn and soybean area. However, soybeans usually do not receive much nitrogen fertilizer in Iowa, with a statewide average of 15.7 kg ha⁻¹ (Jones et al. 2016). Considering this, amounts per corn area alone ranged from 201 kg ha⁻¹ (West Nodaway) to 379 kg ha⁻¹ (Floyd) and averaged 269 kg ha⁻¹ across all watersheds. Interestingly, the commercial N rates in the Rock watershed (116 kg ha⁻¹ to all corn–soybean area) and the Floyd watershed (112 kg ha⁻¹) were similar to the nine-watershed average (117 kg ha⁻¹), this even with abundance of manure N generated by livestock (111 and 117 kg ha⁻¹, respectively). The commercial N rates in the West Nodaway watershed (91 kg ha⁻¹ to all corn–soybean area) were lowest of the nine watersheds, even though the generated manure N was also quite low at 24 kg ha⁻¹, second lowest of the group.

Water quality and hydrology

The annual nitrate (NO₃–N), precipitation, and discharge data for 2017 are shown in Table 3. The precipitation recorded in the Rock (804 mm) and Floyd (759 mm) watersheds was substantially less than the other seven watersheds where the average was 917 mm. Despite lower amounts of rainfall, the Rock and Floyd each had the highest annual NO₃–N yields (24.7 and 30.5 kg ha⁻¹, respectively) and FWA NO₃–N concentration (11.5 and 16.2 mg L⁻¹, respectively) (Fig. 3). The averages for the other seven watersheds were 20.0 kg ha⁻¹ (yield) and 7.3 mg L⁻¹ (FWA concentration). The Monona-Harrison Ditch watershed had the lowest yield of NO₃-N (11.1 kg ha⁻¹) and the West Nodaway River had the lowest FWA concentration (4.9 mg L⁻¹).

Table 3.

2017 hydrology and stream NO₃–N data

Watershed	Annual precipitation	Dischargea	NO3–N measurement days	NO3–N yield	NO3–N yield/precipitation	FWAb NO3–N
	(mm)		(N)	(kg ha−1)	(g ha−1 mm−1)	(mg L−1)
Rock River	804	215	115	24.7	30.7	11.5
Floyd River	759	188	284	30.5	40.2	16.2
Monona-Harrison Ditch	942	172	171	11.1	11.8	7.2
Little Sioux River	816	241	181	17.1	20.9	7.1
Soldier River	846	259	233	21.9	25.9	8.5
Boyer River	1056	312	249	27.1	25.7	8.7
West Nishnabotna River	846	291	293	23.9	28.3	8.2
East Nishnabotna River	974	430	193	23.7	24.3	5.5
West Nodaway River	963	282	288	13.9	14.4	4.9
Average	890	266	223	21.5	24.7	8.6

^aDischarge calculated by dividing total discharge volume at the outlet by watershed area draining to the site

^bFWA is Flow Weighted Average concentration, which is obtained by dividing total river NO₃–N load by total discharge

Fig. 3.

Box plots of 2017 daily average NO₃–N concentrations. The boxes bracket the 25th–75th percentiles; the line in the box indicates the median; the whiskers the 10th and 90th percentiles, and the dots are data points less than (greater than) the 10th (90th) percentiles

Nitrogen budget

An estimated 2017 nitrogen budget was constructed assuming the fertilization rates from 2012 were relevant to 2017, using commercial N and manure data, crop yield data from 2017, and soybean area and yield from 2016 to calculate contributions from nitrogen fixation. This is shown in Table 4 along with the FWA NO₃–N concentrations for comparison. The surplus nitrogen, i.e., the amount applied as commercial fertilizer plus the amount generated by livestock plus the amount fixed by soybeans the previous year minus the amount harvested in the grain, ranged from 55 kg ha⁻¹ (West Nodaway) to 161 kg ha⁻¹ (Floyd) and averaged 99 kg ha⁻¹ across the nine watersheds. The watersheds with the three largest surplus N values (Floyd, Rock, and Boyer) also had the three highest FWA concentrations while the watersheds with the two smallest surpluses also had the two smallest FWA concentrations. The average surplus for the Rock and Floyd (155 kg ha⁻¹) was nearly double the average of the other seven watersheds (83 kg ha⁻¹).

Table 4.

Estimated 2017 nitrogen budget for using commercial N rates, generated livestock manure, soybean fixation from previous year, N harvested in the grain, and stream NO₃–N for comparison

Watershed	Commercial N + generated manure N	Fixation from 2016 soybean crop	Fixation from 2016 soybean crop	Commercial N + manure N + fixation N-grain N	FWA NO3–N
	(Kg NO3–N ha−1 year−1)				(mg L−1)
Rock River	226	100	178	148	11.5
Floyd River	229	107	175	161	16.2
Monona-Harrison Ditch	159	110	169	100	7.2
Little Sioux River	158	111	172	97	7.1
Soldier River	150	96	171	74	8.5
Boyer River	176	98	171	103	8.7
West Nishnabotna River	145	97	159	83	8.2
East Nishnabotna River	138	95	163	70	5.5
West Nodaway River	115	98	159	55	4.9

The FWA concentrations were well correlated with fertilization and crop area (Fig. 4). These concentrations correlated significantly (p < 0.01) with surplus nitrogen (fertilizer + manure + fixation-grain N), commercial + manure + fixation N, commercial + manure N, and manure N, and less significantly with area portion in corn and soybean (p < 0.05). The FWA concentrations did not correlate with commercial N (p > 0.10).

Fig. 4.

Correlations of 2017 watershed Flow Weighted Average (FWA) NO₃–N concentrations with N surplus (a), sum of commercial, manure and fixation nitrogen (b), sum of commercial and manure nitrogen (c), commercial nitrogen (d), generated manure nitrogen (e), and area portion in corn and soybean (f). The dotted portion of the regression line in a is an extrapolation backward to a zero surplus condition. FWA is defined as total NO₃–N load divided by total discharge for 2017

Discussion

Iowa State University (ISU) Extension guidelines for N application rates (kg ha⁻¹) range from 135 to 165 (average 150) for corn following soybean, and 193–221 (average 206) for corn following corn under the current price structure for commercial nitrogen fertilizer and corn grain (Sawyer 2016). When we adjust for statewide N rates to soybeans (15.7 kg ha⁻¹, Jones et al. 2016), commercial fertilizer data for 2012 show average annual commercial N rate to corn in these nine watersheds is 189 kg ha⁻¹, generally in-line-to-slightly-above ISU guidelines. However, these rates do not account for the substantial amounts of manure N generated in the nine watersheds, and especially in the Floyd and Rock watersheds, where the generated manure N is roughly equivalent to commercial N sales. After the generated manure N is added to commercial N, then the amount of N per corn-hectare exceeds ISU guidelines in all watersheds except the West Nodaway, where coincidentally the lowest FWA NO₃-N was recorded in 2017. In the Floyd and Rock watersheds, the commercial N plus N generated from manure sources is ~ 370–380 kg ha⁻¹ of corn (accounting for average N application to soybean), which is about double ISU recommendations. In these two watersheds, the “surplus” N (i.e., commercial + manure + fixation – grain) actually exceeds the ISU recommendations for corn grown after soybeans. It should be pointed out that this is based on 2012 animal populations and that hog and cattle numbers increased substantially in the Rock watershed since then (Fig. 2). The fate of all manure N applied to agricultural fields is not well understood. Some amount of the N in fresh animal manure is lost to volatilization (Kirchmann and Witter 1989) and never becomes available for crop uptake. There is evidence, however, that much of this volatilized N is deposited within 1 km of the confinement (Loubet et al. 2009) and McGinn et al. (2016) reported a 50% decline in deposition 200 m from a cattle confinement. Thus, much of this volatilized N is not lost from the watershed. Additionally, some portion of manure N is often in organic forms and not immediately available to plants after field application, a condition informed by testing the manure and soil for available N (NO₃–N and NH₄–N) (Paul and Beauchamp 1993). This organic N must eventually become available to crops and/or leach into the stream network, and thus it must be considered in watershed N budgets. Finally, it is likely some of the generated manures are being transported beyond watershed boundaries for application elsewhere. Long-range hauling (more than ~ 8 km), however, becomes economically problematic (Fleming et al. 1998) and there is evidence that farmers tend to apply manure on fields nearby confinements (Innes 2000; Jackson et al. 2000). Thus, the majority of manure generated within watersheds of the size studied here is likely to remain in that watershed. All things considered, the amount of purchased (commercial) N plus the amount of N generated by manure is far beyond crop nutrient requirements in some of these watersheds, and this surplus N will accumulate as decaying plant matter, soil organic matter and organisms, and soil water NO₃–N, creating a growing pool of mobile N (Jackson et al. 2000).

When 2017 water quality data are considered alongside these estimates of fertilization and generated manure, the Floyd and Rock watersheds stand out not only for their level of fertilization, but also for stream NO₃–N concentration levels. Despite the relative dryness in these two watersheds compared to the others, their FWA NO₃–N concentrations are nearly double those of the other seven when considered in aggregate (13.9 vs. 7.3 mg L⁻¹). Likewise, the commercial plus manure N in the West Nodaway and East Nishnabotna watersheds is only 60% of that in the Rock and Floyd, and this is reflected in stream water quality where NO₃–N concentrations are only 37% as high as in the Rock and Floyd watersheds. Howarth et al. (2012) estimated when net anthropogenic N inputs (NANI), similar to the surplus N described here, exceeded 1070 kg N km⁻² year⁻¹, 25% of this amount on average was exported to rivers worldwide. The average surplus N for our nine watersheds was 4489 kg N km⁻² during 2017, and an all-watershed average of 33% of this amount exited in the stream network. Our simple N budgets, which do not incorporate pathways such as atmospheric deposition of N and N returned to livestock in animal feed, still produce a value not that different from the Howarth et al. (2012) analysis, and our stream export values could be expected to exceed those of Howarth et al. (2012) because our surplus N is 4 times as large as the threshold in that study.

It is notable that the Rock watershed, with higher livestock densities than the Floyd watershed (Fig. 2), actually has lower levels of stream NO₃–N (Fig. 3). It is important to note that 58% of the Rock River watershed lies outside of the state of Iowa in Minnesota. The state of Minnesota Pollution Control Agency (MPCA) conducts NO₃–N water monitoring on the Rock River at a site about 18 stream-km north of the Iowa border (MPCA 2018). In 2017, six samples were collected by MPCA from April 26 to September 27 and averaged 8.48 mg L⁻¹ NO₃–N. The concentrations downstream at Rock Valley, Iowa, the site of the monitoring conducted for this study, were 10.04 mg L⁻¹ during that period. Thus, we suspect that lower concentrations of NO₃–N in water from Minnesota are diluting higher concentrations of NO₃–N in water contributed by Iowa portions of the Rock watershed.

Recently, the metric NO₃–N yield per unit of precipitation (g NO₃–N ha⁻¹ mm P⁻¹) was used to compare NO₃–N delivery in seven Iowa watersheds (Jones et al. 2018c). In the second 15 years of that study (2002–2016), an average of 22 g NO₃–N ha⁻¹ was mobilized to streams per mm of precipitation. For the 2017 water quality and hydrology data presented here, the nitrogen yield from the Floyd (40.2 g NO₃–N ha⁻¹ mm P⁻¹) and Rock (30.7 g NO₃–N ha⁻¹ mm P⁻¹) watersheds were considerably higher than the other seven, where the aggregated average was 22.0. The West Nodaway watershed received 204 mm more precipitation than the Floyd (27% more) but the NO₃–N yields were less than half, a clear indicator that the supply of loss-vulnerable N was far higher in the Floyd watershed compared to the West Nodaway.

Li et al. (2013) evaluated NO₃–N concentrations and trends for Iowa streams from 1998 to 2012. Of the 48 Iowa streams in that study that had sufficiently long data records for trend analysis, the Rock and Floyd Rivers had the largest positive trends for NO₃–N concentration (0.33 and 0.29 mg L⁻¹ year⁻¹, respectively) during a time when hog and cattle populations in the two watersheds were doubling (Fig. 2). The statistical significance of those trends was strong (p < 0.01). The other sites in that study that were also evaluated here included the West Nodaway River (increasing trend of 0.17 mg L⁻¹ year⁻¹), Boyer River (0.16 mg L⁻¹ year⁻¹), West Nishnabotna River (0.06 mg L⁻¹ year⁻¹), Soldier River (0.03 mg L⁻¹ year⁻¹), and Little Sioux River (0.03 mg L⁻¹ year⁻¹). Without detailed information about manure nitrogen quantities, Li et al. (2013) speculated that manure applications associated with increasing hog populations were a driving factor for the upward NO₃–N trends in western Iowa, and we believe the data presented herein are consistent with that.

There are examples in the literature linking livestock concentration with surplus stream nutrients and degraded water quality in other parts of the world. For example, the northwestern Black Sea was seriously degraded from the 1960s to the 1980s by nutrient runoff from the Danube River, but rapidly improved after 1989 with the closure of many large animal farms as a result of the fall of communist regimes (Mee 2006). Considering that the average animal unit density in our study was 1.60 ha⁻¹, the average magnitude of N surplus we report (99 kg ha⁻¹) is consistent with research from other agricultural regions. For example, Wang et al. (2018) reported average N surpluses of 75-306 kg ha⁻¹ when AU density exceeded 1 ha⁻¹ for several countries in Europe, Asia, and the Americas. Oenema et al. (2007) reported highest levels of NO₃–N leaching in Europe to be in the northwest where livestock densities were highest. Leaching rates 20 to over 50 kg ha⁻¹ were reported in that study, compared to 30.5 and 24.7 kg ha⁻¹ for the Floyd and Rock watersheds, the two highest-density livestock watersheds of the nine assessed here. When considering the NO₃–N transported by these streams, and especially the Rock and Floyd Rivers, it is relevant to consider how this pollutant links to various processes that control stream amounts. At the landscape scale in the U.S. cornbelt, the NO₃–N loading is clearly transport-limited (Sprague et al. 2011; Jones et al. 2017). However, there are years within individual watersheds where supply limitations are controlling (Jones et al. 2017). Furthermore, fertilizer nitrogen has been shown to be a strong predictor and regulator of stream NO₃–N concentrations (David et al. 2010; Li et al. 2013). The Floyd and Rock were the two driest watersheds evaluated here, but still had by far the highest NO₃–N delivery of these nine western Iowa basins. The fact that in the Floyd and Rock watersheds, the commercial N inputs combined with generated manure N were nearly double the other watersheds illustrates the importance of N supply management for water quality improvement. There is ample evidence that U.S. Cornbelt farmers over-apply nitrogen, often in manure forms (Yadav et al. 1997; Jackson et al. 2000; Sheriff 2005; Khanal et al. 2014). This is not necessarily wasteful; rather, the economics of nitrogen can make it more profitable for farmers to concentrate manure applications on nearby fields and purchase chemical fertilizer for the rest of the farm (Letson et al. 1998). In fact, Jackson et al. (2000) concluded that in some scenarios it makes clear economic sense for large livestock confinements to maximize N volatilization losses. In these circumstances, manure becomes a waste product and the practice of squandering manure nutrients itself is not necessarily economically wasteful (Fleming et al. 1998; Sheriff 2005), i.e., the farmer may benefit financially by not fully taking advantage of the fertility benefits available in the generated manure. Many farmers may also manage manure application rates based not on N, but rather phosphorus and/or potassium. Farmers also may apply manure in the fall, followed by commercial fertilizer applications the following spring.

Interestingly, the N inputs in the West Nodaway watershed are in line with ISU recommendations for corn cultivation, and the FWA NO₃–N concentration was a relatively modest 4.9 mg L⁻¹ and the daily concentration never exceeded the safe drinking water standard of 10 mg L⁻¹. This watershed illustrates the obvious opportunity for farmers and policy makers to make progress towards Iowa’s water quality goal of a 45% NO₃–N load reduction (Iowa Nutrient Reduction Strategy 2013). Figure 4a indicates that reducing surplus N by better balancing inputs relative to expected crop needs would reduce stream NO₃–N levels. When considering Fig. 4a, extrapolating the regression backward to a zero surplus N condition results in a FWA NO₃–N concentration of < 1 mg L⁻¹. We acknowledge that legacy N (Van Meter et al. 2017) may elevate stream NO₃–N for prolonged periods after inputs are balanced with crop requirements and that the extrapolation in 4(A) is somewhat speculative. Nonetheless, it is apparent that better management and accounting of manure inputs could generate significant and rapid progress towards Iowa’s water quality objective for stream N. The surplus N relates much more strongly to generated manure N (R² = 0.83) than commercial N inputs (R² = 0.14) among the nine watersheds and therefore this suggests a starting place when assessing inputs on the watershed scale. These findings are consistent with Khanal et al. (2014), who determined that manure-fertilized rotations had a higher net N (i.e., difference between inflows and outflows) statewide in Iowa. While the amount of N generated in livestock manure is not a precise estimate of what will be available to the receiving crop, methods exist to help reduce this uncertainty (Paul and Beauchamp 1993) and integration of commercial fertilizer and manure recommendation systems that account for soil fertility, crop needs, and availability of manure N is needed (Liu et al. 2017). With commercial fertilizer sales seemingly unrelated to the availability of manure N in these watersheds, refinements in planning and manure management hold great potential for producing water quality improvement in areas where livestock has been concentrated. Several policy recommendations were proposed by Jackson et al. (2000) to address similar issues in Central Iowa. These included alternative livestock housing, increased regulatory scrutiny of manure management plans, modification of land zoning rules, and incentivizing extended crop rotations that include small grains and forage legumes. Although now nearly 20 years old, we wish to emphasize that while these recommendations have mostly gone unheeded, they continue to hold potential for more efficient nitrogen use and water quality improvement.

Conclusions

While commercial fertilizer nitrogen input rates are similar among these nine western Iowa watersheds, generated manure N is far higher in two, the Floyd and Rock River watersheds, and the FWA NO₃–N concentrations at the outlets of these watersheds are approximately double that of the other seven. The commercial N inputs plus the generated manure N in these two watersheds total 370-380 kg corn ha⁻¹, which is about double the recommended application rates. The FWA NO₃–N concentration was significantly correlated with total N inputs and generated manure but not with commercial N fertilizer amounts. The only watershed where commercial fertilizer N inputs plus generated manure was consistent with the rate recommendations was the West Nodaway watershed, where the FWA NO₃–N concentrations were lowest and never exceeded 10 mg L⁻¹. Overall, the results from this study strongly suggest that better management of manure holds promise for producing significant water quality improvements at a watershed scale.

Biographies

Christopher S. Jones

is a Research Engineer at the IIHR-Hydroscience and Engineering, University of Iowa. His research interests are contaminant hydrology and water monitoring.

Chad W. Drake

is a Graduate Research Assistant at IIHR-Hydroscience and Engineering, University of Iowa. His research interests include watershed processes and modeling.

Claire E. Hruby

is a Geologist with Iowa Department of Natural Resources. Her research interests are emerging contaminants and fate and transport of manure-related pollutants.

Keith E. Schilling

is the State Geologist at the Iowa Geological Survey. His research interests focus on hydrogeology and watershed processes.

Calvin F. Wolter

is a GIS Analyst with Iowa Department of Natural Resources. His research focuses on agricultural sustainability.