Nh

Sa

b

c

d

a

ARRAA

KNCCS

1

mpM1GwoD

o

UT

0h

Agriculture, Ecosystems and Environment 183 (2014) 21–30

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment

j ourna l h om epage: www.elsev ier .com/ locate /agee

itrogen balance in Iowa and the implications of corn-stoverarvesting

ami Khanala,b,∗, Robert P. Anexa,∗∗, Brian K. Gelderc, Calvin Wolterd

Department of Biological Systems Engineering, University of Wisconsin-Madison, Madison, WI 53706, United StatesDepartment of Forest and Wildlife Ecology, University of Wisconsin-Madison, Madison, WI 53706, United StatesDepartment of Agricultural and Biological Systems Engineering, Iowa State University, Ames, IA 50011, United StatesIowa Geological Survey, Iowa Department of Natural Resources, Iowa City, IA 52242, United States

r t i c l e i n f o

rticle history:eceived 16 November 2012eceived in revised form 7 October 2013ccepted 10 October 2013vailable online 16 November 2013

eywords:itrogen balanceorn-involved crop rotationsorn stover harvestingynthetic-N and manure fertilizers

a b s t r a c t

Increased corn production and removal of corn stover for biofuel production can adversely affect waterquality, soil fertility and productivity due to low nitrogen (N) use efficiency. In this study, the averageannual county-level N balances in Iowa are calculated for three corn-involved rotations: corn-soybean(C-S), corn-corn-soybean (C-C-S) and continuous corn (C-C), receiving either synthetic-N or manure fer-tilizer under 0, 30, 50 and 75% corn stover removal scenarios. Geo-referenced data on soil, crop andlivestock are used to estimate net changes in total N balance in the mineral form after accounting all soilN inflows and outflows. Under a zero stover removal scenario, a state average for net N was 34 kg ha−1 yr−1.Approximately 86% of the land area in the three corn-involved rotations receives synthetic-N fertilizer,and 24% of total synthetic-N treated land is estimated with net N of 24 kg ha−1 yr−1 (i.e., average net N forsynthetic-N treated rotations) or more. Manure-treated rotations are estimated to have 2–6 times highernet N than synthetic-N treated rotations; continuous corn rotations contributing to a higher net N. Thenorthern and central crop districts dominated by animal and corn production have higher net N. Removalof corn stover reduces net N, and synthetic-N treated rotations are estimated to be affected the most. The

−1 −1

percentage of total synthetic-N treated rotations estimated with net N of 24 kg ha yr or more lowersfrom 24% under no stover harvesting to 3% under 75% stover (mass basis) removal scenario. Conversely,the areas with negative net N increase from 1% with no stover harvesting to 10% under 75% harvesting.This study will help prioritize the regions in which management practices that reduce nitrogen loss aremost needed, and those regions most suitable from a nutrient balance standpoint for sustainable stover harvesting.

. Introduction

Row crop production and animal feeding operations are theajor contributors of plant nutrients, particularly nitrogen (N) and

hosphorus (P), exported from the Mississippi River to the Gulf ofexico (David et al., 2010; Alexander et al., 2007; Goolsby et al.,

999). Increased concentration of these nutrients in the northernulf of Mexico has increased the size of the Gulf hypoxic zone

hich has averaged over 15,600 km2 since 1993, making it one

f the largest hypoxic zones in the world (Rabalais et al., 2002).epletion of valuable fisheries and the disruption of ecosystem

∗ Corresponding author at: Department of Forest and Wildlife Ecology, Universityf Wisconsin-Madison, Madison, WI 53706, United States.∗∗ Corresponding author at: Department of Biological Systems Engineering,niversity of Wisconsin-Madison, Madison, WI 53706, United States.el.: +1 608 262 3310; fax: +1 608 262 1228.

E-mail addresses: skhanal2@wisc.edu (S. Khanal), anex@wisc.edu (R.P. Anex).

167-8809/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.agee.2013.10.013

© 2013 Elsevier B.V. All rights reserved.

services are economic and ecological consequences of hypoxiczones. Donner and Scavia (2007) suggested a 50–60% N load reduc-tion is necessary to reach the Gulf hypoxia area goal (i.e., 5-yearrunning average hypoxic zone < 5000 km2) laid out in the HypoxiaAction Plan. Nutrient loss from agricultural land also has humanhealth impacts. Long-term exposure to nitrate (NO3

−) in drinkingwater derived from impacted water sources increases the risk ofcancer in humans (Weyer et al., 2001). The concentration of NO3-Nin agricultural drainage often exceeds the drinking water standardof 10 mg NO3

−-N/L and annual NO3− loadings from farms in the

Midwestern region of U.S. are often over 66 kg NO3−-N ha−1 (Kalita

et al., 2006).It is expected that the implementation of the revised Renew-

able Fuel Standard (RFS2) will put more land in the MississippiRiver Basin into agricultural production either through the transfor-

mation of uncultivated pasture and conservation reserve program(CRP) lands or conversion of other cropping systems to more corn-involved systems (Marshall et al., 2011). In turn, nutrient transportfrom the basin is expected to increase, further compromising the

2 tems a

hcm2iogta(rvMecbwc2W

tptipctRelrcMMos2

acrcpcpncsfa

NiStaca(atl2tu

2 S. Khanal et al. / Agriculture, Ecosys

ypoxic zone reduction goal (Donner and Kucharik, 2008). Higherorn production would also make it more difficult to meet the totalaximum daily load limits (TMDL) under the Clean Water Act (EPA,

012) resulting in additional water bodies not meeting water qual-ty standards. The RFS2 mandates by 2022 the annual productionf 56.8 and 60.5 billion liters (i.e., 15 and 16 billion gallons) of corn-rain and ligno-cellulose based ethanol, respectively. Corn stover,he residues left in fields after corn grain harvest, has been targeteds a readily available feedstock for cellulosic ethanol productionPerlack et al., 2005). Corn stover harvesting has the potential toeduce soil drainage N concentration compared to no stover har-esting in corn-involved cropping systems (Zavattaro et al., 2012;eki et al., 2011). The decision to harvest stover should be consid-

red carefully as it also has the potential to decrease soil organicarbon (SOC); reduce soil microbial activity; increase soil suscepti-ility to compaction, runoff, and soil erosion; reduce nutrient pools,ater retention, and soil fertility, and hence soil productivity and

rop yield (Zavattaro et al., 2012; Blanco-Canqui, 2010; Karlen et al.,009; Graham et al., 2007; Grignani et al., 2007; Andrew, 2006;ilhelm et al., 2007).Iowa is the largest corn, soybean and hog producing state in

he U.S. (NASS, 2012). A large fraction of the total land area islanted with corn and soybean (39 and 26%, respectively). Also,here is a high geographical concentration of large and special-zed confined animal feeding operations (CAFOs), the vast majorityroducing hogs. Most animal waste is disposed through land appli-ation. Large CAFOs typically own less land than would be requiredo use the manure they produce agronomically on their own fields.ecent increases in the number and size of CAFOs have raisednvironmental concerns related to over-application of manure andeaching of excess nutrients from the fields to surrounding wateresources. During the flood year of 1993, Iowa was estimated toontribute about 35% of the total nitrate discharged to the Gulf ofexico although it represents only 4.5% of the total land area ofississippi-Atchafalaya River Basin (Goolsby et al., 1999). Based

n water quality monitoring data, water in the majority of Iowatreams during 2000–2012 was classified as “poor” quality (IDNR,013).

The nutrient balance is often used as an indicator of a system’sbility to maintain soil fertility as well as water quality in adja-ent ecosystems (Grignani et al., 2007). Nutrients in excess of cropequirements are prone to leaching while nutrients in deficit ofrop requirements reduce soil fertility, which in turn lowers croproductivity. Quantifying the nutrient balance of corn-involvedrop rotations under different fertilizer and residue managementractices can help us to develop effective policies addressingutrient-related environmental problems, and achieve sustainableorn stover harvest for biofuel production. The objectives of thistudy are to examine the spatial distribution of N balance under dif-erent corn-based cropping systems, manure management systemsnd corn stover harvesting scenarios in Iowa.

A county level N balance model that links the stocks and flows of in crops, animals, and soil is developed to estimate the N balance

n three major corn-involved crop rotations (i.e., corn-soybean (C-), corn-corn-soybean (C-C-S) and continuous-corn (C-C)), underwo fertilizers (i.e., manure and synthetic-N), and with 0, 30, 50nd 75% corn stover removal. Traditionally, a limited amount oforn stover has been harvested for animal feed and bedding, with

substantial portion of the residues being returned to the soilKarlen et al., 2011). It is assumed that all of the stover residuesre left on the ground, and we consider 0% stover harvesting ashe “business as usual (BAU)” scenario. The model in this study fol-

ows the approach used in previous studies (Burkart et al., 2005,006). Burkart et al. (2005) examined leachable N under alterna-ive land use scenarios that considered an increase in land areasnder perennial cover, integrated livestock with cropping systems,

nd Environment 183 (2014) 21–30

and reduced use of inorganic fertilizer in western Iowa watersheds.In contrast to the work of Burkart et al. (2005), the model devel-oped here is used to estimate leachable mineral N and soil organicN at near steady-state conditions.

Compared to existing models, such as EPIC (Williams et al.,1984), DAYCENT (Del Grosso et al., 2001), DNDC (Li et al., 1992), andAgro-IBIS (Kucharik, 2003; Kucharik et al., 2000), the model used inthis study is simple, needs fewer and more accessible inputs, andcaptures major N dynamics in the agricultural systems at countyand watershed scales. This model includes physical processes, suchas loss of ammonia (NH4) during plant senescence and redeposi-tion of locally derived NH4, which are generally ignored in detailedprocess-based models (Burkart et al., 2005).

The N balance in this study is reported as net mineral N. “NetN” is the difference between total inflows and outflows of N in thesoil system. Positive net N (i.e., surplus N) indicates that the systemhas N inflows in excess of outflows, and surplus N has a potentialto leave agricultural fields through leaching or runoff during wetyears. Negative net N (i.e., deficit N) indicates a system with higherN outflows than inflows. Such a system is losing N over time andwill eventually require additional N to maintain ecosystem functionand crop productivity.

2. Methodology

2.1. Study area

Our study area, the state of Iowa, is located in the north-centralpart of the United States (Fig. 1); it is a core part of the WesternCorn Belt Plains Ecoregion (Omernik, 1987). Soils in the central andnorthern regions have relatively lower clay content, higher bulkdensity, and higher SOC stock levels than other parts of the state.Average annual precipitation varies from 710 mm in the north-east to 965 mm in the southwest, and average annual minimumtemperature is 0.5 ◦C in the northeast and 6.1 ◦C in the southwest.Cropland accounts for more than 85% of the state, and the majorcrops, corn and soybean, dominate the landscape. Figs. S2 and S3 inthe supporting material (SI) provide county level information aboutthe distribution of various cropping systems, animals and fertilizermanagement practices in Iowa.

2.2. N model

Soil nitrogen, both inorganic and organic (i.e., mobile and immo-bile forms of N), is influenced by the addition of crop residues,livestock manure and atmospherically deposited nutrients. Soilorganic nitrogen (SON) is not directly available to plants, and isprotected from loss via leaching, whereas mineral N is availablefor a crop to uptake, and can leave the system through runoff orleaching during wet seasons. The N model developed in this studyrepresents organic N dynamics through the inclusion of three soilorganic pools, which are regulated through the mineralization andimmobilization processes (Burkart et al., 2005), as shown in Fig. 2.The system N balance is calculated as:

Net N = Fert + Mm + Mr + Mson + Fix + Atm − CropU − FertVol

− CropV − Deni (1)

where Net N is total N in the mineral form after accounting allsoil N inflows and outflows, Mm is fertilizer application, Mr is miner-alized manure, Fert is mineralized residue, Msonis the contribution

of mineralized N from three soil organic pools (i.e., Active, Slowand Passive), Fix is atmospheric N fixation, Atm is atmospheric Ndeposition (wet and dry), CropU is crop uptake, FertVol is fertilizerapplication loss via volatilization, CropV is ammonia volatilization

S. Khanal et al. / Agriculture, Ecosystems and Environment 183 (2014) 21–30 23

F ricts ie

fEMDrtce

WGusl

Fea

ig. 1. Distribution of crop rotations as a proportion of total cropland, by crop distach crop districts.

rom crop residues, and Deni is N loss through denitrification.xcept CropV, all N flows vary with crop rotation (i.e., Fert, Mm,r, Mson, Fix and CropU) or with geographic location (i.e., Atm andeni) (see SI for details). The net N is calculated for the entire crop

otation and not for each crop separately, as crops planted in rota-ion can take advantage of residual N from fertilizer applied to arop earlier in the rotation and from N in crop residues (Grignanit al., 2007).

The N model is run until SON reaches a near steady state.e do this because the estimates of SOC from the Soil Survey

eographic database (SSURGO), upon which SON depends, arencertain (Liu et al., 2011), and we wish to determine the newteady state for comparison of net N under various stover col-ection scenarios. An exact steady state condition can rarely be

ig. 2. Soil organic nitrogen pools (A) included in estimation of soil steady-state nitrogstimated after accounting for build-up of N in these pools through immobilization of mre summed to estimate total N in mineral form, (“net N” in B). A surplus indicates that th

n Iowa, USA. Size of pie charts is proportional to the total corn production area in

achieved on an annual basis because of the inter-annual variabilityin soil processes that are driven by changes in climate, atmosphericcarbon-dioxide concentration, crop yield, nitrous-oxide emissionand other soil microbial processes. In a multiyear N budget, inter-annual changes in SON within an acceptably small range (i.e., lessthan 15 kg N ha−1 yr−1) is a useful approximation of soil in a steadystate. This magnitude of change in SON is relatively small com-pared to other uncertainties in N budget components, such as N2fixation, denitrification or the changes in soil inorganic N (Schepersand Raun, 2008). The N model is run until annual change in SON pool

is less than or equal to 15 kg N ha−1 yr−1. Annual changes in bothslow and passive SON pools are considered for this purpose becausethey comprise the largest fraction (i.e., 99%) of the SON pool (TableS5), and have a longer residence time (i.e., greater than 20 years)

en balance (B). Mineralized N from three soil pools: active, slow and passive, isanure and crop residue (A). Total mineralized N from soil pools and other N flowse total inflow is higher than total outflow, while a deficit indicates the opposite.

2 tems a

tcgSeaCan

2

fodycdpmWnPiw

wwllDdyCueAotitaceptal

2al2aIdiwocmCNia

4 S. Khanal et al. / Agriculture, Ecosys

han the active pool (i.e., 1–5 years) and also reflect the long-termhanges in soil due to management practices. Soil organic nitro-en obtained from the SSURGO database is used as the first yearON level. We use the Mson at steady state and the N inflows tostimate net N (Fig. 2). We estimate net N at a county level using

weighted average of the net N of the three crop rotations (C-S,-C-S and C-C) under two fertilizer practices. We compute net N at

crop district level through a weighted average of the county levelet N estimates.

.3. Dataset

We use county information about crop and animal production,ertilizer use, climate, and soil properties in estimating the vari-us N flows in and out of the soil (Fig. S1; SI provides details aboutata sources and N estimation processes). Information about cropields, harvested and planted crop areas, and livestock numbers areollected from the National Agricultural Statistics Service (NASS)atabase. These NASS data are collected on a 5-year cycle. Theeriod of our analysis is 2007–2009 and this was determined by theost recent NASS data (i.e., data of 2007) at the time of the analysis.et N deposition at a county level is estimated based on ammo-

ia deposition measured by the National Atmospheric Depositionrogram (NADP) at different sites in Iowa. Climate data (i.e., precip-tation and mean temperature) obtained from the PRISM database

ere averaged for the period 2007–2009.Data representing the land area in the different crop rotations as

ell as fertilizer management practices (i.e., percent land treatedith manure or synthetic N, and fertilizer application rate in those

ands) are not available at the state or county levels. Cropland dataayers for the period of 2007–2009, available from the United Statesepartment of Agriculture (USDA), are used to derive crop rotationata (i.e., a sequence of crops grown on a field over a period of twoears or more) at the county level. Crop rotations, including C-S, C--S, and C-C rotations dominate the use of Iowa cropland, and thussed in our study. Total land areas under these three rotations arextracted at the county level using the “Zonal Statistics” function ofrcGIS (ESRI, 2009). These three rotations account for 45, 5 and 9%f the total Iowa land area, respectively. Together, these three rota-ions cover 62% of the total corn and soybean planted areas of Iowan 2007. Other rotations that include corn and/or soybean make uphe remaining 38%, such as continuous soybean, corn-soy-wheat,nd corn-soy-alfalfa-alfalfa. The rate of synthetic-N fertilizer appli-ation to these three crop rotations (i.e., C-S, C-C-S, and C-C) isstimated from the state’s total synthetic-N fertilizer use, manureroduction and total cropland. Manure-treated area in each rota-ion is computed at a county level based on total manure-treatedrea, the geographical distribution of CAFOs, and the proportion ofand area in each crop rotation near the CAFOs.

Corn has a higher N requirement than other row crops (DEP,001), so CAFO operators are able to apply manure to corn fieldst a high rate while staying within nitrogen application guide-ines and minimizing manure transportation costs (Ribaudo et al.,003; Khanal et al., 2012). Ribaudo et al. (2003) reported 4.18 kms an average distance that a large hog farm transports manure.t is assumed that manure is applied to corn fields in the imme-iate neighborhood of a CAFO, and that all manure-treated land

n a county is in a corn-involved crop rotation. For each county,e estimate the proportion of area within 5 km of a CAFO that is

ccupied by each crop rotation. This proportion is multiplied by theounty’s total manure-treated land area to estimate the amount ofanure-treated land in each rotation. The geographical locations of

AFOs are obtained from the GIS library of the Iowa Department ofatural Resources (IDNR). The land area in each crop rotation that

s treated with synthetic-N (i.e., non-manured fields) is estimateds the difference between the total area in the rotation and the

nd Environment 183 (2014) 21–30

manure-treated area in the rotation. A manure-treated crop rota-tion is assumed to also be fertilized with synthetic-N if the availablemanure N cannot meet the calculated corn N demand (MacDonaldet al., 2009). The mineral form of the applied manure is comparedwith corn N demand to estimate necessary additional synthetic-Ninput. Corn N demand is estimated as a function of average cornyield and the cropping system, accounting for the role of nitrogen-fixing soybean in the C-S systems (Vitosh et al., 1995). It is estimatedthat 86% of the total land in the three corn-involved crop rotationsare treated with synthetic-N fertilizer only.

2.4. Net N and stream N load

As an evaluation of model prediction accuracy, net N estimatesare compared to measured stream N loads. Since a positive N bal-ance suggests that there is mineral N available for leaching, weexpect that a watershed predicted to have high net N will alsoexhibit a high stream N load. Net N at the watershed scale is com-puted as a weighted average of the net N of the counties thatcomprise the watershed. The watershed net N is compared withthe total stream N loads from the watershed as estimated by theIDNR. This comparison is made for 79 watersheds across Iowa. Thecontribution of each watershed to measured N load in a stream isbased on a regression model (i.e., a LOAD estimator developed bythe U.S. Geological Survey) that uses the annual or seasonal vari-ations in streamflow volume, time, and stream N concentration ofthe watersheds for water-years 2000–2009. The model with thebest fit is used to calculate daily N loads from the stream flow data.The total N load is calculated as NO3-N/0.82 as IDNR monitoring hasshown that, on average, NO3-N concentration is 82% of the total Nconcentration. A multiple linear regression analysis is conducted toestimate the explanatory power of independent variables, includ-ing net N, soil characteristics, including percent of soil, clay, siltand SOM, and percent of tile-drained land in the watershed in pre-dicting stream N loads in a watershed. The Akaike’s informationcriterion (AIC), a way of selecting a model from a set of models(Hilbe, 2009), is used to compare the different models. The modelwith the minimum AIC value is considered to be the best model topredict stream N loads.

2.5. Sensitivity analysis

Previous N balance studies have adopted a range of values forimportant independent variables used to estimate the balance.Some of these variables include the fraction of N in soybean residue,corn and soybean grain, manure and SOM; the percent N thatis re-deposited from the atmosphere; the manure N mineraliza-tion and immobilization rates; the denitrification rate; percent Nloss during fertilizer application; SOM residence time; and C toN ratio in the different soil pools (Table S7). In addition to thesevariables, there are uncertainties and inter-annual variability inagricultural and climate data. For example, corn planted area inIowa decreased from 5.6 million hectares in 2007 to 5.38 millionhectares in 2008, whereas soybean planted areas increased from3.48 million hectares in 2007 to 3.88 million hectares in 2008. TotalN fertilizer use in Iowa varied between 1.2 million metric tons in2007 and 0.9 million metric tons in 2008. Also, there are uncertain-ties in land use/land cover classification in the cropland data layers(Boryan et al., 2011), and in NASS county-level yield information(Cooper et al., 2012). Due to these uncertainties in inputs to the Nmodel, we examine the sensitivity of the N balance to the modelparameter values.

Sensitivity analyses are conducted by individually varying thevalue of each parameter by ±10% of its mean value. Table S7 in the SIprovides the details about the variables used in sensitivity analyses.Several variables, such as manure nutrient content, manure storage

S. Khanal et al. / Agriculture, Ecosystems and Environment 183 (2014) 21–30 25

Fig. 3. Geographical distribution of net N, computed as a weighted average of net N resulting from three crop rotations (C-S, C-C-S and C-C) under two fertilizer practices(i.e., manure and synthetic N) at crop districts level, under 0% stover harvesting scenario. The bar graphs show the relative proportion of net N produced per hectare treatedw entaget l, SWE

arfovuiafic(Nri

sNam(el

3

d

3

3

fStio

ith manure and synthetic-N. The map in the upper right corner indicates the perche crop district boundaries. Crop districts are NW – North West, WC – West CentraC – East Central, and SE – South East.

nd application losses, manure mineralization and immobilizationates, SOM turnover periods and recommended N application ratesor row crops, have subcategories representing the different typesf animal, soil pools, or crops (Table S7). The sensitivity of theseariables that have subcategories are estimated by varying the val-es in all subcategories at once. Nitrogen fixation by soybean is not

ncluded in the sensitivity test because it is contingent upon avail-ble soil N in the model. Soybean is not usually fertilized and its Nxation rate varies across the range of 0–185 kg N ha−1 yr−1 indi-ating that soybean fields usually have near zero or zero N balanceRusselle and Birr, 2004). Soybean only fixes significant amounts of

if available N is insufficient to meet its need, which is extremelyare in a C-S rotation due to the N available from the heavily fertil-zed corn phase of the rotation.

Results of net N sensitivity analysis are presented at both thetate and county levels, focusing mainly on the sensitivity of net

to variables under the two fertilizer management practices andt different geographic locations. The two counties that are esti-ated to have the highest (i.e., Hardin county) and lowest net N

i.e., Clarke) under the zero corn harvesting scenario, are used toxemplify the range of sensitivity results at different geographicocations.

. Results

The nitrogen balance is examined in three cropping systems at county and cropistrict levels to analyze its geographic variability in Iowa.

.1. Cropping systems, nitrogen balance, and harvesting practices

.1.1. Net N under zero stover harvesting scenarioThe county level N balance with no stover harvesting is estimated to range

rom 5 to 76 kg ha−1 yr−1, with a state average of 34 kg ha−1 yr−1 (Fig. S8 and Table9). Manure-treated cropping systems have higher positive net N than synthetic-Nreated systems. Average net N is estimated to be 96 kg ha−1 yr−1 and 24 kg ha−1 yr−1

n the manure and synthetic-N treated crop rotations, respectively (Table S9). Allf the manure-treated crop rotations are found to have net N that is higher than

of tile-drained area in each county. The thick black lines in the large map indicate – South West, NC – North Central, C – Central, SC – South Central, NE – North East,

the average net N in synthetic-N treated rotations. Only 35% of the total landarea in synthetic-N treated crop rotations is estimated to have net N higher than24 kg ha−1 yr−1. The continuous corn rotation is estimated to have 2–4 times highernet N than C-C-S and C-S rotations. Average net N in the manure-treated C-C rota-tion (i.e., 207 kg ha−1 yr−1) is almost 4 times higher than the average net N in thesynthetic-N treated C-C rotation (i.e., 63 kg ha−1 yr−1), mainly due to a higher Napplication rate on manure-treated land (Table S9 and Fig. S3).

Average net N in the South Central crop district is estimated to be about half ofthe net N estimated in crop districts, such as North East, North West, North Cen-tral and Central districts (Fig. 3 and Table S9). While synthetic-N treated croppingsystems in the counties of this district are predicted to have negative net N, manure-treated crop rotations have net N higher than the counties in other districts. Of thetotal land in the three rotations treated with synthetic-N, 1% is predicted to have neg-ative net N, and these lands are concentrated mainly in the South Central crop district(Fig. S8). Due to large fraction of corn involved rotations treated with manure, cropdistricts, such as North West, North East and North Central have higher net N thanother districts. Comparatively low use of synthetic-N fertilizer, but higher manureper unit of crop land, results in a highly disproportionate distribution of net N in theSouth Central crop district.

3.1.2. Nitrogen balance under stover harvesting scenariosRemoval of corn stover lowers net N; the synthetic-N treated corn involved

rotations are impacted more than manure treated rotations. Relative to the baselinescenario (i.e., 0% stover removal), net N is reduced by 28, 47 and 71% in synthetic-Ntreated rotations, and by 8, 14 and 21% in manure-treated rotations, under the 30,50 and 75% stover removal scenarios, respectively. Stover removal also reduces soilN mineralization. Relative to the baseline scenario, mineralized N declines by 14,23 and 34% in the synthetic-N treated rotations, and by 10, 17 and 26% in manure-treated rotations under the 30, 50 and 75% stover removal scenarios, respectively(Fig. S7).

Relative to no stover harvesting, harvesting 75% of stover reduces the total areawith net N higher than the average (i.e., 24 kg ha−1 yr−1) from 24 to 3%, and increasesthe total area in negative net N from 1 to 10% (Fig. 4). Areas with net negative Nunder various stover removal scenarios are mainly in synthetic-N treated rotations,and these areas are concentrated mainly in the South Central crop district (Figs.S9–S11). Even at 75% stover removal, all manure-treated rotations have net N of

24 kg ha−1 yr−1 or more. The spatial distribution of net N at 50% stover harvestingat crop district level is shown in Fig. 5. In districts, such as North East, West Centraland East Central, reduction in net N for C-C rotation under 50% stover harvestingrelative to no harvesting is higher than other rotations (Tables S9 and S13). Sincethe large fraction of corn involved rotations in these districts are composed of C-C

26 S. Khanal et al. / Agriculture, Ecosystems a

Fig. 4. Impact of corn stover harvesting on cropland area with net N higher than24 kg ha−1 yr−1 (left axis) or lower than 0 kg ha−1 yr−1 (right axis). * average net N(i.e., 24 kg ha−1 yr−1) is computed as the average of net N in all synthetic-N treatedria

rTi

3

etidsgp

are high in the watersheds of the North Central, Central, North East and East Central

FCa

otations. Values on the y-axes are estimated as the percentages of the total landn the three synthetic-N fertilized corn-involved crop rotations that have aboveverage (left axis) or negative net N (right axis).

otation, 50% of stover harvesting influences these districts the most than others.he county level distribution of net N at 30, 50 and 75% stover harvesting is providedn the supporting materials (Figs. S9–S11).

.2. N model validation

To validate the model predictions, estimated net N and SON are compared withxperimental mineralized N and SON data from field-scale experiments reported inhe literature. It is, however, important to recognize that the N balance estimatesn this study represent near steady-state conditions in the soil, so these model pre-

ictions cannot be compared directly with empirical field data that represent nonteady-state conditions. Nonetheless, the ability of the model to predict observedeographical trends in mineralized N and SON provides some measure of modelerformance.

ig. 5. Net N under 50% (lower left map) and 0% (upper right map) corn stover harvestin-C-S and C-C) under two fertilizer practices (i.e., manure and synthetic N). The bar graphre the counties used for sensitivity analyses in Fig. 7C and D.

nd Environment 183 (2014) 21–30

3.2.1. Mineralized N and SONMineralized N from C-S and C-C rotations across Iowa are estimated to be in the

ranges of 50–58% and 41–47% of total N required by corn in the C-S and C-C rotations,respectively. These estimates fall within the lower range found by Sawyer et al.(2006) from over 300 field experiments carried out in Illinois, Iowa, Minnesota, andWisconsin. Sawyer et al. (2006) demonstrated that 50–70% of the corn yield in Iowais the result of N contributed by soil N alone, and soil N provides about 45 and 75%of total N required by corn in C-C and C-S rotations, respectively. The slightly lowerestimate of the contribution of mineralized N from soil to corn yield from the modelcompared to the findings of Sawyer et al. (2006) is expected because the corn fieldsin Iowa are not at steady state, and soil organic N is depleted over time despite theincorporation of crop residues and manure into the soil (Paustian et al., 1997). Themodel predicts 11% higher SON in C-C rotations than in C-S rotations. This finding issimilar to other field studies that have shown higher C and N storage (presumablySOC and SON) in C-C than in C-S (Jagadamma et al., 2008; Varvel, 1994), or in cornrotated with other legume crops (Bertora et al., 2009; Grignani et al., 2007). Basedon 8 years of field experiments in Illinois, Coulter et al. (2009) demonstrated 13 and16% higher total N in C-C than in C-S at two soil depths 0–15 and 15–30 cm.

3.2.2. Crop management practice and soil steady stateWe find that manure-treated rotations reach near steady state in less time than

synthetic-N treated rotations. Approximately 58 (standard deviation (SD): 22) and28 (SD: 19) years are required for manure-treated C-S and C-C rotations, and 71 (SD:13) and 48 (SD-11) years for synthetic-N treated C-S and C-C rotations to reach nearsteady state. These times to reach steady soil state fall within the ranges demon-strated by long term field studies. Kubat and Lipavsky (2006) demonstrated that soilplanted with different row crops in Prague-Ruzyne in the Czech Republic took 20–50years to reach steady state. Although not directly comparable with the corn-involvedrotations examined in this study, long term experimental sites planted with barleyand treated with manure in a monoculture system in Rothamsted, England required50 years for total N to reach near steady state (Jenkinson and Rayner, 1977).

3.2.3. Net N and stream N loadThe geographical distribution of net N is found to be similar to the pattern of

IDNR estimates of stream N load (personal communication, Calvin Wolter, IowaDNR). The nitrogen load to streams in the watersheds of Iowa estimated by theIDNR ranges between 2 and 44 kg ha−1 yr−1, and predicted net N ranges between 17and 63 kg ha−1 yr−1 (Fig. 6). While comparing the IDNR estimates of nitrogen load tostreams with net N estimates on watershed basis, both the stream N loads and net N

crop districts. In the South Central crop districts, both the net N and stream N loadsare low. This congruence of net N which is “available for leaching” and estimated Nload to streams suggests that the model captures in general way the spatial patternof the N balance of Iowa. However, estimated net N poorly predicts stream N load in

g scenarios. Net N is a weighted average of the net N of three crop rotations (C-S, shows the relative proportion of net N per ha of fertilizer treatment. Open circles

S. Khanal et al. / Agriculture, Ecosystems and Environment 183 (2014) 21–30 27

F (uppe(

ta

atlvTw

3

suaaccio(actN

pcrp

4

4

sote

ig. 6. Net N based on N-balance by watershed (lower left) and N load to streamspersonal communication, Calvin Wolter, Iowa DNR).

he North West and West Central crop districts. Possible reasons for this discrepancyre discussed in detail in Section 4.

Multiple linear regression analysis indicates that net N, soil texture (mainly, siltnd clay content) and organic matter are the best predictors of stream N loads, andogether explain 52% of its variability. Net N best explains the variability in stream Noads, followed by percent of sand in the watershed soils (Fig. S5). Of the significantariables, only clay has a significant negative relationship with the stream N loads.he regression of stream N loads against net N, percent of silt, clay and SOM in theatershed is shown in Eq. (2):

Stream N Loads : 15.9 + 0.22 ∗ Net N + 0.3 ∗ Silt (%) − 1.46 ∗ Clay (%)

+ 5.08 ∗ SOM (%) (2)

.3. Sensitivity analysis

In Fig. 7, a series of tornado diagrams are used to show the sen-itivity of the N balance to the model parameters. Overall, the totalse of synthetic-N fertilizer in Iowa and the N content in corn grainre the variables to which the N balance is most sensitive (Fig. 7And B). The percentages of land area that are manure-applied or inontinuous corn are the next most influential parameters. The Nontents of soybean and denitrification rate have the next largestnfluence on the N balance of synthetic-N treated rotations. Allther factors are less influential. Soybean residues are more labilei.e., easy to decompose) and, thus, help build up more N in thective pool which results in higher N mineralization. Conversely,orn residues break down slowly and help build up more N inhe slow organic pool than in the active pool, resulting in lower

mineralization from the active pool (Burkart et al., 2005).In Fig. 7C and D, the sensitivities of the N balance to the model

arameters are presented for Hardin and Clarke Counties. Theseounties were chosen to demonstrate the wide range of sensitivityesults among counties as a result of the nature of the agriculturalractices in those counties. Further discussion is provided below.

. Discussion

.1. Crop rotation, fertilizer practices and net N

Manure fertilized rotations tend to have higher net N than

ynthetic-N treated rotations in all parts of the state. The trendf manure-fertilized soils having higher net N than synthetic-Nreated soil is supported by many experimental studies (Hepperlyt al., 2009; Basso and Ritchie, 2005). The difference in the magni-

r right) by watershed as predicted by the Iowa Department of Natural Resources

tude of net N between synthetic-N and manure-treated rotationspredicted here is particularly high in southern tier of the state. Thisresult is primarily related to a high manure-N application rate inmanure-treated crop rotations (Fig. S3).

Higher net N in counties in the North West, West Central, NorthCentral and Central crop districts than counties in all other cropdistricts is related mainly to the dominance of continuous corncropping, manure-treated lands (Figs. S2 and S3), and higher N min-eralization of organic rich soils in these districts. These four cropdistricts compose 64% of total land in the state that is in the threecropping systems and 72% of the state’s total tile-drained cropland.Positive net N in this heavily tile-drained region is more likely toresult in nutrient leaching because tile drains provide a direct pathfor mineral N to reach local surface waters.

The importance of fertilization in the N balance can be seenclearly in sensitivity results. Areas with high net N tend to have alarger proportion of N input from manure and the N balance in theseareas is sensitive to model parameters related to manure applica-tion and manure N. For example, the N balance for Hardin County,which is estimated to have the highest maximum net N, is mostsensitive to the percentage of manure applied acres, the percent-age of continuous corn acres (which receive more manure than C-Sacres), and the percentage N in manure (Fig. 7C). In contrast, forClarke county, the county with the lowest net N, the state N fertil-izer distribution and percentage N in corn grain are more influential(Fig. 7D).

4.2. Net N and stream N load

Differences in the magnitude of net N and stream N loads acrosswatersheds are expected because net N is a measure of excess Nthat has a potential to reach to a stream through leaching and runoffprocesses, whereas N load to a river or stream is an estimate of theN actually deposited in a stream due to leaching and runoff fromnearby agricultural fields. There are several pathways by whichexcess N leaving an agricultural field can fail to reach a local waterbody. For example, runoff N can be absorbed by other crops ornative vegetation before reaching a stream. After leaving the field,

but before reaching a stream, mineral N may be released to theatmosphere through denitrification. These and similar loss path-ways are affected by a range of local environmental characteristicsincluding stream hydrology, tile drainage structure, soil moisture,

28 S. Khanal et al. / Agriculture, Ecosystems and Environment 183 (2014) 21–30

Fig. 7. Sensitivity of net N in: (A) all corn-involved crop rotations under two fertilizers practices; (B) in synthetic-N fertilized corn-involved crop rotations, at a state level;and all corn-involved crop rotations under two fertilizer practices in (C) Hardin County; and (D) Clarke County, with ±10% change in N flows used in N balance model. Blackbars reflect changes in net N when flows are 10% below average; pattern filled bars reflect changes in net N when flows are 10% above average. The numbers next to barsi nges

v kg ha−

5

atmvslthtlebcm

4

eD2fswoato

ndicate the numerical values of the minimum and maximum changes in net N. Chaalues, area-weighted net N is estimated to be 34 kg ha−1 yr−1 for all rotations, 24

kg ha−1 yr−1 in Clarke County.

nd land slope. A higher net N but lower stream N concentration inhe northwest and north central districts may be due to underesti-

ation of N loss through denitrification or absorption by non-cropegetation. Loss of N from soil via denitrification depends on SOM,oil texture, and soil moisture (Jarecki et al., 2008). Although aarge fraction of the land in the corn-producing regions of Iowa isile-drained, soils of these regions are rich in organic matter andave poorly-drained soils. The assumption of a 13% denitrifica-ion rate in these regions might have underestimated the actualoss of nitrogen through denitrification at the site level. Accuratestimation of N loads from agricultural fields to streams demandsetter knowledge of nutrient transport and transformation pro-esses which can only be predicted with very detailed processodels that incorporate highly localized site-specific information.

.3. Manure-treated land and net N

This study shows that the N balance in a county is heavily influ-nced by the amount of land in the county that is manure-treated.espite an increase in the total number of livestock in Iowa between002 and 2007, and the requirement of manure management plansor all farms applying manure from a CAFO of more than 500 live-tock unit, the reported amount of manure-treated land in 2007as lower than in 2002. This increase in total manure nutrient input

n fewer reported manure-treated lands indicates higher manurepplication rates by farmers. The 75th percentile of total manure-reated land in Iowa is estimated to have a manure application ratef approximately 300 kg N ha−1 yr−1. The 2006 USDA’s Agricultural

less than 1 kg ha−1 yr−1 are not shown. When N flows are at their reported average1 yr−1 in synthetic-N treated crop rotations, 75 kg ha−1 yr−1 in Hardin County and

Resource Management Survey (ARMS) reported that nutrients areover applied on more than 95% of corn land in the Corn Belt receiv-ing manure treatment (Ribaudo et al., 2011). Manure is applied atmore than 400 kg N ha−1 yr−1 in the southern tier of counties inIowa (Fig. S3). This high manure application rate suggests the needfor stringent monitoring of manure management practices as wellas the integration of manure into Iowa crop production on a widergeographic scale.

4.4. Fertilizer application rate

This study shows that the rate of N fertilizer application is themain driving force in the estimated N balance. There is, however,a lack of information about N application rate by crop type at acounty level. This study adopts a novel approach to estimate theuse of synthetic-N fertilizer by cropping system at a county level.Prior studies (David et al., 2010; Booth and Campbell, 2007; McIsaacet al., 2002) used information about total N fertilizer distributionin a state and fertilizer expenses at a county level to estimateN fertilizer use in each county, and did not consider manure Nthat must be land applied according to state manure managementrules. Their approach also does not account for variation in fertil-izer application rate based on crop rotation. The limitations in priorstudies were addressed in this study by incorporating information

about the total nutrient distribution (i.e., synthetic N), county levelavailability of fertilizers (i.e., manure N), and the relative N require-ment of different crops in estimating total N use and N applicationrate at a county level (see SI for details).

tems a

fi2ps1n2ai

4

3ysstN2

strcremta

4

tcchesddtbto

5

swsstcgtmpes

rr

S. Khanal et al. / Agriculture, Ecosys

Average synthetic-N application rates estimated for cornelds in C-S (i.e., 176 kg ha−1 yr−1) and C-C rotations (i.e.,06 kg ha−1 yr−1) in this study are close to those observed inrevious field/survey studies (Duffy and Correll, 2006). Iowa exten-ion publications report a rate of 207.4 kg N ha−1 yr−1 in C-C and79.3 kg N ha−1 yr−1 for corn in C-S (Duffy and Correll, 2006). Agro-omic field trials representing standard practice use N at a rate of02.9 kg N ha−1 yr−1 for C-C (Sawyer et al., 2006). Similarly, the N-pplication rate reported for corn in C-S in the USDA ARMS surveys 156.9 kg ha−1 yr−1.

.5. Corn stover harvesting and environmental implications

Our analyses show that, from nutrient balance standpoint,0–75% of corn stover can be harvested, but these rates vary withield, cropping system, and fertilization practice. This analysis con-iders only the N balance. Permissible stover removal thresholdshould, however, be set to avoid soil erosion, soil organic mat-er depletion, and reduced soil productivity (Blanco-Canqui, 2010;ewman et al., 2010; Wilhelm et al., 2007; Wilhelm and Wortmann,004).

Residues play an important role in maintaining soil conditionsuitable for plant establishment and growth. There is feedbackhrough moisture and energy balances between the amount of cropesidue in the soil and plant growing conditions. In wetter climates,onservation tillage practices (i.e., systems that leave more plantesidues on the soil) provide a wetter and cooler environment formerging seedlings, and can result in poor corn crop establish-ent (Linda et al., 2002). Further research needs to be conducted

o determine appropriate residue removal rates for different soilsnd climates.

.6. Tile-drain and net N

It is assumed in this analysis that positive net N usually leaveshe soil either through runoff or leaching, and thus does notontribute significantly to buildup of SON. A large fraction of agri-ultural land in Iowa is tile-trained (Sugg, 2007), and prior studiesave shown that net N in tile-drained watersheds is approximatelyqual to the nitrate N in the stream or river that drains the water-hed (McIsaac and Hu, 2004). While the assumption that net Noes not contribute to buildup of SON is reasonable in heavily tile-rained watersheds, it may not be supportable in other regions inhe study area that are not tile-drained. Underestimation of immo-ilized N in these areas may overestimate net N. It is recommendedhat future studies account for the role of excess N in the formationf SON in both drained and undrained regions.

. Conclusions

The results of this study suggest that N fertilization (i.e., bothynthetic and manure fertilizer) is the primary reason for degradedater quality across large parts of Iowa. Stover harvesting has

ome potential to reduce N leaching from agricultural fields, buttover must not be over harvested to avoid a N deficit in agricul-ural regions. Counties in the North Central, Central and North Eastrop districts comprise the largest fraction of land with high net N;reater stover harvesting (i.e., 75%) in these regions is not expectedo result in a negative N balance, suggesting that these regions

ight be most suitable for stover harvesting from a soil nitrogenerspective. The stover harvest rate at any particular location, how-ver, should be based on site-specific factors relating to erosion and

oil carbon.

The central and northern crop districts in Iowa are N-richegions; leachable N in these regions can be reduced throughemoval of stover to some extent, but stover removal is more

nd Environment 183 (2014) 21–30 29

likely to create N deficit in southern crop districts. Across Iowa,net N in manure-treated rotations is estimated to be higher than insynthetic-N treated crop rotations. For example, net N in manure-treated rotations with 75% stover removal is still higher than the netN predicted in synthetic-N treated rotations with no stover harvest.Understanding of net N in various corn-involved rotations understover removal scenarios will help identify regions where conser-vation efforts and stover removal for biomass should be targeted.This analysis may assist public officials and policy makers in devel-oping effective nutrient management plans in Iowa and other CornBelt states that have adopted more corn based rotations, and aretargeted for increased corn stover collection.

Acknowledgments

We would like to thank the anonymous reviewers whose sug-gestions improved the earlier version of this manuscript. Thismaterial is based upon work supported by the National ScienceFoundation under Grant Nos. CMS-0424700 and CBET-1137677.Any opinions, findings, and conclusions or recommendationsexpressed in this material are those of the authors and do notnecessarily reflect the views of the National Science Foundation.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.agee.2013.10.013.

References

Alexander, R.B., Smith, R.A., Schwarz, G.E., Boyer, E.W., Nolan, J.V., Brakebill, J.W.,2007. Differences in phosphorus and nitrogen delivery to the Gulf of Mexicofrom the Mississippi river basin. Environ. Sci. Technol. 42, 822–830.

Andrew, S., 2006. Crop Residue Removal for Biomass Energy Production: Effectson Soils and Recommendations. White Paper, Published Online. USDA Nat-ural Resources Conservation Service http://soils.usda.gov/sqi/managment/files/AgForum Residue White Paper.pdf

Basso, B., Ritchie, J.T., 2005. Impact of compost, manure and inorganic fertilizer onnitrate leaching and yield for a 6-year maize-alfalfa rotation in Michigan. Agric.Ecosyst. Environ. 14, 321–329.

Bertora, C., Zavattaro, L., Sacco, D., Monaco, S., Grignani, C., 2009. Soil organic matterdynamics and losses in manured maize-based forage systems. Eur. J. Agron. 30,177–186.

Blanco-Canqui, H., 2010. Energy crops and their implications on soil and environ-ment. Agron. J. 102, 403–419.

Booth, M.S., Campbell, C., 2007. Spring nitrate flux in the Mississippi river basin:a landscape model with conservation applications. Environ. Sci. Technol. 41,5410–5418.

Boryan, C., Yang, Z., Mueller, R., Craig, M., 2011. Monitoring US agriculture: the USdepartment of agriculture, national agricultural statistics service, cropland datalayer program. Geocarto Int. 26, 341–358.

Burkart, M., James, D., Liebman, M., Herndl, C., 2005. Impacts of integrated crop-livestock systems on nitrogen dynamics and soil erosion in western Iowawatersheds. J. Geophys. Res. 110, G01009.

Burkart, M., James, D., Liebman, M., van Ouwerkerk, E., Fujii, S., Yamada, K., Kuni-matsu, T., 2006. Integrating principles of nitrogen dynamics in a method toestimate leachable nitrogen under agricultural systems. Water Sci. Technol. 53,289–301.

Cooper, J.S., Kahn, E., Ebel, R., 2012. Sampling error in US field crop unit processdata for life cycle assessment. Int. J. Life Cycle Assess., http://dx.doi.org/10.1007/s11367-012-0454-3.

Coulter, J.A., Nafziger, E.D., Wander, M.M., 2009. Soil organic matter response tocropping system and nitrogen fertilization. Agron. J. 101, 592–599.

David, M.B., Drinkwater, L.E., McIsaac, G.F., 2010. Sources of nitrate yields in theMississippi river basin. J. Environ. Qual. 39, 1657–1667.

Del Grosso, S.J., Parton, W.J., Mosier, A.R., Hartman, M.D., Brenner, J., Ojima, D.S.,Schimel, D.S., 2001. Simulated interaction of carbon dynamics and nitrogen tracegas fluxes using the DAYCENT model. In: Schaffer, M., Ma, L., Hansen, S. (Eds.),Modeling Carbon and Nitrogen Dynamics for Soil Management. CRC Press, BocaRaton, FL, pp. 303–332.

Department of Environmental Protection (DEP), 2001. Field Application of Manure– A Supplement to Manure Management for Environmental Protection. Depart-ment of Environmental Protection, Pennsylvania.

Donner, S.D., Kucharik, C.J., 2008. Corn-based ethanol production compromises goalof reducing nitrogen export by the Mississippi River. PNAS 105, 4513–4518.

3 tems a

D

D

E

E

G

G

G

H

HI

J

J

J

K

K

K

K

K

K

K

L

L

L

M

0 S. Khanal et al. / Agriculture, Ecosys

onner, S.D., Scavia, D., 2007. How climate controls the flux of nitrogen by the Mis-sissippi river and the development of hypoxia in the Gulf of Mexico. Limnol.Oceanogr. 52, 856–861.

uffy, M., Correll, D., 2006. Rotation Profitability Calculator using Opti-mal Nitrogen Rate. Iowa State University Extension Decision Tools,http://www.extension.iastate.edu/agdm/crops/xls/a1-80nratecalculator.xls(accessed 18.02.11).

nvironmental Protection Agency (EPA), 2012. Available from http://water.epa.gov/lawsregs/lawsguidance/cwa/tmdl/intro.cfm

SRI. Available from http://webhelp.esri.com/arcgisdesktop/9.2/index.cfm?(accessed 20.12.09).

raham, R.L., Nelson, R., Sheehan, J., Perlack, R.D., Wright, L.L., 2007. Current andpotential U.S. corn stover supplies. Agron. J. 99, 1–11.

rignani, G., Zavattaro, L., Sacco, D., Monaco, S., 2007. Production, nitrogen andcarbon balance of maize-based forest systems. Eur. J. Agron. 26, 442–453.

oolsby, D.A., Battaglin, W.A., Lawrence, G.B., Artz, R.S., Aulenbach, B.T., Hooper,R.P., Keeney, D.R., Stensland, G.J., 1999. Flux and Sources of Nutrients in theMississippi–Atchafalaya River basin. Topic 3 Report of the Integrated Assess-ment on Hypoxia in the Gulf of Mexico. NOAA Coastal Ocean Program, SilverSpring, MD.

epperly, P., Lotter, D., Ulsh, C.Z., Seidel, R., Reider, C., 2009. Compost, manure andsynthetic-N fertilizer influences crop yields, soil properties, nitrate leaching andcrop nutrient content. Compost Sci. Util. 17, 117–126.

ilbe, J., 2009. Logistic Regression Models. Chapman & Hall/CRC Press.owa Department of Natural Resources (IDNR), 2013. Iowa’s Water Ambient Moni-

toring Program. Water Fact Sheet 2013-1.agadamma, S., Lal, R., Hoeft, R.G., Nafziger, E.D., Adee, E.A., 2008. Nitrogen fertiliza-

tion and cropping system impacts on soil properties and their relationship tocrop yield in the central Corn Belt, USA. Soil Tillage Res. 2 (98), 120–129.

arecki, M.K., Parkin, T.B., Chan, A.S.K., Kaspar, T.C., Moorman, T.B., Singer, J.W., Kerr,B.J., Hatfield, J.L., Jones, R., 2008. Cover crop effects on nitrous oxide emissionfrom manure-treated Mollisol. Agric. Ecosyst. Environ. 134, 29–35.

enkinson, D.S., Rayner, J.H., 1977. The turnover of soil organic matter in some of theRothmasted classical experiments. Soil Sci. 5 (123), 298–305.

alita, P.K., Algoazany, A.S., Mitchell, J.K., Cooke, R.A.C., Hirschi, M.C., 2006. Sub-surface water quality from a flat tile-drained watershed in Illinois, USA. Agric.Ecosyst. Environ. 115, 183–193.

arlen, D.L., Lal, R., Follett, R.F., Kimble, J.M., Hatfield, J.L., Miranowski, J.M., Cam-bardella, C.A., Manale, A., Anex, R.P., Rice, C.W., 2009. Crop residues: the rest ofthe story. Environ. Sci. Technol. 43, 8011–8015.

arlen, D.L., Varvel, G.E., Jane, J.M.F., Baker, J.M., Osborne, S.L., 2011. Monitoring soilquality to assess the sustainability of harvesting corn stover. Agron. J. 103 (1),288–295.

hanal, S., Anex, R.P., Gelder, B.K., Dixon, P., Caragea, P., 2012. Cropping patternchoice with proximity to ethanol production and animal feeding operations.Biofuels Bioprod. Bioref. 6 (4), 431–443.

ubat, J., Lipavsky, J., 2006. Steady state of the soil organic matter in the long-termfields experiments. Plant Soil Environ. 52, 9–14.

ucharik, C.J., Foley, J.A., Delire, C., Fisher, V.A., Coe, M.T., Lenters, J.D., Molling, C.,Ramankutty, N., Norman, J.M., Gower, S.T., 2000. Testing the performance of adynamic global ecosystem model: Water balance, carbon balance, and vegeta-tion structure. Glob. Biogeochem. Cycles 14, 795–825.

ucharik, C.J., 2003. Evaluation of a process-based agro-ecosystem model (agro-IBIS)across the US corn belt: simulations of the interannual variability in maize yield.Earth Interact. 7, 1–33.

i, C., Frolking, S., Frolking, T.A., 1992. A model of nitrous oxide evolution from soildriven by rainfall events: I. Model structure and sensitivity. J. Geophys. Res. 97,9759–9776.

inda, M., Tolbert, V., Cushman, J., 2002. Potential environmental effects of corn (Zeamays L.) stover removal with emphasis on soil organic matter and erosion. Agric.Ecosyst. Environ. 89, 149–166.

iu, S., Tan, Z., Li, Z., Zhao, S., Yuan, W., 2011. Are soils of Iowa USA currently acarbon sink or source? Simulated changes in SOC stock from 1972 to 2007. Agric.

Ecosyst. Environ. 140, 106–112.

arshall, E., Caswell, M., Malcom, S., Motamed, M., Hrubovcak, J., Jones, C., Nickerson,C., 2011. Measuring the Indirect Land-Use Change Associated with Increased Bio-fuel Feedstock Production: A Review of Modeling Efforts: Congressional Report.US Department of Agriculture-ERS, Washington, DC, AP 054.

nd Environment 183 (2014) 21–30

MacDonald, J.M., Ribaudo, M.O., Livingston, M.J., Beckman, J., Huang, W., 2009.Manure Use for Fertilizer and for Energy: Report to Congress. ERS- USDA- AP-037.

McIsaac, G.F., David, M.B., Gertner, G.Z., Goolsby, D.A., 2002. Relating net nitrogeninput in the Mississippi River basin to nitrate flux in the lower Mississippi River:A comparison of approaches. J. Environ. Qual. 31, 1610–1622.

McIsaac, G.F., Hu, X., 2004. Net N input and riverine N export from Illinois agricul-tural watersheds with and without extensive tile drainage. Biogeochemistry 70,251–257.

Meki, M.N., Marcos, J.P., Atwood, J.D., Norfleet, L.M., Steglich, E.M., Williams, J.R.,Gerik, T.J., 2011. Effects of site-specific factors on corn stover removal thresholdsand subsequent environmental impacts in the Upper Mississippi River Basin. J.Soil Water Conserv. 66 (6), 358–399.

National Agricultural Statistics Service (NASS), 2012. Livestock and Animals, Avail-able from http://quickstats.nass.usda.gov/ (accessed 19.09.12).

Newman, J.K., Kaleita, A.L., Laflen, J.M., 2010. Soil erosion hazard maps for corn stovermanagement using National Resources Inventory data and the Water ErosionPrediction Project. J. Soil Water Conserv. 65, 211–222.

Omernik, J.M., 1987. Ecoregions of the conterminous United States. Ann. Assoc. Am.Geographer 77, 118–125.

Paustian, K., Collins, H.P., Paul, E.A., 1997. Management controls on soil carbon.In: Paul, E.A., et al. (Eds.), Soil Organic Matter in Temperate Agroecosystems:Longterm Experiments in North America. CRC Press, Boca Raton, FL, pp. 15–49.

Perlack, R.D., Wright, L.L., Turhollow, A.F., Graham, R.L., Stokes, B.J., Erbach,D.C., 2005. Biomass as Feedstock for a Bioenergy and Bioproducts Industry:The Technical Feasibility of a Billion-Ton Annual Supply DOE/GO-102005-2135 and ORNL/TM-2005/66, Available at http://feedstockreview.ornl.gov/pdf/billion ton vision.pdf

Rabalais, N.N., Turner, R.E., Scavia, D., 2002. Beyond science into policy: Gulf ofMexico hypoxia and the Mississippi River. Bioscience 52, 129–142.

Ribaudo, M., Gollehon, N., Aillery, M., Kaplan, J., Johansson, R., Agapoff, J., Christensen,L., Breneman, V., Peters, M., 2003. Manure Management for Water Quality:Costs to Animal Feeding Operations of Applying Manure Nutrients to Land. U.S.Department of Agriculture, Economic Research Service AER-824.

Ribaudo, M., Delgado, J., Hansen, L., Livingston, M., Mosheim, R., Willamson, J., 2011.Nitrogen in agricultural systems: implications for conservation policy. In: U.S.Department of Agriculture, Economic Research Service ERS-127.

Russelle, M.P., Birr, A.S., 2004. Biological nitrogen fixation. Large-scale assessmentof symbiotic dinitrogen fixation by crops: soybean and alfalfa in the MississippiRiver Basin. Agron. J. 96, 1754–1760.

Sawyer, J., Nafziger, E., Randall, G., Bundy, L., Rehm, G., Joern, B., 2006. Conceptsand rationale for regional nitrogen rate guidelines for corn. In: Iowa State Uni-versity Extension PM-2015, Available from http://www.extension.iastate.edu/Publications/PM2015.pdf

Schepers, J.S., Raun, W.R., 2008. Nitrogen in agricultural systems. In: AgronomyMonograph 49.

Sugg, Z., 2007. Assessing U.S. Farm Drainage: Can GIS Lead to Better Estimates ofSubsurface Drainage Extent? World Resources Inst., Washington, DC.

Varvel, G.E., 1994. Rotation and nitrogen fertilization effects on changes in soil car-bon and nitrogen. Agron. J. 86, 319–325.

Vitosh, M.L., Johson, J.W., Mengel, D.B., 1995. Tri-state Fertilizer Recommendationsfor Corn, Soybeans, Wheat, and Alfalfa. Ohio State University Extension Bul-letin E-2567, Available at http://ohioline.osu.edu/e2567/index.html (verified24.09.12).

Wilhelm, W., Wortmann, C.S., 2004. Tillage and rotation interactions for corn andsoybean grain yield as affected by precipitation and air temperature. Agron. J.96, 425–432.

Wilhelm, W., Johnson, J.M.F., Karlen, D.L., Lightle, D.T., 2007. Corn stover to sustainsoil organic carbon further constrains biomass supply. Agron. J. 99, 1665–1667.

Williams, J.R., Jones, C.A., Dyke, P.T., 1984. A modeling approach to determin-ing the relationship between erosion and soil productivity. Trans. ASAE 27,129–144.

Weyer, P.J., Cerhan, J., Kross, B.C., Hallberg, G.R., Kantamneni, J., Breuer, G., et al.,

2001. Municipal drinking water nitrate level and cancer risk in older women:the Iowa Women’s Health Study. Epidemiology 12, 327–338.

Zavattaro, L., Monaco, S., Sacco, D., Grignani, C., 2012. Options to reduce N loss frommaize in intensive cropping systems in Northern Italy. Agric. Ecosyst. Environ.147, 24–35.