b i om a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 2 3 9 1e2 4 0 0

Avai lab le a t www.sc iencedi rec t .com

ht tp : / /www.e lsev ier . com/ loca te /b iombioe

Land use change in a biofuels hotspot: The case of Iowa, USA

Silvia Secchi a,*, Lyubov Kurkalova b, Philip W. Gassman c, Chad Hart d

aDepartment of Agribusiness Economics, Southern Illinois University, Agriculture Building e Mailcode 4410, 1205 Lincoln Drive,

Carbondale, IL 62901, USAbDepartment of Economics and Finance, Energy and Environmental Systems program, North Carolina A&T State University, Greensboro,

NC 27411, USAcCenter for Agricultural and Rural Development, Iowa State University, 578 Heady Hall, Ames, IA 50011-1070, USAdDepartment of Economics, Iowa State University, 260 Heady Hall, Ames, IA 50011-1070, USA

a r t i c l e i n f o

Article history:

Received 11 July 2008

Received in revised form

12 August 2010

Accepted 18 August 2010

Available online 26 September 2010

Keywords:

Land use change

Economic analysis

Environmental Impact

Energy crop production

Corn-soybean rotation

Land set-aside

* Corresponding author. Tel.: þ1 618 453 171E-mail addresses: ssecchi@siu.edu (S.

chart@iastate.edu (C. Hart).0961-9534/$ e see front matter ª 2010 Elsevdoi:10.1016/j.biombioe.2010.08.047

a b s t r a c t

This study looks at the land use impact of the biofuels expansion on both the intensive and

extensive margin, and its environmental consequences. We link economic, geographical

and environmental models by using spatially explicit common units of analysis and use

remote sensing crop cover maps and digitized soils data as inputs. Land use changes are

predicted via economic analysis of crop rotation choice and tillage under alternative crop

prices, and the Environmental Policy Integrated Climate (EPIC) model is used to predict

corresponding environmental impacts. The study focuses on Iowa, which is the leading

biofuels hotspot in the U.S. due to intensive corn production and the high concentration of

ethanol plants that comprise 28% of total U.S. production. We consider the impact of the

biofuels industry both on current cropland and on land in the Conservation Reserve

Program (CRP), a land set-aside program. We find that substantial shifts in rotations

favoring continuous corn rotations are likely if high corn prices are sustained. This is

consistent with larger scale analyses which show a shift of the current soybean production

out of the Corn Belt. We find that sediment losses increase substantially on the intensive

margin, while nitrogen losses increase less. Returning CRP land into production has

a vastly disproportionate environmental impact, as non-cropped land shows much higher

negative marginal environmental effects when brought back to row crop production. This

illustrates the importance of differentiating between the intensive and extensive margin

when assessing the expansion of biofuel production.

ª 2010 Elsevier Ltd. All rights reserved.

1. Introduction greenhouse gas emissions. However, several studies have also

The environmental impacts of the large scale production of

biofuels are a topic of vigorous debate, which in a short period

of time has generated a relatively large and fast growing body

of literature. The debate has mostly focused on the overall

carbon footprint of biofuels [1e5], because one of the ratio-

nales of policies promoting biofuels is that they reduce

4.Secchi), lakurkal@ncat

ier Ltd. All rights reserved

included other environmental indicators, such as effects on

nitrogen, phosphorus, and pesticide use (see for example

[1,2,4]). Until recently, most of the literature did not include

spatially explicit analyses of the environmental impacts,

mostly because of the complexity and the integrated nature of

the modeling efforts necessary. Such an analysis, however, is

vital if we want to understand whether there are

.edu (L. Kurkalova), pwgassma@iastate.edu (P.W. Gassman),

.

b i om a s s an d b i o e n e r g y 3 5 ( 2 0 1 1 ) 2 3 9 1e2 4 0 02392

disproportional effects in some areas, and to help direct

policy. Some recent papers have incorporated carbon effects

at a global scale [6,7] by linking large scale economic and

environmental models, again with a carbon focus. This paper

takes a regional scale approach, closer to ones used by geog-

raphers and land use scholars (see for example [8e10]). Our

modeling approach consists of a field-level discrete economic

model of crop rotation and land use management that

predicts the impacts of alternative policy scenarios on land

use shifts, which in turn are input into a field-level environ-

mental model to estimate environmental implications, such

as erosion, nutrient losses and carbon sequestration in the

soil. The economic and environmental models are interfaced

using common geographical units of analysis. Because of the

regional scale at which the models operate, the price of crops

and energy sources are given. This bottom-up approach is

complementary to large, top down modeling efforts that

predict global land use and environmental impacts at a coarse

resolution.

The study focuses on the Midwestern U.S. state of Iowa,

which is the leading biofuel hotspot in the U.S. due to inten-

sive corn production and the high concentration of ethanol

plants that comprise over 25% of total U.S. production. Rapid

shifts in crop production are occurring in the state due to

increasing demand for ethanol production. Between 2006 and

2007, corn production in Iowa increased by 26%, bringing the

current Iowa corn crop to 19% of total U.S. production.

Iowa is a very intensively managed agricultural state. The

greatmajority of the land is in the private sector, and it is used

for row crop production. Nevertheless, the state has around

three quarter of a million hectares in the Conservation

Reserve Program (CRP), a land set aside program instituted by

the U.S. federal government. We consider two types of land

use changes: 1) on currently cropped land, because cropland

could be farmedmore intensively to produce more feedstocks

for biofuel production e specifically, in the case of the U.S.,

corn; and 2) on land currently out of agricultural production

(CRP). We refer to the first type of change as a change on the

intensivemargin, because increases in corn production in this

case would not increase the land allocated to crop production,

but would just increase the production of corn per acre, and to

the second as the extensivemargin, because CRP land is in set

aside and using it for crop production again would increase

the area used for agricultural production.

It is important to include both types of effects separately

because, while the initial goal of the CRP program in 1985 was

to reduce crop supply, in more recent years land has been

targeted by the program for its environmental vulnerability.

Indeed, the analysis on the beneficial effects of CRP on the

environment has been wide-ranging, including studies that

have calculated the effects of CRP on sediment losses [11],

carbon sequestration [12], and bird populations [13e15]. While

previous studies have relied on the Natural Resources Inven-

tory database [12] and remote sensing imagery [16,17] to

determine the location of CRP land, we have used the actual

Conservation Reserve Program parcel data obtained from the

Farm Service Agency.

Previous work [18] has also detailed the likely impact of the

biofuel industrye via high crop pricese on thismarginal land,

and the impact on the CRP program. These estimates suggest

that the impact of the ethanol industry could mean that half

the CRP land in the state is returned to row crop production in

the near future. This land was enrolled in the program

because of factors such as high erodibility and impact on

water quality and wildlife. Thus, returning it to agricultural

production has disproportionately high per hectare effects on

soil erosion and nutrient losses. However, because the great

majority of the land in Iowa is already used for crop produc-

tion, the largest impact of the biofuel expansion in absolute

numbers is likely to be on the intensive margin, on already

cultivated cropland. In particular, in Iowa, farmers have

historically tended to plant corn after soybeans in two year

rotations as detailed by Fig. 1. The changes in crop prices

induced by the bioenergy expansion are likely to affect crop

and rotation choices and, indirectly, tillage choices. Some

recent studies have started to assess the impacts that an

increase in corn production could have on surface water

quality in the U.S. Donner and Kucharik [19] assess such

impacts for the whole MississippieAtchafalaya River Basin

using spatially explicit modeling at a county-level scale. The

models are not driven by economics, and there are no differ-

ential impacts of CRP and current cropland. Simpson et al. [20]

differentiate between CRP and current cropland, but do not

use economic analysis to assess how production will respond

to prices, and their study does not include a spatially explicit

analysis. However, all these elements e the responsiveness to

prices, the differentiation of types of land use, and the

spatially explicit analysis, are important to policymakers who

need to assess the environmental impacts of biofuels andwho

devise agricultural, conservation and energy policy. Our

contribution to the literature is to simultaneously include all

these considerations in the analysis.

The paper is structured as followed. First, in the materials

and method section, we detail the datasets used to construct

a baseline land use for Iowa and baseline environmental

indicators. We then construct an economic model that is

based on production costs by tillage and crop rotation and use

forecast prices to predict future land use scenarios. We look at

three price-based scenarios, representing a range of future

market conditions. The land use change maps of these

simulations and their environmental impacts are presented in

the results section. We conclude with a discussion on the

significance of our findings, both for Iowa agriculture and

within a larger context.

2. Materials and methods

2.1. Land use

To construct the baseline land use for Iowa, we use U.S.

Department of Agriculture (USDA) National Agricultural

Statistics Service (NASS) remote sensing crop cover maps [21].

These cropland data layers are published yearly to estimate

crop areas and yields.We combine five years of data, 2002e06,

to construct historical rotations. The remote sensing cropland

data was used to estimate rotations, using a slightly different

methodology, in [22], for the years 2001e07, and comparing

two consecutive years at the time. The data contains some

errors in the coverage and cloud cover obstructed view in

Fig. 1 e Location of 2002e06 crop rotations and CRP land in Iowa.Note: the CRP area is enlarged and not true to scale to better

show its location. CRP areas in Iowa correspond to around 7.9% of cultivated cropland (almost three quarters of a million

hectares of CRP land versus almost 92,000 km2of cultivated cropland).

b i om a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 2 3 9 1e2 4 0 0 2393

some years. In particular, the GIS cover tends to overestimate

the amount of continuous soybeans. This is due to the fact

that the coverage contains a 5e15% error on cropped fields

[21]. The same error is likely to occur for corn. However, in that

case, it is not as easy to determinewhether an error is present,

since continuous corn is a viable crop choice. There is

a widespread consensus that continuous soybeans are not

a common occurrence in Iowa [23] and the literature confirms

this assessment, as two or more years of continuous soybean

create serious problemswith soybean nematodese Heterodera

glycines [24]. Therefore, we created algorithms to re-construct

some of the rotations. Specifically, a three year soybean

sequence is converted to soybean-corn-soybean. As noted

before, the CRP land coverage was obtained from the USDA

Farm Service Agency. Fig. 1 illustrates the rotations and the

location of the CRP land. Tillage choices are an important

determinant of environmental quality in cropped land, and

tillage choice, because of its impact on yields, is linked to the

choice of rotation [25,26]. Unfortunately, there are no recent

tillage data by crop rotation. The Conservation Technology

Information Center (CTIC) maintains a county-level database

of tillage practices, the Crop ResidueManagement Survey [27],

but the database does not identify the previous crop. The

USDA Agricultural Resource Management Survey (ARMS)

includes information on tillage and crop rotations. However,

the survey has a very small sample size and is of limited value

to our analysis [28]. For example, in 2005, information on

tillage practices was not available for about half of the corn

area, and the estimate of no till acres is statistically unreliable

due to the combination of a low sample size and high

sampling error.

Since 1985, farmers wanting to plant crops on highly

erodible land have been required to engage in conservation

activities to participate in farm programs. In Iowa, the most

common conservation practice on highly erodible land (HEL)

is the use of no-till [29]. Therefore, for simplicity, we assume

that in our baseline all highly erodible land is under no-till,

while the rest of the cropped acres are under mulch till. This

means that around 23% of the area planted in corn in 2006 -

corresponding to the HEL area e will be simulated as being in

no-till. A similar percentage area, 25%, is simulated as no-till

soybean. These compare to 14% and 33%, respectively, from

the CTIC survey in the same year, the last one available [29].

Note that the CTIC data estimate that almost 60% of the corn

and 19% of the soybean are not in any reduced tillage

management. Thus, our baseline is likely to be quite conser-

vative in terms of the environmental impact of current land

use. The location of the highly erodible land was determined

by overlaying the rotationmap on the Iowa Soil Properties And

Interpretations Database (ISPAID) map [30].

2.2. Land management practices

Fertilizer application rates are another important determinant

of the environmental impact of agriculture. Fertilizer rates

vary across crop rotations and spatially. In fact, the produc-

tion technologies that farmers use are best seen as bundles

[31]. In particular, there is ample evidence that rotated corn

generates yield benefits over continuous corn, because rota-

tions offer a good defense against some pests [32] and crop

rotations that include legumes improve soil nitrogen levels

[33]. Specifically, there is field-level evidence that corn yields

under a corn-soybean rotation exceed yields under contin-

uous corn [32]. Similar results have been found for corn-

soybean and corn-corn-soybeans rotations [25]. However,

some of these yield effects can be at least partially countered

by increasing nitrogen applications [34]. Therefore, contin-

uous corn tends to be more heavily fertilized than corn in

a two-year rotation with soybeans. Because the dominant

rotation has historically been corn-soybean, there are con-

flicting data on the size of the increase in nitrogen application

rates necessary to maintain yields for continuous corn. The

Fig. 2 e Potential corn yields Note: the CRP area is enlarged and not true to scale to better show the yield potential.

b i om a s s an d b i o e n e r g y 3 5 ( 2 0 1 1 ) 2 3 9 1e2 4 0 02394

latest ARMS survey, conducted in 2005, reports nitrogen rates

of 165.9 kg ha�1 for continuous corn and 156.9 kg ha�1 for corn

after soybeans in Iowa. The latest fertilizer sales data for the

state, going back to 2003, which of course cannot be used to

differentiate across rotations, indicates an average nitrogen

application rate of 145.7 kg ha�1 on corn [35]. Agronomic field

trials suggest application rates of 202.9 kg ha�1 for continuous

corn and 138.9 kg ha�1 for corn after soybeans in Iowa [36].

Finally, a recent Iowa extension publication indicates a rate of

207.4 kg ha�1 for continuous corn and 179.3 kg ha�1 for corn

after soybeans in Iowa [37]. In our analysis, we will assume

that corn after soybeans receives 156.9 kg ha�1 and corn after

corn is fertilized with 213 kg ha�1. This translates into

conservative estimates of the impact of the added nitrogen,

because the baseline nitrogen level is higher than the agro-

nomically necessary rate and reflects the reality of some

“insurance” nitrogen use, while the corn on corn nitrogen

level is close to the agronomical optimum. The empirical

evidence of this behavior is supported by theoretical work.

Uncertainty on the nitrogen soil concentration at the time of

fertilizer application and uncertainty on the amount of rain-

fall expected may increase nitrogen applications over the

economically optimal level of nitrogen applied [38].

As we noted above, the literature is unanimous on the

positiveyieldbenefitsof rotatedcornover continuous cornand

on the differential impact of tillage. Long-term trends on corn

and soybean yields have been studied for plow, chisel, ridge

and no-till systems, and continuous corn and corn-soybean

rotations.Bothcornandsoybeanyieldswere10e11%higheron

average in the non-monoculture rotations across all tillage

regimes, but there were differences across tillage regimes [39].

In particular, corn yield reductions in continuous corn were

muchhigher in no-till corn (17%) than in plowed corn (5%). The

reason for this is that corn residue from no-till regimes can

create problems for germinating and emerging plants. Others

have estimated a 14%decrease in continuous corn yields going

from a plow to a no-till system, while in a corn-soybean rota-

tion the reduction in corn yields going from a plow to a no-till

systemwas 5% [40]. Moreover, corn yields are not significantly

affected by the tillage system of the previous year’s crop if the

crop is soybean. If the crop is corn, however, yields are highest

under conventional tillage and decrease as the tillage type

become less intensive [41]. Finally, there is also someevidence,

based on field work, that two years of corn between soybean

crops boost the soybean crop yield [42].

On the basis of the literature reviewed above, we have

assumed in our calculation of the optimal rotation and tillage

system that yields decrease from their maximum potential

under certain combinations of crop, previous crop, and tillage.

Computationally, this reduction in yields means that we

multiply current year yields by a multiplier less than one, e.g.,

a 10% reduction means multiplication by 0.9. In particular, we

assume that corn after no-till corn will see a reduction of 15%,

corn after mulch till corn will yield 6% less and corn after

conventional till corn will yield 5% less than the potential

yield. Soybeans after no-till corn suffer a 6% yield drag, and

soybeans after two years of corn receive a 4% boost. The

potential yield information is derived from an index of suit-

ability for crop production, the Corn Suitability Rating (CSR),

with a methodology described in [18] (Fig. 2).

2.3. Economic model

We use data from Iowa extension budget publications [43] to

construct costs of production budgets. The budgets include all

the costs of producing crops, from seed to fertilizer application

b i om a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 2 3 9 1e2 4 0 0 2395

and harvesting, and are differentiated by crop and previous

crops. The creation of these budgets, togetherwith the scenario

prices and the yields, allows us to identify the most profitable

crop and rotation choices for a given set of crop prices.

The net returns are used to construct a crop production

cost model. We assume in our analysis that profit maximi-

zation is the driving force in farmers’ behavior [25] and for

simplicity abstract from the impact that the choice of

production technology has on the variance of production.

Farmers optimize by choosing the crop rotation and tillage

system that maximizes their net returns. The crop rotation

budgets by tillage practice take into account some of the

bundled nature of the choices farmers make. Specifically, we

consider three crop rotations, continuous corn, corn-

soybean and corn-corn-soybean, which account for the great

majority of crop production in the state (Fig. 1). We also

consider three tillage systems: conventional till, mulch till

(grouped together with ridge till), and no-till [27], for a total

of 42 possible choices of combined crop choice, rotations and

tillage. The model is static, as we solve for the combination

of rotation and tillage that gives the maximum net returns

for a given set of prices. We abstract in this analysis from

conservation compliance considerations because the future

of such measures is unclear at this point. On the intensive

margin, we assume that farmers will keep the land in

production and will choose the rotation and tillage system

that maximize their profits. For the extensive margin, we use

USDA Soil Rental Rates as the opportunity cost of land

retirement as described in [18]. Thus, returns from farming

must be at least as high as the rental rates for farmers to

return the CRP land to production. This is in line with

previous work that has tied CRP enrollment choices to land

quality and anticipated economic returns to alternative uses

such as cropping [44].

Fig. 3 e Projected rotations on the intensive and e

2.4. Environmental model

The land use data and the land use choice and management

scenarios are combined with digitized soil layer maps to

conduct the environmental modeling. We use the USDA

Natural Resources Conservation Service (NRCS) Soil Survey

Spatial and Tabular Data (SSURGO 2.2). SSURGO’s mapping

scales generally range from 1:12,000 to 1:63,360. SSURGO is the

most detailed level of soil mapping done by NRCS. The

SSURGO database has 10,637 unique soils for Iowa [45]. This

fine grained information allows us to produce very spatially

detailed maps.

The SSURGO soil database is used as the input to the

Environmental Policy Integrated Climate (EPIC) model [46,47],

which was originally called the Erosion Productivity Impact

Calculator and has been applied for a wide range of conditions

worldwide [48]. EPIC is a field-scale model that is designed to

simulate edge-of-field environmental impacts for drainage

areas that are characterized by homogeneous weather, soil,

landscape, crop rotation, and management system parame-

ters. The model operates on a continuous basis using a daily

time step and can perform long-term simulations of hundreds

of years. A wide range of crop rotations, tillage systems, and

other management practices can be simulated with EPIC. The

most recent versions of EPIC feature improved soil carbon

cycling routines that are based on routines used in the Century

model [47,49]. EPIC provides edge-of-field estimates of soil

erosion, nutrient loss, carbon sequestration, and other envi-

ronmental indicators. Geographically appropriate manage-

ment practices such as irrigation, fertilizer rates and tillage

regimes can be fed into the EPIC model. The EPIC model is run

using historical weather data for 30 years, and the results are

reported as the 30 year average for the sediment and nutrient

losses, and as the final carbon pool for carbon.

xtensive margin at corn prices of 167 $ MgL1.

b i om a s s an d b i o e n e r g y 3 5 ( 2 0 1 1 ) 2 3 9 1e2 4 0 02396

2.5. Scenario development

We focus the discussion on three price-based scenarios for

our analysis, with corn prices of 108 $ Mg�1, 142 $ Mg�1, and

167 $ Mg�1. Soybean prices tend to move in unison with corn

prices, both historically and in projected large scale analysis

[50], so we assume a constant difference of $3.50 per bushel, or

around $120 per Mg between the two prices. Themarket place

difference is based on volumes, and since corn is 25.4 kg per

bushel, and soybeans are 27.2 kg per bushel, the actual

difference will change with increasing corn prices.

The three corn prices are derived from large scale partial

equilibrium trade models, and they represent a spectrum

ranging from lower than expected ethanol industry expan-

sion, a current baseline long term price projection, and an oil

industry price shock [51]. More in general, however, our

framework can be used to assess the land use impacts and

costs of price-based policies, such as subsidies, energy price

changes, and changes in technology platform profitability.

Such chances in relative pricesmodify the farmers’ net returns

and consequently the profitmaximizing crop choices and land

use maps. The framework can also be used to assess the

impacts and costs of other types of policies, such as regulation

on crops to be grown in floodplains, within a particular

distance from waterways or along migratory bird routes.

3. Results

According to our analysis, for corn prices of 85 $ Mg�1, and

soybean prices of 208 $ Mg�1, which correspond to average

1996e2006 historical prices, the profit maximizing rotation

was corn-soybean undermulch till. Thus, the budgets and our

baseline (Fig. 3) tillage assumptions represent historical

trends well.

We estimate that, in 2006, cropland in Iowa was farmed

according to the rotations quantified in Table 1. This corre-

sponds to a corn crop area for 2006 of 50,152 km2, which was

very close to the corresponding NASS estimate of 50,990 km2,

of which 49,979 km2 were harvested for grain [52]. Our esti-

mate of corn production for 2006 is 49.886 Mt, which is 4%

lower than the NASS estimate for 2006. With crop prices of

142 $ Mg�1 for corn and 254 $ Mg�1 for soybeans, which

correspond roughly to the crop prices in the spring of 2007, our

model predicts corn production of 59.963Mt. This is 94% of the

NASS estimate for 2007. Thus, it appears that our framework

may somewhat underestimate the production of corn.

The changes in rotations described in Table 2 and shown in

Fig. 3 are directly linked to land productivity. For the most

Table 1 e Percentage of cropped land in Iowa by2002e2006 rotations.

Percent area Rotation

69.94 Corn-soybean

2.78 Continuous corn

2.12 2þ years of continuous soybean

14.06 Corn-corn-soybean

11.10 3þ years of continuous corn

productive land it becomes increasingly profitable to move

towards more corn production as corn prices increase. This

holds true even with increasing yield losses and the added

cost of fertilizer, because corn becomes relatively more prof-

itable than soybeans for highly productive land. As corn prices

increase, the break-even level of potential corn yields that

insure higher profits from more corn-intensive rotations

decreases. Thus, even less productive land shifts first to corn-

corn-soybean rotations and, as corn prices increase, more and

more of that marginal cropland is planted with continuous

corn. The shifts in rotations are tied to the shifts in tillage.

As corn prices exceed 118 $Mg�1, themore productive land

starts to shift into three year rotations. The tillage regimes are

conventional tillage on the first year of corn, tomaximize corn

yields, and mulch till on the second year of corn and on

soybeans, because there are no yield penalties. When corn

prices reach 136 $ Mg�1, the most productive land starts

shifting into continuous corn production. Because the

conservation tillage regimes have higher yields penalties,

continuous corn profits are always maximized using

conventional tillage. Once corn prices climb above 142 $ Mg�1,

marginal land begins shifting from a two-year rotation into

three-year rotations with conventional tillage on the first year

of corn, again to maximize corn yields, and no till on the

second year of corn to capture the cost savings of no-till.

Continuing corn price increases result in an initial shift of

marginal land into a three-year rotation with conventional

tillage on the first year of corn and mulch till on the second

year of corn, and then ultimately into continuous corn. The

CRP land follows a similar pattern except that the three year

rotation that is most profitable is always coupled with

conventional tillage on the first year of corn and mulch till on

the second year of corn. Thus, the tillage implications of the

shifts towardsmore corn production are not as clear cut as the

nitrogen implications described next.

The changes in crop rotations as a function of corn prices

are directly reflected in changes in nitrogen fertilizer appli-

cation. Combining our historical rotations and the nitrogen

rates discussed above, we estimate an historical use of around

814 kt of nitrogen per year, which is 17% higher than 2003

estimates of nitrogen use for corn in Iowa [35]. Our lowest corn

price scenario results in lower nitrogen applied, totaling over

743 kt. However, as three year rotations become more preva-

lent, nitrogen use increases. For the intermediate corn price

scenario, the increase is approximately 35% over our baseline,

and nitrogen use more than doubles in comparison with the

baseline for the very high corn price scenario.

Table 3 details the impact of the land use changes on four

important indicators: sediment losses, nitrogen losses, phos-

phorous and carbon. For sediment losses, we report the sedi-

ment loss fromwater erosionusing theModifiedUniversal Soil

Loss Equation (MUSLE) option provided in EPIC [46,48]. For

nitrogen losses, we report the sum of nitrate losses in runoff,

subsurface flow, and leaching, and the loss of nitrogen in

sediment. We also report total phosphorous losses and the

final carbon pool at the end of 30-year EPIC runs. Five counties

are omitted from the tabulated totals, because SSURGO data

has not yet been compiled for them. This likely causes an

underestimate of the impact since these counties are in the

southernpart of the statewheremore fragile lands are located.

Table 2 e Historical and projected land use on the basis of corn prices.

Rotation area Historic baseline Corn price 108 $ Mg�1 Corn price 142 $ Mg�1 Corn price 167 $ Mg�1

Intensive margin e current cropland (km2)

Corn-soybean 64,389 92,066 38,618 10,717

Corn-corn-soybean 12,944 0 42,784 13,974

Continuous corn 2556 0 10,664 67,375

Extensive margin e current CRP land (km2)

CRP 7087 4189 2492 2027

Corn-soybean 0 2898 2952 1050

Corn-corn-soybean 0 0 1501 1561

Continuous corn 0 0 142 2449

b i om a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 2 3 9 1e2 4 0 0 2397

All edge-of-field indicators worsen. Sediment losses

increase substantially both on the intensive and extensive

margins because of the higher incidence of conventional

tillage and continuous corn as corn prices increase. Nitrogen

losses increase, becausemore continuous corn and three year

rotations including two corn years require higher levels of

nitrogen fertilizer. Phosphorous losses increase as well. The

reason for this is the higher tillage intensity associated with

more corn production e at corn prices of 108 $ Mg�1, 85.30% of

the phosphorus losses are attached to sediment, and at corn

prices of 167 $ Mg�1, that percentage has increased to 93.29%.

Carbon sequestration levels in the soil also decrease as corn

prices increase. On currently cropped land, the decrease is less

than 5% on a per hectare basis from the baseline to the highest

corn price simulated. On the CRP land, on the other hand,

there is a 13% decrease in per hectare carbon from the base-

line to the highest corn price simulated.

This much higher marginal carbon sequestration impact

on CRP illustrates a general point. As cropland in Iowa is

already intensively farmed, the impact of increased corn

planting on environmental burdens is more obvious on the

extensive margin. Table 4 shows the per unit area impact of

farming on both CRP and current cropland, and how returning

CRP land into production has a vastly disproportionate envi-

ronmental impact, as non-cropped land shows much higher

negative marginal environmental effects when brought back

to row crop production. This illustrates the importance of

differentiating between the intensive and extensive margin

when assessing the expansion of biofuel production.

At high corn prices, forcing highly erodible land to no-till

regimes through programs like conservation compliance

would have a sizable impact on the way land currently in CRP

Table 3 e Historical and projected environmental indicators on

Historic baseline Corn price 108 $

Intensive margin e current cropland

Sediment losses (Mg) 20,041,671 36,413,102

Nitrogen losses (Mg) 545,136 609,732

Phosphorus losses (Mg) 19,120 28,020

Soil carbon (Mg) 2,020,321,455 1,978,707,457

Extensive margin e current CRP land

Sediment losses (Mg) 1,023,826 1,646,552

Nitrogen losses (Mg) 6794 20,861

Phosphorus losses (Mg) 533 1167

Soil carbon (Mg) 120,400,225 113,303,996

would be cropped, rather than on whether it stays out of

production. The reason is that for high crop prices, the returns

from farming e even if the types of tillage and consequently

rotations are somewhat restricted e are higher than the soil

rental rates paid by the CRP program. Intuitively, the same

general observation for CRP land holds if there are relative

changes in the price of soybeans. On both the intensive and

extensive margin there is a high sensitivity of the rotations to

the relative profitability of soybeans. Table 5 shows the

acreage responses to increases in soybean prices (each

equivalent to 9.19 $ Mg�1) for given high corn prices.

4. Discussion

The analysis we conduct is very data intensive, but it is

essential in order to assess the economic and environmental

ramifications of biofuel production in hotspots such as Iowa

and to ground truth and complement larger scale modeling.

For example, general and partial equilibrium trade models

only give a large scale geographical determination of the land

use change impacts of biofuels (see for example [7,50,51]). Our

integrated approach uses as inputs price projections obtained

from world-level models, and determines spatially explicit

environmental impacts by soil type at a very fine geographical

scale. Our state level land use projections can also be recon-

ciled with those of world-level models. The fine detail of the

analysis serves several purposes:

� Because we have a very rich set of data and distributions,

this work can help bracket some of the estimates of envi-

ronmental impacts and can provide estimates of their

the basis of corn prices.

Mg�1 Corn price 142 $ Mg�1 Corn price 167 $ Mg�1

45,601,316 64,457,979

741,057 934,464

31,680 34,546

1,956,628,896 1,935,990,115

3,579,939 6,633,448

39,096 57,428

2175 2862

108,297,841 104,843,115

Table 4 e Historical and projected environmental indicators.

Historic baseline Corn price 108 $ Mg�1 Corn price 142 $ Mg�1 Corn price 167 $ Mg�1

Intensive margin e current cropland

Hectares cropped (ha) 8,845,375 8,845,375 8,845,375 8,845,375

Sediment losses (Mg ha�1) 2.27 4.12 5.16 7.29

Nitrogen losses (kg ha�1) 61.63 68.93 83.78 105.64

Phosphorus losses (kg ha�1) 2.16 3.17 3.58 3.91

Soil carbon (Mg ha�1) 228.40 223.70 221.20 218.87

Extensive margin e current CRP land

Hectares cropped (ha) 0 263,703 421,103 462,026

Sediment losses (Mg ha�1) 1.65 2.66 5.78 10.71

Nitrogen losses (kg ha�1) 10.97 33.68 63.13 92.73

Phosphorus losses (kg ha�1) 0.86 1.88 3.51 4.62

Soil carbon (Mg ha�1) 194.41 182.95 174.87 169.29

The CRP results apply to all CRP soils for which EPIC runs were successful, or 619,320 ha.

b i om a s s an d b i o e n e r g y 3 5 ( 2 0 1 1 ) 2 3 9 1e2 4 0 02398

distributions, which can reduce uncertainty and can be used

to enrich Life Cycle Analysis (LCAs), for example by illus-

trating the variability of marginal environmental impacts. It

is difficult to identify the additional crops grown for energy

production, and where they are grown, which determines

their additional (or marginal) environmental impact. Esti-

mating such marginal impacts is important in correctly

quantifying the real impact of the biofuels expansion, but it

also requires data not available at a very large scale [53]. Our

approach allows the estimation of some of the impacts,

such as changes in carbon in the soil. Moreover, as biofuel

LCAs have tended to focus on energy and carbon assess-

ment, a standard methodology to study natural resource

depletion is still lacking [54]. Our framework can provide

valuable information in this area. Our research quantifies

soil erosion, can help assess long term soil fertility, and can

be used as a input for biodiversity analysis, which have all

been identified as good indicators for the environmental

assessment of land use in LCAs [55].

� As the analysis is spatially explicit and GIS based, it can

identify sub-regions of particular interest which can be

missed by world-level models. This can be helpful for

purposes such as planning in the hotspot regions where the

analysis is performed. In particular, the projected land use

changes can be used to identify likely sites where the

construction or expansion of biorefineries will occur. The

information on biomass availability could also be used, for

Table 5 e Sensitivity to soybean prices with corn prices equal

Rotation area

285 $ Mg�1 294 $ Mg�1 303

Intensive margin e current cropland (km2)

Corn-soybean 10,717 14,253 17

Corn-corn-soybean 13,974 19,133 22

Continuous corn 67,375 58,680 51

Extensive margin e current CRP land (km2)

CRP 2931 2906 28

Corn-soybean 233 462 57

Corn-corn-soybean 3167 3325 36

Continuous corn 1337 973 59

example, to plan for second generation lignocellulosic bio-

fuel plants thatmight use corn stover as a feedstock. Finally,

our analysis identifies the opportunity cost that would have

to be met by second generation biofuel crops. For example,

planting switchgrass under a certain market condition

scenario for corn and soybeans would require returns at

least equal to those obtained from planting these crops or,

for land currently retired from production, it would require

returns at least as high as those received from land retire-

ment programs. Besides the information directly derived

from the analysis presented above, our analysis can be used

as input in other studies of land use change assessments.

The maps can be overlaid with information on wildlife

habitat to estimate some of the impacts of the industry on

biodiversity. They can also be used as inputs for both

groundwater and surface water modeling to analyze likely

water quality and quantity impacts. This is going to be of

particular importance in hotspots where the production of

biofuel crops requires irrigation, such as the Great Plains

area in the US.

The main limitation of our work is that, because of its

bottom up approach, it takes world prices as given. However,

given the amount of corn produced in Iowa, it is possible that

world prices could indeed show some responsiveness to large

changes in Iowa corn acreage. We are currently extending the

analysis by linking ourmodel to a large crop productionmodel

to 168 $ MgL1.

Soybean price

$ Mg�1 312 $ Mg�1 322 $ Mg�1 331 $ Mg�1

,737 20,270 24,691 29,748

,553 32,115 41,257 51,654

,777 39,681 26,119 10,664

48 2828 2809 2745

7 885 1146 1332

46 3663 3642 3586

6 291 70 3

b i om a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 2 3 9 1e2 4 0 0 2399

to assess the size of this effect. The environmental impacts

are based on past climate data. Structural changes in precip-

itation patterns, for example, could affect the results. Since all

these cropping systems are well known, the lack of consid-

eration of risk factors is unlikely to bias the results. However,

this could be a bigger issue if the model is extended to new

crops such as perennials for biofuel production, or if corn

stover harvesting is included in the analysis.

The ethanol plants currently existing in Iowa have

a production capacity of 5.73 Mt according to the Renewable

Fuels Association (RFA) [56]. Another 5.22 Mt are planned

according to the latest RFA estimates. Given an ethanol

conversion factor of 0.33 g kg�1 of corn, the current production

requires around 17.8 Mt of corn. An additional 16.1 Mt of corn

would be required to meet currently planned additional

ethanol production capacity. The requirement of corn to meet

current production demand is equivalent to more than 35% of

the 2006 estimated production from our model, and over 27%

of Iowa’s 2007 production estimate of 64Mt fromNASS. To put

these numbers in perspective, the Iowa Corn Growers Asso-

ciation estimated that for the 2005/2006 Iowa corn crop 11 Mt

were used for ethanol production and 14 Mt were fed to Iowa

livestock.

At current yields, we estimate that if all the currently

cropped area (almost 92,000 km2) were converted into

continuous corn, Iowa could produce 88 Mt of corn. We esti-

mate that the conversion of all cropland to continuous corn

would require sustained corn prices of over 197 $ Mg�1. We

estimate that, on the extensive margin, these corn prices

would bring into production over half a million hectares of

CRP land, with corn production levels of 4.6 Mt on the exten-

sive margin. Thus, even for very high corn prices the

production on the extensive margin would equal only 5% of

the total. This type of result is likely to be replicated in other

intensely cropped areas, while the extensive margin is likely

to play a much more important role in providing biofuel

feedstock for more marginal, less productive lands or in areas

of the world where less intensely managed landscapes exist.

These are stark results, showing massive scale changes in

the production of corn in Iowa. They illustrate why Iowa is

such a hotspot for biofuel production. These results are in line

with large scale trade models which forecast that high corn

prices will severely impact soybean acreage in the US, and

that Iowa is likely to remain a net corn exporter, even with its

large livestock industry [51]. The results are also consistent

with a previous study on the environmental impact of corn

expansion in the whole Mississippi River Basin [19].

Acknowledgements

We would like to thank all the participants to the Biofuel

Assessment Conference in Copenhagen, Denmark, June 4e5,

2007, and particularly Henrik Wenzel, for helpful discus-

sions on the role of intensive and extensive margin in land

use impacts. This research was made possible in part by

USDA-CSREES grants 2005-51130-02366 and 2009-10002-

05149, USDA-NIFA grant 2010-65400-20434, NSF grant CDI

CBET-0836607, USEPA Collaborative Agreement CR83371701-

1 and The Nature Conservancy contract C09-025-SIL-P. The

views expressed here are those of the authors and do not

necessarily represent the views or the policies of NSF, USDA,

EPA and The Nature Conservancy.

r e f e r e n c e s

[1] Pimentel D. Ethanol fuels: energy balance, economics, andenvironmental impacts are negative. Nat Resour Res 2003;12(2):127e34.

[2] De Oliveira MED, Vaughan BE, Rykiel Jr EJ. Ethanol as fuel:energy, carbon dioxide balances, and ecological footprint.BioScience 2005;55(7):593e602.

[3] Farrell AE, Plevin RJ, Turner BT, Jones AD, O’Hare M,Kammen DM. Ethanol can contribute to energy andenvironmental goals. Science 2006;311(5760):506e8.

[4] Hill J, Nelson E, Tilman D, Polasky S, Tiffany D.Environmental, economic, and energetic costs and benefitsof biodiesel and ethanol biofuels. Proc Natl Acad Sci USA2006;103(30):11206e10.

[5] Fargione J, Hill J, Tilman D, Polasky S, Hawthorne P. Landclearing and the biofuel carbon debt. Science 2008;319(5867):1235e8.

[6] Searchinger T, Heimlich R, Houghton RA, Dong F, Elobeid A,Fabiosa J, et al. Use of U.S. croplands for biofuels increasesgreenhouse gases through emissions from land-use change.Science 2008;319(5867):1238e40.

[7] Melillo JM, Reilly JM, Kicklighter DW, Gurgel AC, Cronin TW,Paltsev S, et al. Indirect emissions from biofuels: howimportant? Science 2009;326(5958):1397e9.

[8] Veldkamp A, Verburg PH. Modelling land use change andenvironmental impact. J Environ Manage 2004;72(1e2):1e3.

[9] Verburg PH, Eickhout B, van Meijl H. A multi-scale, multi-model approach for analyzing the future dynamics ofEuropean land use. Ann Reg Sci 2008;42(1):57e77.

[10] Hellmann F, Verburg PH. Spatially explicit modelling ofbiofuel crops in Europe. Biomass Bioenergy 2011;35(6):2411e24.

[11] Das C, Capehart WJ, Mott HV, Zimmerman PR,Schumacher TE. Assessing regional impacts of ConservationReserve Program-type grass buffer strips on sediment loadreduction from cultivated lands. J Soil Water Conserv 2004;59(4):134e42.

[12] Feng H, Kling KL, Kurkalova L, Gassman PW. Transfers andenvironmental co-benefits of carbon sequestration inagricultural soils: retiring agricultural land in the UpperMississippi River Basin. Clim Change 2007;80(1e2):91e107.

[13] Herkert JR. Conservation reserve program benefits onHenslow’s Sparrows within the United States. J WildlManage 2007;71(8):2749e51.

[14] Haroldson KJ, Kimmel RO, Riggs MR, Berner AH. Associationof ring-necked pheasant, gray partridge, and meadowlarkabundance to conservation reserve program grasslands.J Wildl Manage 2006;70(5):1276e84.

[15] O’Neal BJ, Heske EJ, Stafford JD. Waterbird response towetlands restored through the conservation reserveenhancement program. J Wildl Manage 2008;72(3):654e64.

[16] Park S, Egbert SL. Remote sensing-measured impacts of theconservation reserve program (CRP) on landscape structurein southwestern Kansas. GISci Remote Sens 2008;45(1):83e108.

[17] Egbert SL, Park S, Price KP, Lee RY, Wu J, Nellis MD. Usingconservation reserve program maps derived from satelliteimagery to characterize landscape structure. ComputElectron Agric 2002;37:141e56.

b i om a s s an d b i o e n e r g y 3 5 ( 2 0 1 1 ) 2 3 9 1e2 4 0 02400

[18] Secchi S, Gassman PW, Williams JR, Babcock BA. Impact ofhigh crop prices on environmental quality: a case of Iowaand the conservation reserve program. Environ Manage 2009;44(4):732e44.

[19] Donner S, Kucharik C. Corn-based ethanol productioncompromises goal for reducing nitrogen export by theMississippi River. ProcNatl Acad Sci U.S.A 2008;105(11):4513e8.

[20] Simpson T, Sharpley A, Howarth R, Paerl H, Mankin K. TheNew Gold Rush: fueling ethanol production while protectingwater quality. J Environ Manage 2008;37(2):318e24.

[21] U.S. Department of Agriculture. National AgriculturalStatistics Service, Research and Development Division.2002e2006 Cropland Data Layer, http://www.nass.usda.gov/research/Cropland/SARS1a.htm; 2010 [accessed 25.03.10].

[22] Stern AJ, Doraiswamy P, Akhmedov B. Crop rotation changesin iowa due to ethanol production. In: Proceedings of theInternational Geoscience and Remote Sensing Symposium.2008 July 6e11, Boston, MA. V-200-03; 2008.

[23] Orr A.U.S. Department of Agriculture DistrictConservationist, personal communication; 2007.

[24] Koenning SR, Schmitt DP, Barker KR, Gumpertz ML. Impactof crop rotation and tillage system on Heterodera glycinespopulation density and soybean yield. Plant Dis 1995;79(3):282e6.

[25] Katsvairo TW, Cox WJ. Economics of cropping systemsfeaturing different rotations, tillage, and management.Agron J 2000;92(3):485e93.

[26] Vetsch JA, Randall GW, Lamb JA. Corn and soybean productionas affected by tillage systems. Agron J 2007;99(4):952e9.

[27] CTIC. National crop residue management survey,conservation tillage data, West Lafayette, IN; 2005.

[28] U.S. Department of Agriculture. Economic research Service.Farm business and household survey data. Customized DataSummaries From ARMS, http://www.ers.usda.gov/Data/ARMS/app/Crop.aspx; 2010 [accessed 26.03.10].

[29] Heimlich R. Agricultural resources and environmentalindicators. 2003 Edition. Washington DC: United StatesDepartment of Agriculture, Economic Research Service; 2003.p. 227, Agriculture Handbook No. AH722.

[30] Iowa cooperative soil survey. Iowa soil properties andinterpretation database, http://icss.agron.iastate.edu/; 2003[accessed 11.12.09].

[31] Sheriff G. Efficient waste? Why farmers over-apply nutrientsand the implications for policy design. Rev Agric Econ 2005;27(4):542e57.

[32] Pikul Jr JL, Hammack L, Riedell WE. Corn yield, nitrogen use,and corn rootworm infestation of rotations in the northernCorn Belt. Agron J 2005;97(3):854e63.

[33] Raimbault BA, Vyn TJ. Crop rotation and tillage effects oncorn growth and soil structural stability. Agron J 1991;83(6):979e85.

[34] Katsvairo TW, Cox WJ. Tillage rotation managementinteractions in corn. Agron J 2000b;92(3):493e500.

[35] Iowa State University Department of Agronomy. IDALS salesdata and NASS survey data, http://extension.agron.iastate.edu/soils/pdfs/Nuse/NBackch5.PDF; 2004 [accessed 16.02.10].

[36] Sawyer J, Nafziger E, Randall G, Bundy L, Rehm G, Joern B.Concepts and rationale for regional nitrogen rateguidelines for corn. Ames, IA: Iowa State UniversityExtension; 2006. p. 26, Report PM 2015.

[37] Duffy M, Correll D. Rotation Profitability Calculator usingoptimal nitrogen rate. Iowa State University ExtensionDecision Tools, http://www.extension.iastate.edu/agdm/crops/xls/a1-80nratecalculator.xls; 2006 [accessed 18.02.10].

[38] Babcock BA. Effects of uncertainty on optimal nitrogenapplications. Rev Agric Econ 1992;14(2):271e80.

[39] Vyn TJ, West TD, Steinhardt GC. Corn and soybean responseto tillage and rotation systems on a dark prairie soil: 25 year

review. In: Proceedings of the 15th Conf. of Int’l. Soil TillageResearch Org.; 2000 July 2e6; Fort Worth, TX. Fort Worth, TX:Int’l. Soil Tillage Research Org. No. 196, pp. 1e10.

[40] Wilhelm WW, Wortmann CS. Tillage and rotationInteractions for corn and soybean grain yield as affected byprecipitation and air temperature. Agron J 2004;96(2):425e32.

[41] Vetsch JA, Randall GW. Corn production as affected bytillage system and starter fertilizer. Agron J 2002;94(3):532e40.

[42] Mallarino AP, Pecinovsky K. Effects of crop rotation andnitrogen fertilization of corn on yields of corn, soybean, andoats. Ames, IA: Iowa State University, Northeast Research andDemonstration Farm; 2006. p. 2, Progress Report ISRF06-13.

[43] Duffy M, Smith D. Estimated costs of crop production inIowad2007. File A1-20, http://www.extension.iastate.edu/agdm/crops/pdf/a1-20.pdf [accessed 19.02.10].

[44] Lubowski RN, Plantinga AJ, Stavins RN. What drives land usechange in the United States? A national analysis oflandowner decisions. Land Econ 2008;84(4):529e50.

[45] US Department of agriculture natural resources conservationservice soil survey spatial and tabular data, http://soildatamart.nrcs.usda.gov/ [accessed 13.01.10].

[46] Williams JR. The erosion productivity impact calculator(EPIC) model: a case history. Philos Trans R Soc London 1990;329(1255):421e8.

[47] Izaurralde RC, Williams JR, McGill WB, Rosenberg NJ.Simulating soil C dynamics with EPIC: model description andtesting against long-term data. Ecol Model 2006;192(3e4):362e84.

[48] Gassman PW, Williams JR, Benson VR, Izaurralde RC,Hauck LM, Jones CA, et al. Historical Development andapplications of the EPIC and APEX models. Ames, IA: Centerfor Agricultural and Rural Development, Iowa StateUniversity; 2005. p. 45, Working paper 05-WP 397.

[49] Kelley RH, Parton WJ, Crocker GJ, Grace PR, Klır J,Korschens M, et al. Simulating trends in soil organic carbonin long-term experiments using the century model.Geoderma 1997;81(1e2):75e90.

[50] FAPRI Staff. 2007 FAPRI U.S. and world agricultural outlook.Ames, IA: Food and Agricultural Policy Research Institute(FAPRI) at Iowa State University; 2007. p. 411, FAPRI StaffReport 07-fsr1.

[51] Tokgoz S, Elobeid A, Fabiosa JF, Hayes DJ, Babcock BA, Yu T,et al. Emerging biofuels: outlook of effects on U.S. grain,oilseed, and livestock markets. Ames, IA: Center forAgricultural and Rural Development, Iowa State University;2007. p. 57, Staff Report 07-SR 101.

[52] U.S. Department of Agriculture. National agriculturalstatistics service. USDA national agricultural statisticsservice e Quick Stats, http://www.nass.usda.gov/QuickStats/Create_Federal_Indv.jsp [accessed 05.03.10].

[53] Delucchi MA. Lifecycle emissions model (LEM): lifecycleemissions from transportation fuels, motor vehicles,transportation modes, electricity use, heating and cookingfuels, and materials. Davis, CA: University of California; 2003.p. 444, Report No. UCD-ITS-RR-03-17.

[54] von Blottnitz H, Curran MA. A review of assessmentsconducted on bio-ethanol as a transportation fuel from a netenergy, greenhouse gas, and environmental life cycleperspective. J Cleaner Prod 2007;15(7):607e710.

[55] Mattsson B, Cederberg C, Blix L. Agricultural land use in lifecycle assessment (LCA): case studies of three vegetable oilcrops. J Cleaner Prod 2000;8(4):283e92.

[56] Renewable Fuels Association. Ethanol biorefinery locations:U.S. fuel ethanol industry biorefineries and productioncapacity, http://www.ethanolrfa.org/industry/locations/[accessed 28.10.07].