ENVIRONMENTAL IMPACT ANALYSIS OF BIOFUEL CROPS EXPANSION IN
MICHIGAN
By
Bradley J. Love

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Biosystems Engineering
2011

ABSTRACT
ENVIRONMENTAL IMPACT ANALYSIS OF BIOFUEL CROPS EXPANSION IN
MICHIGAN
By
Bradley J. Love
The environmental implications of large-scale bioenergy cropping systems are not
well understood, making conservation efforts difficult and inefficient. To address these
issues, the following research objectives were developed: 1) determine if pesticides are
suitable for lignocellulosic ethanol feedstocks, 2) provide critical information to
determine the suitability of bioenergy crops on a variety of types of agricultural land in
Michigan, 3) determine the possible environmental impacts of bioenergy cropping
systems on sediment, nitrogen, and phosphorus yield to surface waters, 4) determine the
implications of bioenergy-associated pesticides on aquatic ecosystems and human
consumption and 5) obtain information to aid in watershed-scale decision making
regarding landuse management. The Soil and Water Assessment Tool (SWAT) was used
to understand the impacts of large-scale bioenergy crop expansions on pollutant loads and
concentrations. Results indicate that intensive bioenergy crops tend to increase pollutant
loads to surface waters, whereas perennial grass species generally mitigate pollutants.
Although herbicide treatments mainly positively influenced biomass accumulation in
corn during field studies, the application of herbicides at a large-scale was found to
increase the impairments of streams in Michigan on the basis of fish ecotoxicity and
human consumption thresholds. In general, this study reinforces the importance of
maintaining a sustainable bioenergy cropping system implementation strategy to avoid
environmental and human health risks.

Copyright by
BRADLEY J. LOVE
2011

This thesis is dedicated to my family, friends, and colleagues for their generous love and
support.

iv

ACKNOWLEDGMENTS
I would first like to thank my family, my mom and dad, Jon and Gail, my sister and
brother-in-law, Katie and Chad for their love and support. Katie and Chad, thank you for
going through college first and leading the way and being an ear for phone conversations
wondering why we do this to ourselves, dad, for being my best friend and best hunting
buddy, and mom for your love, kindness, and for feeding my dog when my research kept
me from being home.
I would also like to thank Dr. Amir Pouyan Nejadhashemi for being an extraordinary
advisor, mentor, and friend. Without your invaluable research advice, our random
conversations, and your persistent encouragement to complete the work in a timely
manner, this work and my personal and professional growth would not have been
possible. I would also like to thank my committee members Dr. Kurt Thelen, Dr. Yan
(Susie) Liu, and Dr. Lizhu Wang for their support, guidance, and knowledge each has
provided throughout my research.
I would like to thank my labmates Sean, Matt, Edwin, Yaseen, Subhasis, and Mike, for
not only their help, but also for the great memories. Sean for being a great friend
throughout our time since BE 130, Matt for being the only fellow hunter and for all of the
hysterical laughs, Edwin for bringing out your Air Jordans, Yaseen and Subhasis for
changing my perspective of your great cultures, and Mike for… well thanks Mike. Also
to my instructor and friend, Dr. Fred Arkema-Bakker for his commitment to provide us
with global current events and our great conversations. Finally I would like to give thanks
to the greatest university on the planet, Go Green!
v

TABLE OF CONTENTS
LIST OF TABLES ........................................................................................................... ix
LIST OF FIGURES ......................................................................................................... xi
LIST OF ABBREVIATIONS ....................................................................................... xiii
1. INTRODUCTION......................................................................................................... 1
2. LITERATURE REVIEW ............................................................................................ 4
2.1 Overview ................................................................................................................... 4
2.2 Bioenergy .................................................................................................................. 4
2.2.1 Feedstock Sources .............................................................................................. 6
2.2.2 Benefits of Bioenergy ........................................................................................ 7
2.2.3 Problems of Bioenergy....................................................................................... 8
2.2.4 Pesticide Use with Bioenergy .......................................................................... 10
2.3 Ecosystem Services ................................................................................................ 12
2.3.1 Definition ......................................................................................................... 12
2.3.2 Types ................................................................................................................ 13
2.3.3 Valuation .......................................................................................................... 20
2.4 Watershed Models and Their Applications in Ecology .......................................... 21
2.4.1 Annualized Agricultural Non-Point Source Model (AnnAGNPS) .................. 22
2.4.2 Hydrologic Simulation Program-FORTRAN (HSPF) ..................................... 23
2.4.3 Soil and Water Assessment Tool (SWAT) ...................................................... 24
2.4.4 Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) ........... 25
3. INTRODUCTION TO METHODOLOGY AND RESULTS ................................. 27
4. THE EFFECTS OF POSTEMERGENCE HERBICIDES ON BIOMASS
ACCUMULATION AND BIOENERGY QUALITY OF FIELD AND SILAGE
CORN (Zea Mays) .......................................................................................................... 30
4.1 Introduction ............................................................................................................. 31
4.2 Materials and Methods ............................................................................................ 33
4.3 Results and Discussion ........................................................................................... 35
4.3.1 Crop Injury ....................................................................................................... 35
4.3.2 Variation by Corn Type ................................................................................... 35
4.3.3 Variation by Herbicide Treatment ................................................................... 38
4.4 Conclusion .............................................................................................................. 41
5. ENVIRONMENTAL IMPACT ANALYSIS OF BIOFUEL CROPS EXPANSION
IN THE SAGINAW RIVER WATERSHED ............................................................... 43
5.1 Introduction ............................................................................................................. 44
5.2 Materials and Methods ............................................................................................ 48
5.2.1 Study Area ....................................................................................................... 48
5.2.2 General Water Quality Issue ............................................................................ 49
5.2.3 SWAT Model ................................................................................................... 50
vi

5.2.4 Crop Rotations and Management Operations .................................................. 52
5.2.5 Model Setup ..................................................................................................... 55
5.3 Results and Discussion ........................................................................................... 74
5.3.1 Watershed-level Environmental Impacts of Bioenergy Crop Expansion ........ 74
5.3.2 Assessment of Marginal Land Conversion Impact on Water Quality ............. 77
5.3.3 Importance of Using Local Information on Model Predictions ....................... 79
5.3.4 Monthly Pollution Load Variation ................................................................... 81
5.3.5 Determining Practicality of Scenarios for Landuse Change ............................ 91
5.4 Conclusion .............................................................................................................. 99
APPENDIX .................................................................................................................... 102
6. WATER QUALITY IMPACT ASSESSMENT OF LARGE SCALE BIOFUEL
CROPS EXPANSION .................................................................................................. 110
6.1 Introduction ........................................................................................................... 111
6.2 Materials and Methods .......................................................................................... 114
6.2.1 Study Area ..................................................................................................... 114
6.2.2 General Water Quality Issue .......................................................................... 115
6.2.3 SWAT Model ................................................................................................. 116
6.2.4 Crop Rotations and Management Operations ................................................ 117
6.2.5 Model Setup ................................................................................................... 120
6.3 Results and Discussion ......................................................................................... 130
6.3.1 Basin-wide Impact of Bioenergy Cropping Rotations ................................... 130
6.3.2 Basin-level Priority Areas for Targeting Conservation Efforts ..................... 135
6.3.3 Suitability of Bioenergy Cropping Rotations on Different Scenarios ........... 142
6.3.4 Statistical Significance of Bioenergy Cropping Rotations Changes from Base
....................................................................................................................... 147
6.4 Conclusion ............................................................................................................ 149
7. EFFECTS ON AQUATIC AND HUMAN HEALTH DUE TO LARGE SCALE
BIOENERGY CROP EXPANSION ........................................................................... 154
7.1 Introduction ........................................................................................................... 155
7.1.2 Modeling Pesticide Fate and Transport ......................................................... 160
7.1.3 Pesticides........................................................................................................ 163
7.2 Materials and Methods .......................................................................................... 171
7.2.1 Study Area ..................................................................................................... 171
7.2.2 SWAT Model ................................................................................................. 172
7.3 Results and Discussion ......................................................................................... 181
7.3.1 Basin-wide Concentration Fluctuation .......................................................... 181
7.3.2 Basin-wide Impact of Pesticides on Stream Impairment ............................... 194
7.3.3 Basin-wide Temporal Variation of Stream Impairment ................................ 198
7.3.4 Basin-wide Spatial Variation of Stream Impairment ..................................... 202
7.4 Conclusions ........................................................................................................... 205
8. CONCLUSIONS ....................................................................................................... 210
8.1 GENERAL CONCLUSIONS ...................................................................................... 210
8.2 RECOMMENDATIONS FOR FUTURE RESEARCH ................................................... 212
vii

REFERENCES .............................................................................................................. 215

viii

LIST OF TABLES
Table 1. Corn maturity 28 days after postemergence treatment, dry biomass yield, percent
glucose, and ethanol yield in 2008 and 2009 by corn type averaged by herbicide
treatment. .............................................................................................................. 37
Table 2. Growing degree days (GDD) accumulated in East Lansing, MI from 4/1-10/31.
............................................................................................................................... 37
Table 3. Regression values for Figure 1. .......................................................................... 38
Table 4. Dry biomass yield by herbicide treatment and corn type averaged over years. . 41
Table 5. Sensitivity analysis result.................................................................................... 60
Table 6. Final parameter values (CDL landuse without management operations applied).
............................................................................................................................... 67
Table 7. Results of statistical analysis for calibration and validation periods. ................. 68
Table 8. Final parameter values (CDL landuse with management operations applied). .. 69
Table 9. Results of statistical analysis for calibration and validation periods (CDL
landuse with management operations applied). .................................................... 74
Table 10. Long-term average monthly sediment load (Saginaw Bay watershed). ........... 87
Table 11. Land use scenarios ranking based on environmental benefits. ......................... 95
Table 12. Fertilizer application rates for each rotation used in SWAT. ......................... 102
Table 13. Corn-soybean-canola rotation management operations. ................................. 104
Table 14. Calibration and validation periods for each basin. ......................................... 126
Table 15. Results of statistical analysis for calibration and validation periods. ............. 128
Table 16. Final parameter values from calibration. ........................................................ 130
Table 17. Length of priority stream change from base (km) for Scenario 4 (the values
with gray background represent the percent changes from the Base or current
landcover scenario). ............................................................................................ 140
2

Table 18. Area of priority subbasins change from base (km ) for Scenario 4 (the values
with gray background represent the percent changes from the Base or current
landcover scenario). ............................................................................................ 141

ix

Table 19. Total combined pollutant load of all watersheds. ........................................... 145
Table 20. Significant difference (p-value) between bioenergy rotation and base scenario.
............................................................................................................................. 147
Table 21. Bluegill and human toxicity thresholds. ......................................................... 169
Table 22. Corn-soybean-rye rotation management operations. ...................................... 177
Table 23. Pesticide application rates for each rotation. .................................................. 178
-1

Table 24. Concentrations of Atrazine (µg L ) ............................................................... 186
-1

Table 25. Concentrations of Bromoxynil (µg L ) ......................................................... 187
-1

Table 26. Concentrations of Glyphosate (µg L ) .......................................................... 188
-1

Table 27. Concentrations of Metolachlor (µg L ) ......................................................... 189
-1

Table 28. Concentrations of Pendimethalin (µg L )...................................................... 190
-1

Table 29. Concentrations of Sethoxydim (µg L ) ......................................................... 191
-1

Table 30. Concentrations of Trifluralin (µg L ) ............................................................ 192
-1

Table 31. Concentrations of 2,4-D (µg L ) ................................................................... 193
Table 32. Kilometers of stream segments exceeding threshold toxicity levels for the
Bluegill and humans (the values with gray background represent Bluegill and
white represents human threshold exceedances). ............................................... 196

x

LIST OF FIGURES
Figure 1. Corn height 28, 42, and 56 days after treatment with error bars representing one
standard deviation. ................................................................................................ 38
Figure 2. Location of the Saginaw River Watershed. For interpretation of the references
to color in this and all other figures, the reader is referred to the electronic version
of this thesis. ......................................................................................................... 49
Figure 3. Saginaw Weather Stations. ................................................................................ 51
Figure 4. Gauging Stations and 303(d) listed waterbodies. .............................................. 52
Figure 5. Precipitation deviation from 19-year mean for study period............................. 70
Figure 6. Uncalibrated Flow vs. Observed Flow at USGS Station 04157000.................. 71
Figure 7. Calibrated Flow vs. Observed Flow at USGS Station 04157000...................... 72
Figure 8. Calibrated Flow vs. Observed Flow at USGS Station 04157000...................... 73
Figure 9. 19 Year Annual Sediment/Nutrient Load at Main Channel Outlet (Scenario 4).
............................................................................................................................... 76
Figure 10. 19 Year Annual Sediment/Nutrient Load at Main Channel Outlet (Scenario 3).
............................................................................................................................... 78
Figure 11. Annual Average Sediment Load for Three Corn Simulations. ....................... 80
Figure 12. Corn Stover vs. Switchgrass Scenario 4 Monthly Sediment Load Change from
Base. ...................................................................................................................... 83
Figure 13. Corn Stover vs. Switchgrass Scenario 4 Monthly Nitrogen Load Change from
Base. ...................................................................................................................... 84
Figure 14. Corn Stover vs. Switchgrass Scenario 4 Monthly Phosphorus Load Change
from Base. ............................................................................................................. 85
Figure 15. Scenario 1 (a), 2 (b), 3 (c), and 4 (d) implemented in the SWAT model. ..... 106
Figure 16. Basins included in study area. ....................................................................... 115
Figure 17. Precipitation deviation from 19-year mean for study period......................... 126
Figure 18. 19-year annual average basin-wide sediment load at the watershed outlet. .. 133
Figure 19. 19-year annual average basin-wide total nitrogen load at the watershed outlet.
............................................................................................................................. 134

xi

Figure 20. 19-year annual average basin-wide total phosphorus load at the watershed. 134
Figure 21. Priority streams for base scenario sediment, nitrogen and phosphorus (a,b,c),
continuous corn sediment, nitrogen and phosphorus (d,e,f), and switchgrass
-1
sediment, nitrogen and phosphorus (g,h,i) (in µg L ) ....................................... 138
Figure 22. Priority areas for base scenario sediment, nitrogen and phosphorus (a,b,c),
continuous corn sediment, nitrogen and phosphorus (d,e,f), and switchgrass
-1
sediment, nitrogen and phosphorus (g,h,i) (in kg ha ) ...................................... 139
Figure 23. Basins included in study area. ....................................................................... 172
Figure 24. Basinwide variability of Glyphosate concentrations in log-scale (Dotted line
represents the Bluegill ecotoxicity threshold of 5600 ppb and solid line represents
the human toxicity threshold of 700 ppb). .......................................................... 185
Figure 25. Basinwide temporal variation in stream length impairments of three crop
rotation scenarios based on fish thresholds ......................................................... 200
Figure 26. Basinwide temporal variation in stream length impairments of three crop
rotation scenarios based on human thresholds .................................................... 201
Figure 27. Streams exceeding Bluegill toxicity threshold (top) and human threshold
(bottom) for base, continuous corn, and switchgrass (in order left-to-right) ...... 204

xii

LIST OF ABBREVIATIONS
Parameter

Description

Ae

Acid equivalent

Ai

Active ingredient

Alpha_Bf

Baseflow alpha factor (days)

AnnAGNPS

Annualized Agricultural Non-Point Source Model

Biomix

Biological mixing efficiency

Blai

Maximum potential leaf area index

Canmx

Maximum canopy storage (mm)

CDL

Crop Data Layer

Ch_K2

Channel effective hydraulic conductivity (mm/hr)

Ch_N2

Manning’s nvalue for main channel

Ch_Cov

Channel cover factor

Ch_Erod

Channel erodibility factor

Cn2

Initial SCS CN II value

DEM

Digital Elevation Model

Epco

Plant uptake compensation factor

Esco

Soil evaporation compensation factor

GIS

Geographic Information Systems

Gw_Delay

Groundwater delay (days)

Gwgmn

Threshold water depth in the shallow aquifer for flow (mm)

HRU

Hydrologic Response Unit

HSPF

Hydrologic Simulation Program-FORTRAN

xiii

HUC

Hydrologic Unit Code

MUSLE

Modified Universal Soil Loss Equation

Nperco

Nitrogen percolation coefficient

NSE

Nash-Sutcliffe Efficiency

N_Updis

Nitrogen uptake distribution parameter

Phoskd

Phosphorus soil portioning coefficient

Pperco

Phosphorus percolation coefficient

Ppm

Parts per million

Ppb

Parts per billion

R

2

Coefficient of determination

Rchrg_Dp

Deep aquifer percolation fraction

RMSE

Root mean square error

RUSLE

Revised Universal Soil Loss Equation

Smtmp

o
Snowmelt base temperature ( C)

Sol_Awc

Available water capacity (mm H2O/mm soil)

Sol_Z

Soil depth (mm)

Sol_No3

Initial NO3 concentration (mg/kg)

Sol_Orgn

Initial Organic N concentration (mg/kg)

Sol_Orgp

Initial Organic P concentration (mg/kg)

Spcon

Linear re-entrainment parameter for channel sediment routing

Spexp

Exponential re-entrainment parameter for channel sediment routing

xiv

STATSGO

State Soil Geographic Database

Surlag

Surface runoff lag time (days)

SWAT

Soil and Water Assessment Tool

Timp

Snow pack temperature lag factor

TKN

Total Kjeldahl Nitrogen

TN

Total nitrogen

TP

Total phosphorus

USDA

United States Department of Agriculture

USEPA

United States Environmental Protection Agency

USGS

United States Geological Survey

Usle_P

USLE support practice factor

xv

1. INTRODUCTION
It is an accepted fact that the availability of fossil fuels is finite and that there will
be a peak of oil production. It is suggested that fossil fuel availability is likely currently
in decline. In fact, there has been more oil consumed than found for longer than the past
two decades (Hanlon and McCartney, 2008). Affordable and accessible oil is therefore
beginning to become less obtainable. As domestic crude oil production has decreased
over the past decades, imports have filled the gap. For example, Canada nearly doubled
their oil production between 2001 and 2007, mainly from oil sands requiring large
amounts of water for extraction, to accommodate the immense oil demands of the United
States (Wu et al., 2009). Fossil fuels account for about 80 percent of global energy
supply, and will be exhausted in a matter of decades at current consumption rates
(Goldemberg, 2007). The instability of the global energy sector has led to recent
increases in the demand for alternatives, especially in developed countries with a
dependence on fossil fuels. Prior to the mid-1990s, Brazil was the only country
employing large-scale production of energy through biological means, whereas today the
spread of bioenergy production has swept the entire globe (Mol, 2007). Public awareness
of United States dependence on oil and the insecurity of the energy sector, climate
change, and the current economic downswing has led to great support of renewable
energy sources (Wu et al., 2009). In order to meet the demands of a growing global
energy budget in a sustainable approach, increased energy use efficiency of existing
carbon-based fuels and an unfathomable supply of carbon-neutral energy sources will be
required (Lewis and Nocera, 2006). Model-based assessments examining bioenergy
expansion rarely account for the competition of land for production of food and animal

1

feed (Schaldach et al. 2010). The implications resulting from a large-scale shift to
bioenergy cropping systems to fulfill these needs are not well-known, in which
unintended consequences to water resources and ecosystem services such as biodiversity
may result (Groom et al., 2008; Stone et al., 2010).
One of the major challenges facing scientists is to determine the associated
environmental risks that energy production from agricultural biomass may provoke
(Carroll and Somerville, 2009). In addition, determining the vulnerabilities of production
spatially is necessary prior to implementing bioenergy crops into systems currently used
mainly for food production (Mol, 2007). Understanding the environmental consequences
of a production system based on bioenergy crops is needed to develop cost-effective and
sustainable management strategies to negate irreversible water quality impacts (Thomas
et al., 2009). However, the impact of large-scale production of bioenergy crops on water
resources is not well-known (Wu et al., 2009), which is one of the goals of the research
presented in this manuscript.
The research in this paper evaluates the long-term environmental implications,
specifically water quality, of large-scale bioenergy cropping system expansion using the
Soil and Water Assessment Tool (SWAT) for four watersheds in Michigan over a variety
of hypothetical landuse and crop rotation scenarios.

The specific objectives of this study are:
Determine the suitability of pesticides for application on lignocellulosic ethanol
feedstocks to avoid yield loss
Provide critical information to aid in determining the suitability of bioenergy
crops on a variety of types of agricultural land in Michigan

2

Determine the possible environmental impacts of bioenergy cropping systems on
sediment, nitrogen, and phosphorus yield to surface waters
Determine the implications of bioenergy-associated pesticides on aquatic
ecosystems and human consumption
Obtain information to aid in watershed-scale decision making regarding landuse
management

3

2. LITERATURE REVIEW
2.1 OVERVIEW
This literature review is divided into sections describing recent global energy
demand issues and potential alternatives, bioenergy and the associated water quality and
quantity implications, pesticide use repercussions of bioenergy cropping systems,
ecosystem services and their role in the valuation of environmental benefits, watershed
models useful in determining the potential alterations in the availability of ecosystem
services, modeling bioenergy cropping system expansion, and the gaps in knowledge to
be addressed by the research presented in the subsequent chapters.

2.2 BIOENERGY
The increased demand of global and domestic energy has led to the start of a
transition to alternative sources. Renewable fuel sources originating in the agricultural
sector have become a very important means for the replacement, or at least partial
substitution, of fossil fuels. Domestic renewable energy sources have the potential to
reduce dependence on foreign resources, mitigate climate change, diversify domestic
energy sources, and allow for economic growth and rural development (Scheffran and
BenDor, 2009). Bioenergy, or energy obtained through biological sources, is the focus of
much political debate and research efforts in recent times. Energy from biomass could
potentially provide a large portion of the sustainable global energy demands, yet
currently only accounts for 7 percent of total global energy supplies (Berndes et al., 2003;
de Fraiture et al., 2008). Policy that has been introduced as targets to increase the
production of alternative fuels include:
4

Biomass Research and Development Act of 2000 (Chum and Overend, 2001)
Energy Policy Act of 2005 (domestic use of renewable fuels to 28.4 billion liters
by 2012) (Balat and Balat, 2009)
Energy Independence and Security Act (EISA) of 2007 (136 billion liters of
biofuels by 2022) (Balat and Balat, 2009)
Farm Bill (2002 and 2008) (Johnson 2010)
The United States Department of Energy (USDOE) has enacted the Biofuels Initiative
(BFI) that sets a target goal of making fuels derived from cellulosic biomass cost
competitive by 2012 and replacing 30 percent of current petroleum-based transportation
fuels with biofuels by 2030 (USDOE 2007). By the year 2025, global energy
consumption is expected to increase by over 57 percent from 2002 values (Rooney et al.,
2007). Meanwhile, renewable energy accounts for only about 7 of the nation’s 99.3
Quadrillion Btu consumption (USEIA 2009).
To hastily meet demands for biomass and the commitment guidelines set forth by
policymakers, significant alterations to current landuse leading to increases in cultivated
acreage will result (Rowe et al., 2009). However, large-scale production of bioenergy
feedstocks are limited due to landuse competition with traditional food crops (Scheffran
and BenDor, 2009). In addition, global population continues to grow, which relies on
agriculture to produce more amounts of food on less land (Carroll and Somerville, 2009).
In order to avoid sharp inclines in food prices, biofuel crops must also be considered to
be grown on agriculturally marginal land or surplus lands with degraded production
capabilities (Lal and Pimentel, 2007). Marginal or degraded lands account for 1460 Mha
of the 5700 Mha used for agricultural purposes worldwide, in which the production of

5

biofuels may retard the rate of further degradation through conventional, intensive
agricultural practices (Carroll and Somerville, 2009). However, the implications of
expanding cropland to accommodate demands (both energy and food) are not adequately
understood at a large-scale (Rowe et al., 2009).
2.2.1 Feedstock Sources
Biofuels are liquid fuels, such as ethanol and biodiesel, produced from biological
feedstocks and mainly used for transportation and heating (de Fraiture et al., 2008). Lynd
et al. (2009) specifies four main categories of feedstocks that have potential in a
bioenergy-based markets including:
Lignocellulosic materials (e.g. perennial grasses, woody biomass, agricultural
residues)
Starch-rich crops (e.g. corn and wheat)
Sucrose-rich crops (e.g. sugarcane and sugarbeets)
Oilseed crops (e.g. soybeans, palm, and canola)
The feedstocks that are considered a realistic source for biofuels in the United States
include native prairie grasses, corn, corn stover, wheat straw, sugar cane, switchgrass,
soybeans, rapeseed, oil palm, poplar, willow, wood and crop residues, and algae (Groom
et al., 2008). Meanwhile, one of the major goals of the USDOE is to increase the
economic stability and improve competiveness of cellulosic feedstocks (primarily
switchgrass and agricultural/woody residues) for liquid transportation fuels (Lorenz et al.,
2009). Crop residues, unlike biomass from grains that compete with food and feed
markets, have great potential as a bioenergy feedstock for both direct combustion and for
the synthesis of ethanol from refineries capable of converting lignocellulose (Moebius-

6

Clune et al., 2008). The removal of large-quantities of crop residue from farmland
oftentimes reduces soil health and depletes organic matter, reduces the carbon
sequestering abilities of the soil, and increases soil erosion potential (Lal and Pimentel,
2007). However, by removing excess residue from the soil surface, soils warm faster in
the spring, have better seed germination, and provides less available habitat for unwanted
diseases or parasites (Moebius-Clune et al., 2008).
2.2.2 Benefits of Bioenergy
The benefits associated with the production of biofuels are commonly referenced.
Biofuels are looked to as an answer to reducing greenhouse gases responsible for climate
change and the nation’s vast addiction with fossil fuels, which are politically risky and
environmentally harmful (Solomon, 2010). Reducing fossil fuel dependency has the
potential to cut consumer costs, mitigate climate change, create new jobs, promote the
supply of domestic energy, and diversify energy sources ensuring domestic energy
security (Goldemberg, 2007). Climate change is arguably one of the most important
issues of the planet and to the sustainability of human practices. Over the long-term past,
the mean concentration of atmospheric CO2 increased from 280 ppm to 380 ppm
between the 1700s and 2005, respectively (Raupach et al., 2007). Abrupt climate change
resulting from this enormous increase of greenhouse gases will likely be responsible for
altering both frequency and magnitude of natural phenomena such as drought, floods, and
extreme wind and temperature (Gregory et al., 1997; Alley, 2004). Bioenergy crops show
promise in providing carbon sequestering benefits and being a carbon neutral energy
source or a net carbon sink (Adler et al., 2007).

7

In addition, long-rotation perennial bioenergy crops such as C4 grasses (e.g.
switchgrass and miscanthus) and woody biomass (hybrid poplar) have higher water use
efficiencies (WUE), which is crucial not only to freshwater resources but also to be able
to provide sustainable cropping systems under future conditions predicted to have less
available water for cultivation (Oliver et al., 2009). Liebig (2008) has shown that grass
species such as switchgrass have potential to reduce soil erosion and maintain or enhance
soil organic carbon levels.
2.2.3 Problems of Bioenergy
Unfortunately, the negative aspects of bioenergy production are ever-present and
must be considered prior to any significant implementation of such practices. In order for
biofuels to be a sustainable solution to fossil fuels, their cultivation must avoid additional
negative environmental impacts. The expansion of bioenergy systems to meet future
demands will require tremendous land requirements, making it essential to examine the
large-scale environmental impacts these practices may generate (Nyakatawa et al., 2006).
It is likely that the increased demand for bioenergy feedstocks may ultimately transform
current agricultural landscapes as production systems evolve and new species of crops
become commercially available. When landuse change is considered, the benefits of
biofuels become difficult to quantify and may be negated due to possible environmental
impacts (McLaughlin and Walsh, 1998). For example, greenhouse gas emissions
associated with landuse change from current landcover to large-scale corn production for
ethanol are estimated to double (Searchinger et al., 2008).
Climate change will undoubtedly affect water resources. Higher temperatures and
less precipitation associated with climate change will ultimately force agriculture utilize

8

more freshwater for irrigation, even though this sector already accounts for 80 percent of
all water consumed in the United States (Stone et al., 2010). In order to meet the Billion9
10 3
Ton Vision of the USDOE, 8.64 x 10 and 5.01 x 10 m of water for current and future

(2030) goals is required, a 6-fold increase (Stone et al., 2010). The increased water use
for both cultivation of feedstock sources and the refining processes associated with
biofuels may negatively impact biodiversity and other ecosystem services (Groom et al.,
2008). Increasing the yields of bioenergy crops to meet demands will require increased
amounts of agricultural inputs (Searchinger et al., 2008). Corn, the current crop supplying
majority of biofuels in the United States, requires more fertilizer and pesticides than any
other crop grown in the nation (Pimentel and Patzek 2005). However agriculture is
already responsible for the largest contribution of non-point source pollution to
freshwater in the United States (Foley et al., 2005). The increased inputs associated with
many bioenergy crops may lead to a higher number of impaired waterbodies, resulting in
risk of human and aquatic ecosystem health hazards (Nyakatawa et al., 2006). Row crop
production is the largest contributor of nitrogen and phosphorus fluxes to the Mississippi
River basin resulting in eutrophication, reduced dissolved oxygen, algal blooms, and
impaired aquatic habitat of surface waterbodies, and hypoxic zones in the Gulf of Mexico
(Carpenter et al., 1998; Powers, 2007). Eutrophication is commonly responsible for
impaired waterbodies and contaminated drinking water supplies, which increases the cost
of treatment for human consumption (Carpenter et al., 1998; Nyakatawa et al., 2006).
The supply of adequate and clean freshwater is vital to human existence and is a
fundamental requirement for sustainable agricultural production of bioenergy feedstocks
(Gleick, 1998).

9

Agricultural expansion, most likely as input-intense monoculture cropping
systems, decreases available habitat for many species, which may inhibit population
dynamics and healthy reproduction (Groom et al., 2008). This could be avoided in part
by considering the cultivation of cellulosic feedstocks including perennial, polycultured
species such as native grasses. However, crops such as these must be economically
competitive with traditional crops currently grown for food and fuels and with other
fossil fuel sources, which is unlikely without large incentives and subsidies (Khanna et
al., 2008). Even with economic incentives, the cost of losing ecosystem services such as
biodiversity, carbon sequestration, crop pollination, erosion control, water quality
enhancements, flood mitigation, and food production must be considered, which is an
extremely daunting task (Groom et al., 2008).
2.2.4 Pesticide Use with Bioenergy
As previously mentioned, an increase in land under agricultural production is
expected with an increased demand for biomass for renewable energy. This causes
conflicts in multiple ways. For example, competition for land used to produce food and
animal feed is likely to inflate food prices and the amount of arable land under production
will increase to accommodate demands. In addition, water resources are expected to
experience negative consequences due to increased water consumption and pollution
resulting from an increased amount of commercial fertilizers and pesticides used as a
means to increase yields (Siber et al., 2009). Most pesticides can be harmful to humans,
animals, insects, aquatic life, and other organisms due to their hormone disrupting
behavior and ability to bioaccumulate in organisms as the predator progresses down the
food chain, making consumption toxic and oftentimes leading to chronic illness or

10

reproductive harm (Holvoet et al., 2008). Therefore, there has been a growing concern
over the damage to ecosystems and to human health when using surface waters as a
drinking water source in agricultural watersheds (Huber et al., 2000; Vazquez-Amabile et
al., 2006). Pesticides can be released into rivers and aquifers during and shortly after
precipitation events that occur after chemical application (Siber et al., 2009). The most
significant pathway for agricultural chemicals to enter streams is through transportation
in surface runoff, in which dissolved pesticide losses in runoff generally exceed that of
sediment-bound chemicals (Huber et al., 1998; Luo et al., 2008). Aquatic species are
especially vulnerable to high levels of soluble pesticide in surface waters (Holden, 1972;
Serrano et al., 1995).
Due to the possible detrimental unintended consequences of extensive agricultural
chemical use, it is vital to investigate pesticide transport and loss on a large scale.
However, decisions regarding landuse management have traditionally been based on
rough estimates of pollutant loading, which fails to incorporate the complex relationships
between climate, soil and hydrologic characteristics (Kannan et al., 2006). By modeling
the fate and transport of pesticides in a watershed with a hydrological model, an
evaluation of environmental impacts resulting from various management practices and
landuse scenarios is more effective and representative of the true physical and chemical
interactions (Du et al., 2006). By having a better understanding of pesticide fate and
transportation throughout a watershed, decisions regarding priority areas to target
conservation efforts are made more effectively and efficiently, providing greater care of
protecting fragile and valuable ecosystems. Several models exist that are useful in

11

determining the effects of agricultural practices on watershed health, which are discussed
in detail below.

2.3 ECOSYSTEM SERVICES
2.3.1 Definition
Ecosystem services are the resources, products, and processes the environment
provides that promote sustainable and comfortable human life. By identifying and
quantifying these services, the measurement of these contributions and worth of their loss
is possible (Ruffo and Kareiva, 2009). Many processes, such as crop pollination and
carbon sequestration, go unnoticed on a day-to-day basis. Resources such as clean water,
energy, and pharmaceuticals are often taken for granted. It is relatively easy to place
value on tangible items. Valuing the loss of an intangible service that is assumed as finite
becomes considerably difficult.
Services can be classified into public and private services. Public services are
considered nonrivalry (the use/consumption of a service by one person does not reduce
the opportunity for others to receive the same benefits) and nonexcludability (no one is
excluded from the use/consumption of any given service) (Maler et al., 2009). Climate
change is an example that satisfies each; if climate changes for one, it will change for all
utilizing the same service, and climate change excludes no one from its use (Maler et al.,
2009). Services that do not fall into these typologies are termed private services (Maler et
al., 2009). Determining services as public or private influences the value given and the
method of valuation techniques, as public services are much more difficult to value than
private services that are commonly traded or sold (Maler et al., 2009).

12

2.3.2 Types
2.3.2.1 Biodiversity
Traditional views of conservation have focused on biodiversity, or the variety of
life ranging from plants and animals to genes and biomes (Naidoo et al., 2008).
Understanding the overlap of biodiversity and ecosystem services is a fairly new concept.
Ecosystem services are provided from biodiversity itself. The intermediate-disturbance
hypothesis is a common means of explaining the maintenance of biodiversity in
ecosystems (Roxburgh et al., 2004). When disturbance is between extreme levels, the
biodiversity in the affected region should be at its highest. However, negative biological
impacts have been an increasing concern of anthropogenic activity. Land-use changes
and poor land management practices have altered ecosystems in ways consequently
resulting in harmful effects to the well-being of humans and natural systems (Nelson et
al., 2008).
Biodiversity as a term is ubiquitous in environmental literature and policy.
However, ecological, social, and economic sustainability, in which biodiversity provides
the very underpinnings, is relatively unknown among the general public (Cork, 2001). In
recent years, after the American National Forum on Biodiversity held in 1986, a spike in
global awareness of the effects of loss of organisms, communities, and ecosystems has
increased the usage and understanding of biodiversity (Thompson and Starzomski, 2007).
In science, biodiversity is commonly measured with methods of species richness. By
linking species richness and ecosystem function, developing an argument for the
conservation of biodiversity is further strengthened (Schwartz et al., 2000). However, the
effects of biodiversity on the functioning of ecosystem services are explained more
thoroughly with the attributes of species rather than only species richness (Giller et al.,

13

2004).

These methods, in a sense, determine the sensitivity of an eco-region to

disturbance. Biological indicators, such as macroinvertebrates, are also good sources in
assessing the health and level of disturbance in a given ecological area (Walsh, 2006).
Indicators are vital in the proper assessment in determining biodiversity protection and
maintenance of ecosystem processes (Karr, 1991). Most of the causes of decreased
biodiversity, such as forced interactions of native and human-introduced species, can be
regulated and reversed with policy and landuse management changes (Chapin et al.,
2000). Without careful and adequate preservation efforts, the very “nature” we take for
granted will continue to be at risk of great harm. As a result of global environmental
change, biodiversity is decreasing a thousand times faster than that indicated by fossil
records (Balvanera et al., 2006). Bioenergy cropping systems, with an expected increase
in monocropping of a single crop type being cultivated consecutive years without proper
rotation, reduces biodiversity in the plant, herbivore, and decomposer ecosystems (Swift
and Anderson, 1993). However, coupled with increased global energy-use efficiency and
careful environmental considerations, renewable energy sources can potentially expand
the diversity of non-fossil, sustainable energy sources (Koh and Ghazoul, 2008). Crops
with relatively low agricultural input, such as mixtures of perennial native grasses,
cultivated on degraded lands avert habitat destruction and enhance biodiversity over
traditional monocultures such as corn and soybeans (Tilman et al., 2006).
2.3.2.2 Carbon Storage
Carbon dioxide is a main contributing greenhouse gas (GHG) to suspected global
climate change. Reducing carbon dioxide emissions has become an extremely central
topic in governments worldwide (Guthrie and Kumareswaran, 2009). The Kyoto Protocol

14

is one of the largest attempts at promoting sustainable carbon-related processes
worldwide, setting GHG emission limits on participating countries (Laurijssen and Faaij,
2009). Three main Kyoto mechanisms were developed – emissions trading, clean
development mechanism, and joint implementation (Garcia-Quijano et al., 2007). From
these provisions, the trading of “carbon credits” has emerged as an additional means to
stimulate cooperation. Projects that sequester carbon are eligible to earn credits that are
sold to entities that require additional credits to compensate for their carbon emissions.
The Chicago Climate Exchange and the European Climate Exchange recognize carbon
sequestration as a market and actively trade GHG credits. Furthermore, terrestrial-based
carbon sequestration has become an extremely valuable ecosystem service (Lippke and
Perez-Garcia, 2008). The economy, environment, and society benefit from the market
value given to the continued sequestering and long-term storage of carbon.
When there is a landuse change from natural vegetation to farmland or urban land,
carbon sequestered in the soil and tied-up in above ground biomass is released to the
atmosphere, contributing to climate change (Chan et al., 2006). More than a third of
carbon dioxide emission since 1850 has been a result of land conversion, thus natural
ecosystems provide a valuable service of storing carbon (Chan et al., 2006).
According to the Intergovernmental Panel on Climate Change (IPCC), trees
sequester significant amounts of carbon for the atmosphere as stored carbon-based
biomass (Guthrie and Kumareswaran, 2009). Emission trading programs provide an
incentive for lowering the cost of emission mitigation programs and technology, and
promote environmentally-friendly land use changes that support economic sustainability
(Ruddell 2006).

15

In

order

to

avoid

unintended

negative

consequences

that

may

be

counterproductive in promoting participation in reduced-carbon footprint developments
and production practices, cap and trade programs need to include all markets that impact
carbon emission or storage and sequestration (Lippke and Perez-Garcia, 2008). Carbon
sequestration is regulated via plant growth, plant death, and plant oxidation (Huston and
Marland, 2003). These factors determine the biomass produced, the size and thus amount
of carbon available to be stored in the plant, and the amount of carbon reintroduced into
the land (Huston and Marland, 2003). Converting lands, especially native vegetation and
forestland, to bioenergy crops could potentially create a carbon debt by releasing stored
carbon to the atmosphere (Fargione et al., 2008). On the contrary, implementing these
crops on degraded lands can provide GHG mitigation and simultaneous economic
advantage (Fargione et al., 2008). The greenhouse gas impact of renewable energy is one
of its most attractive qualities. When bioenergy crops, especially low-input grasses, are
placed on suitable lands and properly managed, they are capable of sequestering carbon
while avoiding competition for land for food production (Tilman et al., 2006).
2.3.2.3 Crop Pollination
Pollination is the natural transfer of pollen to the ovaries of a flowering plant in
order for fertilization to occur. Pollination by animal means is required for 15-30 percent
of the US food supply and many other insects other than honeybees contribute to
pollination of numerous crops (Chan et al., 2006). The stability and amount of crop
pollination services that native bee species provide depends on the proportion of natural
upland habitat and vegetation surrounding the site, rather than farm management
practices (Kremen, 2005). Many farmers temporarily import colonies of European

16

honeybees during bloom of crops to ensure adequate pollination occurs (Kremen, 2005).
However, native, unmanaged bee populations near natural habitat are generally more
diverse and abundant, which provides enhanced stability, quality, and quantity of
pollination services (Kremen, 2005). Conserving wild, native pollinators near agricultural
habitats improves pollination, which increases crop yield (Zhang et al., 2007). It is also
important to note that competition for pollinators from flowering weeds and other noncrop plants may decrease yields in some instances (Zhang et al., 2007). The success of
crop cultivation directly relates to the caloric and micronutrient availability for human
consumption (Klein et al., 2007). Furthermore, the success of pollinators completing their
tasks effects the production of food, promotes increased biodiversity, provides the
necessary reproduction for plants to continue the food chain, allows for the growth of
vegetation for carbon sequestration, amends soil organic material, acts as a source of
water treatment, and provides many other ecosystem services.
Not all plants require pollinators, such as wind pollinated or self-pollinating
flowering plants. However, 87 of 115 globally important crops require pollination from
some animal source (Klein et al., 2007). With habitat conservation considered,
maintaining pollination is a valuable ecosystem service. The delivery of this service
depends on habitat conservation and is influenced by changing landuse (Lonsdorf et al.,
2009). Just as landuse change could potentially negate or worsen GHG emissions of
bioenergy production, crop pollination could experience negative impacts resulting from
loss of insect habitat and diversity (Norris, 2008). In addition, large-scale landuse change
into wind-pollinated transgenic species could potentially provide a means of transgene
leaking into native species (Moon et al., 2010). Therefore it is imperative to understand

17

the environmental consequences of such a shift and to have the appropriate biotechnology
policy in place to maintain sustainable genetic containment (Moon et al., 2010).
Therefore, mixtures of native grass species remain the most suitable bioenergy crop to
maintain suitable pollinator habitat and diversity. Research has shown positive
correlation with the diversity of pollinating insects and birds with plant diversity in
grasslands in Wisconsin, further solidifying the benefits of perennial polycultures and
native species over traditional intensive crops for insect and animal diversity (Fargione et
al., 2009). Landuse intensification may ironically result in reduced productivity due to a
loss of important insect species, thus a reduction in vital pollination services (Bos et al.,
2007).
2.3.2.4 Water Quality
Maintaining the supply of freshwater to agricultural, industrial, and residential
lands requires both limiting the degradation of water quality and maintaining active water
purification through natural means such as wetlands (Chan et al., 2006). Sediment is the
largest pollutant of waterways in the United States (Baker, 1992). Furthermore,
agriculture is the largest contributor to excess nutrients, mainly nitrogen and phosphorus,
in freshwater in the United States (Foley et al., 2005). Soil erosion and the deposition of
soil, or sedimentation, occur naturally. However, through the use of modern agricultural
techniques and development with dramatic soil disturbance, soil erosion has accelerated,
leading to soil fertility and crop production issues. Sediment that accumulates in wetlands
can affect the functioning of the wetland and the ecosystem services provided, such as
floodwater storage and habitat, resulting in lower wetland productivity and decreased
biodiversity (Gleason, 2008). Excessive sedimentation can also sacrifice the integrity and

18

operation of infrastructure such as dams, treatment basins, wetlands, and flood mitigating
reservoirs. The added maintenance to avoid such issues is costly and increases where
sediment removal is difficult, which can lead to ecosystem degradation. The use of best
management practices (BMPs) has reduced wind and water soil erosion on sensitive
cropland and grasslands. Conservation practices through land use and management
practices can significantly reduce soil erosion and nutrient loss from soils, thus improving
the ability for the delivery of ecosystem services (Gleason, 2008). Natural landscape acts
as a sediment retention service by retaining soil and decreasing the velocity of overland
flow. Increased sediment transport associated with intensifying agricultural production
for bioenergy feedstocks can potentially cause nutrient loading issues as bound nutrientsoil particles are carried in overland flow during rain events (Nyakatawa et al., 2006).
Nutrient loading of nitrogen and phosphorus in surface waters results in algal blooms,
reduced dissolved oxygen levels, loss of biodiversity, and impaired aquatic habitat and
reproduction (Carpenter et al., 1998). In addition, excessive loading of these pollutants
oftentimes result in eutrophic conditions, leading to impaired waterbodies and drinking
water contamination, jeopardizing human safety and the health of aquatic ecosystems,
and reducing the overall aesthetics of the waterbody (Carpenter et al., 1998; Nyakatawa
et al., 2006). Therefore, the sustainability of bioenergy systems is immensely reliant on
the decisions made for landscape management to provide economic benefits while
maintaining environmentally-friendly landuse practices. As a result of the tremendous
uncertainty in water resource allocation during a time of great agricultural expansion, this
topic is the defining focus of the presented research.

19

2.3.3 Valuation
Valuation is the last of three steps in decision making (Daily et al., 2000). The
first step, identifying any alternatives, is an extremely important part of decision making,
as well as the second step of identifying and measuring all possible impacts for each
scenario (Daily et al., 2000). However, valuation aids in determining the extent of any
and all consequences a decision has on current and future well-being (Daily et al., 2000).
Placing a quantifiable value on a system provided by an intangible being is difficult at
best, and involves human judgment, which is in no way the absolute answer to the worth
of losing or changing such a service (Daily et al., 2000). The economic value of the
environment and the associated reasons for its conservation are complex and unclear.
Current estimations for the value of ecosystem services $33 trillion per year compared to
a global GNP of $18 trillion per year (Costanza et al., 1998). However, ecosystem
services are often given little weight in terms of policy making, which may ultimately
perturb the sustainability of human life (Costanza et al., 1998). The concept of placing
value on intangible pieces of human life and the environment is difficult to fathom.
Changes in the quality or quantity of services influence the benefits delivered to humans
or the costs humans face with the removal of such services (Costanza et al., 1998). A
large portion of positive contributions that the environment makes on human well-being
benefits people directly, rather than using the market as a vector for translocation of
benefits (Costanza et al., 1998). Clean air and water, climate regulation, and recreation
are examples of services that are oftentimes overlooked and undervalued. In recent years,
with a decline in the honey bee population, the value of ecosystem services such as these
has become a topic of substantial interest. It is estimated that pollination and pest control
from native insect species was worth in excess of $240 billion at 1994 prices (Norris,

20

2008). It is becoming increasingly apparent that maintaining a sustainable ecosystem is
crucial both economically and environmentally. Although bioenergy cropping systems
have the potential to provide cost-effective, domestic energy, it is clear that without the
conservation of ecosystem services, the long-term outlook for the agroeconomy could
become disastrously bleak. By taking ecologically-informed approaches to landscape
management that include ecosystem service provision, agriculture provides a means of
effectively increasing ecosystem services (Porter et al., 2009).

2.4 WATERSHED MODELS AND THEIR APPLICATIONS IN ECOLOGY
Due to the complex physical and chemical relationships present in ecosystems,
computer simulation models are useful in determining long-term impacts of altering
agricultural land management (Thomas et al., 2009). Evaluating hypothetical situations
that may be present in the future and predicting the impacts on ecosystem services at a
large-scale requires sophisticated computer models. Proper model selection depends on
many factors including the availability data, quality of data, desired model capabilities
and the scope of research objectives. Three main types of watershed models exist:
empirical, conceptual, and physically based. Physically based models account for
physical characteristics of the region such as landcover, soil type, topography and
climatic data. Empirical and conceptual models usually do not include these factors.
Complex physically based models are useful in describing processes as a result of natural
phenomena, but oftentimes require much more data and input parameters. Multiple
computer models exist that are suitable for watershed modeling and that have been used
in previous studies for both water quantity and quality predictions.

21

2.4.1 Annualized Agricultural Non-Point Source Model (AnnAGNPS)
The AnnAGNPS model is a single event, continuous version, watershed-scale
model designed to estimate the NPS pollution generation watersheds that are devoted
mainly to agricultural production (Bingner et al., 2007). AnnAGNPS was developed by
the United States Department of Agriculture (USDA) Agricultural Research Service
(ARS) and the USDA Natural Resources Conservation Service (NRCS) (Wang et al.,
2009). The model performs distributed modeling by dividing a watershed into
homogenous units in order to predict surface water loadings of pollutants such as
sediment, nitrogen, phosphorus, and agricultural chemicals. Each unit has relatively
uniform physical and hydrologic characteristics (i.e. soil type, landuse, and topography).
2.4.1.1 AnnAGNPS Components
AnnAGNPS uses the Soil Conservation Service (SCS) curve number method and
the Revised Universal Soil Loss Equation (RUSLE) to generate the daily runoff and daily
sheet and rill erosion from fields, respectively (Bingner et al., 2007). In-stream routing
simulates the transport of sediment downstream. The Hydro-geomorphic Universal Soil
Loss Equation (HUSLE) is used to provide a delivery ratio of sediment yield from
erosion to the reach (Wang et al., 2009). Channel runoff uses Manning’s equation and
lateral, subsurface flow is calculated using Darcy’s equation (Borah and Bera, 2003). A
modfified Einstein equation and the Bagnold equation are used to calculate the sediment
transport capabilities of the reach (Bingner et al., 2007). Precipitation can be in the form
of irrigation, snowmelt, or rainfall.
2.4.1.2 Previous AnnAGNPS Research
AnnAGNPS has been used by Parker et al. (2008), Yuan et al. (2007), Wang et
al. (2009) and Mankin et al. (2003) to evaluate the effects on water quality as a result of

22

agricultural practices, conservation practices, landuse change, and Best Management
Practice (BMP) effectiveness.
2.4.2 Hydrologic Simulation Program-FORTRAN (HSPF)
HSPF is a continuous, lumped-parameter watershed model that produces a time
history of water quality and quantity data for mixed agricultural and urban watersheds
(Borah and Bera, 2003; Holvoet et al., 2007). The model is commonly used to evaluate
the impacts on these processes due to landuse change, point and non-point source
pollution, and other physical phenomena and was developed by the Environmental
Protection Agency (EPA) National Exposure Research Laboratory (Bicknell et al., 1997).
An advantage of HSPF over other models is its hourly temporal scale capable of making
long-term predictions during periods of available continuous precipitation data. However,
due to its theoretical nature and the empirical equations used for model calculation, HSPF
requires a number of parameters, adding to the complexity of the underlying model
complexity.
2.4.2.1 HSPF Components
The HSPF model predicts both runoff and constituent transport processes on
pervious and impervious land areas and in-stream processes in the main channel as well
as mixed reservoirs (Borah and Bera, 2003). Runoff is calculated using the ChezyManning equation and subsurface flow is calculated using empirical relationships of
interflow, percolation, and groundwater outflow. Overland sediment is calculated using
rainfall splash detachment, in which the wash-off of detached sediment particles is based
on transport capacity of overland flow.

23

2.4.2.2 Previous HSPF Research
The HSPF model has been used by Diaz-Ramirez et al. (2005), Al-Abed and AlShariff (2008), Choi and Deal (2008), and Nasr et al. (2007) to evaluate landuse change
and management practices on watershed hydrological characteristics and nutrient
transport.
2.4.3 Soil and Water Assessment Tool (SWAT)
SWAT is a physically based model developed to estimate streamflow and stream
reach loadings of nutrients, pesticides, and sediments on a watershed-scale on a
continuous daily time-step in predominantly agricultural watersheds (Spruill et al., 2000).
The SWAT model was developed by the USDA-ARS. The SWAT model is used to
predict the long-term impact of management practices on constituent yields in complex
watersheds with varying soils, landuse, weather conditions, and management operations
(Arnold et al., 1998). The model bases its predictions on a complex structure of routing
runoff and constituents throughout a defined watershed and is one of the most
comprehensive models of its kind (Saleh and Du, 2004; Powers, 2007). The watershed
delineation tool in SWAT allows a watershed to be divided into subwatersheds based on
topography, which are further divided into HRUs (hydrologic response unit) based on
land use, soil type, and slope. By defining HRUs, the location of constituent load
origination can be identified based on site-specific (field-scale) characteristics. This is
useful for determining the best location for implementing conservation practices and for
identifying high-priority areas in need of further evaluation.
2.4.3.1 SWAT Components
The major components of the model include weather data, soils data, slope
classifications, landuse data, nutrient and pesticide applications, and management

24

practices such as tillage, fertilization, and planting/harvesting dates. The hydrologic
component of the SWAT model uses the modified SCS curve number method to
determine surface runoff during precipitation events (Saleh and Du, 2004). Soil erosion
and sediment yields for each HRU are determined by the Modified Universal Soil Loss
Equation (MUSLE) (Neitsch, 2005). Lateral subsurface and groundwater flow is
calculated based on a kinematic storage model and empirical equations. Channel routing
and flow are based on the variable storage coefficient method Manning’s equation,
respectively (Borah and Bera, 2003). Channel bed degradation and deposition, nitrogen
and phosphorus cycles, and pesticide fates are also simulated in SWAT.
2.4.3.2 Previous SWAT Research
A substantial number of studies have used SWAT to evaluate a wide variety of
environmental applications. As of 2007, 250 peer-reviewed published articles had been
identified that either used SWAT or reviewed SWAT components (Gassman et al.,
2007). For example, Kannan et al. (2006), Vazquez-Amabile et al. (2006), Du et al.
(2006), Ng et al. (2010), Ullrich and Volk (2009), Maski et al. (2008), Kamble et al.
(2005), Arnold et al. (1998), Saleh et al. (2000), Saleh and Du (2004), Galvan et al.
(2009), and Santhi et al. (2006), Luo et al. (2008), Tong and Naramngam (2007), and
many others, have used SWAT to evaluate landuse change, BMPs, agricultural practices,
spatial scaling effects, and bioenergy production on hydrology, sediment, nutrient, and
pesticide yields.
2.4.4 Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST)
InVEST is a model developed by the Natural Capital Project, among others, to
support environmental decision-making regarding the change in ecosystem services

25

provided by the environment due to landuse change (Daily et al., 2009). The InVEST
model is spatially explicit GIS interfaced program that is based on ecological production
functions and methods for assessing the economic value of services, conservation of
biodiversity, and value of commodities provided by the landscape, as reported in either
biophysical or monetary terms (Nelson et al., 2009). The model operates at different
levels of complexity, or tiers, to ensure model adaptability across a variety of applications
(Daily et al., 2009). However, the variation in data availability, data quality, and
understanding of the underlying model processes results in model uncertainty (Nelson et
al., 2009).
2.4.4.1 InVEST Components
The InVEST model relies heavily upon user-defined input parameters such as
threats to biodiversity, carbon pools, landcover harvest rates, and market prices of
commodities. Due to the subjectivity of this data, the model has an inherent uncertainty
associated with the results. However the model does require data layers that are readily
accessible and accepted such as landcover, soil depth, precipitation, and the digital
elevation model. The model operates on an annual average time step, although future
versions are expected to include daily to monthly time steps (Tallis et al., 2010).
2.4.4.2 Previous InVEST Research
The InVEST model has been used by Nelson et al. (2009) to predict changes in
ecosystem services based on stakeholder-defined scenarios of landuse change.

26

3. INTRODUCTION TO METHODOLOGY AND RESULTS
This thesis is in the form of four research papers that have been submitted to
scientific journals. The first paper entitled “The Effects of Postemergence Herbicides on
Biomass Accumulation and Bioenergy Quality of Field and Silage Corn (Zea mays)”
aims to determine the suitability of pesticide applications in the case that large-scale
implementation of lignocellulosic biomass production becomes widespread. This
research is the result of field studies conducted at the Michigan State University
Agronomy Farm in East Lansing, MI in 2008 and 2009 to evaluate the effects of
postemergence herbicide treatments on a) field and silage corn above-ground biomass
and b) cell wall digestibility (translated into ethanol yield). The herbicides that were
evaluated were 2,4-D amine, atrazine, bromoxynil, dicamba +diflufenzopyr, mesotrione,
and nicosulfuron with silage and field corn treated with the same application rates.
The second paper titled “Environmental Impact Analysis of Biofuel Crops
Expansion in the Saginaw River Watershed” aims to examine the long-term
environmental implications, specifically water quality, of bioenergy cropping system
expansion, to determine the environmental suitability of bioenergy crops, and to obtain
information to aid in the decision making process regarding landuse management in the
Saginaw River watershed. The Soil and Water Assessment Tool (SWAT) model was
used to model the effects of four landuse scenarios, 17 different bioenergy crops under 42
rotations on sediment load, total nitrogen load, and phosphorus load during a 19 year
(1990 to 2008) period on a daily basis. The SWAT model was calibrated and validated
for daily streamflow, sediment load, total Kjeldahl nitrogen load, and total phosphorus
load using observed water quality and climate data. Management operations, including

27

crop rotation schedules, fertilizer and pesticide applications, and tillage regimes were
applied to provide the most practical and accurate representation of actual agricultural
practices in the region. The research presented in this paper was performed at a watershed
scale and includes the implications of converting marginal land to bioenergy crops, the
importance of using local agricultural management practices in model setup, the temporal
variation of pollutant loads as compared to the base scenario (current landcover), and the
suitability of bioenergy crops grown on four landuse scenarios.
The third paper titled “Water Quality Impact Assessment of Large Scale Biofuel
Crops Expansion” aims to evaluate the environmental impacts of bioenergy cropping
systems on a larger scale with extensive spatial variability of landuse and climatic,
physiographic, and geographic conditions. The same procedure was followed as the
previous paper, with the only difference being a monthly time step and a more expansive
study area. The Soil and Water Assessment Tool (SWAT) model was again used to
model the effects of four landuse scenarios and 17 different bioenergy crops have on
sediment load, total nitrogen load, and phosphorus load during a 19 year (1990 to 2008)
period on a monthly basis. The SWAT model was calibrated and validated for monthly
streamflow, sediment load, nitrogen load (various nitrogen formulations depending on
watershed), and total phosphorus load using observed water quality and climate data. The
research presented in this paper assists in land management decision making at a largescale by showing the spatial variability of pollutant loads for each bioenergy rotation, by
identifying high priority areas for the implementation of conservation practices to reduce
sediment, nitrogen, and phosphorus contribution to surface waters, determining the

28

suitability of bioenergy crops on various landuse scenarios, and determining the statistical
significance of the differences of each rotation from the base scenario.
The fourth and final paper titled “Effects on Aquatic and Human Health due to
Large Scale Bioenergy Crop Expansion” aims to determine the impacts of agricultural
practices (specifically herbicide application in bioenergy cropping systems) on aquatic
ecotoxicity and human health. The Soil and Water Assessment Tool (SWAT) model was
used to predict pesticide concentration in each stream reach during a six year (20002005) study period on a daily basis. The herbicides atrazine, bromoxynil, glyphosate,
metolachlor, pendimethalin, sethoxydim, triflualin, and 2,4-D amine were incorporated
into the bioenergy crop rotation management practices in SWAT. Threshold toxicity
levels were obtained for the bluegill and for human consumption for all pesticides being
evaluated through an extensive literature review. The resulting data collection was used
to demonstrate the enormous amount of basin-wide pesticide concentration variability
and determine the length of stream impairments due to the application of each herbicide.
With this data, policy makers and land managers will be able to identify areas where
pesticide application will have negative impacts, and therefore, where conservation
efforts should be focused.

29

4. THE EFFECTS OF POSTEMERGENCE HERBICIDES ON BIOMASS
ACCUMULATION AND BIOENERGY QUALITY OF FIELD AND SILAGE
CORN (ZEA MAYS)
a
b,
b
Brad J. Love , Wesley J. Everman *, Calvin F. Glaspie , A. Pouyan

Nejadhashemi
a

a

, Nicholas

Santoro

c

Department of Biosystems and Agricultural Engineering, Michigan State University,
216 Farrall Hall, East Lansing, MI 48824 USA

b

Department of Crop and Soil Science, Michigan State University, 464 Plant and Soil
Sciences, East Lansing, MI 48824 USA
c
Department of Energy, Great Lakes Bioenergy Research Center, Michigan State
University, 162 Food and Safety Toxicology Building, East Lansing, MI 48824 USA

* Author who correspondence should be addressed. Tel.: (517) 355-0271 Ext. 1223.
Email address: everman@msu.edu (Wesley J. Everman)
Abstract. Energy derived from renewable, biological sources is crucial in mitigating
negative environmental impacts and ensuring future energy needs. However, despite
innumerable advancements in bioenergy research, an improved understanding of the
agronomic effects of herbicide treatments on both corn development and ethanol
feedstock quality is necessary for the development of successful full-scale feedstock
production. Field studies were conducted at the Michigan State University Agronomy
Farm in East Lansing, MI in 2008 and 2009 to evaluate the effects of postemergence
herbicide treatments on a) grain and silage corn above-ground biomass and b) cell wall
digestibility (translated into ethanol yield). Biomass samples were analyzed for glucose
release by enzymatic cell wall digestion and potential cellulosic ethanol production was
calculated. Injury was observed two weeks after treatment with 2,4-D and bromoxynil in
-1
2008, this resulted in a 1.8 t ha decrease of grain corn biomass for bromoxynil

30

treatments only. In contrast, herbicide treatments had no significant effects on the
biomass accumulation of silage corn. Bromoxynil, when sprayed on silage corn, yielded
greater biomass than the non-treated. The greatest biomass and glucose release by
-1
enzymatic cell wall digestion was observed for the 2008 silage corn at 20 t ha and 17.6

percent glucose release, respectively. Cool weather resulted in poor growing conditions
for the 2009 season. The 187 fewer growing degree days in turn exaggerated the effects
of herbicide treatment on plant growth and glucose percentages. Silage corn produced
significantly greater biomass, glucose release by enzymatic cell wall digestion and
ethanol yield than the field variety with the same nutrient inputs. The results suggest that
the selection of both corn type and herbicide treatment significantly influences the
biomass accumulation as well as the bioenergy feedstock quality.

4.1 INTRODUCTION
Increasing demand for renewable, carbon-neutral energy sources to replace
current fossil fuels has led to an increase in the production of fuels derived from
biological sources (Pimentel et al., 2009). Biofuels in the United States are currently
produced primarily from agricultural feedstocks, specifically from starch-rich corn grain
(Mol, 2007). Although corn grain provides a basis for transition from fossil fuels to
renewable sources, the current fermentation process excludes much of the sugars located
in structural fibers, known as lignocellulose (Himmel et al., 2007). Cellulosic biomass
can potentially provide a sustainable source of energy, liquid transportation fuels, and
biomaterials, while minimizing competition for food production (Himmel et al., 2007;
Antizar-Ladislao and Turrion-Gomez, 2008). As the biofuel industry evolves, the

31

scientific knowledge base supporting agricultural feedstock production must follow suit.
However, many aspects of large-scale bioenergy crop production are adapted from
current agronomic practices emphasized on improved grain yields, not cellulosic ethanol
yield (Carroll and Sommerville, 2009). In order to allow growers to maximize
productivity and further support the need for biofuel production, information regarding
the agronomic aspects of bioenergy crop production must be more available, especially
relating to herbicides (McLaughlin and Walsh, 1998). It has been shown that herbicides
have the potential to affect secondary metabolism and other physiological processes,
resulting in altered plant growth, especially during early growth stages (Alla and Younis,
1995; Vencill et al., 1989). In addition, reports have documented injury resulting from
herbicide applications of bromoxynil, nicosulfuron, and dicamba on corn, which often
leads to only minor grain yield loss (Kunkel et al., 1996; Krausz et al., 1999; Green and
Ulrich, 1993). A number of studies have been conducted to evaluate the affect herbicides
have on crop yield, based on their effectiveness at suppressing weeds that compete for
moisture, sunlight, and nutrients (Vanheemst, 1985; Harvey and Wagner, 1994; Lybecker
et al., 1991; Askew and Wilcut, 1999; Hall et al., 1992). However, after an extensive
literature review, the authors are unaware of any studies that examine the interaction
between herbicides, corn biomass and ethanol yield, based solely on the chemical’s
influence on cellular and physiological behavior that control plant growth factors.
Therefore a study was conducted to evaluate the effect postemergence (POST) herbicides
labeled for use in corn have on biomass yield and potential cellulosic ethanol yield. Grain
and silage corn hybrids were harvested and dried in the same manner as forage maize,
mimicking conditions for treating feedstock that would be used in large-scale cellulosic

32

ethanol production. The results presented in this study are compared over a two year
period and are derived from measurements of above-ground biomass, glucose release by
enzymatic cell wall digestion, and theoretical ethanol yield.

4.2 MATERIALS AND METHODS
Experiments were conducted in 2008 and 2009 at two separate field locations at
the Michigan State University Agronomy Farm in East Lansing, MI on a sandy-loam
(fine-loamy, mixed, mesic, Aeric Ochraqualfs) soil with 2.3 percent soil organic matter
(SOM) and pH of 7.1 and 3.3 percent SOM and pH of 6.2 in 2008 and 2009, respectively.
Tillage at the site included fall chisel-plowing and spring field cultivation. Two corn
hybrids were examined; Dekalb® DKC 46-60 (grain corn hybrid) and NK® N49E3
-1
(silage corn hybrid) were planted at seed populations of 79,074 and 76,603 seeds ha ,

respectively, on May 15, 2008 and May 12, 2009. The entire study received soil-applied
S-metolachlor {2-chloro-N-(2-ethyl-6-methylphenyl)-N-[(1S)-2-methoxy-1-methylethyl]
acetamide} at 1400 g ai ha-1 and atrazine [6-chloro-N-ethyl-N'-(1-methylethyl)-1,3,5-1
triazine-2,4-diamine] at 1400 g ai ha prior to planting to reduce weed pressure and aid

in hand-weeding. Plots were maintained weed-free by hand hoeing throughout the
growing season. Each plot (3.05 m x 10.67 m) was planted with four rows of seed spaced
76 cm apart. The experimental design was a split-plot, randomized complete block with
four replications of each treatment. Main plot factors were corn type of either grain or
silage corn with sub-plot factors consisting of seven herbicides (Table 4). Herbicide

33

treatments were applied at the V4 growth stage with a tractor-mounted sprayer calibrated
to deliver 187 L/ha at 207 kPa using TeeJet® 8003 flat-fan nozzles.
Crop injury was assessed visually 2 weeks after treatment (WAT). Growth stage,
corn height, and leaf number were recorded from four representative plants within the
two innermost corn rows every two weeks following application until silking. A selfpropelled harvester was used to harvest and measure the biomass from the inner two rows
of each plot in a manner likely to be used for large-scale bioenergy feedstock production.
Forage samples were collected from each plot, weighed, and oven-dried to determine the
total dry matter. Samples from each plot were then ground (1-2 mm particle size) and
analyzed for percent dry matter, forage quality, and percent glucose.
Glucose release by enzymatic digestion of cell walls for ethanol production was
determined using the method described by the Department of Energy Great Lakes
Bioenergy Research Center (GLBRC) at Michigan State University (Santoro et al. 2010).
Microtubes were loaded with 2 mg of thoroughly mixed, dried and ground plant material
for each sample of corn biomass. Samples were analyzed in triplicate with biological
replicates included (Santoro et al. 2010). A dilute NaOH (6.25 mM) pretreatment
o
solution was added to each tube and incubated at 90 C for 3 hours. After cooling, 50mM

citrate buffer (pH 4.5) and 0.01 percent sodium azide were added to the solutions.
Digestions were initiated using Accellerase 1000 (a gift from Genencor) at a
-1

concentration of 30 mg protein g glucan . Samples were incubated at 50 C for 20 hours.
The microtubes were centrifuged at 1,500 x g for 3 minutes, separating the solid residue
from the digested biomass. Glucose was determined by the glucose oxidase-peroxidase
method as described (Santoro et al. 2010). In this paper “glucose release” will refer to
34

the amount of glucose released from cell walls resulting from the enzymatic action of
Accellerase 1000. The amount of glucose released is expressed as a percent; w is derived
from the mass of glucose detected per total starting dry weight of plant biomass.
All data were subjected to analysis of variance using the MIXED procedure in the
SAS Statistical Software Package (SAS, 2009).

4.3 RESULTS AND DISCUSSION
4.3.1 Crop Injury
Bromoxynil and 2,4-D showed significant injury in early growth stages two
weeks after treatment. Injury was characterized by twisting and leaning of the stem, fused
brace roots, and poor development of brace roots in plots treated with 2,4-D and necrotic
tissue in plots treated with bromoxynil. However, no late season corn injury was
observed with any of the POST treatments in either year. Significant biomass yield
reduction was not observed in silage corn as a result of this injury. However, biomass of
grain corn was reduced when treated with bromoxynil, which may be a result of the
injury observed in early growth stages.
4.3.2 Variation by Corn Type
A year by corn type interaction was observed for corn maturity (measured in vstage) (Table 1). Silage corn in 2009 was less mature than grain corn in 2008 and 2009
and silage corn in 2008. This is likely due to the N49E3 silage corn variety being longer
maturity (106 day) than the DKC 46-60 grain corn variety (96 day), and to adverse
growing conditions in 2009. In addition, a year by corn type interaction was observed for
biomass yield. Both silage and grain corn yielded significantly less biomass in 2009,
likely due to the 187 fewer growing degree days relative to 2008 (Table 2), resulting in

35

delayed maturity for both hybrids. Silage corn in 2008 had the greatest biomass, and
biomass within varieties was greater in 2008 than in 2009 (Table 1). Percent glucose
released and ethanol yield values followed similar trends as observed for maturity; both
values were greater for silage corn in 2008 than in 2009 and also both values were greater
for silage corn than grain corn in the two years. The 2009 growing season accumulated
fewer heat units, leading to a decrease in vegetative growth and poorly developed plant
tissue prior to physiological maturity (Nielsen and Hinkle, 1996). As a result, the 2008
growing season provided sufficient heat units for the silage variety to maximize its yield
potential and digestibility for ethanol fermentation. Silage corn, generally used for
ruminant dairy feed, is assessed for forage quality using multiple techniques, one of
which being neutral detergent fiber digestibility (NDFD). NDFD relates the convertibility
of cell wall structures in plant tissue to ethanol and may be an efficient method for
determining bioenergy feedstock quality (Lorenz et al., 2009). Silage corn hybrids
generally accumulate more biomass than grain corn hybrids (Figure 1). Silage corn
surpasses the vegetative growth of grain corn during early growth stages in typical
production years like 2008; however, in 2009 no biomass advantage was observed.

36

Table 1. Corn maturity 28 days after postemergence treatment, dry biomass yield, percent
glucose, and ethanol yield in 2008 and 2009 by corn type averaged by herbicide
treatment.
b
Corn Type
Year
Maturity
Yield
Glucose
Ethanol Yield
-1
-1
%
t ha
L ha
Grain
2008
9.0
a
17
b
15.3 b
1599
b
Silage
2008
9.0
a
20
a
17.6 a
2160
a
Grain
2009
8.6
a
15.6 c
15.3 b
1467
b
Silage
2009
7.9 b
15.9 bc
15.5 b
1507
b
LSD
0.5
1.1
1
170
a

Means within columns followed by the same letter are not significantly different p ≤
0.05.
b
Represents 95 percent of the theoretical ethanol yield (0.647 mL ethanol g glucose-1)
(Thomas et al., 1996).
Table 2. Growing degree days (GDD) accumulated in East Lansing, MI from 4/1-10/31.
Year
2008
2009
Normal
a

GDD
2467
2654
2726

Source: (MSUEP, 2010)

37

Figure 1. Corn height 28, 42, and 56 days after treatment with error bars representing one
standard deviation.
Table 3. Regression values for Figure 1.
Intercept
Grain Corn 2008
Silage Corn 2008
Grain Corn 2009
Silage Corn 2009

-0.174
-0.737
0.066
-0.454

2
r
0.990
0.979
0.822
0.898

Slope
0.051
0.069
0.030
0.043

4.3.3 Variation by Herbicide Treatment
An herbicide by corn type interaction was observed for grain and silage corn
biomass yields as affected by herbicide treatment. No year effect was observed; therefore
data are over averaged years (Table 4).

In addition, bromoxynil and mesotrione

treatments on silage corn resulted in the greatest yields. Atrazine and dicamba +
diflufenzopyr treated silage corn produced significantly less yield than the bromoxynil
and mesotrione treated silage corn. However the yield of Atrazine and dicamba +
38

diflufenzopyr treated silage corn was not significantly different than the remaining silage
corn treatments. Silage corn had a greater biomass yield overall than grain corn
regardless of herbicide application. The treatments in grain corn that produced the
greatest biomass were the non-treated, atrazine, and mesotrione. The greatest biomass
reduction was observed when bromoxynil was applied to grain corn, yielding less than
the non-treated grain corn, this is the direct opposite of what was observed for silage
corn. Bromoxynil treatments in silage corn resulted in the largest yield increase. It is
suggested that many herbicides can actually alter the activity of enzymes that control the
secondary metabolism, possibly explaining the increase in yield compared to the nontreated corn (Alla and Younis, 1995; Cottingham and Hatzios, 1992)). As previously
discussed Table 1 shows a year by corn type interaction in ethanol yield. Although there
appeared to be slight interactions with herbicide treatments, the differences are not
statistically significant. This is likely due to the noise in the collected data. However, the
interaction between corn type and herbicide treatment is compelling to support careful
selection of herbicide treatment for corn to be grown for bioenergy feedstock. It is
important to note that this study was conducted on only one variety for each corn type;
further research on the differential response to herbicides by hybrid variety may provide
additional useful information for future bioenergy feedstock production. Many studies
have evaluated the tolerance of various corn hybrids to herbicide treatments and observed
a general large span of differing tolerances (Green, 1998; Rowe et al., 1990). Although
those comparisons were not performed in this study, the gathered data and evidence of
possible herbicidal effects on biomass yield and glucose release from lignocellulose
suggest this information is vital for producers considering an herbicide regime for corn

39

production when optimizing total above-ground biomass while maintaining essential
weed control.

40

Table 4. Dry biomass yield by herbicide treatment and corn type averaged over years.
b
a
Corn
Biomass
Rate
Herbicide
c
Type
Yield
-1
g ai (ae) ha

-1
t ha

Grain
2,4-D amine
atrazine + COC
bromoxynil
dicamba + diflufenzopyr + NIS
+ AMS
mesotrione + COC + AMS
nicosulfuron + COC
non-treated

530
1120 + 1.25 % v/v
420
200 + 0.25 % v/v +
1070
110 + 1 % v/v + 540
30 + 1 % v/v

15.8
16.9
14.9
16.1

2,4-D amine
atrazine + COC
bromoxynil
dicamba + diflufenzopyr + NIS
+ AMS
mesotrione + COC + AMS
nicosulfuron + COC
non-treated

530
1120 + 1.25 % v/v
420
200 + 0.25 % v/v +
1070
110 + 1 % v/v + 540
30 + 1 % v/v

17.5
17.1
18.9
17.2

de
bcd
e
cde

17.1 bcd
16.5 cde
16.7 bcd

Silage

b
c

19.0 a
18.2 ab
17.4 abc
1.6

LSD
a

abc
bcd
a
bcd

Abbreviations: AMS, ammonium sulfate; COC, crop oil concentrate; NIS, non-ionic surfactant.
-1
Treatment rates for dicamba are expressed in g ae ha .
Means within columns followed by the same letter are not significantly different p ≤ 0.05.

4.4 CONCLUSION
The goal of the research done in this study was to determine the impact that
postemergence herbicide treatments applied to grain and silage corn varieties has on
aboveground biomass accumulation, glucose release from lignocellulose, and ethanol
yield. Cellulosic biomass has the potential to be a viable feedstock for liquid fuel
production, namely ethanol. However, it is essential to investigate any negative

41

consequences the use of herbicides may have on the biomass yield of corn stover for
commercial feedstock production practices. According to observations and data acquired
in this study, silage corn hybrids yield more biomass and release greater amounts of
glucose per dry weight (resulting in greater ethanol yield) when favorable growing
conditions are present. In addition, although visible physiological plant injury was
observed in early growth stages, herbicide treatment resulted in increased biomass yield
and glucose release compared to non-treated silage corn. Although there was a significant
corn type by year interaction, there were no statistically significant interactions with
herbicide treatments. In summary, we conclude that postemergence herbicides can have
potentially beneficial effects when applied to corn that will be used as a feedstock for
cellulosic ethanol production.

Acknowledgments
The authors would like to thank Andrew Chomas, Keith Dysinger, Jacob
Gebhardt, Bryce Conklin, and Dan Tratt for their assistance.

42

5. ENVIRONMENTAL IMPACT ANALYSIS OF BIOFUEL CROPS EXPANSION
IN THE SAGINAW RIVER WATERSHED
Brad Love, A. Pouyan Nejadhashemi
Department of Biosystems and Agricultural Engineering,
Michigan State University, East Lansing, Michigan 48824, USA
2
Abstract. The Saginaw River watershed (15,262 km ) located in the east central portion

of Michigan’s Lower Peninsula is the State’s largest and is largely used for agricultural
production. Non-point source pollutant loadings in the form of sediment, nitrogen, and
phosphorus are a major issue in this watershed and have led to 22 303(d) listed
waterbodies and multiple Total Maximum Daily Loads (TMDL). The environmental
implications of expanding agricultural land into bioenergy crop production are not wellknown. This study implements the Soil and Water Assessment Tool (SWAT) to predict
the effects landuse conversion will have at a large watershed-scale. The model generally
performed satisfactory on a daily basis during the calibration and validation periods for
flow (2001-2008) and sediment and nutrients (2000-2005). Management operations,
including crop rotation schedules, fertilizer and pesticide applications, and tillage regimes
were applied to provide the most practical and accurate representation of actual
agricultural practices in the region. Four scenarios including 15 crop rotations were
developed to reflect current and future trends of producing first and second generation
bioenergy crops, respectively. The results from SWAT showed that environmental effects
varied greatly between crops based on the landscape being converted. Continuous corn
with stover removal, when implemented on all land that can be tilled, has the potential to
increase total nitrogen, total phosphorus and sediment loads by up to 1375, 792, and 493

43

percent, respectively. However, when applied on land classified as “other” crops, the
same rotation experiences reductions in both sediment and phosphorus. Perennial grasses
(miscanthus, switchgrass, native grasses) experience similar behavior of varied pollutant
loading dependent on landscape.

5.1 INTRODUCTION
Increasing global and domestic energy demands have led to an enormous push for
a transition to alternative sources. Despite vast increases in traditional energy technology
(i.e. petroleum based fuels, nuclear, solar), a viable solution to reliance on fossil fuels has
yet to accommodate growing demand. Fossil fuels are available in a finite amount,
accounting for about 80 percent of global energy supply, and will be exhausted in a
matter of decades at current consumption rates (Goldemberg, 2007). Consequently,
renewable fuel sources originating in the agricultural sector have become a very
important means for the replacement, or at least partial substitution, of fossil fuels.
Reducing fossil fuel dependency potentially cuts consumer costs, mitigates climate
change, creates new jobs, promotes the supply of domestic energy, and diversifies energy
sources. By the year 2025, global energy consumption is expected to increase by over 57
percent from 2002 values (Rooney et al., 2007). Meanwhile, renewable energy accounts
for only 7 percent of the nation’s 99.3 Quadrillion Btu consumption (USEIA, 2009).
Biofuels are liquid petroleum replacement fuels, such as ethanol and biodiesel,
derived from biomass conversion. Biomass provided 53 percent of the nation’s renewable
energy in 2008 and is the sole sustainable solution for fuel, chemicals, and materials
generally produced from petroleum (USEIA, 2009; Lynd et al., 2009). These “biofuels”

44

are particular appealing to environmentalists and policymakers for their promise as a
solution to foreign oil dependency, increasing greenhouse gases, and rising energy costs
(Groom et al., 2008). However, evidence of negative consequences are prevalent, as for
example, greenhouse gas emissions from landuse change towards corn-based ethanol are
estimated to nearly double (Searchinger et al., 2008). The National Renewable Fuel
Standard program (RFS) was created as a provision of the United States Energy Policy
Act of 2005 (USEPA, 2009a). Revisions to this program in the 2010 RFS have increased
the requirements for renewable fuel volume to 15.2 and 30 billion gallons by 2012 and
2020, respectively (USEPA, 2009a). In order to achieve these milestones, the agricultural
production of the nation is going to have to expand and policy must support this sector.
Unfortunately, the implications of bioenergy cropping expansion are not adequately
understood or accepted especially at large scales.
The aspects of biofuel production, whether positive or negative, are an extremely
necessary consideration in the sustainability of such a practice. Biofuel crop production
shows promise in reducing the United States’ dependency on fossil fuels, adding political
and economic stability to the nation, and providing reliable energy sources for future
generations (de Friature et al., 2008). However, the long-term environmental
repercussions resulting from a shift towards bioenergy cropping systems are of imperious
concern. Carbon emissions contributing to global warming through greenhouse gas
accumulation are of key significance for the success of biofuels. Energy from bioenergy
crops are largely carbon neutral, especially when compared relative to fossil fuels, as the
carbon emitted for production is regained during the growth of the replacement crop
(Rowe et al., 2009; Gillingham et al., 2008). However, when landuse change is taken into

45

consideration, many of the benefits associated with biofuels are uncertain, and in some
cases offset by the negative environmental implications (McLaughlin and Walsh, 1998).
It is possible that the conversion of agricultural land for production of bioenergy
feedstock will possibly threaten environmental quality, specifically water quality
(Nyakatawa et al., 2006). Therefore, in order for biofuels to be a sustainable solution and
to prevent degradation to the natural biodiversity, their existence must avoid intensifying
agricultural impacts and the displacement of native habitats or cropland necessary for
food production (Groom et al., 2008). Ecosystem services such as biodiversity, carbon
sequestration, crop pollination, erosion control, water quality enhancements, flood
mitigation, and food production may experience loss (Groom et al., 2008).
Effectively addressing the impacts of bioenergy cropping systems should include
water quality and water quantity aspects.
a) Water quality: Increasing the acreage under cultivation to produce bioenergy crops,
many of which are input-intensive, may produce negative water quality implications.
Agriculture is the largest contributor to excess nutrients, mainly nitrogen and phosphorus,
in freshwater in the United States (Foley et al., 2005). The current biofuel crop of choice,
corn, is input-intensive, requiring more nitrogen fertilizer, herbicide, and insecticide than
any other crop in the United States (Pimentel et al., 2005). Nutrient loading of nitrogen
and phosphorus in surface waters results in algal blooms, reduced dissolved oxygen
levels, loss of biodiversity, and impaired aquatic habitat and reproduction (Carpenter et
al., 1998). During precipitation events, phosphorus binds to the soil particles of sediment
carried by overland flow. These pollutants along with nitrogen are responsible for
eutrophication, which is a main cause of impaired surface waters, and drinking water

46

contamination (Nyakatawa et al., 2006). The results are negative effects on human safety
and aquatic ecosystems such as death of coral reefs, fish kills, reductions in harvestable
fish, and reduced aesthetic value of the water body (Carpenter et al., 1998).
b) Water quantity: With freshwater necessary for human life, the quality of ground and
surface water is imperative for survival. However, the quantity of surface waters is an
extremely essential topic as well. Agriculture is already responsible for 70 percent of the
world’s freshwater withdrawals (Stone et al., 2010). Irrigation of agricultural land
accounts for 65 percent of total freshwater withdrawals in the U.S. and has increased 60
percent since 1960 (Hutson et al., 2004; Berndes, 2002). A significant amount of water is
required for the refining portion of ethanol production (4-6 gallons water per gallon
ethanol) as well as for plant growth, especially under irrigation (de Fraiture et al., 2008).
This makes it difficult to warrant the expansion and intensification of agricultural
production when water use is highly stressed to a point passed sustainable, especially
when an increase in irrigation withdrawals by 20 percent are required to meet future
global food demand by 2050 (de Fraiture et al., 2008; Berndes, 2002). However, crops
such as switchgrass, a C4 perennial crop, have been shown to produce substantial yields
without irrigation; and they have a positive effect on water quality and quantity and soil
erosion (McLaughlin and Walsh, 1998).
The objectives of this study were to examine the long-term environmental
implications, specifically water quality, of bioenergy cropping system expansion,
determine the environmental suitability of bioenergy crops, and obtain information to aid
in the decision making process regarding landuse management. The Soil and Water
Assessment Tool (SWAT) was used to model the effects of four scenarios, 17 different

47

bioenergy crops under 42 rotations on sediment load, total nitrogen load, and phosphorus
load in the Saginaw River Watershed during a 19 year (1990 to 2008) period on a daily
basis. This watershed, which drains into the Saginaw Bay and subsequently Lake Huron,
is intensively cultivated with crops.

5.2 MATERIALS AND METHODS
5.2.1 Study Area
The Saginaw River Watershed (hydrologic unit code 040802) is located in the
east central portion of Michigan’s Lower Peninsula, draining into the Saginaw Bay of
2
Lake Huron. The watershed is the largest in Michigan (22,556 km ) and is home to the

nation’s largest contiguous freshwater coastal wetland system (USEPA, 2009b). This
study focuses on the six watersheds located within HUC 040802 (Figure 2), including the
Tittabawassee (04080201), Pine (04080202), Shiawassee (04080203), Flint (04080204),
Cass (04080205), and the Saginaw (04080206) watersheds. Consequently, the total
2
watershed area in this study is about 15,262 km (1,526,200 hectares) with approximately

44.6 percent being used for agricultural purposes, 15 percent developed, 25 percent
forest, 14 percent wetlands, and the rest water. The long-term average annual
precipitation that was obtained from the National Climatic Data Center (NCDC)
precipitation stations selected (Figure 3) is 82.1 centimeters (32.32 inches). The
minimum elevation (above the sea level) in the watershed is 177 meters whereas the
maximum elevation is 472 meters, resulting in a 245 meter mean elevation.

48

Figure 2. Location of the Saginaw River Watershed. For interpretation of the references to
color in this and all other figures, the reader is referred to the electronic version of this thesis.
5.2.2 General Water Quality Issue
Non-point source pollution is a major contributor to the deteriorated water quality
conditions of the region. The watershed has many threats to water quality, and due to
contaminated sediments, degraded fisheries, fish consumption advisories, and impairment
to recreational activities, the watershed has been designated as one of the 26 Great Lakes
Areas of Concerns (AOCs) listed in the United States (USEPA, 2009c). In order for a
waterbody or watershed to qualify as an AOC, it must have at least one beneficial use
impairments, such as beach closings or degradation of benthic habitat (GLIN, 2009). The
main contributors to these impairments are excessive nutrient loading, soil erosion, and
contaminated sediments. There are currently 22 records on the Michigan Section 303(d)

49

listing, and a number of Total Maximum Daily Loads (TMDLs) have been developed for
the Saginaw River Watershed, including sediment (Pine), total suspended solids (Cass),
E.Coli (Flint, Cass, Tittabawassee and Shiawassee) and pathogens (Flint) (USEPA,
2010). The 303(d) listed waterbodies are shown in Figure 4 along with the United States
Geological Survey (USGS) flow gauging station 04157000 and Michigan Department of
Environmental Quality (MDEQ) water quality gauging station 090177.
5.2.3 SWAT Model
The Soil and Water Assessment Tool (SWAT) is a physically based model
developed to estimate flow generation and nutrient, pesticide, and sediment loadings on a
watershed scale. These estimations are based on landuse practices in complex watersheds
with varying soils, landuse, weather conditions, and management operations over long
time periods (Arnold et al., 1998). ArcSWAT, the ArcGIS extension of the SWAT 2005
program, is a graphical user interface for the SWAT model used in this study. The major
components of the model include weather data, soils data, slope classifications, landuse
data, nutrient and pesticide applications, and management practices. The SWAT model
uses the modified SCS curve number method to determine surface runoff during
precipitation events (Neitsch et al., 2005). Soil erosion and sediment yields are
determined by the Modified Universal Soil Loss Equation (MUSLE). If properly
calibrated, the SWAT model can efficiently predict the impacts of future land use
conditions on flow and water quality. The accuracy of this model depends on complete
understanding of the regional processes and of internal model processes. Therefore,
setting up the model requires considerable time commitment for both data collection and
model familiarization (Gassman et al., 2007).

50

Figure 3. Saginaw Weather Stations.

51

Figure 4. Gauging Stations and 303(d) listed waterbodies.

5.2.4 Crop Rotations and Management Operations
5.2.4.1 Crops
The main focus of this paper is to examine the effects of landuse change into
bioenergy crops, therefore the crops had to be identified for their suitability as a potential
use in the region for biofuel feedstock. The bioenergy crops considered for conversion
from the current landuse were barely, canola, cassava, corn, corn stover, jatropha,

52

miscanthus, native grasses, potato, rice, rye, sorghum, soybean, sugar cane, sugarbeet,
switchgrass, and wheat. However, after determining the suitability for growth in the
Saginaw Watershed region, the amount of inputs, management practices, climate, and
cost of conversion from typical farming practices to the proposed crops, the crops
considered were reduced to corn, canola (fall), cereal rye, sorghum, soybean, miscanthus,
corn stover (40 percent residue removal), switchgrass, and native grasses.
5.2.4.2 Rotations and Operation Schedules
Crop rotation is a historical cultural practice that effectively controls noxious
weeds, pests and disease, increases soil fertility and healthy physical structure, enhances
root-zone biological activity, and increases productivity (NCSU, 2005). In this study,
possible crop rotations that would be likely implemented were identified. The rotations
were assembled using data obtained from the Michigan State University Extension
(MSUE) for the bioenergy crops used in the simulations. Based on this information, 15
rotations were developed, including corn-soybean-canola (Crn-Soy-Can), corn-soybeanrye (Crn-Soy-Rye), corn-soybean (Crn-Soy), Corn Stover (40 percent residue removal)soybean (CrnS-Soy), sorghum-soybean (Srg-Soy), continuous canola (CCan), continuous
corn (CCrn), continuous corn with stover removal (CCrnS), continuous rye (CRye),
continuous sorghum (CSrg), continuous soybean (CSoy), miscanthus (Mis), Switchgrass
(Crn-Soy-Swg), and native grasses (NaGr). A summary of the rotations and their
associated fertilizer application rates are provided in Table 12. All rotations repeat after
the sixth year except miscanthus, switchgrass and native grasses. The miscanthus and
native grasses repeat for the entire study period and the switchgrass rotation continues as
switchgrass until year ten when the cycle is repeated. Typically, SWAT applies its default

53

values for crop inputs unless otherwise instructed. Thus, in order to accurately simulate
vegetation cover and biomass, cropping systems, and management practices consistent
with local standards, management operations were introduced to the model. The
operations were based on data obtained from the MSUE and typical farm practice in the
Saginaw Watershed region. For example, the corn-soybean-canola (Crn-Soy-Can)
operation schedule is provided in Table 13. Each rotation followed the same format with
tillage practices, fertilizer and pesticide applications, and planting and harvesting dates
developed specifically for each crop.
5.2.4.3 Scenarios
The crop rotations were integrated into the SWAT model and tested under four
scenarios. The scenarios were developed for the bioenergy crops, based on suitability and
practicality. The scenarios are provided in Figure 15.
•

Scenario 1: The first scenario considers all land currently cultivated for row crops
and grains, hays, and seeds (i.e. corn, soybeans, wheat, hay, etc.), which
represents the majority of current agricultural land in the watershed.

•

Scenario 2: The second scenario considers only the other crops (i.e. sugarbeets,
potatoes, dry beans, fruit crops, etc.) for conversion to bioenergy crops.

•

Scenario 3: The third scenario considers farmland uses and other land (i.e. fallow
cropland, pasture, wasteland, etc.), which can be considered marginal land.

•

Scenario 4: The final scenario is the combination of the previous scenarios, and
represents the overall most significant landuse change from current state to the
bioenergy cropping rotations.

54

The scenarios range in amount of landuse change from very little (1.42 percent of
watershed area) to the worst case scenario with 63 percent of the watershed area being
converted to the bioenergy rotation. The landuse sections (i.e. Row Crops, Farmland
Uses) are based on definitions of landuse classification (CDL). Woodland was not
considered for landuse change to bioenergy crops. In addition, the rotations that were
acceptable for producing bioenergy crops on the land chosen for each scenario are based
on the current landuse, typical production value for the land into bioenergy crops, the
feasibility of the expected yield, and the difficulty for farmers to acquire the planting and
harvesting equipment and the capital required to invest in the operations not already
included the farm’s practices. For example, marginal land has generally low levels of soil
organic matter, poor soil structure and fertility, and tends to produce low yields for
traditional crops. Therefore, crops such as wheat and soybeans are not practical for
cultivation. Corn stover was not selected as suitable for this landuse due to the already
low levels of soil organic matter in marginal soils, which cannot afford to have additional
residue removed post-harvest. Each scenario went through the same steps in order to
provide the most realistic rotations that can be grown on each landuse.
5.2.5 Model Setup
5.2.5.1 Input Data Sources
Successful use of the ArcSWAT model relies on acquiring suitable and reliable
data The State Soil Geographic Database (STATSGO) developed by the Natural
Resources Conservation Service (NRCS) National Cartography and Geospatial Center
(NCGC) was used to define the soil characteristics in the basin. The slope classification
was performed using two slope classes (0-5 percent and >5 percent). The topography was

55

based on the USGS Digital Elevation Model (DEM 90 meter resolution), which was used
in delineation of the watershed boundaries. Subsequently, the watershed was divided into
254 subwatersheds, which allowed the model user to spatially characterize areas of the
watershed regarding spatial impacts of land use change.
The watershed was further divided into individual hydrologic response units
(HRUs) based on threshold land percentage values in the SWAT HRU definition. The
threshold values were the percentage of landuse over the subbasin area, soil class
percentage over the landuse area, and slope class percentage over the landuse area. For
the purpose of this project, these thresholds were set to 20, 10 and 20 percent,
respectively.
The USGS National Hydrography Dataset (NHD) was used to define the stream
network. The latest Cropland Data Layer (2008 CDL 56 meter resolution) developed by
the USDA National Agricultural Statistics Service (NASS) was used as the landuse
representation. This is useful for this study having the agricultural land split into
individual crop classes. The landuse values in the CDL were converted to the closest
alternative landuse definition in SWAT, resulting in an input table redefining the CDL
table grid values in SWAT. As discussed previously, the precipitation and temperature
data were collected from the NCDC database at the locations described in Figure 4. Other
meteorological data (wind speed, relative humidity, and solar radiation) required by the
model was estimated using the SWAT weather generator.
5.2.5.2 Operation of SWAT Model
The period of study for analysis was January 1, 1990 to December 31, 2008. For
this time period, daily flow data was available from the USGS without breaks in data.

56

Model output from a period of this length accurately represents the long-term effects of
watershed behavior, especially changes resulting from bioenergy crop production. The
model was run 41 times, based on the different scenarios and rotations. The SWAT
model output that was compared for this study was total nitrogen (NH4 + NO2 + NO3 +
Organic N) in kilograms, total phosphorus (Mineral P + Organic P) in kilograms, and
Sediment in metric tons. The output data was obtained at the watershed outlet, and on a
subbasin level.
5.2.5.3 Sensitivity Analysis
Model performance is dependent on the sensitivities of parameters’ influence on
the processes inherent in SWAT. A sensitivity analysis identifies the parameters that are
most influential on flow, sediment, and nutrients in the model. The result from the
sensitivity analysis is useful in determining the parameters to be used for model
calibration. A sensitivity analysis was performed for the Saginaw watershed using four
different landuse and management combinations: National Land Cover Data (NLCD
2001) without local management practices applied (NLCDN), CDL landuse without
management operations applied (CDL_NoMgmt), CDL landuse with management
operations applied (CDL_Mgmt), and CDL landuse with management operation applied
and observed data (CDL_Mgmt (Obs). The previously developed management operations
were applied to the CDL in order to provide a comparison of model output sensitivities
based solely on varied practices. It is important to note that the CDL is essentially the
NLCD map with higher detail given to agricultural land. For example, the NLCD does
not subdivide land used for agricultural purposes into detail, whereas the CDL gives each

57

landuse a class of its own, such as corn, soybean, sugarbeets, and onions. The 10 most
influential parameters for each set of model outputs (out of 41) are presented in Table 5.
The importance of using observed data during the sensitivity analysis procedure is
evident in Table 5. The cells highlighted in grey denote parameters that influence the
model output differently, based on the landuse, management practice, and observed data
combination used. By comparing (CDL_Mgmt) and (CDL_Mgmt (Obs), it is observed
that a number of parameters show very different levels of influence on the model
predictions. For example, landuse and management combinations have little influence on
the outcome of the sensitivity analysis for flow. However, when using observed data, the
arrangement and composition of the top ten parameters are significantly different. By
omitting the use of observed data, the integrity and quality of further calibrations could
be sacrificed.
Similarly, parameters for the sediment sensitivity analysis experienced
reorganization. By not using the observed data, the influence Alpha_Bf has on flow is
omitted, whereas Blai and Biomix are quite possibly exaggerated in their influence on
model predictions. In addition, Rchrg_Dp was ranked 10 for NLCD, yet was ranked 18
and 13 for the CDL_NoMgmt and CDL_Mgmt scenarios, respectively. This could
possibly be explained by the inherent differences of the NLCD and CDL landuse
classifications. Rchrg_Dp, the deep aquifer percolation fraction, is highly dependent upon
land cover, which is in much higher detail in the CDL map. The model sensitivity to
nutrient parameters follows a similar trend, in which the variation between landuse,
management, and observed data combinations has the potential to produce inconsistent

58

model predictions, which could in turn adversely affect the confidence in the value of
model output.
Sensitivity analysis is merely a tool in helping to identify parameters useful in
calibration of the model. A reasonable value of parameters during calibration with
reasonable justification for the use of each is imperative to produce results representative
of true watershed behavior. Site-specific knowledge and data are helpful in determining
which

of

the

most

sensitive

parameters

59

are

practical

in

application.

Table 5. Sensitivity analysis result.
Flow
Rank

NLCDN

CDL_NoMgmt

CDL_Mgmt

Sediment
CDL_Mgmt

NLCDN

CDL_NoMgmt

CDL_Mgmt

(Obs)

CDL_Mgmt
(Obs)

1

Cn2

Cn2

Cn2

Cn2

Spcon

Spcon

Spcon

Cn2

2

Rchrg_Dp

Alpha_Bf

Alpha_Bf

Alpha_Bf

Ch_N2

Usle_P

Ch_N2

Spcon

3

Esco

Rchrg_Dp

Rchrg_Dp

Surlag

Cn2

Ch_N2

Cn2

Alpha_Bf

4

Sol_Z

Esco

Esco

Rchrg_Dp

Ch_K2

Cn2

Spexp

Ch_K2

5

Sol_Awc

Timp

Timp

Ch_N2

Spexp

Ch_K2

Usle_P

Ch_N2

6

Alpha_Bf

Gwqmn

Gwqmn

Esco

Alpha_Bf

Timp

Alpha_Bf

Surlag

7

Canmx

Canmx

Sol_Awc

Ch_K2

Usle_P

Surlag

Ch_K2

Usle_P

8

Timp

Sol_Awc

Canmx

Sol_Awc

Blai

Blai

Timp

Spexp

9

Gwqmn

Ch_K2

Sol_Z

Timp

Esco

Biomix

Esco

Timp

10

Ch_K2

Sol_Z

Ch_K2

Sol_Z

Rchrg_Dp

Canmx

Surlag

Blai

Total Nitrogen
Rank

NLCDN

CDL_NoMgmt

CDL_Mgmt

Total Phosphorus
CDL_Mgmt

NLCDN

CDL_NoMgmt

CDL_Mgmt

(Obs)

CDL_Mgmt
(Obs)

1

Cn2

Canmx

Cn2

Canmx

Cn2

Canmx

Cn2

Sol_Awc

2

Surlag

Cn2

Alpha_Bf

Cn2

Sol_Awc

Cn2

Alpha_Bf

Cn2

60

Table 5 (cont’d)
3

Sol_Awc

Alpha_Bf

Canmx

Alpha_Bf

Canmx

Sol_Awc

Ch_K2

Alpha_Bf

4

Rchrg_Dp

Sol_Awc

Timp

Sol_Awc

Surlag

Ch_K2

Timp

Sol_Z

5

Sol_Z

Blai

Sol_Awc

Surlag

Gwqmn

Blai

Usle_P

Canmx

6

Alpha_Bf

Sol_Z

Nperco

Sol_Z

Sol_Z

Alpha_Bf

Sol_Awc

Surlag

7

Ch_K2

Nperco

Blai

Epco

Alpha_Bf

Sol_Z

Canmx

Ch_K2

8

Nperco

Ch_K2

Ch_K2

Nperco

Esco

Timp

Surlag

Blai

9

Esco

Esco

Rchrg_Dp

Blai

Blai

Surlag

Blai

Esco

10

Canmx

Timp

Usle_P

Esco

Timp

Esco

Smtmp

Gwqmn

61

5.2.5.4 Calibration/Validation
Calibration is a necessary component of modeling, especially in hydrologic
modeling (Moriasi et al., 2007). This process is the adjustment of input parameter values
to best match the observed data, reducing model uncertainty. Calibration is performed
either manually or by automatic calibration. Once a period for calibration data has been
selected, the model is run with initial values in place, and the agreement between model
output and observed data is measured using an objective function, while an optimization
procedure finds the parameter values that optimize the objective function (McCuen et al.,
2006). Validation follows calibration to establish model credibility. Statistical analysis
was performed for the validation period, leading to both qualitative and quantitative
verification of the model’s degree of accuracy.
Oftentimes, a single evaluation measure may indicate model prediction
performance in a biased or exceedingly positive/negative manner. Therefore, model
predictions were evaluated using three statistical methods: Nash-Sutcliffe’s efficiency
2
(NSE), root mean squared error (RMSE), and the coefficient of determination (R ). The

Nash-Sutcliffe efficiency determines how well the model output conforms to a 1:1
relationship of observed data, effectively determining goodness-of-fit (McCuen et al.,
2006). The NSE is computed with the relationship in Equation 1:

NSE=1-

∑ቀYobs -Ypred ቁ

2

∑൫Yobs -Ymean ൯

62

2

(1)

where Yobs is the observed value for the evaluated component, Ypred is the model
predicted value for the evaluated component, Ymean is the mean of observed data for the
evaluated component, and n is the total number of data points. NSE ranges from - ∞ to 1.
If the value for NSE is 1, the model exactly predicts the observed values for the evaluated
component. If the value for NSE is 0, the sum of squares of the difference between the
observed and model predicted is equal to the difference between the observed and the
mean of the observed values. Typically, NSE values greater than 0.65 are very good,
greater than 0.50 are satisfactory, and values less than 0.5 are considered unsatisfactory
on monthly basis.28
The root mean squared error (see Equation 2) is the vertical deviation of a data point from
a fitted line, providing a value in measureable units. The RMSE has important advantages
over traditional correlation methods, such as indicating error in the measured units
(Moriasi et al., 2007).

ܴ‫ ܧܵܯ‬ൌ ට

మ

∑൫௒೚್ೞ ି௒೛ೝ೐೏ ൯
௡

(2)

where n is the total sample size. Values of RMSE are that of the predicted variable, such
as cubic meters per second for flow. This gives an indication of quantitative differences
in the regression prediction from the observed (fitted) line.
2
The coefficient of determination, R , is a statistical measure for goodness-of-fit.

In this case, it is the percent of total variation in the y variable explained by the regression
2
equation (Menard, 2000). R values range from 0 to 1, where values close to one indicate

63

that the regression equation (model) is predicting the y variable with confidence and that
the dispersion of the predicted value is equal to the observed value. One of the main
drawbacks in using only this method for analysis of statistical fit is that the R

2

relationship fails to address under- or overpredictions (Krause et al., 2005). Values
greater than 0.5 are generally considered acceptable in monthly basis (Santhi et al.,
2
2001). The R is calculated with Equation 3:

ܴ ൌ൬
ଶ

∑ሺை೔ ିை೘೐ೌ೙ ሻሺ௉೔ ି௉೘೐ೌ೙ ሻ

ඥ∑ሺை೔ ିை೘೐ೌ೙ ሻమ ඥሺ௉೔ ି௉೘೐ೌ೙ ሻమ

൰

ଶ

(3)

where O are the observed values and P are the predicted values.
In order to adjust the model outputs for flow, sediment, and nutrients, a
combination of manual calibration and the autocalibration feature imbedded in SWAT
were used to enhance the overall effectiveness of parameter estimation. The simulation
period for flow calibration was chosen for having periods of high, low, and moderate
precipitation years compared to the mean rainfall for the previous 19 years (Figure 5).
Flow was calibrated for an 8 year period (2001-2008), and sediment and nutrients for a 6
year period (2000-2005). The first year for each of the simulation periods were treated as
a warm-up year, and was not included in the statistical comparisons. Calibration was
performed on the first half of the available data periods and validation was performed on
the second half.
Calibration and validation was performed once with SWAT default crop values
and once with management operations derived from local data. Calibration for each of the
model outputs (flow, sediment, nutrients) was performed sequentially, ensuring unbiased

64

and autonomous results. The first calibration was for flow, with the parameters chosen
based on the sensitivity analysis and the specific knowledge of the parameters’ effect on
flow. Sediment calibration followed, again with parameters chosen in a similar fashion
while omitting any that influence flow. Nutrient calibration was the final series of
simulations, omitting parameters which influence sediment and/or flow. The parameters
for each calibration were chosen based on the sensitivity analysis results and on the
influence each has on the model output.
Calibration and validation were performed for the watershed twice on a daily
basis, once without management operations applied prior to calibration (Tables 6 and 7)
and once with them applied (Tables 8 and 9). The goal is to compare the results obtained
from these scenarios and emphasis the importance of using local management practice
information while setting-up the model.
The model performed poorly with the default values, estimating flow NSE of 1.566 (Table 7). However, the calibrated flow NSE for the calibration and validation
periods was 0.647 and 0.726, respectively. For the first set-up calibration (without
considering management operations), the default parameter values in the SWAT model
resulted in low model efficiency for flow. Kjeldahl nitrogen and sediment performed
better with the default values (Table 7) and these results are acceptable for model
performance and support the use of calibrated parameter values for predictions. Similarly,
evaluations of the sediment and nutrient calibrations were completed. Sediment
calibration considerably increased the model performance. However, efforts to calibrate
nutrients yielded poor NSE results compared to uncalibrated, even though RMSE and R

2

were improved. Since nitrogen and phosphorus performed best under default values,
65

further calibration was not necessary and these values were applied in subsequent model
simulations. The poor nutrient calibration can likely be attributed to the low availability
of monitoring data in the watershed (less than 60 points during 6 years of model
simulation). It should be noted that the analyses were performed on a daily basis. In
general, monthly or annual evaluations improve statistical values considerably when used
in place of daily evaluations (Spruill et al., 2000; Gassman et al., 2007).

66

Table 6. Final parameter values (CDL landuse without management operations applied).
Parameter

Change By Method

Parameter input to Swat

Alpha_Bf

Replace Value

0.999

Ch_K2

Replace Value

38.937

Ch_N2

Replace Value

0.845

Cn2

Multiply By

1.001

Epco

Replace Value

0.840

Esco

Replace Value

0.457

Gw_Delay

Add

-9.797

Rchrg_Dp

Replace Value

0.006

Sol_Awc

Multiply By

0.875

Surlag

Replace Value

2.204

Biomix

Multiply by (-50 to 50)

0.554

Ch_Cov

Replace Value

0.696

Ch_Erod

Replace Value

0.467

Spcon

Replace Value

0.010

Spexp

Replace Value

1.156

Nperco

Replace Value

0.200

Phoskd

Replace Value

175.000

Pperco

Replace Value

10.000

Sol_No3

Multiply by (-50 to 50)

Default *

Sol_Orgn

Multiply by (-50 to 50)

Default *

Sol_Orgp

Multiply by (-50 to 50)

Default *

Flow Calibration

Sediment Calibration

Nutrient Calibration

* When initial values not user defined (default), SWAT initializes levels based on
relationship as function of depth

67

Table 7. Results of statistical analysis for calibration and validation periods.
Parameter

Uncalibrated

Calibration

Validation

Combined

NSE

-1.545

0.647

0.726

0.692

RMSE
2
R

297.5

77.27

67.80

102.8

0.086

0.660

0.735

0.699

NSE

0.613

0.462

0.933

0.548

RMSE
2
R

2066

2226

177.3

3111

0.589

0.873

0.297

0.861

Kjeldahl

NSE

0.428

0.382

0.209

0.329

Nitrogen

RMSE
2
R

5552

3683

3241

4906

0.366

0.786

0.526

0.692

Total

NSE

0.619

0.657

0.813

0.619

Phosphorus

RMSE
2
R

769.1

707.3

302.2

769.1

0.671

0.600

0.812

0.671

Flow

Sediment

As mentioned previously, the calibration process was performed for the watershed
by applying the management operations/schedules that were developed based on local
data. The parameters that were used in calibration and their associated final values are
provided in Table 8. The uncalibrated hydrograph is provided in Figure 6, showing the
extent of discontinuity to the observed data. The hydrograph based on model output after
calibration was completed is provided in Figure 7. This figure represents both the
calibration and validation periods. As observed in Figure 8, showing a 1:1 comparison
between predicted and observed flow, the calibrated model tends to predict flows well at
low –flow periods (less than 350 cms), yet tends to have a larger span of prediction with
flows greater than this point. The calibration statistics for this method are provided in
Table 9. The calibrated model performed well for flow, sediment, and phosphorus.
However, nitrogen performed poorly after model calibration. This is likely due to the lack

68

of available data and the interrelationship between the parameters affecting both nitrogen
and phosphorus. Since both nutrients were calibrated simultaneously, the model found
the best prediction for both, which in turn reduced the nitrogen prediction efficiency.

Table 8. Final parameter values (CDL landuse with management operations applied).
Parameter

Change Method

Parameter input to Swat

Alpha_Bf

Replace Value

0.937

Canmx

Replace Value

6.336

Ch_K2

Replace Value

41.194

Ch_N2

Replace Value

0.312

Cn2

Multiply By

1.068

Esco

Replace Value

0.523

Rchrg_Dp

Replace Value

0.159

Sol_Awc

Multiply By

0.646

Spcon

Replace Value

0.006

Timp

Replace Value

0.068

Usle_P

Replace Value

0.093

69

30
25

Deviation from Mean (%)

20
15
10
5
0
-5
-10
-15
2002

2003

2004

2005

2006

Year

Figure 5. Precipitation deviation from 19-year mean for study period.

70

2007

2008

3000

2500

Flow (cms)

2000

1500

1000

500

0
1/1/2002

5/16/2003

9/27/2004
Observed

2/9/2006

Uncalibrated

Figure 6. Uncalibrated Flow vs. Observed Flow at USGS Station 04157000.

71

6/24/2007

1.40E+03

1.20E+03

Calibration

Validation

Flow (cms)

1.00E+03

8.00E+02
Observed
Calibrated
6.00E+02

4.00E+02

2.00E+02

0.00E+00
1/1/2003

5/15/2004

9/27/2005

2/9/2007

6/23/2008

Date

Figure 7. Calibrated Flow vs. Observed Flow at USGS Station 04157000.

72

1200

Calibrated Flow (cms)

1000

800

600

400

200

0
0

200

400

600

800

Observed Flow (cms)

Figure 8. Calibrated Flow vs. Observed Flow at USGS Station 04157000.

73

1000

1200

Table 9. Results of statistical analysis for calibration and validation periods (CDL
landuse with management operations applied).
Parameter

Uncalibrated

Calibration

Validation

Combined

NSE

-1.633

0.766

0.666

0.723

RMSE
2
R

297.1

62.94

74.88

69.43

0.031

0.771

0.733

0.744

NSE

0.474

0.422

0.500

0.467

RMSE
2
R

2902

1977

2152

2922

0.496

0.842

0.693

0.712

Kjeldahl

NSE

0.126

-0.293

-0.068

-0.207

Nitrogen

RMSE
2
R

5599

5367

3808

6581

0.370

0.108

0.649

0.193

Total

NSE

-4.852

0.326

0.815

0.452

Phosphorus

RMSE
2
R

3016

882.3

271.7

923.2

0.305

0.475

0.839

0.541

Flow

Sediment

5.3 RESULTS AND DISCUSSION
5.3.1 Watershed-level Environmental Impacts of Bioenergy Crop Expansion
The effect of landuse change from current cropland practices to a significant
increase in land cover to a bioenergy cropping system is expected to alter the physical
and chemical processes in the watershed, leading to variations in pollution load. This is
due to increased tillage and wide row spacing with crops such as corn, sorghum, and
canola, increased fertilizer use, and variations in seasonal soil cover.
Average annual sediment, total nitrogen, and total phosphorus load for the study
period was estimated at a watershed level, reflecting the overall load at the basin main
channel outlet into the Saginaw Bay. The results for Scenario 4 are presented in Figure 9.
It is important to note that this figure is based on average annual loads, which tend to

74

smooth any radical spikes in constituent loading. The total pollution loads for each
constituent were used as a baseline for comparison from the current landuse (2007 CDLBase scenario). Nearly all crops except miscanthus, native grasses, rye, and switchgrass
experience increased pollution to the waterbody outlet. Among the remainder of the
studied crops, corn stover removal (40 percent residue removal) scenario experiences the
worst overall environmental impact. The predicted sediment load ranged from 93 percent
reduction for the miscanthus rotation to a 233 percent increase for the continuous
sorghum rotation. Similarly, nitrogen loads varied significantly from a 228 percent
increase with introduction of the continuous corn stover rotation to a 14 percent reduction
for the switchgrass rotation. The variation of phosphorus load is less significant for the
evaluated crop rotations. However the values range from 62 percent increase to 27
percent reduction for continuous corn stover and continuous rye, respectively. Based on
the given information, removal of corn stover for bioenergy production has the potential
to significantly impact the environment, due to a lower amount of ground cover in the
event of a rainstorm. Conversely, the expansion of land to native grasses and switchgrass
for

bioenergy

production

can

have

75

positive

environmental

consequences.

Figure 9. 19 Year Annual Sediment/Nutrient Load at Main Channel Outlet (Scenario 4).

76

5.3.2 Assessment of Marginal Land Conversion Impact on Water Quality
A similar display of pollutant loads for Scenario 3 (Marginal Land) is provided in
Figure 10. There is a clear trend as the amount of sediment load decreases with closergrown crops, with less tillage practices, longer growing seasons, and perennial growth
cycles. When native grasses are considered for marginal land conversion, nitrogen load
increases. This is likely due to the nitrogen cycling methods used in SWAT at the field
and in-stream. Marginal land at its current state (prior to landuse change to native
grasses) is fallow or under conservation programs to a large extent. The native grass
rotation for bioenergy production considers harvesting the above-ground biomass at the
end of the growing season. This leaves little ground cover, which allows for the loss of
nitrogen in overland flow to a much greater extent than under natural conditions. Without
the removal of crop biomass and residue, the carbon:nitrogen ratio results in nitrogen
immobilization, decreasing the soil ammonium and nitrate (Knops and Tilman, 2000). By
interrupting this natural cycle of nitrogen accumulation as bound particles in the shallow
soil layers, it is possible for nitrogen mobilization.

77

Figure 10. 19 Year Annual Sediment/Nutrient Load at Main Channel Outlet (Scenario 3).

78

5.3.3 Importance of Using Local Information on Model Predictions
As discussed previously, a sensitivity analysis and model calibration was
performed under four different landuse and management combinations: NLCD without
local management practices applied, CDL landuse without management operations
applied, CDL landuse with management operations applied, and CDL landuse with
management operation applied and observed data used. The goal in this discussion is to
evaluate the importance of incorporating data of local agricultural practices and observed
data on model predictions.
In this study, the results from model calibration for each of the three CDL landuse and
management combinations were evaluated and the model results for each of the 42
bioenergy crop rotations were compared. As an example, Figure 11 shows the variation
of sediment, nitrogen, and phosphorus loads at watershed scale under three combinations:
(MCM) that was calibrated while considering local management information, (MC) that
was calibrated without the use of local management operations, but this information was
incorporated during model examination of different crop scenarios (sub-scenarios), and
(DC) that used the SWAT default crop information for both model calibration and crop
sub-scenarios. Since MCM uses the local cropping information, it is expected to generate
the most realistic result for comparison with the two other combinations, because it is
representative of actual crop production practices in the study area. Based on this
analysis, the results of model predictions may be misleading if the practices best suited
for the study area aren’t addressed. In this case, sediment could be over-predicted by a
factor of 11, nitrogen by a factor of 1.2-1.9, and phosphorus by a factor of 2.1-2.4
resulting in possible misinformed decisions of landuse policy.

79

Figure 11. Annual Average Sediment Load for Three Corn Simulations.

80

5.3.4 Monthly Pollution Load Variation
Analysis of constituent loads and their resulting change from the base values were
also performed on a monthly time-step. This allows for the direct visualization of
variation between wet, dry, and average climatological periods. In addition, this can assist
in watershed management efforts to mitigate pollution under extreme pollution load
discharge to waterbodies. For example, for large-scale bioenergy crop expansion, it is
extremely important to determine the optimal landuse combinations and rotations that
allow sustainable production of biomass while maximizing economic return. Figure 12
represents the comparison of two severe cases: Continuous Corn Stover (top) and
Switchgrass (bottom) and their respective change from the base scenario for sediment
load to the main channel outlet. Here it is observed that continuous corn with 40 percent
stover removal results in substantial increases in sediment generation at watershed scale,
with up to a 493 percent increase over current landuse scenario. The switchgrass rotation
also resulted in large increases in sediment loads, with up to a 381 percent increase.
However, with the exception of 4 smaller spikes, these increases can be attributed to the
corn and soybean years of each rotation cycle (years 1 and 2 of each schedule, as
discussed previously and shown in Table 12), and the establishment year of switchgrass
in which the crop experiences lower productivity and fraction of soil cover than
subsequent growing seasons. Corn and soybeans were used prior to the 8 years of
switchgrass for sustainability purposes, as it is generally recommended to incorporate
such practices to reduce the chances of disease and poor stands of switchgrass cultivation.
The remainder of the years in this rotation experience significant sediment load
reductions with up to a 95 percent monthly reduction.

81

Monthly load changes from base were also calculated for total nitrogen and total
phosphorus represented in Figures 13 and 14, respectively. A similar trend is present for
total nitrogen as there was in sediment, continuous corn stover experiences overall large
spikes in nitrogen loads with up to a 1375 percent increase of nitrogen to the main
channel outlet, whereas switchgrass experienced reductions throughout the majority of
the study period. Nitrogen load increases during the first few years of each cycle is the
result of corn and soybean cultivation, however additional sharp increases outside of
these years may be caused by extreme climatological events. Therefore, we can conclude
that switchgrass production at large-scale can help in reducing sediment and nitrogen
loads within the study area.
Phosphorus loadings consistently increased (up to 792 percent) in the continuous
corn stover rotation, while the switchgrass rotation experienced reductions (up to 91
percent) in the periods outside of corn, soybean and establishment years. However, the
number and magnitude of significant phosphorus load increases during years under
switchgrass cultivation (up to 134 percent increase) are more frequent than both sediment
and nitrogen of the same periods. This may be due to the time of fertilizer application
(early spring), which usually occurs during the wet months. Another factor possibly
contributing to the increases is the collection and storage of nutrients in the soil after
application during corn and soybean years. Consequently, the nitrogen and phosphorus is
transported when the soil has minimal cover from switchgrass early in the growing
season during its recovery from the fall biomass harvest. In regards to corn, the extreme
increases in nutrient loads may be due to the increase in intensive practices such as
primary and secondary tillage, and increase in fertilizer requirements.

82

Figure 12. Corn Stover vs. Switchgrass Scenario 4 Monthly Sediment Load Change from Base.

83

Figure 13. Corn Stover vs. Switchgrass Scenario 4 Monthly Nitrogen Load Change from Base.

84

Figure 14. Corn Stover vs. Switchgrass Scenario 4 Monthly Phosphorus Load Change from Base.

85

Long-term monthly average sediment is provided in Table 10. Similar tables were
constructed for nitrogen and phosphorus and are available upon request. This information
is important because large magnitude increases of pollutant loading to waterbodies during
short periods of time may result in severe and detrimental effects on aquatic ecosystems,
violate total maximum daily load (TMDL) requirements, and threaten the sustainability
of bioenergy cropping systems in agricultural watersheds.
Based on the entire 19-year study period, the highest sediment and phosphorus
loads are consistently produced during March. However, the largest amount of nitrogen
load is produced in April. The months of December through June have the highest
sediment and phosphorus load, while March through May have the highest nitrogen load.
The correlation between sediment and phosphorus is expected due to their physical
properties of particle binding. However, nitrogen is mainly transported to stream
networks through groundwater systems from infiltration. Both extreme precipitation and
common time of nitrogen fertilizer application during the months of April and May can
explain this phenomenon. The period of second largest contribution of nitrogen is
December and January, where fall-applied nitrogen through either fertilizer or manure
applications is conveyed to the stream-network via lateral flow. By observing the column
providing the ratio of outlet sediment load to subbasin sediment load in Table 10, the
outlet produces 15.2-53.7 percent more sediment than at subbasin level. This is likely
attributed to bank and streambed erosion or in-stream processes. However, this is
important when evaluating the mitigation effects of land management practices. If the
pollutant loads are evaluated on subbasin level, the total contribution of load to the
watershed may be underestimated.

86

Table 10. Long-term average monthly sediment load (Saginaw Bay watershed).
Monthly Sediment Load (metric tons)

Outlet load/
subbasin load
(Percent)

Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

Base

53.7

4679

5076

12265

5164

7438

4790

1833

1014

1593

1338 2148

3215

Scen1
CCan

48.1

6794

5876

15126

5580

5328

2871

1332

1913

2633

1803 3146

4012

Scen1
CCrn

51.0

7327

7783

17200

7052

9566

5145

1794

1212

2563

2015 3183

4875

6863

7452

16198

6570

8469

5006

1802

1221

2594

2063 3270

4752

2944

2288

4615

2251

2716

6139

1562

770

1414

635

918

1310

13591

13742

32062

11520

13329

8156

2240

1341

2527

1934 4635

8987

8205

9476

21822

8291

11473

8981

3367

1403

1939

1760 3052

5543

7897

8663

19822

7488

10831

6173

2447

1177

2285

1776 3193

5211

9278

8856

20560

8309

9375

6594

2924

1708

2584

2106 3594

5602

4325

4858

8580

4231

5919

6148

2658

1010

2071

1280 1477

2776

7467

8415

19139

7107

10375

5762

2435

1179

2284

1777 3175

5119

Subscenario

Scen1
CCrnS
Scen1
CRye

43.9

Scen1
CSrg

47.7

Scen1
CSoy

50.1

Scen1
Crn Soy

50.4

Scen1
Crn Soy
Can
Scen1
CrnSoy
Rye
Scen1
CrnSSoy

51.0

50.7

47.8
50.5

87

Table 10 (cont’d)
Scen1
Misc

18.9

Scen1
NaGr

19.8

Scen1
SorSoy

49.1

Scen1
Swch

42.7

Scen2
CCrnS

52.7

Scen2
CrnSSoy

52.9

Scen2
Misc

52.0

Scen2
NaGr

52.0

Scen2
Swch

52.2

Scen3
CCan

41.9

Scen3
CCrn

42.5

244

269

504

310

347

325

332

291

297

225

212

184

264

292

542

340

404

353

341

295

309

239

226

203

10509

10980

25325

9578

12403

8504

2753

1268

2331

1847 3590

6710

1421

3258

5874

3515

5413

1734

755

498

721

800

496

944

3799

3828

9080

3827

4468

3331

1404

937

1566

1155 1601

2450

3830

3864

9235

3868

4595

3366

1435

932

1546

1144 1596

2472

3518

3532

8409

3535

4110

3124

1351

882

1461

1068 1455

2265

3518

3533

8409

3536

4113

3123

1351

880

1461

1065 1455

2267

3568

3651

8630

3705

4340

3187

1364

893

1475

1088 1469

2298

8003

7711

18986

7299

7616

5078

2410

2388

3390

2637 3598

5037

8668

8954

20266

8310

10110

6571

2552

1721

3176

2613 3647

5667

88

Table 10 (cont’d)
Scen3
CRye

45.2

Scen3
CrnSoy
Can
Scen3
CrnSoy
Rye
Scen3
Misc

49.5

Scen3
NaGr

49.5

Scen3
Swch

45.0

Scen4
CCan

40.8

Scen4
CCrn

43.4

Scen4
CCrnS

43.9

Scen4
CRye

39.7

Scen4
CSor

37.4

40.9
43.5

5437

5197

11718

4934

5727

6886

2586

1430

2327

1499 2064

3113

9847

9566

22388

9037

10202

7471

3301

2282

3208

2693 3888

6054

6642

6850

14191

6242

7897

7025

3046

1679

2749

1843 2427

4089

3510

3536

8318

3557

4163

3124

1352

912

1460

1080 1477

2277

3537

3546

8344

3579

4231

3166

1359

911

1463

1075 1477

2284

4296

5731

12078

5560

7462

4164

1678

1116

1794

1639 1740

2879

10879

9873

24791

9021

8454

4850

2370

3242

4451

3423 5195

6728

12139

13064

28443

11574

14892

8347

3022

2042

4196

3528 5286

8223

11573

12562

27091

10993

13454

8222

3044

2066

4202

3618 5457

8026

4862

4094

8004

3648

4082

9306

2678

1273

2242

1083 1546

2274

19687

20581

45920

17037

19628

12607

3799

2351

4183

3391 6646

12451

89

Table 10 (cont’d)
Scen4
CSoy

41.0

Scen4
CrnSoy

42.0

Scen4
CrnSoy
Rye
Scen4
CrnSSoy

41.8
42.3

Scen4
Misc

15.2

Scen4
NaGr

17.0

Scen4
SorSoy

39.2

Scen4
Swch

38.4

13263

15615

34572

12831

17333

13810

5616

2533

3146

3185 4965

9051

12847

14420

31744

12093

16555

9863

4039

2049

3692

3148 5220

8619

7369

8217

14409

6904

9152

9669

4276

1812

3323

2002 2426

4602

12343

14068

30802

11612

15791

9332

4006

2052

3699

3158 5216

8461

255

292

552

322

363

313

292

244

254

192

195

191

297

332

624

382

475

375

312

260

286

227

223

221

16071

17722

39238

14568

18639

13214

4549

2238

3790

3193 5600

10500

2170

5391

9431

5432

8374

2702

979

694

1045

1338

1546

90

774

5.3.5 Determining Practicality of Scenarios for Landuse Change
Table 11 provides a comparison of all rotations based on contribution to annual
average sediment, nitrogen, and phosphorus load. This orientation allows for a direct
evaluation of the potential environmental impacts for each of the four scenarios, and thus
different amounts of landuse change on different landscape classes. This is important for
policy making and decisions for landuse change in agricultural region. In order to
accommodate the growing demand for available land to produce bioenergy crops, the
assessment of environmental implications is imperative.
The scenarios, as presented previously, are as follows: Scenario 1 considers all
land currently cultivated for row crops and grains, hays, and seeds (i.e. corn, soybeans,
wheat, hay, etc.) for conversion to bioenergy crops, Scenario 2 considers only the other
crops (i.e. sugarbeets, potatoes, dry beans, fruit crops, etc.), Scenario 3 considers
farmland uses and other land (i.e. fallow cropland, pasture, wasteland, etc.), which in all
practical purposes may be deemed as marginal land, and Scenario 4 is the combination of
all previous scenarios, and represents the overall most significant landuse change from
current state to the chosen bioenergy cropping rotations.
By examining the list based on sediment, a number of rotation placements are of
particular interest. As expected, the majority of sediment mitigating crops includes the
perennial grasses such as miscanthus, native grasses, and switchgrass. However,
continuous rye (Scen1_ContinuousRye) when applied to land currently being cultivated
for row crops results in a surprising 45.5 percent reduction in sediment load. Continuous
rye under Scenario 4 shows similar results with a 10.8 percent reduction. As rye is most
commonly planted in the fall, acting as a cover crop for winter months and begins

91

growing extensively as soon as the soil warms in the spring. Consequently, this period is
during times of heavy rainfall, acting as a cover to what would otherwise be bare soil.
When coupled with corn and soybeans, rye fails to mitigate the sediment pollution these
two crops produce as efficiently as the continuous rye rotation. Interestingly enough,
when land currently under the classification of other crops (Scenario 2) is converted to a
corn rotation with stover removal, there is a significant reduction (25.1-25.9 percent) in
sediment load. This is extremely important when determining the sustainability of corn
stover as a potential bioenergy crop. In most cases, rotations with corn stover removal are
considered a risky practice, but when applied to land in this scenario the effects are
beneficial. In addition, conservation programs such as the Conservation Reserve Program
(CRP) encourage land owners to allow the land currently under cultivation to return to a
native vegetation state (USDA-NRCS, 2009). The goal of this is to reduce pollutants in
agricultural runoff. However, in this study miscanthus mitigates a higher percentage of
sediment than native grasses in every scenario that was evaluated. This suggests that the
miscanthus may be a suitable and more economically sustainable practice than would be
a native grass landscape. As the data suggests, these long-rotation perennial grasses
extend the growing season, require soil disturbance for planting only on the establishment
year, and have dense root spacing and canopy cover. The result is consistent and ample
ground cover year-round, decreasing erosion potential. Sediment reduction is expected
for perennial crops or crops traditionally grown with low-input intensive practices, but
not for crops such as continuous corn. Crops such as this are input-intensive and require
substantial tillage. Close grown crops and rotations including these, such as continuous
rye, also experience reduced sediment yield for similar reasons to the perennial grasses.

92

Corn stover removal substantially increases the amount of soil loss, accompanying the
loss of crop residue, which is valuable in maintaining adequate soil cover until the
following growing season.
The organization of the rotations in respect to their nitrogen values compared to
base presents yet more unexpected results. Continuous sorghum on Scenarios 1 and 4
result in 20.6 and 2.8 percent reductions in nitrogen, respectively. However sorghum
when accompanied with soybeans on Scenario 4 results in a 32.4 percent increase in
nitrogen load. Continuous soybean on Scenario 4 has an 86.6 percent increase as well,
which helps to explain the increase from continuous sorghum to sorghum soybean. When
sorghum is cultivated for five years followed by a year of soybeans, the adverse effect of
soybeans is negated to a large extent, unlike the short two year rotation of sorghumsoybean. In either case of continuous sorghum or the sorghum-soybean rotations,
nitrogen load is increased when these rotations are implemented on scenario 4. Here, the
only difference is more land in production of these rotations (other crops and marginal
land). Thus it can be concluded that these lands should not be used for sorghum
cultivation at a large extent. The perennial grass species reverse their standings from
sediment when considered on a nitrogen basis. Except for native grass and switchgrass on
Scenarios 1 and 4, these species exhibit increases in nitrogen load. Miscanthus, the crop
with the most positive influence on sediment mitigation, experiences 12.7-72.2 percent
increases of nitrogen load. This result makes it difficult to justify converting large
amounts of land to miscanthus. However, based on watershed health, miscanthus may be
suitable in certain applications. For example, if sediment is of utmost importance,

93

miscanthus is an excellent mitigation tool, but if nitrogen levels are important,
miscanthus may not be the most suitable crop.
Phosphorus loads are closely related to sediment load results. Nearly all of the
crops that experience sediment mitigation also show decreases in phosphorus loadings.
The exception to this rule is miscanthus on Scenarios 3 and 4. Again, as observed with
nitrogen loadings, when miscanthus is considered for marginal land conversion there is
an adverse effect on phosphorus load. This further suggests miscanthus is poorly suited
for cultivation on marginal lands, as both nitrogen and phosphorus pollutant loads
increase. Canola, a crop not yet discussed here, experiences sediment, nitrogen, and
phosphorus load increases on every scenario when considered for conversion. This also
suggests that canola has poor suitability for extensive cultivation in this watershed.

94

Table 11. Land use scenarios ranking based on environmental benefits.

Sub-scenario

Sediment
% Change
from Base

Sub-scenario

Total N %
Change
from Base

Sub-scenario

Total P %
Change
from Base

Scen4_Miscanthus

-93.1

Scen1_Switchgrass

-26.2

Scen1_Continuous Rye

-33.1

Scen1_Miscanthus

-93

Scen1_Continuous
Sorghum

-20.6

Scen4_Continuous Rye

-27.2

Scen1_NativeGrass

-92.5

Scen1_Continuous Rye

-17.8

Scen1_NativeGrass

-27.1

Scen4_NativeGrass

-92.1

Scen1_NativeGrass

-16.2

Scen1_Switchgrass

-21.5

Scen1_Switchgrass

-49.7

Scen4_Switchgrass

-14.2

Scen4_NativeGrass

-21

Scen1_Continuous Rye

-45.5

Scen1_SorghumSoy

-6.4

Scen2_Switchgrass

-12.9

Scen2_Miscanthus

-31.3

Scen4_Continuous
Sorghum

-2.8

Scen2_Miscanthus

-12.8

Scen2_NativeGrass

-31.3

Scen4_NativeGrass

-1.8

Scen2_NativeGrass

-11.9

Scen3_Miscanthus

-31.2

Scen1_CornSoyRye

-0.6

Scen4_Switchgrass

-11.8

Scen3_NativeGrass

-30.8

Scen1_Continuous Corn

5.3

Scen2_Continuous
Corn Stover

-11.4

Scen2_Switchgrass

-29.4

Scen1_CornSoy

9.9

Scen1_CornSoyRye

-11.2

Scen2_Continuous Corn
Stover

-25.9

Scen4_Continuous Rye

10.1

Scen2_CornStover Soy

-11.1

Scen2_CornStover Soy

-25.1

Scen1_Miscanthus

12.7

Scen3_Continuous Rye

-7.2

95

Table 11 (cont’d)
Scen4_Switchgrass

-21.1

Scen4_Continuous Rye

-10.8

Scen1_CornSoyRye

Scen1_Continuous
Soybean

23.5

Scen3_NativeGrass

-6.8

25

Scen1_Miscanthus

-6.2

-10.3

Scen1_Continuous
Canola
Scen2_Switchgrass

29.9

Scen3_Switchgrass

-2.7

Scen3_Switchgrass

-0.8

Scen1_CornStover Soy

30.3

Scen3_Miscanthus

2.6

Scen3_Continuous Rye

4.7

Scen2_NativeGrass

30.5

Scen1_SorghumSoy

3.3

Scen1_Continuous
Canola
Scen3_CornSoyRye

11.6

Scen2_Miscanthus

31.1

Scen3_CornSoyRye

3.8

27.9

Scen2_CornStover Soy

32.1

Scen4_CornSoyRye

5.4

Scen1_Continuous Corn
Stover
Scen1_Continuous Corn

31.1

Scen4_SorghumSoy

32.4

5.4

37.9

Scen1_CornSoy Canola

33.2

Scen1_Continuous
Canola
Scen4_Miscanthus

Scen3_Continuous
Canola

46.7

Scen2_Continuous
CornStover

34.8

Scen4_CornSoyRye

46.7

Scen4_Continuous Corn

43.1

Scen1_CornStover Soy

46.8

Scen3_Switchgrass

47.1

Scen1_CornSoy

52.2

Scen4_CornSoyRye

47.9

96

Scen1_Continuous
Sorghum
Scen1_Continuous
Soybean
Scen1_CornSoy
Scen3_Continuous
Canola

7.9
11.1
13.3
14.2
15.4

Table 11 (cont’d)
Scen1_CornSoyCanola

61.2

Scen3_NativeGrass

50.4

Scen1_Continuous
Corn

17.1

Scen3_Continuous Corn

62.7

Scen4_Miscanthus

52.9

Scen1_CornStover Soy

18.3

Scen1_Continuous
Soybean

68.8

Scen3_Continuous Rye

60.3

Scen3_Continuous
Corn

20.6

Scen3_CornSoyCanola

77.9

Scen4_CornSoy

60.5

Scen4_Continuous
Canola

84.5

Scen3_Continuous Corn

69.3

Scen1_Continuous
22
Corn Stover
Scen3_CornSoy Canola 23.9

Scen1_SorghumSoy

89.5

Scen3_Miscanthus

72.2

Scen1_CornSoy Canola 24.1

Scen4_Continuous Corn
Stover

118.2

Scen3_CornSoyRye

78.7

Scen4_SorghumSoy

30.9

Scen1_Continuous
Sorghum

125.6

Scen4_Continuous
Soybean

86.6

Scen4_Continuous
Canola

34.5

Scen4_Continuous Corn

127

Scen4_CornStover Soy

138.4

Scen4_CornSoy

145.9

Scen4_CornStover Soy

Scen4_Continuous
Soybean

168.9

Scen1_Continuous Corn
Stover

Scen4_Continuous
Canola
Scen3_Continuous
Canola

Scen4_Continuous
Sorghum
Scen4_Continuous
Soybean

43

98.1

Scen4_CornSoy

53.7

106.7

Scen4_Continuous
Corn

53.9

95.6
96.8

97

46.3

Table 11 (cont’d)
Scen4_SorghumSoy

195.4

Scen3_CornSoy Canola

111

Scen4_CornStover Soy

58.6

Scen4_Continuous
Sorghum

232.9

Scen4_Continuous Corn
Stover

227.6

Scen4_Continuous
Corn Stover

62.3

98

5.4 CONCLUSION
The SWAT model is a powerful tool in assessing the environmental implications
of the expansion of bioenergy crops, as expressed in the scenarios developed for this
project. The objective of this study is to examine the environmental implications,
specifically water quality, of bioenergy cropping system expansion. The Soil and Water
Assessment Tool (SWAT) was used to model the effects of four scenarios, 17 different
bioenergy crops under 42 rotations on sediment load, total nitrogen load, and phosphorus
load in the Saginaw River Watershed (hydrologic unit code 040802) during a 19 year
(1990 to 2008) period on a daily basis. This watershed, which drains into the Saginaw
Bay and subsequently Lake Huron, is intensively cultivated for crops.
The results obtained therein show that there is a direct correlation between
landuse change to bioenergy crops and water quality. Significant increases in sediment
(up to 233 percent) occurred when converting current landcover to wide-row intensive
crops such as a continuous sorghum rotation. Nitrogen and phosphorus loads increased
with such crops as well with 228 percent and 62 percent increases of total nitrogen and
total phosphorus, respectively, when applied to all currently-available land for cultivation
(Scenario 4). However, important second-generation bioenergy grass crops such as
miscanthus, native, grasses, and rye experienced admirable pollutant load mitigation,
with reductions up to 93 percent, 26 percent, and 33 percent of sediment, nitrogen, and
phosphorus respectively. These values reflect average annual reductions, though, which
are far less drastic than the fluctuations observed on a monthly time step. For example,
continuous corn with stover removal resulted in up to a 493 percent increase of sediment
load, up to a 1375 percent increase in total nitrogen load, and up to 792 percent increase

99

in total phosphorus load over the current landuse scenario. Increases of pollutant loading
to the channel of such large magnitudes may result in direct impairment to aquatic
ecosystems, total maximum daily load (TMDL) requirements, and threaten the
sustainability of bioenergy cropping systems in agricultural watersheds. The results of
this study show that oftentimes bioenergy crops are not suitable for cultivation on certain
landscapes. However, multiple crops that were once thought of as detrimental to the
environment show promise in certain applications, such as continuous corn with stover
removal cultivated in lands previously classified as other crops, which resulted in both
sediment and phosphorus mitigation. Equally important is the potential implications that
the grass species may have when grown on certain landscapes, especially marginal land,
where substantial increases in nitrogen and phosphorus were exhibited.

Acknowledgements
The authors would like to acknowledge Mr. Chaopeng Shen, and Mr. Sean
Woznicki and Dr. Fred Bakker-Arkema for their added project support, and Dr. Kurt
Thelen, Dr. Wesley Everman, Mr. Dennis Pennington, Dr. Scott Swinton, and Ms. Laura
James for their generous contribution of information.

100

APPENDIX

101

APPENDIX
Table 12. Fertilizer application rates for each rotation used in SWAT.

Rotation

Rate
(kg/ha)

Crop in Rotation
Year 6

Canola

Canola

Rye

Canola

Canola

Nitrogen

28

140

140

28

140

140

39.2

39.2

39.2

39.2

39.2

39.2

Corn

Corn

Corn

Corn

Soybean

Nitrogen

194

194

194

194

194

-

59.5

59.5

59.5

59.5

59.5

45

Corn

Corn

Corn

Corn

Corn

Soybean

Nitrogen

194

194

194

194

194

-

Phosphorus

59.5

59.5

59.5

59.5

59.5

45

Soybean

Rye

Rye

Soybean

Rye

Rye

Nitrogen

-

28

28

-

28

28

Phosphorus

45

39.2

39.2

45

39.2

39.2

Sorghum
Continuous
Sorghum

Year 5

Phosphorus

Continuous
Rye

Year 4

Corn

Continuous
Corn
Stover

Year 3

Phosphorus
Continuous
Corn

Year 2

Rye
Continuous
Canola

Year 1

Sorghum

Sorghum

Nitrogen

112

112

112

112

112

-

Phosphorus

45

45

45

45

45

45

102

Sorghum Sorghum

Soybean

Table 12 (cont’d)
Soybean
Continuous
Soybean

Soybean

Corn

Soybean

Soybean

Corn

Nitrogen

-

-

194

-

-

194

Phosphorus

45

45

59.5

45

45

59.5

Miscanthus (Continuous)
-

84

84

84

84

84

-

17

17

17

17

17

Corn

Soybean

Switchgrass Switchgrass Switchgrass Switchgrass

Nitrogen

194

84

-

84

84

84

Phosphorus

Switchgrass

Nitrogen
Phosphorus

Miscanthus

59.5

17

-

17

17

17

Native Grasses (Continuous)
Native
Grasses

CornSoybeanRye

-

-

-

-

-

-

Phosphorus
CornSoybeanCanola

Nitrogen

-

-

-

-

-

-

Corn

Soybean

Canola

Corn

Soybean

Canola

Nitrogen

194

-

140

194

-

140

Phosphorus

59.5

45

39.2

59.5

45

39.2

Corn

Soybean

Rye

Corn

Soybean

Rye

Nitrogen

194

-

28

194

-

28

Phosphorus

59.5

45

39.2

59.5

45

39.2

103

Table 12 (cont’d)

SorghumSoybean

Soybean

Corn

Soybean

Corn

Soybean

Nitrogen

194

-

194

-

194

-

59.5

45

59.5

45

59.5

45

Nitrogen

Corn
194

Soybean
-

Corn
194

Soybean
-

Corn
194

Soybean
-

Phosphorus

59.5

45

59.5

45

59.5

45

Nitrogen

Sorghum
112

Soybean
-

Sorghum
112

Soybean
-

Sorghum
112

Soybean
-

Phosphorus

Corn
StoverSoybean

Corn

Phosphorus

CornSoybean

45

45

45

45

45

45

Table 13. Corn-soybean-canola rotation management operations.
Date

Practice

SWAT Practice

1-May

Soil Finish

Field Cultivator Ge15ft

4-May

Nitrogen Application (Urea)

Urea

4-May

Soil Finish

Field Cultivator Ge15ft

5-May

Phosphorus Application (P2O5)

Elemental Phosphorus

5-May

Plant Corn Seed

Plant/Begin Growing Season

5-May

Bicep II Magnum ® (PRE)

Atrazine

Amount/ha

173 lb

194 kg

53 lb

59.5 kg

0.5 qt
104

Amount/acre

1.39 kg

Year

1

Table 13 (cont’d)
5-May

Bicep II Magnum ® (PRE)

Metolachlor

0.39 qt

1-Nov

Combine Harvest Corn Grain

Harvest and Kill

15-Nov

Fall Chisel

Coulter-chisel Plow

14-May

Soil Finish

Field Cultivator Ge15ft

14-May

Phosphorus Application (P2O5)

Elemental Phosphorus

14-May

Soil Finish

Field Cultivator Ge15ft

15-May

Plant Soybean Seed

Plant/Begin Growing Season

7-Jun

Spray Roundup ®

Glyphosate Amine

1-Oct

Combine Harvest Soybean Grain

Harvest and Kill

4-Oct

Soil Finish

Field Cultivator Ge15ft

7-Oct

Nitrogen Application (Urea)

Urea

125 lb

140 kg

7-Oct

Phosphorus Application (P2O5)

Elemental Phosphorus

35 lb

39.2 kg

7-Oct

Treflan ® Pre-Plant Incorporated
Herbicide

Trifluralin

2 pt

1.1 kg

7-Oct

Plant Canola Seed

Plant/Begin Growing Season

25-Apr

Post-Emergence Poast ® Herbicide

Sethoxydim

1.5 pt

0.315 kg

15-Jul

Combine Harvest Canola

Harvest and Kill

30-Oct

Fall Chisel

Coulter-chisel Plow

105

1 kg

40 lb

45 kg

22 fl oz

0.87 kg
2

3

All Row Crops and
Small Grains/Hays

A

Bioenergy
Crops
Barley
Cassava
Corn
Potato
Canola
Rice
Rye
Sorghum
Soybean
Sugarbeet
Sugar Cane
Wheat
Jatropha
Miscanthus
Switchgrass
Corn Stover
Native
Grasses

Barley
Corn
Potato
Canola
Rye
Sorghum
Soybean
Sugarbeet
Wheat
Miscanthus
Switchgrass
Corn Stover
Native
Grasses

Bioenergy
Crops
Considered
Corn
Canola
Rye
Sorghum
Soybean
Miscanthus
Corn Stover
Switchgrass
Native
Grasses

Figure 15. Scenario 1 (a), 2 (b), 3 (c), and 4 (d) implemented in the SWAT model.

106

Associated
Rotations
Considered
(Crn-Soy-Can)
(Crn-Soy-Rye)
(Crn-Soy)
(CrnS-Soy)
(CCan)
(CCrn)
(CCrnS)
(CRye)
(Srg-Soy)
(CSrg)
(Csoy)
(Mis)
(SwG)
(NaGr)

Other Crops

B

Bioenergy
Crops
Barley
Cassava
Corn
Potato
Canola
Rice
Rye
Sorghum
Soybean
Sugarbeet
Sugar Cane
Wheat
Jatropha
Miscanthus
Switchgrass
Corn Stover
Native
Grasses

Barley
Corn
Potato
Canola
Rye
Sorghum
Soybean
Sugarbeet
Wheat
Miscanthus
Switchgrass
Corn Stover
Native
Grasses

Bioenergy
Crops
Considered
Miscanthus
Switchgrass
Corn Stover
Native
Grasses

Figure 15 (cont’d)

107

Associated
Rotations
Considered
(Mis)
(SwG)
(CrnS-Soy)
(CCrnS)
(NaGr)

Marginal Land
(Farmland Uses and
Other Land)

C

Bioenergy
Crops
Barley
Cassava
Corn
Potato
Canola
Rice
Rye
Sorghum
Soybean
Sugarbeet
Sugar Cane
Wheat
Jatropha
Miscanthus
Switchgrass
Corn Stover
Native
Grasses

Barley
Corn
Potato
Canola
Rye
Sorghum
Soybean
Sugarbeet
Wheat
Miscanthus
Switchgrass
Corn Stover
Native
Grasses

Figure 15 (cont’d)

108

Bioenergy
Crops
Considered
Corn
Canola
Rye
Switchgrass
Miscanthus
Native
Grasses

Associated
Rotations
Considered
(Crn-Soy-Can)
(Crn-Soy-Rye)
(CCrn)
(CCan)
(CRye)
(Mis)
(SwG)
(NaGr)

Row Crops, Small
Grains/Hays, Other
Crops and Marginal
Land

D

Bioenergy
Crops
Cassava
Corn
Potato
Canola
Rice
Rye
Sorghum
Soybean
Sugarbeet
Sugar Cane
Wheat
Jatropha
Miscanthus
Switchgrass
Corn Stover
Native
Grasses

Barley
Corn
Potato
Canola
Rye
Sorghum
Soybean
Sugarbeet
Wheat
Miscanthus
Switchgrass
Corn Stover
Native
Grasses

Bioenergy
Crops
Considered
Corn
Canola
Rye
Sorghum
Soybean
Miscanthus
Switchgrass
Corn Stover
Native
Grasses

Associated
Rotations
Considered
(Crn-Soy-Rye)
(CrnS-Soy)
(Crn-Soy)
(Srg-Soy)
(CCrn)
(CCrnS)
(CCan)
(CRye)
(CSrg)
(CSoy)
(Mis)
(SwG)
(NaGr)

Figure 15 (cont’d)

* Cells with yellow fill denote crops determined to be unsuitable in the scenario due to the regions physical attributes, climate,
and preexisting harvesting equipment. Cells with blue fill denote crops that would further be defined as “poor suitability”
based on excessive capital cost for equipment, low productivity on common soil types in the watershed, and the ease of
farmers to alter preexisting management practices to accommodate new cropping systems.

109

6. WATER QUALITY IMPACT ASSESSMENT OF LARGE SCALE BIOFUEL
CROPS EXPANSION
a
a,*
Brad J. Love , A. Pouyan Nejadhashemi
a

Department of Biosystems and Agricultural Engineering, Michigan State University,
216 Farrall Hall, East Lansing, MI 48824 USA

* Author who correspondence should be addressed. Tel.: (517) 432-7653 fax: 517-4322892. Email address: pouyan@msu.edu (A. Pouyan Nejadhashemi)
Abstract. The challenges we face in transitioning to global production of biomass as
renewable feedstock sources in a way that is both economically feasible and
environmentally sustainable are ubiquitous. In this study, the Soil and Water Assessment
Tool (SWAT) was used to predict the possible long-term environmental implications,
specifically water quality, due to large-scale bioenergy cropping system expansion based
on four landuse scenarios and 15 bioenergy crop rotations for four watersheds, totaling
244 model simulations. The study area consists of four large watersheds totaling 53,358
2
km located in Michigan’s Lower Peninsula used intensively for agricultural production

all directly contributing to the Great Lakes. Non-point source pollutant loadings of
sediment, nitrogen, and phosphorus are a major issue throughout the study area,
contributing to 139 records on the Michigan Section 303(d) list for waterbody
impairments. The results from SWAT suggest that in general, perennial grass species are
most suitable for large-scale implementation in this study area, whereas traditional
intensive row crops should be implemented with caution on such a broad scale. In
addition, bioenergy row crops such as continuous sorghum, continuous corn stover, and

110

continuous canola exhibit dramatic pollution load variation caused by differences in
climate and physiographic characteristics throughout the study area. Therefore, additional
attention should be made when selecting locations for implementation of these crops.
Row crops also had the highest increases of high priority areas for sediment, nitrogen,
and phosphorus of 15.82 percent (continuous sorghum), 26.0 percent (continuous corn
stover), and 3.35 percent (continuous soybean), respectively. Based on the data from this
study, it is not recommended that marginal land be converted to any row crop bioenergy
rotation in areas with preexisting high nitrogen levels. Statistical analyses demonstrate
that perennial grass species significantly reduce sediment on all lands except marginal
lands. Meanwhile, only bioenergy crops grown on row crop land experiences significant
decreases in total nitrogen load. With the exception of row crops cultivated on marginal
lands and all agricultural land, the majority of bioenergy crops significantly reduce total
phosphorus loads.

6.1 INTRODUCTION
Biofuels are liquid fuels, such as ethanol and biodiesel, produced from biological
feedstocks and are mainly used for transportation and heating (de Fraiture et al., 2008).
The benefits associated with the production of biofuels are commonly referenced.
Biofuels are looked to as an answer to reducing greenhouse gases responsible for climate
change and the nation’s vast addiction with fossil fuels, which are politically risky and
environmentally harmful (Solomon, 2010). Renewable energy sources are readily
available, ensuring domestic energy security rather than relying on oftentimes politically
volatile foreign countries (Goldemberg, 2007). However, negative aspects are equally

111

important to examine prior to considering large-scale implementation of bioenergy crops
as a sustainable practice. When landuse change is considered, the benefits of biofuels
become difficult to quantify and may be negated due to possible environmental impacts
(McLaughlin and Walsh, 1998). Greenhouse gas emissions associated with landuse
change from current landcover to large-scale corn production for ethanol are estimated to
double (Searchinger et al., 2008). Bioenergy cropping systems are likely to increase
pollution, further impair already damaged soils in consequence of intense agricultural
practices, and accumulate harmful greenhouse gases from their production and use
(Groom et al., 2008). In addition, the expansion of bioenergy crop production will
undoubtedly increase the amount of water use through evapotranspiration and reallocate
water resources from available water for human consumption to the atmosphere as water
vapor (Berndes, 2002). However, the daunting fact of biofuel production is that landuse
change into agricultural production is inevitable.
A revision to the National Renewable Fuel Standard program has mandated 15.2
and 30 billion gallons of renewable fuel by 2012 and 2020, respectively (USEPA 2009a).
To hastily meet demands for biomass, significant alterations to current landuse leading to
increases in cultivated acreage will result (Rowe et al., 2009). However, large-scale
production of bioenergy feedstocks are limited due to landuse competition with
traditional food crops (Scheffran and BenDor, 2009). Therefore, biofuel crops must also
be considered to be grown on agriculturally marginal land or surplus lands with degraded
production capabilities (Lal and Pimentel, 2007). Unfortunately, the implications of
expanding cropland to accommodate demands are not adequately understood at a largescale (Rowe et al., 2009).

112

In order for biofuels to be a sustainable solution to fossil fuels, their cultivation
must avoid intensifying any negative environmental impacts. Freshwater is vital to
human existence and maintaining its health is a fundamental requirement for sustainable
agricultural production (Gleick, 1998). Increased intensive agricultural practices
associated with current bioenergy crops such as corn, soybeans, sugarcane, etc. may lead
to impairments of waterbodies. Agriculture is responsible for the largest contribution of
non-point source pollution to freshwater in the United States (Foley et al., 2005). Corn,
the current crop supplying majority of biofuels in the United States, requires more
fertilizer and pesticides than any other crop grown in the nation (Pimentel and Patzek
2005). Biofuel production requires increased amounts of fertilizer use and irrigation to
increase crop yields to meet demands (Searchinger et al., 2008). However, even under
current practice, row crop production is the largest contributor of nitrogen and
phosphorus fluxes to the Mississippi River basin resulting in eutrophication, reduced
dissolved oxygen, algal blooms, and impaired aquatic habitat of surface waterbodies, and
hypoxic zones in the Gulf of Mexico (Carpenter et al., 1998; Powers, 2007).
Eutrophication is many times responsible for impaired waterbodies and contaminated
drinking water supplies, which increases the cost of treatment for human use (Carpenter
et al., 1998; Nyakatawa et al., 2006). As a direct consequence, human safety and aquatic
ecosystem health are at risk.
The research in this paper evaluates the long-term environmental implications,
specifically water quality, of large-scale bioenergy cropping system expansion. This aids
in determining the suitability of bioenergy crops and obtaining information to aid in
watershed-scale decision making regarding landuse management. The Soil and Water

113

Assessment Tool (SWAT) was used to predict the effects of four future landuse scenarios
and 15 bioenergy crop rotations combined to form 42 total cropping scenarios on
sediment, nitrogen, and phosphorus loads in four watersheds in agricultural regions of
Michigan during a 19 year (1990 to 2008) period. The model was calibrated and
validated on a monthly basis and model output was evaluated on an annual time step.

6.2 MATERIALS AND METHODS
6.2.1 Study Area
The study area is located in the lower portion of Michigan and includes the
following four watersheds (Figure 16):
The Saginaw River basin (hydrologic unit code (HUC) 040802) is located in the
east central portion of Michigan’s Lower Peninsula and includes six HUC 8-digit
watersheds (Tittabawassee, Pine, Shiawassee, Flint, Cass, and Saginaw River).
The size of drainage area is 1,526,242 ha in which about 40 percent being used
for agricultural productions, 6 percent developed, 42.4 percent forest, and 12
percent wetlands.
The St. Clair-Detroit basin (HUC 040900) is located in southeast Michigan and
includes four HUC 8-digit watersheds (St. Clair, Clinton, Detroit, and Huron
River). The size of drainage area is 818,303 ha with approximately 30.3 percent
being used for agricultural production, 39.4 percent developed, 28 percent forest,
and 2.3 percent wetlands.
The Southeastern Lake Michigan basin (HUC 040500) is located in southwest
Michigan and includes five HUC 8-digit watersheds (Kalamazoo, Upper Grand,
Maple, Lower Grand, and Thornapple River). The size of drainage area is

114

1,889,426 ha with approximately 53.3 percent being used for agricultural
production, 10.6 percent developed, 28.8 percent forest, and 7.3 percent wetlands.
The St. Joseph basin (HUC 04050001) is located in the southwest portion of
Michigan and the northeast portion of Indiana. The size of drainage area is
1,101,781 ha with approximately 70.9 percent being used for agricultural
productions, 6.3 percent developed, 16.4 percent forest, and 6.4 percent wetlands.

Figure 16. Basins included in study area.
6.2.2 General Water Quality Issue
In general, the main source of non-point source pollution in the study region is
agricultural production, which is a major contributor to deteriorating water quality

115

conditions. Multiple waterbodies, including the Kalamazoo, Saginaw, Clinton, St. Clair,
Detroit, and Rouge Rivers, in the study area have been classified as part of the 26 Great
Lakes Areas of Concerns (AOCs) listed in the United States (USEPA, 2009c). In order to
be classified as an AOC, the waterbody must have at least one beneficial use impairment
such as beach closings, excessive nutrient loading, or contaminated sediment (USEAPA,
2009b). There are currently 139 records listed on the Michigan Section 303(d) of the
Federal Clean Water Act for impairments including mercury, nutrient exceedances,
endosulphan, total dissolved and total suspended solids, turbidity, polychlorinated
biphenyls (PCB), polybrominated biphenyls (PBB), dichlorodiphenyltrichloroethane
(DDT), dissolved oxygen, dioxin (TCDD), pathogens, E. coli, poor fish and
macroinvertebrate ratings, impaired habitat, nuisance plant and algal growths, and
untreated sewage discharge (MSUIWR, 2010).
6.2.3 SWAT Model
Pollution transport estimation requires large amounts of complex datasets in order
to accurately represent the chemical, physical, and biological processes, in which highly
sophisticated hydrologic models excel (Powers, 2007). Therefore, the Soil and Water
Assessment Tool (SWAT) was used in this study. This is a physically based model
developed to estimate streamflow and stream reach loadings of nutrients, pesticides, and
sediments on a watershed scale (Spruill et al., 2000). The model bases its predictions on a
complex structure of routing runoff and constituents throughout a defined watershed and
is one of the most comprehensive models of its kind (Saleh and Du, 2004; Powers, 2007).
SWAT generates flow and constituent load estimates based on landuse practices in
complex watersheds with varying soils, landuse, weather conditions, and management

116

operations over long time periods (Arnold et al., 1998). ArcSWAT version 2005 is an
ArcGIS extension and graphical user interface for the SWAT model that was used in this
study. The major components of the model include weather data, soils data, slope
classifications, landuse data, nutrient and pesticide applications, and management
practices including extensive agricultural operations. The hydrologic component of the
SWAT model uses the modified SCS curve number method to determine surface runoff
during precipitation events (Saleh and Du, 2004). Soil erosion and sediment yields are
determined by the Modified Universal Soil Loss Equation (MUSLE) (Neitsch et al.,
2005). With successful model calibration, SWAT is an extremely useful tool in predicting
the environmental implications of land use change.
6.2.4 Crop Rotations and Management Operations
6.2.4.1 Rotations and Operation Schedules
Crop rotations have traditionally been used throughout the history of modern
agriculture as a method to control noxious weeds, pests and disease, and to enhance soil
fertility, physical structure, root-zone biological activity, productivity (NCSU, 2005). In
this study, crops suitable for bioenergy feedstock and their associated crop rotations that
would be likely implemented were identified. The bioenergy crops considered for
conversion from the current landuse were barely, canola, cassava, corn, corn stover,
jatropha, miscanthus, native grasses, potato, rice, rye, sorghum, soybean, sugar cane,
sugarbeet, switchgrass, and wheat. However, after determining the suitability for growth
throughout the study region the crops considered were reduced to corn, canola (fall),
cereal rye, sorghum, soybean, miscanthus, corn stover, switchgrass, and native grasses.
The rotations that were deemed as acceptable for producing bioenergy crops on a specific
landuse were based on the current landuse and its typical production capability, the
117

difficulty for farmers to acquire the planting and harvesting equipment, and the capital
investment required to invest in the operations not already included the farm’s practices.
For example, marginal land has generally low levels of soil organic matter, poor soil
structure and fertility, and tends to produce low yields for traditional crops. Therefore,
crops such as wheat and soybeans are not practical for cultivation. Additionally, due to
low soil organic matter in marginal soils, corn stover was not selected as suitable for this
landuse. Each scenario was subject to a series of reviews in order to provide the most
realistic rotations for the region of study.
The rotations were assembled using data obtained from the Michigan State
University Extension (MSUE) for the bioenergy crops used in the simulations. However,
it is important to note that the information obtained from MSUE is general for the study
area and may vary from field-to-field. Based on this information, 15 rotations were
developed, including corn-soybean-canola (Crn-Soy-Can), corn-soybean-rye (Crn-SoyRye), corn-soybean (Crn-Soy), Corn Stover (40 percent residue removal)-soybean (CrnSSoy), sorghum-soybean (Srg-Soy), continuous canola (CCan), continuous corn (CCrn),
continuous corn with stover removal (CCrnS), continuous rye (CRye), continuous
sorghum (CSrg), continuous soybean (CSoy), miscanthus (Mis), Switchgrass (Crn-SoySwg), and native grasses (NaGr). For example, the corn-soybean-canola (Crn-Soy-Can)
operation schedule is provided in table (detail information for the rest of agricultural
rotations can be found in Love and Nejadhashemi (2010a). Each rotation followed the
same format with tillage practices, fertilizer and pesticide applications, and planting and
harvesting dates developed specifically for each crop.

118

6.2.4.2 Scenarios
The crop rotations were integrated into the SWAT model and tested under four
scenarios. The scenarios were developed for the bioenergy crops, based on suitability and
practicality.
Scenario 1: The first scenario considers all land currently cultivated for row crops
and grains, hays, and seeds (i.e. corn, soybeans, wheat, hay, etc.), which
represents the majority of current agricultural land in the watershed. This scenario
is the most likely if and only if the commodity prices for bioenergy feedstock are
competitive with traditional crops and the equipment needed for their cultivation
is similar to preexisting machinery requiring justifiable capital investment.
Scenario 2: The second scenario considers only the other crops (i.e. sugarbeets,
potatoes, dry beans, fruit crops, etc.) for conversion to bioenergy crops. This
scenario represents the least amount of land considered for conversion and
represents a case in which bioenergy crops may not gain momentum on typical
agricultural land.
Scenario 3: The third scenario considers cultivation in lands that are not currently
used for agricultural productions (i.e. fallow cropland, pasture, wasteland, etc.),
which can be considered marginal land. This scenario would be typical if
bioenergy feedstock prices were not competitive with traditional commodities, but
there was still enough demand to drive farmers to turn marginal land into
production they would otherwise not use due to low productivity of row crops.
Scenario 4: The final scenario is the combination of all previous scenarios, and
represents the overall most significant landuse change from current state to the
bioenergy cropping rotations. This scenario would likely occur if demand for

119

bioenergy feedstock increased to economically justify the displacement of land
for traditional crops for livestock and human consumption.
In the above scenarios, the range converted to the bioenergy rotation varies from
0.2 percent to 54 percent of the watershed area. The data for the above analysis was
acquired from the US Department of Agriculture- National Agricultural Statistics Service
cropland data layer. In addition, the conversion of woodlands to bioenergy crop
productions was not considered as a viable option in this region and therefore was not
included in this study.
6.2.5 Model Setup
6.2.5.1 Input Data Sources
Successful use of the ArcSWAT model relies on acquiring suitable and reliable
data. The soil characteristics of the four basins were defined by the State Soil Geographic
Database (STATSGO) developed by the National Cartography and Geospatial Center
(NCGC) of the Natural Resources Conservation Service (NRCS). The topography was
based on the USGS Digital Elevation Model (DEM 90 meter resolution), which was used
in delineation of the watershed boundaries. The USGS National Hydrography Dataset
(NHD) was imposed on the DEM maps to define the stream network. Subsequently, each
watershed was divided into subbasins, which allowed characterizing different areas of the
watershed regarding potential environmental impacts of land use changes.
Each watershed was further divided into individual hydrologic response units
(HRUs) by defining thresholds based on the percentage of landuse, soil class, and slope
over the subbasin area. The slope classification was performed using four slope classes
(0-2, 2-5, 5-10 and >10 percent). For the purpose of this project, these thresholds were set
to 20, 10 and 20 percent, respectively (Winchell et al., 2009).
120

The Cropland Data Layer (2008 CDL 56 meter resolution) developed by the
USDA National Agricultural Statistics Service (NASS) was used as the landuse
representation. A table was created to relate CDL landuse values to corresponding SWAT
landuse values, which was referenced in the “lookup table” feature in SWAT. In the case
when a crop did not exist in the SWAT database, such as miscanthus, a new crop was
defined based on agronomic aspects found in literature and consultation with bioenergy
crop experts. For basin-wide climatic data, long-term precipitation (74 stations) and
temperature (56 stations) data was collected from the National Climatic Data Center
(NCDC) for a 19-year period (1990-2008). Other meteorological data (wind speed,
relative humidity, and solar radiation) required by the model was estimated using the
SWAT weather generator.
6.2.5.2 Operation of SWAT Model
The period of study for analysis was January 1, 1990 to December 31, 2008.
Model output from a period of this length accurately represents the long-term effects of
watershed behavior. This is vital when considering the effects that bioenergy cropping
systems may have on watershed characteristics, which often takes extended time to
develop. The SWAT model outputs that were compared for this study were total nitrogen
(NH4 + NO2 + NO3 + Organic N) in kilograms, total phosphorus (Mineral P + Organic
P) in kilograms, and sediment in metric tons. The output data was obtained at 4
watershed outlets, 879 subbasins, and over 5,970 km of stream networks.
6.2.5.3 Sensitivity Analysis
A sensitivity analysis identifies and ranks the parameters that are most influential
on model output (flow, sediment, and nutrients) and is useful for determining the

121

parameters for successful model calibration (Moriasi et al., 2007). In this study, each
watershed was subjected to a sensitivity analysis in which observed daily flow rate,
sediment, and nutrients load were used. In addition, management operations were applied
to provide the most accurate representation of physical and chemical interactions that
may influence model output. Although a sensitivity analysis aids in identifing parameters
useful in calibration of the model, a reasonable value and reasonable justification (sitespecific knowledge and data) for the use of each parameter is imperative to produce
results representative of true watershed behavior. The importance of using both observed
data and local management operations during the sensitivity analysis procedure was
discussed in Love and Nejadhashemi (2010a).
6.2.5.3 Calibration/Validation
Model calibration is a necessary component of hydrologic modeling to increase
confidence of model predictions (Moriasi et al., 2007). The calibration process estimates
model input parameter values by comparing the generated output for a simulation to an
observed dataset. Calibration is performed manually or automatic or combination of
manual and auto-calibration. Once a period for calibration data is set, the model is run
with the model initial parameter values. The agreement between model output and
observed data is measured using a mathematical relationship, or objective function, and
an automated optimization procedure finds the values for each parameter that optimize
the objective function (Gupta et al., 1999). Validation follows calibration to establish the
capability of the calibrated model to make accurate predictions by comparing model
output to an observed dataset of a period not used in calibration (Moriasi et al., 2007).
The calibration and validation periods are provided in Table 14.

122

In order to avoid providing biased, exceedingly positive or negative evaluation of
model performance, multiple model evaluation statistics should be used (Krause et al.,
2005). For this study, the Nash-Sutcliffe efficiency (NSE), root mean squared error
2
(RMSE), and the coefficient of determination (R ) were selected to determine model

prediction efficiency. The Nash-Sutcliffe efficiency determines conformity of the model
output to the observed dataset based on a 1:1 relationship, effectively determining
goodness-of-fit (McCuen et al., 2006). The NSE is computed with the relationship in
Equation 4:

NSE=1-

∑ቀYobs -Ypred ቁ

2

∑൫Yobs -Ymean ൯

2

(4)

where Yobs is the observed value for the evaluated component, Ypred is the model
predicted value for the evaluated component, Ymean is the mean of observed data for the
evaluated component, and n is the total number of data points. NSE ranges from - ∞ to 1.
If the value for NSE is 1, the model exactly predicts the observed values. If the value for
NSE is 0, the sum of squares of the difference between the observed and model predicted
is equal to the difference between the observed and the mean of the observed values.
Typically, NSE values greater than 0.75 are very good, greater than 0.65 are good,
greater than 0.5 are satisfactory, and values less than 0.5 are considered unsatisfactory on
monthly basis (Moriasi et al., 2007). However, statistical ratings such as NSE are projectspecific, and oftentimes are difficult to obtain satisfactory results in situations when a
complete time series of data are not available (Moriasi et al., 2007). The NSE is at a
disadvantage due to the fact that the differences between predicted and observed values

123

are squared, resulting in overestimation of model performance for large values (i.e. peak
flows) and underestimation for lower values (i.e. low flow conditions) (Krause et al.,
2005).
The root mean squared error (eq. 5) is the vertical deviation of a data point from a
fitted line, providing a value in measureable units. The RMSE has important advantages
over traditional correlation methods, such as indicating error in the measured units
(Moriasi et al., 2007).
మ

∑൫௒೚್ೞ ି௒೛ೝ೐೏ ൯
ܴ‫ ܧܵܯ‬ൌ ට
௡

(5)

where n is the total sample size. Values of RMSE are that of the predicted variable, such
as cubic meters per second for flow. This gives an indication of quantitative differences
in the regression prediction from the observed (fitted) line in the squared units of the
constituent being evaluated.
2
The coefficient of determination, R , is a statistical measure for goodness-of-fit.

In this case, it is the percent of total variation in the dependent variable explained by the
2
regression equation (Menard, 2000). R values range from 0 to 1, where values close to 1

indicate that the regression equation (model) is predicting the dependent variable with
confidence and that the dispersion of the predicted value is equal to the observed value.
One of the main drawbacks of the R

2

method is its failure to address under- or

overpredictions (Krause et al., 2005). Values greater than 0.5 are generally considered

124

2
acceptable when the model operates on a monthly basis (Santhi et al., 2001). The R is

calculated with Equation 6:

ܴ ൌ൬
ଶ

∑ሺை೔ ିை೘೐ೌ೙ ሻሺ௉೔ ି௉೘೐ೌ೙ ሻ

ඥ∑ሺை೔ ିை೘೐ೌ೙ ሻమ ඥሺ௉೔ ି௉೘೐ೌ೙ ሻమ

൰

ଶ

(6)

where O are the observed values and P are the predicted values.
Combination of manual and autocalibration were used for flow, sediment, and
nutrient calibration. Calibration should include data from a 3-5 year period during high,
low, and moderate precipitation years to effectively activate all model processes for a
diverse range of hydrologic events (Moriasi et al., 2007). Therefore, based on a
comparison with the 19-year mean of precipitation for all watersheds, the period of 20002005 was used for the calibration and validation (figure 17). However, due to periods
lacking observed data, calibration periods varied slightly between watersheds (Table 14).
Constituent samples were available only in a limited amount. Therefore LOADEST, a
FORTRAN program developed by the USGS, was used in this study to estimate monthly
sediment, total phosphorus, and total nitrogen loads for calibration (Runkel et al., 2004).
One year prior to each period listed in Table 14 was treated as a warm-up period
and was not included in the statistical comparisons. Calibration was performed on the
first half of the period with available data and validation was performed using data from
the second half of the available data period.

125

Deviation from 19-year mean (%)

15
10
5
0
-5
-10
-15
-20

2000
2001 2002 2003
2004 2005
Figure 17. Precipitation deviation from 19-year mean for study period.
Table 14. Calibration and validation periods for each basin.
Basin
040500

Sediment

040802

040900

Calibration

1/20006/2002

6/2001-9/2003

1/200212/2003

1/200012/2002

Validation

Flow

04050001

7/200212/2004
1/20006/2002
7/200212/2004
1/20006/2002
7/200212/2004
1/20006/2002
7/200212/2004

10/200312/2005
6/2001-9/2003

1/200412/2005
1/200212/2003
1/200412/2005
6/20039/2004
10/200412/2005
1/200212/2003
1/200412/2005

1/200312/2005
1/20016/2003
7/200312/2005
1/200212/2003
1/200412/2005
1/20016/2003
7/200312/2005

Calibration
Validation

Nitrogen[a]

Calibration
Validation

Phosphorus Calibration
Validation

10/200312/2005
6/2001-3/2003
4/200312/2005
6/2001-9/2003
10/200312/2005

[a] Nitrogen was evaluated with Total Nitrogen (040500), Nitrate (04050001), Total Kjeldahl
Nitrogen (040802), and Total Nitrogen (040900).

126

According to the evaluation criteria discussed before, the uncalibrated model
performed poorly using the model default values for all watersheds (table 15). However,
the combined NSE for the calibrated model performed well above satisfactory for flow,
sediment, and phosphorus, yet failed to meet the NSE of 0.5 requirement for nitrogen.
However, due to the very low availability of sampling data in each watershed and the
difficulty in obtaining suitable nitrogen calibration, these statistical values were deemed
as satisfactory for the purposes of this project. The intended use of the model is to
evaluate changes of constituent loads on a large-scale for a variety of scenarios requiring
high flexibility. Overall, calibration resulted in satisfactory statistical analysis indicating
the more than adequate prediction capabilities of the model, allowing for confidence in
model output for comparison. NSE values for flow, sediment, nitrogen, and phosphorus
ranged from 0.628 to 0.803, 0.589 to 0.843, 0.232 to 0.398, and 0.628 to 0.900,
2
respectively. Values for R followed a parallel trend, with adequate levels of correlation

with values for flow, sediment, nitrogen, and phosphorus ranging from 0.676 to 0.809,
0.590 to 0.867, 0.508 to 0.680, and 0.656 to 0.904, respectively. The values obtained for
the RMSE statistic are well within acceptable limits for the given scope of this project.
The final parameter values that were obtained during calibration are provided in table 16
and were applied prior to each subsequent model simulation. As a result of the calibration
and validation process, the model is able to effectively capture the complex physical and
chemical interactions taking place to provide confident, representative predictions of
future pollution generation due to the large-scale application of bioenergy cropping
systems.

127

Table 15. Results of statistical analysis for calibration and validation periods.
Parameter
Flow

Period

Statistic

Uncalibrated NSE
2
R
RMSE
Calibrated

Validated

11.82

0.718
0.753

0.652
0.629

0.628
0.730

0.541
0.674

[b] 35.20

45.63

47.53

7.381

0.843
0.838

0.603
0.769

0.714
0.716

0.695
0.702

[b] 30.24

31.24

60.38

6.205

0.803
0.809

0.628
0.692

0.698
0.713

0.689
0.676

[b] 35.85

39.11

54.34

5.950

-0.717

-18.763

0.388

0.046

0.367

0.430

0.407

0.475

[c] 563.3

1003

510.7

266.9

0.782

0.621

0.544

0.332

0.809

0.601

0.668

0.488

[c] 201.7

149.6

245.9

121.1

0.878
0.904

0.560
0.565

0.702
0.766

0.785
0.955

[c] 183.6

140.5

629.9

175.0

0.843

0.589

0.693

0.763

0.867

0.590

0.743

0.755

[c] 193.0

145.1

478.2

150.5

-15.91

-0.159

-174.5

Uncalibrated NSE
2
R
RMSE
Calibrated

NSE
2
R
RMSE

Validated

NSE
2
R
RMSE

Combined

NSE
2
R
RMSE

Nitrogen

52.98

NSE
2
R
RMSE

Sediment

51.22

NSE
2
R
RMSE

Combined

[b] 61.88

NSE
2
R
RMSE

[a]

Basin
040500 04050001 040802 040900
0.229
0.098
0.472
-0.136
0.571
0.579
0.544
0.655

Uncalibrated NSE

-1.157

128

Table 15 (cont’d)
2

0.238

0.391

0.273

0.309

[d] 39986

469.9

6615.5

34929

0.121
0.331

0.278
0.634

0.445
0.552

-0.044
0.534

[d] 25608

14356

7917

2066

0.596
0.740

0.189
0.627

-0.603
0.750

0.318
0.654

[d] 18199

0.398

7730
0.320

5281
0.315

2633
0.232

0.508

0.549

0.680

0.619

Uncalibrated RMSE [d] 22277
NSE
-5.047

11530
-3.960

6729
0.146

2367
-0.054

2

0.553

0.114

0.346

0.222

[d] 2994

0.793

31711
0.624

1089
0.581

402.4
0.450

0.853

0.659

0.635

0.471

[d] 487.9

0.947

502.5
0.632

558.3
0.712

184.3
0.964

0.959

0.643

0.746

0.948

[d] 368.7

0.900

384.1
0.628

1073
0.699

90.6
0.839

0.904

0.656

0.734

0.813

[d] 433.5

447.3

855.6

145.2

R
Calibrated

RMSE
NSE
2
R

Validated

RMSE
NSE
2
R

Combined

RMSE
NSE
R

Phosphorus

R
Calibrated

RMSE
NSE
R

Validated

2

RMSE
NSE
R

Combined

2

2

RMSE
NSE
R

2

RMSE

[a] Nitrogen was evaluated with Total Nitrogen (040500), Nitrate (04050001), Total
Kjeldahl Nitrogen (040802), and Total Nitrogen (040900).
-1
[b] The units for flow RMSE are meters 3 s .

129

-1
[c] The units for sediment RMSE are tons day .
-1
[d] The units for nutrients RMSE are kilograms day .

Table 16. Final parameter values from calibration.
Basin
04050001
040802

Parameter Change Method

040500

040900

Alpha_Bf
Biomix
Canmx

Replace Value
Replace Value
Replace Value

0.32509
-

0.58501
-

0.69280
-

0.071630
0.1
0.27201

Ch_K2
Ch_N2

Replace Value
Replace Value

115.94
0.075737

44.735
0.44107

87.931
0.58573

25.828
0.081347

Cn2
Esco
Nperco
N_Updis

Multiply By
Replace Value
Replace Value
Replace Value

1.20925
0.42962
1.0
-

1.20038
0.33532
-

1.13267
0.34490
-

1.15602
0.37760
0.5
10

Rchrg_Dp
Smtmp
Sol_Awc
Spcon

Replace Value
Multiply By
Multiply By
Replace Value

Surlag
Timp
Usle_P

Replace Value
Replace Value
Replace Value

0.31961
0.33342
0.19753
0.87389
1.0079364
1.29317
1.079115
0.8747
0.00055413 0.0044107 0.0074351 0.0091561
8.7156
0.37496
0.35780

9.7813
0.58155
0.23878

6.1303
0.16357
0.63399

9.7684
0.39598
0.95221

6.3 RESULTS AND DISCUSSION
6.3.1 Basin-wide Impact of Bioenergy Cropping Rotations
Altering the landscape with the introduction of agriculturally-intensive bioenergy
crops is expected to result in changed pollutant loadings to waterbodies. In this study,
average annual sediment, total nitrogen, and total phosphorus loads were obtained for
each watershed at the outlet for the study period (1990-2008). In order to examine the
impacts of landuse conversion on water quality, the model results obtained from the four

130

cropping scenarios (Scenario 1 to 4) and throughout the study area were evaluated
(Figure 18, Figure 19, and Figure 20). The statistical analysis provided in this section was
performed in STATISTICA data analysis software (StatSoft, Inc., 2010). This allowed
for a large-scale assessment encompassing all watersheds and all possible landuse
scenarios, providing insight to the potential extreme pollutant load variation. Traditional,
intensive crops such as corn, sorghum, canola and soybean experience increases in
sediment and nitrogen load, yet in certain cases have the potential to reduce phosphorus
loads. In general, sorghum is not a viable option for cultivation in regions with concerns
of sedimentation and phosphorus loadings, yet this crop has nearly zero median impact
on nitrogen loads. In addition, the perennial grass species (miscanthus, native grasses,
and switchgrass) extensively mitigate sediment and phosphorus, yet have the potential to
increase nitrogen slightly. By adding rye, a close-grown annual grass, to a rotation (i.e.
into a corn-soybean rotation) pollutant loadings experience an overall decrease. When rye
is grown in a continuous rotation, nitrogen and phosphorus loads experience a decrease,
while sediment is nearly zero impact. It is important to note that in general, all cropping
rotations have both maximum and minimum values of increased and decreased pollutant
loadings, respectively. This is expected due to evaluating an extremely large region with
diverse climatological, geological, and agricultural characteristics.

However, some

rotations exhibit the magnitude of this behavior in different amounts. For example,
miscanthus and switchgrass have the least sediment load variability (141 and 142 percent
range, respectively), whereas continuous sorghum has the highest variability (321 percent
range) as well as the highest maximum change from the base scenario (312 percent).
Ironically, continuous sorghum has the smallest range of total nitrogen load, with

131

continuous corn stover having the largest span of values. The perennial grass species
have low variability of sediment and phosphorus load, further supporting their use in
watersheds with loading issues regarding these pollutants. In general, perennial grass
species are most suitable for large-scale implementation in this study area, whereas
traditional intensive row crops should be implemented with caution on such a broad
scale. The highest ranges for sediment, total nitrogen, and total phosphorus, as expected,
were the result of the row crop rotations continuous sorghum, continuous corn stover, and
continuous canola, respectively. The large ranges observed show the variability that
bioenergy crop rotations are likely to exhibit. This is largely in part to their unsuitability
in various locations in such a large study area, which also has large variation in
climatological factors. Small ranges show how certain crops are less dependent on these
drivers.

132

Figure 18. 19-year annual average basin-wide sediment load at the watershed outlet.

133

Figure 19. 19-year annual average basin-wide total nitrogen load at the watershed outlet.

Figure 20. 19-year annual average basin-wide total phosphorus load at the watershed.

134

6.3.2 Basin-level Priority Areas for Targeting Conservation Efforts
In order to effectively target agricultural areas for the implementation of
conservation efforts and Best Management Practices (BMPs), and to determine the areas
in greatest need for further research and investigation, priority maps were created for the
entire basin. Three classes (low, medium, and high) of priority concerns were formed by
dividing the study area based on what was essentially a quantile classification for each
constituent. As it was discussed earlier, Scenario 4 of the landuse conversion represents
the most extreme landuse conversion scenario among the studied scenarios. Therefore, in
this section of study, the impacts of landuse conversion on both priority concerns areas
and streams were evaluated by comparing Base Scenario and Scenario 4. The overall
changes in the length of river network priority streams, based on stream concentration, at
different locations of the study area are provided in Table 17 with the corresponding
percentages change from low, medium, and high compared to the base scenario.
Identifying, these streams is important since changes in stream concentration can directly
or indirectly endanger aquatic organisms living in those streams. Figure 21 provides a
visualization on priority concerns streams and the impacts of intensive bioenergy crop
expansion. In this figure, the stream concentration under Base scenario is compared with
continuous corn (Scenario 4), and switchgrass (Scenario 4) rotations. This indicates that
converting all agricultural land to a continuous corn rotation will result in a substantial
increase in stream reaches classified as high priority as well as nearly all low priority
streams shifting to medium priority for all constituents. These findings support those
discussed in the previous section (3.1) for pollutant loads found at the main channel
outlet under a continuous corn rotation (for all four scenarios). Switchgrass experiences a
mixed change of sediment concentration, with some streams being reduced to low or

135

medium priorities, whereas others experience a jump to the high priority. Also with
switchgrass, a substantial amount of stream reaches (34 percent) change to medium or
high priorities for total nitrogen concentration. However, when total phosphorus
concentration is considered, there is a general reduction in high priority streams. In
general, the perennial grasses, although mixed benefits are present, are more suitable for
implementation than intensive annual bioenergy crops. Miscanthus had the least amount
of high priority streams of 7.1 and -23.1 percent change from the base scenario for
nitrogen and phosphorus, respectively (Table 17). By adding rye into a corn-soybean
rotation, the amount of high priority streams is decreased by 10.8, 6.4, and 20.9 percent
for sediment, total nitrogen, and total phosphorus, respectively. The sorghum-soy rotation
is suitable for implementation in regions where stream concentration of sediment and
phosphorus are of concern, as it reduced the most high priority streams by 21.8 percent
over base, but not in regions with high preexisting levels of nitrogen. The agriculturally
intensive row crops had the highest increases of high priority streams for sediment,
nitrogen, and phosphorus of 39.1 percent (continuous sorghum), 67.9 percent (continuous
corn stover), and 25.5 percent (continuous corn stover), respectively.
As it was discussed earlier, identifying the area of concerns in a watershed is the
first step in developing the control strategy to mitigate negative impacts of agricultural
expansions. The changes in priority areas at subbasin level are provided in Figure 22 with
the corresponding percentages provided in Table 18. The results for this analysis are
comparable to the stream, yet have some subtle differences. In general, when a
continuous corn rotation is applied to all agricultural land, there is a considerable increase
of medium and high priority areas by 44.7, 45.7, and 42.1 percent for sediment, total

136

nitrogen, and total phosphorus load, respectively. However, some subbasins that were
originally high priority were reduced to medium priority, resulting in a shift of priority
area focus to other sensitive regions. Switchgrass appears to have a mitigation effect on
sediment and total phosphorus loads (Figure 22). Again, a small number of previously
low priority subbasins have been shifted to medium or high priority for sediment load.
Meanwhile, high priority areas were reduced by 8.9 percent due to large-scale
switchgrass production implementation, although the amount of high priority areas is
reduced for total nitrogen (5.1 percent), the number of medium priority areas is increased
by 29.2 percent over the base scenario. Corn stover removal has the potential to increase
the amount of high priority streams and subbasins, both over the base scenario and over
continuous corn and corn soybean rotations. As was observed in the priority streams
previously mentioned, the agriculturally intensive row crops had the highest increases of
high priority areas for sediment, nitrogen, and phosphorus of 15.8 percent (continuous
sorghum), 26.0 percent (continuous corn stover), and 3.35 percent (continuous soybean),
respectively. However, adding a close-grown grass specie (i.e. rye) into the rotation is
advantageous. Continuous rye decreases the amount of high priority areas by 9.0, 6.0 and
8.2 percent for sediment, total nitrogen, and total phosphorus, respectively. The sorghumsoybean rotation is in general highly suitable for large-scale implementation based on this
analysis. The perennial grass species are overall very suitable for large-scale
implementation. Miscanthus reduced high priority areas for total nitrogen and total
phosphorus by 7.7 and 8.4 percent, respectively. This analysis is merely a tool to help
decision makers identify the possible locations that may be of high concern to monitor in

137

the event of large-scale expansion of bioenergy cropping rotations as discussed in this
paper.

Figure 21. Priority streams for base scenario sediment, nitrogen and phosphorus (a,b,c),
continuous corn sediment, nitrogen and phosphorus (d,e,f), and switchgrass sediment,
-1
nitrogen and phosphorus (g,h,i) (in µg L )

138

Figure 22. Priority areas for base scenario sediment, nitrogen and phosphorus (a,b,c),
continuous corn sediment, nitrogen and phosphorus (d,e,f), and switchgrass sediment,
-1
nitrogen and phosphorus (g,h,i) (in kg ha )

139

Table 17. Length of priority stream change from base (km) for Scenario 4 (the values with gray background represent the percent
changes from the Base or current landcover scenario).
Low
-1944
Continuous Canola
-32.56
-2438
Continuous Corn
-40.83
Continuous Corn stover -2571
-43.06
-2281
Continuous Rye
-38.20
-2628
Continuous Sorghum
-44.02
-2588
Continuous Soybean
-43.34
-2588
Corn Soy
-43.34
-2399
Corn Soy Rye
-40.18
-2588
Corn stover Soy
-43.34
-2623
Sorghum Soy
32.40
1934
Miscanthus
30.93
1846
Native Grass
-43.94
-1034
Switchgrass
-17.31

Sediment
Medium
998
16.72
836
14.00
749
12.54
1218
20.41
291
4.87
425
7.13
556
9.32
1010
16.92
602
10.08
318
-10.58
-632
-10.40
-621
5.32
915
15.33

High
946
15.84
1602
26.83
1822
30.53
1062
17.79
2338
39.15
2162
36.22
2031
34.03
1389
23.26
1985
33.26
2306
-21.82
-1303
-20.53
-1226
38.62
118
1.98

Low
-2437
-40.82
-2321
-38.88
-2472
-41.40
-2120
-35.52
-2242
-37.55
-2418
-40.50
-2389
-40.01
-2373
-39.74
-2457
-41.15
-2343
-38.11
-2275
-22.20
-1326
-39.25
-2030
-34.00

140

Nitrogen
Medium
-810
-13.56
-263
-4.40
-1584
-26.52
259
4.34
-102
-1.70
-936
-15.67
-732
-12.26
-366
-6.12
-966
-16.18
-525
-5.74
-342
15.10
901
-8.79
264
4.43

High
3247
54.38
2584
43.28
4055
67.93
1861
31.18
2343
39.25
3354
56.17
3121
52.28
2738
45.87
3422
57.33
2868
43.85
2618
7.10
424
48.04
1766
29.57

Phosphorus
Low Medium High
-1638
981
658
-27.44 16.43
11.02
-1820
894
926
-30.48 14.97
15.52
-2323
801
1521
-38.90 13.42
25.48
111
333
-443
1.85
5.57
-7.42
-2199
706
1493
-36.83 11.82
25.00
-2209
780
1429
-37.00 13.06
23.94
-2155
755
1400
-36.09 12.65
23.44
-1116
964
152
-18.69 16.15
2.55
-2266
774
1492
-37.95 12.96
25.00
-2080
852
1227
11.13
3.75
-14.88
665
224
-889
32.09
-9.00
-23.09
1916
-537
-1379
-34.84 14.28
20.56
724
-138
-586
12.12
-2.31
-9.81

2
Table 18. Area of priority subbasins change from base (km ) for Scenario 4 (the values with gray background represent the percent
changes from the Base or current landcover scenario).

Low
-16529
Continuous Canola
-30.97
-23877
Continuous Corn
-44.74
Continuous Corn stover -18135
-33.98
-13255
Continuous Rye
-24.84
-22896
Continuous Sorghum
-42.91
-21179
Continuous Soybean
-39.69
-19561
Corn Soy
-36.66
-15036
Corn Soy Rye
-28.18
-19108
Corn stover Soy
-35.81
-22257
Sorghum Soy
22.21
11854
Miscanthus
20.86
11133
Native Grass
-41.71
-369
Switchgrass
-0.69

Sediment
Medium
16375
30.69
21025
39.40
15976
29.94
14515
27.20
14455
27.09
15560
29.16
15479
29.01
15780
29.57
15571
29.18
14566
-12.44
-6636
-11.08
-5915
27.30
5121
9.60

High
154
0.29
2852
5.34
2159
4.05
-1260
-2.36
8441
15.82
5619
10.53
4082
7.65
-744
-1.39
3537
6.63
7691
-9.78
-5218
-9.78
-5218
14.41
-4752
-8.90

Low
-24645
-46.18
-24371
-45.67
-29693
-55.64
-14262
-26.73
-22340
-41.86
-27775
-52.05
-26973
-50.55
-22675
-42.49
-28523
-53.45
-24734
-35.94
-19177
-7.50
-4001
-46.35
-12866
-24.11

141

Nitrogen
Medium
22346
41.87
21041
39.43
15821
29.65
16986
31.83
21553
40.39
22371
41.92
23153
43.39
21616
40.51
20169
37.80
23007
35.88
19149
15.16
8089
43.11
15590
29.21

High
2299
4.31
3330
6.24
13872
26.00
-2724
-5.10
787
1.47
5404
10.13
3820
7.16
1059
1.98
8354
15.65
1727
0.05
28
-7.66
-4088
3.24
-2724
-5.10

Phosphorus
Low Medium
-18995 20906
-35.60
39.18
-22485 22062
-42.14
41.34
-24788 23815
-46.45
44.63
4511
0
8.45
0.00
-23627 21887
-44.28
41.01
-23447 21657
-43.94
40.58
-22577 21765
-42.31
40.79
-9455
13014
-17.72
24.39
-23593 22120
-44.21
41.45
-21842 21177
25.38
-16.93
13543
-9032
29.19
-20.74
15579 -11068
-40.93
39.68
7672
-3161
14.38
-5.92

High
-1911
-3.58
423
0.79
974
1.82
-4511
-8.45
1741
3.26
1790
3.35
812
1.52
-3559
-6.67
1473
2.76
665
-8.45
-4511
-8.45
-4511
1.25
-4511
-8.45

6.3.3 Suitability of Bioenergy Cropping Rotations on Different Scenarios
A major intent of this paper is to provide decision makers insight as to the
changes of pollutant loadings that could be expected from changes to bioenergy crops. In
order to accommodate the growing demand for available land to produce bioenergy crops
sustainably, the assessment of environmental implications is imperative. As previously
mentioned, a total of 4 landuse scenarios were developed in this study, resulting in a
broad range of land conversion amounts. The scenarios are as follows: Scenario 1
considers all land currently cultivated for row crops and grains, hays, and seeds (i.e. corn,
soybeans, wheat, hay, etc.) for conversion to bioenergy crops, Scenario 2 considers only
the “other crops” land (i.e. sugarbeets, potatoes, dry beans, fruit crops, etc.), Scenario 3
considers farmland uses and other land (i.e. fallow cropland, pasture, wasteland, etc.),
which in all practical purposes may be deemed as marginal land, and Scenario 4 is the
combination of all previous scenarios, and represents the overall most significant landuse
change. Table 19 provides a comparison of all rotations based on their contribution to
annual average sediment, nitrogen, and phosphorus loads at the study area. By arranging
the data in this manner, direct evaluation of the potential environmental impacts for each
of the four scenarios is made apparent.
By examining Table 19, there are a number of points that need further discussion.
As expected the perennial grass species reduced sediment, nitrogen, and phosphorus
loadings in Scenario 1 with the exception of miscanthus, which slightly increased total
nitrogen by 4.62 percent. Adding canola to the corn-soybean rotation in Scenario 1
increased all constituents, whereas adding rye has the opposite effect, which is also true
of Scenario 4. This is expected as adding rye, a close-grown annual cereal grass, into the
rotation postpones the cultivation of corn by a year every 3 years. This provides
142

vegetative cover during vulnerable winter and early spring months with extensive growth
occurring during late spring and early summer, a period usually experiencing large
amounts of rainfall.
Based on the data from this study, it is not recommended to convert land
previously cultivated with “other crops” (Scenario 2) to any bioenergy rotation in areas
with preexisting high nitrogen levels. The same argument can be made for converting
marginal land (Scenario 3) as all rotations on this landuse increases nitrogen levels and
are not suitable replacements for the current landcover here, though miscanthus and
native grasses are suitable when nitrogen levels are of less concern. Only land that is
currently cultivated for row crops (Scenario 1) experiences any form of nitrogen
mitigation, in which continuous rye, sorghum-soybean, native grass, and switchgrass
decrease total nitrogen from the base scenario. The removal of corn stover is also not
recommended as it increases the levels of total nitrogen and total phosphorus loadings at
the watershed outlets when compared to the base scenario and the continuous corn
rotation.
The perennial grass rotations (switchgrass, miscanthus, native grasses) are the
saviors of bioenergy cropping rotations in general. They reduce sediment up to 86.76
percent except when grown on marginal land, which experiences a 17.87 percent increase
of sediment when switchgrass is applied. This is most likely due to this rotation having
corn and soybeans during the first two years of each rotation cycle. Love and
Nejadhashemi (2010a) observed drastic increases of pollutants during these years when
evaluated on a monthly time step. However, during periods the following years of
switchgrass as the landcover, a 93 percent reduction in sediment occurred. Therefore, it is

143

suggested that switchgrass be considered as a suitable bioenergy crop in this case, as long
as conservation tillage or no-till management operations be considered for corn and
soybeans in this rotation. In general, the perennial grasses and rye decrease phosphorus
and the intensive crops such as corn and sorghum increase phosphorus loads.

144

Table 19. Total combined pollutant load of all watersheds.
Sub-scenario

Base
Scen1_Continuous Canola
Scen1_Continuous Corn
Scen1_Continuous CornStover
Scen1_Continuous Rye
Scen1_Continuous Sorghum
Scen1_Continuous Soybean
Scen1_Corn Soy
Scen1_Corn Soy Canola
Scen1_Corn Soy Rye
Scen1_CornStover Soy
Scen1_Sorghum Soy
Scen1_Miscanthus
Scen1_Native Grass
Scen1_Switchgrass
Scen2_Continuous CornStover
Scen2_CornStover Soy
Scen2_Miscanthus
Scen2_Native Grass
Scen2_Switchgrass
Scen3_Continuous Canola
Scen3_Continuous Corn
Scen3_Continuous Rye
Scen3_Corn SoyCanola
Scen3_Corn Soy Rye
Scen3_Native Grass

Sediment
Total Nitrogen
Total Phosphorus
Load
% Change
% Change
% Change
(tons)
from Base Load (kg) from Base Load (kg) from Base
473470
- 20062390
1606130
474950
0.31 21992080
9.62
1212830
-24.49
569040
20.19 21163620
5.49
1370890
-14.65
552470
16.69 27515330
37.15
1410920
-12.15
399790
-15.56 17053860
-15.00
853790
-46.84
662380
39.90 18350560
-8.53
1421670
-11.48
616680
30.25 22616070
12.73
1402420
-12.68
590170
24.65 21549760
7.41
1374450
-14.42
597820
26.26 23932710
19.29
1460160
-9.09
458230
-3.22 20540200
2.38
1070030
-33.38
577190
21.91 23786210
18.56
1393350
-13.25
654390
38.21 19910470
-0.76
1345990
-16.20
91110
-80.76 20989570
4.62
802720
-50.02
119280
-74.81 15993390
-20.28
677780
-57.80
282200
-40.40 17579150
-12.38
852380
-46.93
436020
-7.91 28071700
39.92
1175200
-26.83
438320
-7.42 27960600
39.37
1174410
-26.88
421080
-11.07 27893180
39.03
1157450
-27.94
421400
-11.00 27850050
38.82
1158060
-27.90
426690
-9.88 27851420
38.82
1161520
-27.68
674100
42.37 40197350
100.36
1580440
-1.60
723000
52.70 38603750
92.42
1668360
3.87
629560
32.97 36226820
80.57
1382760
-13.91
741700
56.65 41356770
106.14
1710840
6.52
657240
38.81 38558350
92.19
1493810
-6.99
475010
0.33 35592530
77.41
1282170
-20.17

145

Table 19 (cont’d)
Scen3_Miscanthus
Scen3_Switchgrass
Scen4_Continuous Canola
Scen4_Continuous Corn
Scen4_Continuous CornStover
Scen4_Continuous Rye
Scen4_Continuous Sorghum
Scen4_Continuous Soybean
Scen4_Corn Soy Rye
Scen4_Corn Soy
Scen4_CornStover Soy
Scen4_Sorghum Soy
Scen4_Miscanthus
Scen4_Native Grass
Scen4_Switchgrass

462470
558090
649600
772300
755900
538090
887700
824700
620860
794200
780300
872500
91850
135390
366360

-2.32
17.87
37.20
63.11
59.65
13.65
87.49
74.18
31.13
67.74
64.80
84.28
-80.60
-71.40
-22.62

39169110
36533490
34461140
32483590
44351540
25395310
27805260
35339620
31481110
33302840
37242850
30532740
32404910
23694470
26310980

146

95.24
82.10
71.77
61.91
121.07
26.58
38.59
76.15
56.92
66.00
85.64
52.19
61.52
18.10
31.15

1344380
1376500
1674150
1963300
2038100
1010140
2047510
2016760
1406110
1956950
2002750
1896260
913240
679540
994880

-16.30
-14.30
4.24
22.24
26.90
-37.11
27.48
25.57
-12.45
21.84
24.69
18.06
-43.14
-57.69
-38.06

6.3.4 Statistical Significance of Bioenergy Cropping Rotations Changes from Base
A key question is whether the changes in pollution load are statistically different
from the current landuse scenario. To accomplish this, t-tests were performed to
determine the statistical significance levels. A p-value of 0.05 or less rejects the
hypothesis, in which there are no significant differences in pollution generation between
the bioenergy crop rotation and the current landuse scenario (Base Scenario). The results
from the t-tests are provided in Table 20. A cell highlighted in green denotes a significant
decrease in pollutant load from the base scenario, where a red cell denotes an increase in
pollutant load. Based on the analysis, it is apparent that in general perennial grass species
significantly reduce sediment. The exception to this is when these crops are cultivated on
marginal lands (Scenario 3). As discussed previously, only bioenergy crops grown on
land typically devoted to cultivating row crops (Scenario 1) experiences any significant
decrease in total nitrogen load. It is recommended that bioenergy crops be placed
sparingly in areas that have preexisting high levels of nitrogen load to streams. With the
exception of traditional, intensive row crops cultivated on marginal land (Scenario 3) and
all agricultural land (Scenario 4), the majority of bioenergy crops significantly reduce
total phosphorus loads.
Table 20. Significant difference (p-value) between bioenergy rotation and base scenario.
Sub-scenario
Scen1_Continuous Canola
Scen1_Continuous Corn
Scen1_Continuous CornStover
Scen1_Continuous Rye
Scen1_Continuous Sorghum
Scen1_Continuous Soybean
Scen1_Corn Soy

[a]

p-value for given constituent
Sediment (tons)
Total N (kg)
Total P (kg)
0.96590
0.00000
0.00000

0.00156
0.02167
0.00000

0.00009
0.00000
0.00000

0.09421
0.00000
0.00000
0.00000

0.00029
0.00318
0.00011
0.00286

0.00000
0.00001
0.00000
0.00000

147

Table 20 (cont’d)
Scen1_Corn Soy Canola
Scen1_Corn Soy Rye

0.00000
0.66593

0.00000
0.32899

0.00033
0.00000

Scen1_CornStover Soy
Scen1_Sorghum Soy
Scen1_Miscanthus
Scen1_Native Grass

0.00000
0.00000
0.00000
0.00000

0.00000
0.00659
0.00000
0.73903

0.00000
0.00000
0.00000
0.00000

Scen1_Switchgrass
Scen2_Continuous CornStover
Scen2_CornStover Soy
Scen2_Miscanthus

0.00024
0.00000
0.00000
0.00000

0.00187
0.00000
0.00000
0.00000

0.00000
0.00000
0.00000
0.00000

Scen2_Native Grass

0.00000
0.00000
0.00000

0.00000
0.00000
0.00000

0.00000
0.00000
0.65024

0.00000
0.00004
0.00000
0.00000

0.00000
0.00000
0.00000
0.00000

0.05087
0.00115
0.01173
0.04407

0.80892
0.12460
0.00163
0.00228

0.00000
0.00000
0.00000
0.00000

0.00001
0.00001
0.00112
0.60469

0.00000
0.00000
0.26324
0.00000

0.00000
0.00000
0.00303
0.00003

0.00000
0.00000
0.00007
0.00000

0.00000
0.00828
0.00000
0.00000
0.00000

0.00000
0.00000
0.00000
0.00000
0.00000

0.00000
0.08777
0.00000
0.00000
0.00000

0.00000
0.00000
0.07133

0.02608
0.00000
0.00171

0.00000
0.00000
0.00008

Scen2_Switchgrass
Scen3_Continuous Canola
Scen3_Continuous Corn
Scen3_Continuous Rye
Scen3_Corn SoyCanola
Scen3_Corn Soy Rye
Scen3_Native Grass
Scen3_Miscanthus
Scen3_Switchgrass
Scen4_Continuous Canola
Scen4_Continuous Corn
Scen4_Continuous CornStover
Scen4_Continuous Rye
Scen4_Continuous Sorghum
Scen4_Continuous Soybean
Scen4_Corn Soy Rye
Scen4_Corn Soy
Scen4_CornStover Soy
Scen4_Sorghum Soy
Scen4_Miscanthus
Scen4_Native Grass

Scen4_Switchgrass
[a] P-values less than 0.05 indicate significant difference in change from base

148

[b] Cells highlighted in Green denote a significant decrease in pollutant load over base
[c] Cells highlighted in Red denote a significant increase in pollutant load over base

6.4 CONCLUSION
Instability of the global energy sector has led to recent increases in the demand for
alternatives to fossil fuels to meet the energy and transportation needs of the nation. In
addition, domestic renewable energy sources have the potential to reduce dependence on
foreign resources, mitigate climate change, diversify domestic energy sources, and allow
for economic growth and rural development (Scheffran and BenDor, 2009). Biomass is
expected to play a key role in developing alternative fuel sources, but only accounts for 7
percent of total global energy supplies (de Fraiture et al., 2008). Therefore, a global shift
in biomass production is expected to occur. The challenges we face in transitioning to a
system that is both economical feasible and low-impact on the environment are
ubiquitous. In order to evaluate the environmental consequences of such expansion of
bioenergy crops, specifically on water quality, a large-scale assessment using models is
imperative.
The Soil and Water Assessment Tool (SWAT) was used in this study to model the
effects of bioenergy cropping expansion on water quality based on four landuse scenarios
and 15 bioenergy crop rotations for four large watersheds, totaling 244 model runs. The
output data that was evaluated in this study was sediment, total nitrogen, and total
phosphorus for a 19 year (1990-2008) period on an annual basis. The watersheds
included in this study are intensively cultivated for agricultural production and all flow
into the Great Lakes.

149

The results obtained from this study were arranged in three sections and can be
summarized in the following:
6.3.1 Basin-wide Impact of Bioenergy Cropping Rotations: Traditional, intensive
crops such as corn, sorghum, canola and soybean experience increases in
sediment and nitrogen load, yet in certain cases have the potential to reduce
phosphorus loads. Miscanthus and switchgrass have the least sediment load
variability (141 and 142 percent range, respectively), whereas continuous
sorghum has the highest variability (321 percent range) as well as the highest
maximum change from the base scenario (312.3 percent). In general, sorghum is
not a viable option for cultivation in regions with concerns of sedimentation and
phosphorus loadings, yet has nearly zero median impact on nitrogen loads. In
addition, the perennial grass species (miscanthus, native grasses, and switchgrass)
extensively mitigate phosphorus, yet have the potential to increase nitrogen
slightly.
6.3.2 Basin-level Priority Areas for Targeting Conservation Efforts: It is evident
that converting all agricultural land (Scenario 4) to a continuous corn rotation will
result in a considerable increase of medium and high priority areas by 44.74,
45.67, and 42.13 percent for sediment, total nitrogen, and total phosphorus load,
respectively. Also with switchgrass, a substantial amount of stream reaches (34
percent) change to medium or high priorities for total nitrogen concentration. The
agriculturally intensive row crops had the highest increases of high priority
streams for sediment, nitrogen, and phosphorus of 39.15 percent (continuous
sorghum), 67.93 percent (continuous corn stover), and 25.48 percent (continuous

150

corn stover), respectively. As was observed in the priority streamss, the
agriculturally intensive row crops had the highest increases of high priority areas
for sediment, nitrogen, and phosphorus of 15.82 percent (continuous sorghum),
26.0 percent (continuous corn stover), and 3.35 percent (continuous soybean),
respectively. Miscanthus reduced high priority areas for total nitrogen and total
phosphorus by 7.66 and 8.45 percent, respectively. In general, the perennial
grasses, although mixed benefits are present, are more suitable for implementation
than intensive annual bioenergy crops.
6.3.3 Suitability of Bioenergy Cropping Rotations on Different Scenarios: As
expected the perennial grass species reduced sediment, nitrogen, and phosphorus
loadings when applied on row-crop land (Scenario 1) with the exception of
miscanthus, which slightly increased total nitrogen by 4.62 percent. Adding
canola to the corn-soybean rotation in Scenario 1 increased all constituents,
whereas adding rye has the opposite effect, which is also true of Scenario 4.
Based on the data from this study, it is not recommended to convert land
previously cultivated with “other crops” (Scenario 2) to any bioenergy rotation in
areas with preexisting high nitrogen levels. The same argument can be made for
converting marginal land (Scenario 3) in which all rotations increased nitrogen
levels and are not suitable replacements for the current landcover here, though
miscanthus and native grasses are suitable when nitrogen levels are of less
concern. Only land that is currently cultivated for row crops (Scenario 1)
experiences any form of nitrogen load reduction, in which continuous rye,
sorghum-soybean, native grass, and switchgrass decrease total nitrogen from the

151

base scenario. The removal of corn stover is also not recommended as it increases
the levels of total nitrogen and total phosphorus loadings at the watershed outlets
when compared to the base scenario and the continuous corn rotation. The
perennial grass rotations reduce sediment up to 86.76 percent except when grown
on marginal land, which experiences a 17.87 percent increase of sediment when
switchgrass is applied.
6.3.4 Statistical Significance of Bioenergy Cropping Rotations Changes from
Base: Based on a series of t-test analyses, it is apparent that in general perennial
grass species significantly reduce sediment. The exception to this is when these
crops are cultivated on marginal lands (Scenario 3). As discussed previously, only
bioenergy crops grown on land typically devoted to cultivating row crops
(Scenario 1) experiences any significant decrease in total nitrogen load. With the
exception of traditional, intensive row crops cultivated on marginal land (Scenario
3) and all agricultural land (Scenario 4), the majority of bioenergy crops
significantly reduce total phosphorus loads.
Based on the results aforementioned, it is evident that the suitability of bioenergy crops
for large-scale implementation is dependent upon the application in question, specifically
the landuse it is intended to replace. Additionally, rotations that are deemed unsuitable
for cultivation may be tailored to meet the requirements of a specific landuse. For
example, corn that is included in the switchgrass rotation may have to include
conservation practices such as conservation tillage or no-till and reduced fertilizer
application rates in order to reduce sedimentation and nutrient transport to surface waters.
In order to expand our understanding of large-scale bioenergy crop expansion, it is

152

suggested that additional studies in different climatological and physiographic regions are
performed to validate the findings discussed in this paper.

Acknowledgements
The authors would like to acknowledge Mr. Sean Woznicki and Dr. Fred BakkerArkema for their added project support, and Dr. Kurt Thelen, Dr. Wesley Everman, Mr.
Dennis Pennington, Dr. Scott Swinton, and Ms. Laura James for their generous
contribution of information.

153

7. EFFECTS ON AQUATIC AND HUMAN HEALTH DUE TO LARGE SCALE
BIOENERGY CROP EXPANSION
a
b
a,*
Brad J. Love , Matt D. Einheuser , A. Pouyan Nejadhashemi

a Department of Biosystems and Agricultural Engineering, Michigan State University,
216 Farrall Hall, East Lansing, MI 48824 USA
b Department of Crop and Soil Sciences, Michigan State University, 159 Plant and Soil
Science Building, East Lansing, MI 48824 USA

* Author who correspondence should be addressed. Tel.: (517) 432-7653 fax: 517-4322892. Email address: pouyan@msu.edu (A. Pouyan Nejadhashemi)
Abstract. The environmental consequences of a shifting agricultural production system
are not well known, especially in regards to aquatic ecosystem and human health. In this
study, the environmental impacts of large scale bioenergy crops were evaluated using the
Soil and Water Assessment Tool (SWAT). Daily pesticide concentration data for a study
2
area consisting of four large watersheds located in Michigan (totaling 53,358 km ) was

estimated over seven year period (2000-2006). Atrazine, bromoxynil, glyphosate,
metolachlor, pendimethalin, sethoxydim, triflualin, and 2,4-D model output was used to
predict the possible long-term implications that large-scale bioenergy crop expansion
may have on the bluegill (Lepomis macrochirus) and humans. Threshold toxicity levels
were obtained for the bluegill and for human consumption for all pesticides being
evaluated through an extensive literature review. Model output was compared to each
toxicity level for the suggested exposure time (96-hr for bluegill and 24-hr for humans).
The results suggest that traditional intensive row crops such as canola, corn and sorghum
may negatively impact aquatic life, and in most cases do affect the availability of safe
drinking water. The continuous corn rotation, the most representative rotation for current

154

agricultural practices for a starch-based ethanol economy, delivers the highest
concentrations of glyphosate to streams. In addition, continuous canola contributed to a
96-hr concentration of 1.11 ppm of trifluralin, a highly toxic herbicide, which is 8.7 times
the ecotoxicity of bluegills and 21 times the safe drinking water level. Continuous corn
also resulted in the impairment of over 541,152 kilometers of stream. However, there is
promise with second-generation lignocellulosic bioenergy crops such as switchgrass,
which resulted in a 171,667 kilometers reduction in total stream length that exceeds the
human threshold criteria, as compared to the base scenario. Results of this study may be
very useful in determining the suitability of bioenergy crop rotations and aid in decision
making regarding the adaption of large-scale bioenergy cropping systems.

7.1 INTRODUCTION
Biofuels are energy sources, mainly liquid petroleum replacement fuels such as
ethanol and biodiesel, that are derived from the conversion of biological resources
(Hoogeveen et al., 2009). In 2008, biomass provided 53 percent of the nation’s renewable
energy (USEIA, 2009) while production of ethanol derived from biological sources
tripled between 2000 and 2007 (Hoogeveen et al., 2009).
It is expected that global energy consumption will increase by 57 percent in the
next 25 years (Rooney et al., 2007). Meanwhile, rising oil prices and increased political
pressure to develop carbon-neutral fuels have led to an increase in demand for alternative
energy sources. As for the United States, reducing dependency on foreign petroleum
imports may provide many benefits such as lower consumer costs, climate change
mitigation, job creation, and diversification of domestic energy sources. As a possible

155

scapegoat for the global petroleum feeding frenzy, the popularity of biofuels has
increased as they may provide the means for a shift away from conventional petroleum
sources (Solomon, 2010). There is, however, evidence of negative consequences are
prevalent, as for example, greenhouse gas emissions from landuse change towards cornbased ethanol are estimated to nearly double (Searchinger et al., 2008). Therefore, all
negative and positive aspects of large-scale bioenergy crop production are important to
consider in order for this approach to be a sustainable and cost-effective replacement of
current production methods (Nyakatawa et al., 2006).
Agriculture is the largest contributor to excess nutrients, mainly nitrogen and
phosphorus, in freshwater in the United States (Foley et al., 2005). In addition, The
demand for higher yields and more acreage in production will consequently increase
pressure to apply more pesticide likely resulting in more frequent occurrences of
pollution of these types in waterbodies (Landis et al., 2008). For example, the current
biofuel crop of choice, corn, is agrichemicals input-intensive, requiring more nitrogen
fertilizer, insecticide, and herbicide than any other crop in the U.S. (Pimentel, 2005).
Many of the other first-generation biofuel feedstocks follow suit requiring large amounts
of inputs in order to produce a viable crop. Oftentimes, this results in increased nutrient
and chemical loadings in water, leading to negative effects on human safety and aquatic
ecosystems such as death of coral reefs, fish kills, reductions in harvestable fish, and
reduced aesthetic value of the water body (Carpenter et al., 1998). Therefore, surface
waters are at a considerable risk of impairment due to the amount of pesticides that may
be present in surface runoff and infiltration to groundwater resources (Hoogeveen et al.,
2009). However, second generation bioenergy crops, including switchgrass, miscanthus,

156

and native grasses, require less intensive practices and less commercial fertilizers and
pesticides (Oliver et al., 2009).
Pesticides are applied to agricultural fields in the effort to control and eliminate
pests (Pimentel, 2005), often minimizing crop loss and leading to more efficient
production (Helfrich et al. 2009). There are three main types of pesticides utilized in
agriculture, including herbicides, used to control noxious and invasive vegetation;
insecticides, used to protect crops against insects; and fungicides, used for protecting
crops from diseases (Helfrich et al. 2009). Of the three types, herbicides are the most
commonly used (Gilliom, 2007; Helfrich et al., 2009) and account for 70 percent of
national used pesticides (USGS, 1999). Some commonly applied and detected herbicides
include atrazine, metolachlor, alachlor, cyanazine, simazine, and 2,4-D (Gilliom, 2007;
USGS, 1999).
Pesticides are most frequently detected within streams in areas of agriculture, as
seen in a ten year U.S. Geological Survey’s National Water-Quality Assessment program
(USGS NAWQA) study who observed that 97 percent of streams samples and 61 percent
of shallow groundwater samples taken at agricultural sites had detections of one or more
pesticides (Gilliom 2007). However, it is difficult to measure their occurrence since they
are often in mixtures with several other compounds (USGS, 1999; Belden et al., 2007;
Gilliom, 2007). While, contamination from pesticides is more of a threat to streams and
other surface water, groundwater contamination can be more difficult to reverse and can
take longer in recovering (USGS, 1999).
Transport of pesticides can occur through numerous modes including, overland
runoff, lateral movement, soil erosion, leaching, groundwater, tile drainage, and drift

157

during application (Pimentel, 2005; USGS, 1999; Solomon et al., 2008). Meanwhile,
there are several factors that have been shown to determine and influence this
transportation. Precipitation, storm water runoff, soil drainage, slope, irrigation practices,
groundwater connections, and application timing are some of these influences
(Domagalski et al., 2008; USGS, 1999; MSU Weeds, 2010). The pesticide itself can also
effect transportation through characteristics such as its solubility, stability, and ability to
be broken down by climatic conditions and microorganisms within soils (USGS, 1999;
Helfrich et al., 2009). Once pesticides reach a channel they can travel to downstream
unpolluted waters (Schulz and Liess, 1999) and continue to persist in both the water and
bed sediments (Gilliom, 2007), eventually being exposed to biota through respiration,
absorption through the skin, or consumption (Helfrich et al., 2009).
The presence of pesticides within a lotic system can have both direct and indirect
influences on aquatic biota, leading to lethal and sub-lethal effects (Scott and Sloman,
2004; Fleeger et al., 2003; Cook and Moore, 2008; Relyea, 2009). These effects can be
controlled by toxicity and persistence of a pesticide, exposure time, and the quantity that
an organism is exposed to (Helfrich et al., 2009). Studies have shown that pesticides can
influence and impede on behavioral and physiological characteristics of aquatic
organisms (Cook and Moore, 2008; Scott and Sloman, 2004). More specifically, sensory,
hormonal, neurological, and metabolic systems can be affected, leading to reductions in
reproduction, biomass, and health of individuals and communities (Cook and Moore,
2008; USGS, 1999). With fish in particular, Baldwin et al. (2009) observed that
threatened salmon species recovery could be constrained due to pesticides, based on
reductions in survival and productivity. Goodbred et al. (1997) demonstrated that sites

158

with higher pesticide concentrations showed lower hormone ratios in common carp,
which could indicate further potential effects (USGS, 1999). Impacts can even reach to
the point of mortality and fish kills have been reported due to pesticides. According to
Pimentel et al. (1993), between six and fourteen million fish are killed by pesticides each
year, which is often an underreported statistic (Pimentel, 2005). Macroinvertebrates also
have shown similar direct responses to pesticides, especially insecticides, for which they
are very sensitive to (Overmeyer, 2010). Schulz and Liess (1999) observed 11 common
macroinvertebrate species disappear and others’ populations reduce due to insecticide
contamination within a stream. Reductions in relative abundance and number of sensitive
species have also been shown to be related to the stress of pesticides (Schafer et al.,
2007). Although herbicides are not generally as toxic to aquatic biota as insecticides are,
they have still been shown to have similar effects (Helfrich et al., 2009; USGS, 1999).
This is especially apparent when it comes to indirect influences. This includes herbicides’
tendency to reduce aquatic vegetation, effecting food sources, habitat, and dissolved
oxygen conditions for both fish and macroinvertebrates (Pimentel, 2005; Overmeyer,
2010). For example, in a review looking at atrazine and its effects on fish, Solomon et al.
(2008) found little consistency on any direct affects from the pesticide, however did
recognize a consistency with indirect affects through food supply and habitat loss.
Macroinvertebrates have also shown similar indirect relationships with herbicides as well
(Overmeyer, 2010). By considering the negative impacts that pesticides can have on
aquatic ecosystems, the U.S. Environmental Protection Agency (USEPA, 2008)
performed ecological risk assessments to investigate the threats involved with the use of
specific pesticides based on scientific measurements and judgment.

159

Along with aquatic health, human health can also be jeopardized by exceedingly
high levels of pesticides within water resources. Several studies have shown that
exposure to pesticides can contribute or lead to cancer and other harmful human illnesses
(Safe, 2004; Fan et al., 2007). While some pesticides may be carcinogenic, others may
affect humans in other ways including risks to the nervous system and hormones, along
with less severe risks including skin irritations (Omalley, 1997). These risks are often
only seen at high levels of exposure, yet some pesticides have been banned from use due
to their extreme toxicity, including DDT (1,1,1-trichloro-2,2-di(4-chlorophenyl)ethane)
within the United States and atrazine (2-chloro-4-ethylamine-6-isopropylamino-Striazine) in Germany (Helfrich et al., 2009; Takacs et al., 2002).
Specific water quality benchmarks or thresholds are also often set to recognize
areas where individual pesticides are present at concentrations that can harm aquatic biota
or even human health. These benchmarks are commonly presented in LC50 measures that
reflect lethal concentrations at which 50 percent of tested animals are killed in a given
period of time (USEPA, 2009d; Gilliom, 2007; Helfrich et al., 2009). It is important to
note that the LC50 is an extreme case of lethality and does not consider any other biotic
factor such as nutrient loading, dissolved oxygen, and turbidity. It is possible that there
may be impairments to aquatic health well in advance of reaching these LC50 toxicity
levels.
7.1.2 Modeling Pesticide Fate and Transport
Simulation models have been shown to be a valuable tool when looking at
pesticide fate and transport and can be useful in executing ecological risk assessments
both at a field and watershed scales (Parker et al., 2007; Vazquez-Amabile et al., 2006;

160

Luo and Zhang, 2009a; Quilbe et al., 2006; Luo and Zhang, 2009a). Several models that
were used in the past to study the fate and transport of pesticides including the
Hydrological Simulation Program-HSPF (Laroche et al., 1996), Groundwater Loading
Effects of Agricultural Management Systems Model-GLEAMS (Vazquez-Amabile et al.,
2006; Thomas et al., 2009), Agricultural Policy / Environmental eXtender model-APEX
(Harman et al., 2004; Mudgal et al., 2008), Annualized Agricultural Non-Point Source
model-AnnAGNPS (Heathman et al., 2008) and Soil and Water Assessment Tool-SWAT
(Larose et al., 2007; Luo and Zhang, 2009b; Holvoet et al., 2007). Due to the
comprehensive nature and ability to include in-stream processes, SWAT was chosen as
the best model for the scope of this project.
SWAT utilizes equations adopted from GLEAMS to simulate pesticide movement
on land (Neitsch et al., 2005). The processes involved in the model include plant foliage
interaction and wash-off, degradation, leaching, percolation, as well as surface runoff and
sediment transport of soluble and sorbed pesticides. SWAT differs from GLEAMS in that
it can only model one pesticide at a time; however, its capability to simulate in-stream
processes, gives it an advantage (Neitsch et al., 2005; Vazquez-Amabile et al., 2006).
SWAT has been evaluated and implemented in multiple studies looking at the fate of
pesticides. A Vazquez-Amabile et al. (2006) study used SWAT model to estimate
atrazine level with varying corn planting season scenarios. The study further applied the
results to perform risk analysis by calculating exceedance probabilities and comparing
average monthly levels with US Environmental Protection Agency (USEPA) drinking
water standards. Some have proposed and implemented some modifications to SWAT

161

code, including enabling the code to account for direct loss of pesticides during
application and have observed more accurate simulated results (Holvoet et al., 2008).
Determination of which biofuel crops and management strategies may have the
least unfavorable consequences to environments is vital for future decision making
(Thomas et al., 2009). In this regard, the use of models including SWAT, have been
demonstrated to be valuable in obtaining the information needed to perform risk analysis
(Thomas et al., 2009; Vazquez-Amabile et al., 2006), however, based on an extensive
literature review the authors are unaware of any study using modeling tools and risk
analysis in determining pesticide levels dealing with conversions of lands to biofuel crops
in a large-scale. In addition, recent modeling studies have been limited to one to four
pesticides at small scales (field to small watershed) (Kannan et al., 2006; Luo et al.,
2008; Luo and Zhang, 2009b; Vazquez-Amabile et al., 2006; Larose et al., 2007; Holvoet
et al., 2007; Neitsch et al., 2002).
The study conducted here evaluates the long-term environmental implications,
specifically on fish and humans from pesticides, of large-scale bioenergy cropping
system expansion. This aids in determining the suitability of bioenergy crops and
obtaining information to aid in watershed-scale decision making regarding landuse
management. The Soil and Water Assessment Tool (SWAT) was used to predict the
effects of 14 future landuse/bioenergy crop rotations scenarios in four large watersheds in
agricultural regions of Michigan over six years (2000 to 2005). The model was calibrated
and validated for flow, sediment, and nutrients and the model outputs on seven pesticides
loads and concentrations were estimated on a daily time step.

162

7.1.3 Pesticides
Based on information obtained from the Michigan State University Extension
(MSUE), pesticide application rates for different bioenergy crops were estimated and
later used for modeling purposes. In addition, both human consumption and aquatic
ecotoxicity thresholds were obtained from literatures. The pesticides included in analysis
are presented in the subsequent sections.
7.1.3.1 Atrazine
One of the most commonly and widely applied agricultural herbicides,
particularly in corn production, is Atrazine [2-chloro-4-ethylamine-6-isopropylamino-Striazine] (Ribaudo and Bouzaher, 1994; EXTOXNET, 1996). With corn and other crops
being able to metabolize and detoxify atrazine (Ribaudo and Bouzaher, 1994), it
selectively targets and controls broadleaf weeds by inhibiting photosynthesis (MSU
Weeds, 2010). Atrazine is a Chloro-triazine herbicide, whose transport, based on its
chemical properties, is often through leaching and runoff. A secondary means of transport
can also be through its absorption to sediments. Atrazine can persist in environments with
a half-life of 60 to 150 days in soils and exceptionally longer in anaerobic conditions
along with in surface and ground water. Because atrazine and its degradates can be
moderately toxic at high concentrations and for long periods of time to both aquatic
animals and humans, the USEPA has established maximum concentration levels
(Ribaudo and Bouzaher, 1994; Vazquez-Amabile et al., 2006). Standards for drinking
water for humans have been established at 0.003 mg/L (USEPA, 2010d; VazquezAmabile et al., 2006).

163

7.1.3.2 Bromoxynil
Bromoxynil (3,5-dibromo-4-hydroxybenzonitrile) is a widely used postemergence, contact herbicide selective to broad-leaved weeds (Topp et al., 1992).
Bromoxynil enters the environment through many pathways such as to the atmosphere by
plant evapo-transpiration and volatilization from soil surface, incorporation with surface
runoff, especially during precipitation events immediately following application, and
released into soil by treated vegetation (Baxter and Cummings, 2008). Bromoxynil has a
low persistence in soil with a half-life of several days, with microbial activity being the
primary degradation mechanism (Topp et al., 1992; Baxter and Cummings, 2008). It is
also suspected that bromoxynil has teratogenic effects on mammals, inhibiting or altering
the reproductive and developmental capabilities (EXTOXNET, 1996).
7.1.3.3 Glyphosate
Glyphosate [N-(phosphonomethyl) glycine] is a nonselective, post emergence
herbicide that is one of the most widely used (USEPA, 1993; Duke and Powles, 2008;
USDA, 1997). It is taken up by plants rapidly through leaves and then prevents the plant
from making amino acids, which restricts the making of protein and eventually leads to
death. Glyphosate and surfactants associated with the pesticide are strongly absorbed in
soils and degraded by microorganisms leading to a soil half-life, depending on
conditions, that can range from 3 to 249 days. This strong absorption also limits its ability
to be transported by runoff and its movement into groundwater (USDA, 1997; Duke and
Powles, 2008). Glyphosate, once in water, is easily dissolved and has a half-life of 35 to
63 days (USDA, 1997). Although Glyphosate can be toxic to aquatic non-targeted
vegetation, which can indirectly can affect biota, it is relatively non-toxic to fish and
invertebrates (USEPA, 1993; Duke and Powles, 2008; USDA, 1997; EXTOXNET,

164

1996). Drinking water standards for Glyphosate have been established by the USEPA at
0.7 mg/L (USEPA, 2010d).
7.1.3.4 Metolachlor
Metolachlor [2-chloro-6'-ethyl-N-(2-methoxy-1-methylethyl)acet-o-toluidide] is
an herbicide used on corn, soybeans, sorghum and other crop lands in effort to have preemergent control of grasses and broadleaf weeds. Metolachlor, a chloroacetanilide
herbicide, controls these unwanted species through the inhibition of protein synthesis. It
has a soil half-life of 15 to 70 days and an even higher persistence in water, leading it to
be detected, along with its degradates, in groundwater and surface waters. Metolachlor,
although not much of a threat in bioaccumulation, is moderately toxic to aquatic biota
(CDC, 2010; EXTOXNET, 1996).
7.1.3.5 Pendimethylin
Another selective herbicide is Pendamethlin [N-(1-ethylpropyl)-3,4-dimethyl-2,6dinitrobenzenamine], which is used in controlling both broadleaf weeds and grassy weed
species. This dinitroanililine herbicide can be used in both pre-emergence and early postemergence scenarios, when managing crop productions such as corn, soybeans,
winterwheat, and several others (Triantafyllidis et al., 2009; USEPA, 1997;
EXTOXNET, 1996). Pendimethalin restricts seedling growth by inhibiting cell division
and elongation (Triantafyllidis et al., 2009). It’s moderately persistent in the environment
with a field half-life of 40 days. Due to being highly absorbed by soils and insoluble in
water, Pendimethalin has limited risk to groundwater and the transport to surface waters
is mainly through soil erosion (EXTOXNET, 1996; Triantafyllidis et al., 2009).
Pendimethalin has a low acute toxicity, yet is highly toxic to aquatic species, including
fish, and is a possible carcinogen in humans (Megadi et al., 2010).
165

7.1.3.6 Sethoxydim
Sethoxydim (2-[1-(ethoxyiamino)]butyl-5-[2-ethyl-thio

propyl]-3-hydroxy-2-

cyclohexen-1-one) is a postemergence herbicide selective against annual and perennial
grass species in broadleaf crops (Holshouser and Coble, 1990). It acts as an inhibitor of
the plastidic enzyme acetylcoenzyme A carboxylase (Tardif et al., 1993). Sethoxydim has
low persistence with reported half-lives of 5 to 25 days in soil, in which photodegradation
on soil takes only several hours (Wauchope et al., 1992). Sethoxydim has been observed
to be slightly toxic to aquatic species and less toxic to humans, with unlikely chronic
effects (Tu et al., 2001).
7.1.3.7 Trifluralin
Like Pendamethlin,

Trifluralin

[a,a,a-trifluoro-2,6-dinitro-N,N-dipropyl-p-

toluidine] is a dinitroaniline herbicide and is applied pre-emergence for inhibiting growth
of grasses and broadleaf weeds in agricultural areas (Grover et al., 1997; EXTOXNET,
1996; USEPA, 1996). Trifluralin is highly absorbed by soil and low solubility in water,
leading it to become immobile often and less of a threat to groundwater. The persistence
of Trifluralin is dependent on several soil conditions and its half-life can range from 45
days to 8 months. It can be degraded through several processes including microbial
activity, volatilization, and photolysis (Grover et al., 1997; EXTOXNET, 1996).
Trifluralin is highly toxic to fish, macroinvertebrates, and aquatic vegetation
(EXTOXNET, 1996).
7.1.3.8 2,4-D
In continuing efforts to control broadleaf weeds, 2,4-D [(2,4-dichlorophenoxy)
acetic acid] is a phenoxyacetic herbicide that is used post emergence and widely applied
throughout many land uses, agricultural lands being the dominant application setting

166

(USEPA 2008). 2,4-D can be in several forms, including salts, acids, amines, or esters
(Munro et al., 1992; EXTOXNET, 1996). Controlling weed growth is achieved through
the absorption of 2,4-D by roots and leaves, ultimately affecting metabolism, growth
patterns, and food transport (Munro et al., 1992). 2,4-D, although lacking persistence in
soils and aerobic aquatic systems with a half-life of 6.2 days and 15 days respectively,
has been shown to be moderately persistent in anaerobic aquatic conditions with a halflife of up 333 days (USEPA, 2008). 2,4-D has been shown to be detected in both
groundwater and surface waters despite its shorter half-life (EXTOXNET, 1996). Its
toxicity varies with form, esters being highly toxic to aquatic biota while amines and
acids are far less toxic. 2,4-D is also toxic to aquatic vascular plants, which can further
their risk to biota. In addition, human health risks are a potential and standards of 0.07
mg/L have been set by the USEPA (USEPA, 2010b).
7.1.3.9 Aquatic and Human Toxicity Thresholds
Threshold levels were used for both humans and fish, in attempt to evaluate the
potential harmful impacts of pesticide concentrations within streams. Measures often
used to describe levels of concern for humans, are based on USEPA set maximum
concentration levels (MCL) for drinking water. MCLs are defined as the “maximum
permissible level of a contaminant in water which is delivered to any user of a public
water system” (USEPA, 2010a). In the absence of a MCL for specific pesticides, a
maximum permissible level goal (MCLG) was calculated based on the reference dose
(RfD), which is the amount a human can be exposed to on a daily basis with no harmful
effects. The RfD is multiplied by an assumed weight of an adult (70 kg) and then divided
by the average amount of daily water consumption (2 Liters). This is then multiplied by a

167

drinking water contribution which is commonly assumed as 20 percent (Lin and Lee,
2007). This technique has been documented and is used by the USEPA and many other
governments (USEPA, 2010a; Massachusetts Department of Environmental Protection,
2010).
For fish a standard measurement commonly used, when dealing with pesticides, is
the 96 hour LC50. LC50s are concentrations for a set period of time that are lethal to 50
percent of a test population (Helfrich, 2009). Due to the inherent complexity of pesticide
fate in the environment and maintaining overall impact of these agrochemicals
applications of aquatic ecosystems, an umbrella specie should be selected. In this study,
the Bluegill (Lepomis macrochirus), was selected as a biological indicator acting as
representation of aquatic health and habitat needs of other aquatic life (Fargione et al.,
2009). In addition, Bluegill is a native fish and historically widely distributed throughout
the study area. In the next step, the median LC50 levels for Bluegill and different
herbicides were obtained from thousands of studies available on the USEPA’s ECOTOX
database (USEPA, 2010b). The previously discussed threshold values used in this study
are provided in Table 21.

168

Table 21. Bluegill and human toxicity thresholds.
2,4 - D Atrazine Bromoxynil Glyphosate Metolachlor Pendimethalin Sethoxydim Trifluralin
13500
5600
10000
980
133300
127.5
Bluegill, 96 hour LC50 263000 24000
70
Human, MCL
* MCLG – based off of RfD

3

140*

700

169

700*

280*

630*

52.5*

The use of models, including SWAT, have been demonstrated to be valuable in
obtaining the information needed to perform risk analysis (Thomas et al., 2009; VazquezAmabile et al., 2006). However, based on an extensive literature review the authors are
unaware of any studies using SWAT and risk analysis in determining pesticide levels
dealing with conversions of lands to biofuel crops. In addition, SWAT simulations within
2
recent studies have been limited to one to four pesticides at small scale (1.42 km to
2
14983 km ) (Kannan et al., 2006; Luo et al., 2008; Luo and Zhang, 2009b; Vazquez-

Amabile et al., 2006; Larose et al., 2007; Holvoet et al., 2007; Neitsch et al., 2002).
The research in this paper evaluates the long-term environmental implications,
specifically on fish and humans from pesticides, of large-scale bioenergy cropping
system expansion. This aids in determining the suitability of bioenergy crops and
obtaining information to aid in watershed-scale decision making regarding landuse
management. SWAT was used to predict the effects of 14 future landuse/bioenergy crop
rotations scenarios in four large watersheds in agricultural regions of Michigan over six
year (2000 to 2005) period. The model was calibrated and validated for flow, sediment,
and nutrients and the model outputs on pesticides loads and concentrations were
evaluated on a daily time step.

170

7.2 MATERIALS AND METHODS
7.2.1 Study Area
The study area consists of four large watersheds totaling 53,358 km

2

located in

Michigan’s Lower Peninsula used intensively for agricultural production. Meanwhile,
these watersheds, described below, directly discharge to the Great Lakes (Figure 23):
The Saginaw River basin (hydrologic unit code (HUC) 040802) is located in the
east central portion of Michigan’s Lower Peninsula and includes six HUC 8-digit
watersheds (Tittabawassee, Pine, Shiawassee, Flint, Cass, and Saginaw River).
The region drains 1,526,242 ha with 40 percent used for agricultural production, 6
percent developed, 42.4 percent forest, and 12 percent wetlands.
The St. Clair-Detroit basin (HUC 040900) is located in southeast Michigan and
includes four HUC 8-digit watersheds (St. Clair, Clinton, Detroit, and Huron
River). The region drains 818,303 ha with 30.3 percent used for agricultural
production, 39.4 percent developed, 28 percent forest, and 2.3 percent wetlands.
The Southeastern Lake Michigan basin (HUC 040500) is located in southwest
Michigan and includes five HUC 8-digit watersheds (Kalamazoo, Upper Grand,
Maple, Lower Grand, and Thornapple River). The region drains 1,889,426 ha
with 53.3 percent used for agricultural production, 10.6 percent developed, 28.8
percent forest, and 7.3 percent wetlands.
The St. Joseph basin (HUC 04050001) is located in the southwest portion of
Michigan and the northeast portion of Indiana. The region drains 1,101,781 ha
with 70.9 percent used for agricultural production, 6.3 percent developed, 16.4
percent forest, and 6.4 percent wetlands.

171

Figure 23. Basins included in study area.
7.2.2 SWAT Model
In order to effectively model the dynamic relationships of pesticides at a
watershed scale, a reliable hydrodynamic model must be used (Holvoet et al., 2004).
SWAT is a hydrodynamic and physically based model developed to predict the impacts
of land use scenarios on flow and nutrient, pesticide, and sediment loadings on a
watershed scale. The model can be setup based on landuse practices in complex
watersheds with varying soils, landuse, weather conditions, and management operations
over long time periods (Arnold et al., 1998). ArcSWAT, the ArcGIS extension of the
SWAT 2005 program, is a graphical user interface for the SWAT model used in this

172

study. The major inputs to the model include weather datasets, soils data, elevation map,
landuse data, nutrient and pesticide applications, and management practices. The SWAT
model uses the modified SCS curve number method to determine surface runoff during
precipitation events (Neitsch et al., 2005). Soil erosion and sediment yields are
determined by the Modified Universal Soil Loss Equation (MUSLE). The accuracy of
this model depends on complete understanding of the regional processes and of internal
model processes. Therefore, setting up the model requires considerable time commitment
for both data collection and model familiarization (Gassman et al., 2007).
7.2.2.1 Pesticides in SWAT
The fate and transport of pesticides in the SWAT model include, plant foliage
interaction and wash-off, degradation, leaching, percolation, as well as surface runoff and
sediment transport of soluble and sorbed pesticides.
During a precipitation event, a portion of the pesticide applied onto the foliage
may be washed off. The solubility of the applied chemical, the timing of application and
precipitation, rainfall intensity, and plant morphology determine the fraction that is lost
(Eq. 7).

‫ݐݏ݌‬௙,௪௦௛ ൌ ݂‫ݎ‬௪௦௛ ∗ ‫ݐݏ݌‬௙

(7)

where, pstf,wsh is the amount of pesticide washed from foliage to the soil surface (kg
pesticide/ha), frwsh is the pesticide wash-off fraction, and pstf is the amount of pesticide
on the foliage (kg pesticide/ha).
Pesticides also experience degradation, in which chemical conversion of
compounds into simpler forms occurs. The time it takes for a compound concentration to

173

be reduced by half is termed the half-life. The half-life in soil layers (Eq. 8) is a
combination of effects caused by volatilization, photolysis, hydrolysis, biodegradation,
and chemical reactions.

‫ݐ‬ଵ⁄ଶ,௦ ൌ

଴.଺ଽଷ

௞೛,ೞ೚೔೗

(8)

where, t1/2,s is the half-life of the pesticide in the soil (days) and kp,soil is the soil
degradation rate constant (1/day). The corresponding foliar half-life is presented below in
Eq. 9:

‫ݐ‬ଵ⁄ଶ,௙ ൌ

଴.଺ଽଷ

௞೛,೑೚೗೔ೌೝ

(9)

where, t1/2,foliar is the half-life of the pesticide on foliage (days) and kp,foliar is the
foliar degradation rate constant (1/day). In SWAT, pesticides that are highly soluble in
water have the potential to be transported with percolation throughout the shallow soil
layers. Pesticides in the soil are transported as a dissolved solution or attached to
sediment particles.
7.2.2.2 Movement of soluble pesticide
Pesticide that is soluble in water is highly mobile in surface runoff, lateral flow,
and percolation through shallow surface layers. The amount of pesticide in each type of
movement is calculated by a number of equations representing the physical and chemical
processes that control the flow or surface and soil layer water, the properties of the soil
and pesticide, and the pesticide percolation coefficient.

174

7.2.2.3 Transport of sorbed pesticide
Sediment that is delivered to the main stream channel due to surface runoff may
also be transporting pesticides attached to soil particles. The amount of sorbed pesticides
depends on the sediment load delivered from the agricultural fields. Since the sediment
load in overland flow delivered to the stream is enriched by colloidal clay particles, the
sediment load contains a larger proportion of pesticides than that of the soil layer itself.
This relationship is defined by the enrichment ratio, or the ratio of pesticide transported
with the sediment to the concentration on the soil surface, developed by Menzel (1980).
7.2.2.4 Pesticide routing
It is important to note that SWAT incorporates routing and reservoir and instream pesticide transformations, leading to a fundamentally superior model (VazquezAmabile et al., 2006). A limitation to the capabilities of pesticide routing, though, is that
the model can only route a single pesticide through the stream network at a time. The
model uses a mass balance approach to determine the pesticide in a stream segment by
determining any additions from inflow and resuspension and diffusion of pesticide from
the sediment layer. Since SWAT can essentially model only one pesticide during each
simulation, 168 model runs were performed to study the impacts of eight herbicides
including

atrazine,

metolachlor,

glyphosate,

2,4-D,

pendimethylin,

trifluralin,

bromoxynil, and sethoxydim on water quality and aquatic and human health.
7.2.2.5 SWAT Model Setup
In order to provide the most realistic and accurate representations of flow and water
quality, the SWAT model was calibrated for flow, sediment, and nutrients (total nitrogen
and phosphorus) on a monthly time step. Calibration should include data from three to
five years during both dry and wet periods to effectively activate all model processes for

175

a diverse range of hydrologic events (Moriasi et al., 2007). Therefore, based on a
comparison with the 19-year mean of precipitation for all watersheds, the period of 20002005 was used for the calibration and validation. Detailed information about model
calibrations and validations were presented in Love and Nejadhashemi (2010).
Management operation schedules were incorporated into the SWAT model to provide a
more realistic representation of the true agricultural practices in the study area. Bioenergy
cropping rotations were developed to provide a depiction of the practices that would be
likely associated with these crops in Michigan. A schedule was developed for each
rotation based on data collected from the Michigan State University Extension. The
schedule for the corn-soybean-rye rotation is provided in Table 22 as an example. Table
23 represents the pesticide application rates that were used in all of the rotations in this
study. Each rotation was applied on all agricultural land in the study area, providing the
most extreme scenario of landuse conversion representing future bioenergy crop
expansion in the region.

176

Table 22. Corn-soybean-rye rotation management operations.
Date

Practice

SWAT Practice

Amount/acre Amount/ha Year

1-May Soil Finish

Field Cultivator Ge15ft

4-May Nitrogen Application (Urea)

Urea

4-May Soil Finish

Field Cultivator Ge15ft

173 lb

5-May Phosphorus Application (P2O5) Elemental Phosphorus

194 kg

53 lb

59.5 kg
1

5-May Plant Corn Seed

Plant/Begin Growing Season

5-May Bicep II Magnum (PRE)

Atrazine

0.5 qt

1.39 kg

5-May Bicep II Magnum (PRE)

Metolachlor

0.39 qt

1 kg

1-Nov Combine Harvest Corn Grain

Harvest and Kill

40 lb

45 kg

15-Nov Fall Chisel

Coulter-chisel Plow

14-May Soil Finish

Field Cultivator Ge15ft

14-May Phosphorus Application (P2O5) Elemental Phosphorus
14-May Soil Finish

Field Cultivator Ge15ft

15-May Plant Soybean Seed

Plant/Begin Growing Season

7-Jun Spray Roundup Weathermax ® Glyphosate Amine

2
22 fl oz

25 lb

1-Oct Harvest Soybean Grain

Field Cultivator Ge15ft

7-Oct Nitrogen Application (Urea)

Urea

28 kg

Harvest and Kill

4-Oct Soil Finish

0.87 kg

177

Table 22 (cont’d)
7-Oct Phosphorus Application (P2O5) 18-46-0
7-Oct Plant Rye Seed

35 lb

39.2kg

Plant/Begin Growing Season

15-Apr Nitrogen Application (28% N)

Elemental Nitrogen

25 lb

28 kg

25-Apr Spray Frontline 2,4-D

2,4-D Amine

0.5 lb a.i.

0.56 kg

15-Jul Harvest (0.7 Efficiency)

Harvest and Kill

30-Oct Fall Chisel

3

Coulter-chisel Plow

Table 23. Pesticide application rates for each rotation.
-1

Rotation

Pesticide Rates (kg ha )
Atrazine

Bromoxynil

Glyphosate

Metolachlor

Pendimethalin Sethoxydim

Trifluralin

2,4-D

Base

1.39

-

0.87

1.00

-

-

-

-

Continuous Canola

-

-

-

-

-

0.315

1.10

0.56

Continuous Corn

1.39

-

0.87

1.00

-

-

-

-

Continuous Corn
Stover
Continuous Rye

1.39

-

0.87

1.00

-

-

-

-

-

-

0.87

-

-

-

-

0.56

Continuous Sorghum

1.39

0.42

0.87

1.00

-

-

-

-

Continuous Soy

1.39

-

0.87

1.00

-

-

-

-

Corn Soy

1.39

-

0.87

1.00

-

-

-

0.56

178

Table 23 (cont’d)
Corn Soy Rye

1.39

-

0.87

1.00

-

-

-

-

Corn stover Soy

1.39

-

0.87

1.00

-

-

-

-

Miscanthus

1.39

0.42

0.87

1.00

-

-

-

-

Native Grass

-

-

-

-

1.12

-

-

0.28

Sorghum Soy

-

-

-

-

-

-

-

Switchgrass

1.39

-

0.87

1.00

-

-

0.28

179

1.12

7.2.2.6 Pesticide Monitoring Dataset
All pesticide monitoring datasets were obtained from the USEPA STORET
(http://www.epa.gov/storet/dbtop.html), and the USGS surface-water data collection
(http://water.usgs.gov/osw/data.html) for each watershed. However, the available data
was limited to a fraction of both the herbicides and watersheds. Out of the 32 total
watershed/herbicide combinations needed for the model calibration, only 12 were found.
Furthermore, three of the four watersheds were missing data for six of the eight
herbicides. The data at these locations were generally limited to less than a dozen grab
samples, making calibration impractical and unreliable. It is important to note that a
number of the pesticides are not widely used in Michigan. For example, pendimethalin
was included in this study as a possible herbicide to be used in switchgrass and
miscanthus, yet is not labeled for their use currently. Therefore, the best alternative is to
adjust model parameters using datasets presented in literatures. This is the common
practice and others, including Coffey et al. (2010), have used literature values in place to
adjust model parameters for calibration when adequate observed data is not available.
Based on an extensive literature review, it was determined that the pesticide percolation
coefficient (PERCOP) was the most influential parameter in SWAT for pesticide
transport (Du et al., 2006; Vazquez-Amabile et al., 2006; Larose et al., 2007; Luo et al.,
2008). From Du et al. (2006) and Neitsch et al. (2002), the PERCOP value for atrazine,
metolachlor, and trifluralin were obtained to be 0.025, 0.3 and 0.3, respectively. Values
for pesticides not included in previous studies were modeled using for the default values
included in SWAT (Neitsch et al., 2005). It is also important to note that pesticide
transport is influenced most by the surface runoff and sedimentation pathways, which

180

were successfully calibrated in this study (Pionke and Chesters, 1973; Karickhoff et al.,
1979; Weston et al., 2004).

7.3 RESULTS AND DISCUSSION
In the following sections, the results obtained in this study are presented in a
manner to highlight the possible implications of landuse change towards bioenergy
cropping systems on surface water quality. The concentration fluctuation that is likely to
be observed over a large basin with varying geographic and climatic characteristics are
highlighted in the first section, followed by a discussion accentuating the possibility of
impairing stretches of streams by exceeding the thresholds of toxicity for aquatic biota
(represented by the Bluegill fish) and human health as a function of direct consumption
from these waterbodies. The final two sections, temporal and spatial variation of stream
impairment identify the location in time and space that require consideration prior to
implementing bioenergy crops on a large-scale to refrain from damaging aquatic
ecosystems and water quality.
7.3.1 Basin-wide Concentration Fluctuation
By altering the landscape into large-scale agricultural production focused on
bioenergy feedstocks, an expected change in pollutant concentrations in surface waters
may occur. In this study, the daily concentrations of pesticides in stream reaches were
collected within the study area for the period (2000-2006). In order to demonstrate the
immense variability that occurs over a large scale, a box plot was created for one of the
pesticides studied here (Glyphosate) showing the basin wide stream concentration
variations based on different crop rotations. Due to large variability in pesticide

181

concentration, log transformations were used (Figure 24). Glyphosate was chosen as it is
commonly applied throughout the entire study area and is included in all rotations but the
continuous canola, miscanthus, and native grass landuse scenarios. Similar concentration
variations were observed in other pesticides, which are presented in Tables 24 through
31. However, only maximum and median values are presented in these tables because
they are more important to aquatic and human toxicity. Below these variations will be
discussed for each pesticide:
By examining the fluctuation present over the entire basin, it is apparent that
many factors play a role in determining the actual stream concentration of pesticides
upon application to agricultural areas. Continuous corn, the most representative rotation
for current agricultural practices for a starch-based ethanol economy, delivers the highest
concentrations of Glyphosate to the stream (Table 26). In essence, this rotation has the
greatest possibility of containing the highest levels of glyphosate, well above both the
bluegill and human toxicity levels. The base scenario (current landcover and practices)
actually did not reach the threshold for bluegill ecotoxicity, which is an indicator that
current trends in agricultural production may not endanger aquatic species or human
health. Meanwhile, switchgrass has the lowest median values indicating that this rotation
is the safest of any of the rotations examined. In general, the close-grown grass species of
rye and switchgrass have the lowest negative impacts on water quality, whereas the
intensive crops corn and sorghum show the worst cases of glyphosate concentrations. In
general, all of the bioenergy cropping scenarios have the possibility of jeopardizing the
safety of surface waters for human consumption due to glyphosate contamination but at
different levels.

182

Atrazine, another common herbicide is represented in Table 24 and experiences
similar results. Here, the most intensive rotations, continuous corn and continuous corn
stover, experience increased maximum and median concentrations over base, whereas the
majority of other rotations experience negligible or decreased levels of atrazine in the
stream reach.
Metolachlor, another extensively used agricultural pesticide, yields mixed results
(Table 27). In general, metolachlor concentrations increase over base with intensive
cropping rotations that includes corn and sorghum, indicating the low suitability of these
crops to be adopted at a large scale.
The remainder of presented pesticides (bromoxynil, pendimethalin, sethoxydim,
triflualin, and 2,4-D) cannot be compared directly to the base scenario since these
pesticides are not currently used in crop rotations. However, these pesticides will be
introduced in the basin upon implementation of second generations of bioenergy crops.
The results from bromoxynil application indicate that continuous sorghum tends to
increase pesticide pollution, whereas including soybeans in a two-year rotation reduces
the amount of this chemical in surface waters (Table 25). Pendimethalin is applied on
switchgrass only and is not currently labeled for this specific crop, yet does suggest that
large-scale adoption of a bioenergy system based on switchgrass may have negative
water quality effects due to the use of this chemical (Table 28). Sethoxydim and
trifluralin (Table 29 and Table 30) are only applied to continuous canola. Again this is
useful in simply showing the possibility of introducing two chemicals that may lead to
further degradation of waterbodies. In the case that switchgrass is implemented to a large
scale, the results in Table 31 suggest that switchgrass has the potential to reduce

183

concentrations of 2,4-D over continuous rye and corn-soybean-rye rotations. Although
the aforementioned pesticides cannot be compared to the base scenario, they can be
compared based on the amount of stream length that exceeds toxicity threshold criterion
presented in the next section (7.3.2).

184

Figure 24. Basinwide variability of Glyphosate concentrations in log-scale (Dotted line represents the Bluegill ecotoxicity threshold of
5600 ppb and solid line represents the human toxicity threshold of 700 ppb).

185

-1
Table 24. Concentrations of Atrazine (µg L )

Base
Continuous
Canola
Continuous
Corn
Continuous
Corn stover
Continuous
Rye
Continuous
Sorghum
Continuous
Soy
Corn Soy
Corn Soy Rye
Corn stover
Soy
Miscanthus
Native Grass
Sorghum Soy
Switchgrass

40500
Max
Median
1.04E+03
1.18E-02
-

4050001
Max
Median
3.16E+02
4.23E-03
-

40802
Max
Median
1.58E+02
2.23E-03
-

40900
Max
Median
5.80E+03
1.20E-03
-

1.24E+03

2.23E-02

6.40E+02

4.19E-03

2.28E+02

2.47E-02

2.22E+06

1.64E-02

1.16E+03

2.34E-02

6.51E+02

4.34E-03

2.27E+02

2.48E-02

7.47E+03

1.02E-02

-

-

-

-

-

-

-

-

1.38E+03

1.59E-02

6.33E+02

2.79E-03

1.80E+02

1.65E-02

8.44E+03

6.66E-03

1.59E+03

6.99E-04

2.35E+02

2.23E-05

2.02E+02

1.54E-04

4.00E+03

3.29E-04

1.59E+03
6.59E+02
1.59E+03

9.07E-05
1.25E-06
9.54E-05

4.03E+02
3.40E+02
4.03E+02

6.87E-06
3.15E-07
6.80E-06

2.30E+02
1.18E+02
2.30E+02

1.40E-04
8.60E-07
1.42E-04

4.33E+03
7.03E+02
4.32E+03

4.95E-05
1.12E-06
4.93E-05

1.37E+03
6.69E+00

3.78E-05
2.27E-07

3.25E+02
3.93E+00

3.21E-06
5.44E-08

1.86E+02
4.28E+00

4.58E-05
5.96E-08

3.49E+03
9.75E+00

1.49E-05
2.15E-07

186

-1
Table 25. Concentrations of Bromoxynil (µg L )

40500
Base
Continuous
Canola
Continuous
Corn
Continuous
Corn stover
Continuous
Rye
Continuous
Sorghum
Continuous
Soy
Corn Soy
Corn Soy Rye
Corn stover
Soy
Miscanthus
Native Grass
Sorghum Soy
Switchgrass

Max
-

Median
-

4050001
Max
Median
-

Max
-

Median
-

Max
-

Median
-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

6.63E-06

1.85E+02

8.03E+01

40802

2.02E-06

40900

4.19E+01

4.74E-06

4.44E+02

1.04E-05

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

1.26E+02
-

7.37E-07
-

3.97E+01
-

2.35E-07
-

4.74E+01
-

187

1.55E-06
-

4.48E+02
-

1.18E-05
-

-1
Table 26. Concentrations of Glyphosate (µg L )

Base
Continuous
Canola
Continuous
Corn
Continuous
Corn stover
Continuous
Rye
Continuous
Sorghum
Continuous
Soy
Corn Soy
Corn Soy Rye
Corn stover
Soy
Miscanthus
Native Grass
Sorghum Soy
Switchgrass

40500
4050001
40802
40900
Max
Median
Max
Median
Max
Median
Max
Median
6.31E+02
4.51E-03
2.02E+02
7.95E-03
5.11E+02
6.19E-04
4.74E+03
1.67E-02
-

-

-

-

-

-

-

-

4.14E+02

2.19E-01

2.66E+02

1.17E-01

5.42E+01

3.32E-03

3.10E+05

3.22E-01

4.15E+02

2.21E-01

2.30E+02

1.17E-01

5.39E+01

3.34E-03

3.71E+03

2.04E-01

2.46E+03

4.56E-03

4.94E+01

6.94E-04

3.68E+02

8.18E-04

3.14E+03

1.43E-03

4.13E+02

2.17E-01

3.40E+02

1.18E-01

5.74E+01

4.64E-02

4.59E+03

2.00E-01

1.05E+03
9.24E+02
9.24E+02

1.18E-02
9.09E-03
2.41E-03

1.89E+02
2.59E+02
1.72E+02

7.34E-03
5.06E-03
1.19E-03

5.76E+02
5.52E+02
5.52E+02

4.86E-03
2.44E-03
6.31E-04

1.13E+04
9.62E+03
9.62E+03

1.26E-02
6.27E-03
3.15E-03

8.71E+02

9.18E-03

2.18E+02

5.10E-03

5.37E+02

2.44E-03

9.16E+03

6.28E-03

-

1.05E+03
9.24E+02

9.17E-03
1.34E-03

3.36E+02
1.72E+02

5.00E-03
1.16E-03

188

5.73E+02
5.52E+02

2.46E-03
6.03E-04

1.13E+04
9.62E+03

6.58E-03
1.30E-03

-1
Table 27. Concentrations of Metolachlor (µg L )

Base
Continuous
Canola
Continuous
Corn
Continuous
Corn stover
Continuous
Rye
Continuous
Sorghum
Continuous
Soy
Corn Soy
Corn Soy Rye
Corn stover
Soy
Miscanthus
Native Grass
Sorghum Soy
Switchgrass

40500
4050001
40802
40900
Max
Median
Max
Median
Max
Median
Max
Median
1.87E+03
3.39E-02
1.08E+03
1.75E-02
9.04E+02
6.97E-03
1.11E+04
1.48E-02
-

-

-

-

-

-

-

-

2.19E+03

6.35E-02

1.72E+03

1.79E-02

7.47E+02

6.64E-02

3.75E+06

1.01E-01

2.19E+03

6.45E-02

1.72E+03

1.80E-02

8.81E+02

6.65E-02

1.24E+04

6.03E-02

-

-

-

-

-

-

-

-

2.34E+03

7.30E-02

1.81E+03

2.13E-02

7.94E+02

7.25E-02

2.06E+04

7.24E-02

1.75E+03
2.06E+03
1.38E+03

4.14E-04
1.23E-03
1.00E-05

8.66E+02
1.06E+03
8.26E+02

8.44E-05
2.20E-04
3.49E-06

6.13E+02
7.53E+02
4.42E+02

7.07E-04
2.49E-03
8.80E-06

5.86E+03
6.20E+03
1.58E+03

9.03E-04
3.66E-03
1.68E-05

2.06E+03

1.25E-03

1.06E+03

2.18E-04

7.53E+02

2.51E-03

6.19E+03

3.66E-03

-

2.21E+03
9.34E+00

1.17E-03
1.32E-07

1.14E+03
4.83E+00

1.98E-04
8.17E-08

189

8.06E+02
6.87E+00

2.47E-03
6.14E-08

6.64E+03
1.20E+01

3.09E-03
3.43E-07

-1
Table 28. Concentrations of Pendimethalin (µg L )

40500
Base
Continuous
Canola
Continuous
Corn
Continuous
Corn stover
Continuous
Rye
Continuous
Sorghum
Continuous
Soy
Corn Soy
Corn Soy Rye
Corn stover
Soy
Miscanthus
Native Grass
Sorghum Soy
Switchgrass

Max
-

Median
-

4050001
Max
Median
-

Max
-

Median
-

Max
-

Median
-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

4.30E+01

3.42E-03

5.33E+01

7.73E-04

190

40802

2.92E+01

40900

4.79E-04

2.65E+02

2.49E-03

-1
Table 29. Concentrations of Sethoxydim (µg L )

40500
Base
Continuous
Canola
Continuous
Corn
Continuous
Corn stover
Continuous
Rye
Continuous
Sorghum
Continuous
Soy
Corn Soy
Corn Soy Rye
Corn stover
Soy
Miscanthus
Native Grass
Sorghum Soy
Switchgrass

Max
-

Median
-

4050001
Max
Median
-

4.82E-05

6.85E+02

4.80E+02

40802
Max
-

7.46E-07

40900
Median
-

4.17E+02

Max
-

2.16E-04

Median
-

0.00E+00

0.00E+00

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

191

-1
Table 30. Concentrations of Trifluralin (µg L )

40500
Base
Continuous
Canola
Continuous
Corn
Continuous
Corn stover
Continuous
Rye
Continuous
Sorghum
Continuous
Soy
Corn Soy
Corn Soy Rye
Corn stover
Soy
Miscanthus
Native Grass
Sorghum Soy
Switchgrass

Max
-

Median
-

4050001
Max
Median
-

9.68E-02

9.59E+02

1.30E+02

40802
Max
-

4.58E-02

40900
Median
-

5.00E+02

Max
-

2.58E-02

Median
-

1.11E+03

7.65E-02

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

192

-1
Table 31. Concentrations of 2,4-D (µg L )

40500
Base
Continuous
Canola
Continuous
Corn
Continuous
Corn stover
Continuous
Rye
Continuous
Sorghum
Continuous
Soy
Corn Soy
Corn Soy Rye
Corn stover
Soy
Miscanthus
Native Grass
Sorghum Soy
Switchgrass

Max
-

Median
-

4050001
Max
Median
-

Max
-

Median
-

Max
-

Median
-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

3.78E-03

8.68E+04

6.35E+04

40802

3.82E-04

40900

4.19E+04

6.51E-03

4.31E+04

1.01E-02

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

8.51E+02
-

7.07E-04
-

3.02E+02

6.38E+02
-

3.37E-05

6.14E-06
-

3.08E+02

5.66E+02
-

2.42E-06

193

2.02E-03
-

3.19E+02

5.42E+02
-

8.82E-05

7.98E-04
-

3.05E+02

1.76E-04

7.3.2 Basin-wide Impact of Pesticides on Stream Impairment
The concentration of agricultural chemicals in streams is highly variable, making
it difficult to predict possible waterbody impairments. This variability is important as it
oftentimes leads to excessive chemical concentrations that potentially degrade the quality
of the waterbody for human consumption and aquatic health. In order to better understand
the impacts of agricultural chemical applications on surface water contaminations, the
total length of impaired streams that exceeded the predefined threshold values for aquatic
and human health were calculated. Table 32 provides the total basinwide length of
impaired waterbodies for each rotation and herbicide. By finding the total length of
streams that exceeded the toxicity levels for the exposure limit, each rotation can be
compared to one another when they share the same pesticides applied.
By examining Table 32, it is important to note that the base scenario does have a
large number of stream reaches that exceeds the human threshold level for atrazine,
glyphosate, and metolachlor of 180597, 1033, and 1348 km, respectively, totaling
181,630 km of the 11,438,253 km total stream length (1.6 percent of total streams’
lengths). As discussed previously, continuous corn was observed to have the highest
basinwide glyphosate concentration. This was again observed in this analysis as
continuous corn yielded the most extensive length of streams exceeding the human
threshold (8572 km, an increase of 7719 km over base) and the only rotation to exceed
the bluegill threshold (1063 km of stream). This rotation also has atrazine and
metolachlor exceedance lengths of 657 and 2906 km, respectively, yielding a grand total
of 3966 km of streams exceeding the bluegill threshold criteria. Human thresholds were
also exceeded for the continuous corn rotation for atrazine, glyphosate, and metolachlor
totaling 541,141 km of stream impairments. The result of such extensive waterbody
194

impairments, whether bluegill or human, strongly challenges the suitability of continuous
corn rotations for large-scale implementation. Continuous sorghum experienced a small
amount of streams exceeding the metolachlor threshold criteria for bluegills, resulting in
6.3 km. The continuous canola rotation resulted in 17152 and 110790 km of streams
exceeding the bluegill and human thresholds for trifluralin, respectively.
Not all rotations experienced increases of pesticide impairments, though. Perhaps
the most significant decreases of threshold exceedances can be credited to the perennial
grass species. Switchgrass resulted in a 171,667 km reduction in total stream length hat
exceeds the human threshold criteria, and miscanthus and native grasses do not have any
impaired stream length. Miscanthus requires only a single application of pendemethalin
and 2,4-D during the establishment year only and native grasses do not require herbicide
application. Switchgrass, however, is not suggested to be cultivated for indefinite time
limits like miscanthus and native grass, and is suggested to incorporate corn and soybean
in the first two years of each rotation cycle, which is typically around 10 years. Thus,
switchgrass does have atrazine, metolachlor, and glyphosate applied in the first two
rotation years, followed by pendimethalin and 2,4-D during the establishment year.
However, even with these pesticide applications, switchgrass still manages to mitigate
overall pesticide contribution to streams. It is important to note that no herbicides are
currently labeled for application on miscanthus and switchgrass, the results presented in
this paper are therefore hypothetical scenarios in which are expected to occur if these
crops are adapted into a large-scale production scheme.

195

Table 32. Kilometers of stream segments exceeding threshold toxicity levels for the Bluegill and humans (the values with gray
background represent Bluegill and white represents human threshold exceedances).
Scenario
Base
Continuous
Canola
Continuous
Corn
Continuous
Corn stover
Continuous
Rye
Continuous
Sorghum
Continuous
Soybean
Corn
Soybean
Corn
Soybean
Rye
Corn stover
Soybean
Miscanthus

Atrazine

Bromoxynil Glyphosate

Metolachlor

Pendimethylin Sethoxydim

Trifluralin

2,4-D

Total

0
180597
0
0
656.9
532056
0
399423
0
0
0
339631
0
154049
0
173001

0
0
0
0
0
0
0
0
0
0
0
104.9
0
0
0
0

0
1033.1
0
0
1063.4
8571.9
0
422.3
0
1327.4
0
473.1
0
4727.4
0
3536.0

0
1348.4
0
0
2906
34121
0
4376.7
0
0
6.3
7769.7
0
1900.8
0
2733.5

0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

0
0
0
11.1
0
0
0
0
0
0
0
0
0
0
0
0

0
0
17151.8
110790
0
0
0
0
0
0
0
0
0
0
0
0

0
0
0
0
0
0
0
0
0
137677
0
0
0
0
0
0

0
181630
17152
108842
3969.5
541152
0
399846
0
139004
6.3
340198
0
158777
0
176537

0

0

0

0

0

0

0

0

0

81117
0
173169
0

0
0
0
0

3690.1
0
3392.5
0

181.9
0
2737.2
0

0
0
0
0

0
0
0
0

0
0
0
0

12972
0
0
0

97779
0
176561
0

196

Table 32 (cont’d)

Native
Grass
Sorghum
Soybean
Switchgrass

0
0
0
0
130941
0
212.1

0
0
0
0
84.4
0
0

0
0
0
0
4082.9
0
2589.8

0
0
0
0
3492.6
0
0

Note: Total basinwide stream length is 11,438,253 km

197

0
0
0
0
0
0
0

0
0
0
0
0
0
0

0
0
0
0
0
0
0

0
0
0
0
0
0
7161.4

0
0
0
0
158033
0
9963.2

7.3.3 Basin-wide Temporal Variation of Stream Impairment
As previously discussed in Section 7.3.1, the concentration of pesticides in the
stream fluctuates depending on many drivers such as application timing and amount,
precipitation, etc. This is important in determining the proper management practices to
attempt to minimize the amount of agricultural chemicals introduced to surface water
bodies. However, temporal variation is needed to adjust agricultural practices and
determine proper chemical application timing to minimize losses to streams and to
properly allocate conservation resources and perform monitoring to detect possible
threats to aquatic ecosystems and humans health. Figures 25 and 26 are provided to give
an example of the temporal variability of stream impairments due to pesticide
concentrations exceeding thresholds set forth to represent safe levels for aquatic
ecosystems (Bluegill) and humans uses, respectively. These analyses are presented only
for year 2 of the base, continuous corn, and switchgrass cropping rotations for all
pesticides. This year was chosen as pesticides were only applied to switchgrass during the
establishment year, which was the second modeled year.
The length of streams in exceedance of the threshold level for the base scenario
includes glyphosate, atrazine, and metolachlor, continuous corn includes atrazine,
glyphosate, metolachlor, and the switchgrass rotation scenario contains pendimethalin
and 2,4-D. Overall, June appears to be the month that the most streams (in length) exceed
the threshold limit for the Bluegill, followed by July with 452.4 and 441 kilometers,
respectively. Meanwhile, May experiences a considerably less amount (58.6 kilometers)
yet this marks the start of the pesticide concentration increase in surface waters. This is
expected as this period has substantial precipitation and it follows the planting dates of

198

corn, which has a pre-emergent application of atrazine and metolachlor on the day of
planting. A similar trend was observed for human threshold exceedances (Figure 26)
where May, June and July had the most stream segments exceeding the human health
thresholds for atrazine, glyphosate, and metolachlor. Again, base and continuous corn
have pre-emergent applications of atrazine and metolachlor in May, whereas switchgrass
has pendimethalin and 2,4-D applications in June, explaining why switchgrass has the
most exceedances in this month. The decreasing length of exceeded streams following
the month of application for each rotation reflects the trend of decreasing available
chemical concentrations in soil and on foliage. These chemicals have short half-lives (3
to 249 days) and degrade quickly in aerobic conditions present throughout the study area.
By determining the most susceptible times of the year, appropriate conservation planning
can follow, effectively decreasing the amount of impaired waterbodies. With a proper
assessment of spatial variation in compliment with this temporal variation could provide
a more comprehensive understanding of the effects bioenergy cropping systems may
have on surface water impairments.

199

Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
0

50

100

150

200

250

300

350

400

450

500

Kilometers of Stream Length in Exceedance
Switchgrass

Continuous Corn

Base

Figure 25. Basinwide temporal variation in stream length impairments of three crop rotation scenarios based on fish thresholds

200

Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
0

10000

20000

30000

40000

50000

60000

70000

Kilometers of Stream Length in Exceedance
Switchgrass

Continuous Corn

Base

Figure 26. Basinwide temporal variation in stream length impairments of three crop rotation scenarios based on human thresholds

201

7.3.4 Basin-wide Spatial Variation of Stream Impairment
In order to control and mitigate the negative impacts of bioenergy crop expansion,
it is important to first identify the locations within the watershed in which the impairment
occur. This will allow us to not only identify the source of impairment, but also, helps in
the decision making processes by identifying the most appropriate control practices that
can be implemented. Based on the previous section of this study (Section 7.3.3), the
months with the largest length of stream impairments were identified. May was observed
as the worst month for human threshold exceedances for the base and continuous corn
scenarios, whereas June was the worst for the human threshold exceedances in the
switchgrass rotation and aquatic exceedances in the continuous corn rotation (only
rotation to exceed bluegill thresholds). In the next step, the concentrations for each
stream segment for the day in this month with the worst amount of exceedance were
collected and mapped (Figure 27). It is important to note that Figure 27 presents the worst
case scenario; however, the risk associated with impairment is significant. For example,
the continuous corn rotation is the only rotation to exceed bluegill ecotoxicity levels
while both base and switchgrass scenarios did not cause impairment concerning bluegill
health. When comparing these rotations based on human health thresholds, it is observed
that all have the potential to drastically impair surface waters for human consumption.
Continuous corn has an extensive amount of impairment (77.8 percent of total stream
length impaired), due to the increased pesticide use associated with the large-scale
implementation of this intensive row crop. However, impairment on the given day for
switchgrass results in only 38 percent of total stream length exceeding the human toxicity
threshold.

202

It is important to note here that base does have a fair amount of impairment itself
(34.5 percent of stream lengths), indicating that the current practices in this study area are
inadequate at mitigating pollutants and have the potential to continue to create hazardous
conditions for drinking water. This is, of course, in the absence of water treatment, which
is costly and not available throughout the entirety of the study area. These findings
suggest the great importance of taking caution when implementing agricultural practices
or altering landscapes on a large-scale. Shifts of landuse towards bioenergy cropping
rotations have the potential to alter the safety of surface waters in regards to pesticide
concentration, which could potentially increase with intensive row crops or decrease with
second generation crops such as switchgrass.

203

Figure 27. Streams exceeding Bluegill toxicity threshold (top) and human threshold (bottom) for base, continuous
corn, and switchgrass (in order left-to-right)

204

7.4 CONCLUSIONS
Recent shifts in the global energy sector have led to an increased demand for
bioenergy feedstock production for renewable energy. However, the environmental
consequences of this shift in agricultural production system are not well known,
especially in regards to aquatic and human health. Therefore, determination of which
biofuel crops and management strategies may have the least unfavorable consequences to
environments is vital for future decision making (Thomas et al., 2009). In order to
evaluate these consequences, a large-scale assessment using models is imperative.
The Soil and Water Assessment Tool (SWAT) was used in this study to evaluate
the effects that large-scale bioenergy crop production may have on pesticide fate and
transport and subsequently on aquatic and human health. The evaluation included four
2
large watersheds (totaling 53,358 km ), 14 bioenergy cropping rotations, and 8 pesticides

including atrazine, bromoxynil, glyphosate, metolachlor, pendimethalin, sethoxydim,
trifluralin, and 2,4-D. The watersheds included in this study are intensively cultivated for
agricultural production and all flow into the Great Lakes. Due to the manner in which
SWAT simulate pesticide, each pesticide had to be individually evaluated. Therefore, 168
model runs were performed and the model output was evaluated for a six year period
(2000-2005) on a daily time step.
The results obtained from this study were arranged in three sections and can be
summarized in the following:
•

7.3.1 Basin-wide Concentration Fluctuation: Among all studied crop rotations,
continuous corn caused the highest concentrations of Glyphosate to the stream. In

205

essence, this rotation has the greatest possibility of containing the highest levels
of glyphosate, well above both the bluegill and human toxicity levels. The result
of such extensive waterbody impairments strongly challenges the suitability of
continuous corn rotations for large-scale implementation. However, the base
scenario (current landcover and practices) actually did not reach the threshold for
bluegill ecotoxicity, which is an indicator that current trends in agricultural
production may be tolerable for aquatic ecosystems. In general, all of the
bioenergy cropping scenarios have the possibility of jeopardizing the safety of
surface waters for human consumption due to glyphosate contamination. In
addition, atrazine and metolachlor, which are both very commonly applied in the
study area, experience similar results with increased maximum and median
concentration over the base scenario, indicating the low suitability of these crops
to be adopted at a large scale. Crop rotations that incorporate close-grown grass
species, especially switchgrass and native grasses, are the most suitable for largescale implementation.
•

7.3.2 Basin-wide Impact of landuse conversion on Stream Impairment: By finding
the total length of streams that exceeded the toxicity levels for the exposure limit,
each rotation can be compared to one another when the same pesticides are
applied. The base scenario does result in a large amount of stream impairments
for either human or aquatic bluegills for atrazine, glyphosate, and metolachlor.
Meanwhile, continuous corn rotation caused water body impairment (exceeding
the human consumption threshold) from 181629.9 km (under current landuse
scenario) to 541151.5 km. In addition, this rotation also degraded water quality

206

for aquatic ecosystems in 3969.5 km of stream in which atrazine, glyphosate, and
metolachlor concentration exceeded the bluegill threshold of LC50. Meanwhile,
in many cases the results were positive. Meanwhile, the pollution mitigation
exhibited by the perennial grass species. For example, switchgrass resulted in a
171,667 km reduction in total impaired stream that exceeds the human threshold
criteria, and miscanthus and native grasses rotation did not cause any water
quality impairment.
•

7.3.3 Basin-wide Temporal Variation of Stream Impairment: Temporal variation
in water quality is important to identify threats to aquatic ecosystems and humans
health during various time periods to focus conservation efforts. Overall, June
appears to be the month that the most streams (in length) exceed the threshold
limit for the Bluegill. A similar trend was observed for human threshold
exceedances where May, June and July had the most stream segments exceeding
the human health thresholds for atrazine, glyphosate, and metolachlor. By
determining the most susceptible times of the year, appropriate conservation
planning can follow, effectively decreasing the amount of impaired waterbodies.

•

7.3.4 Basin-wide Spatial Variation of Stream Impairment: Identifying locations
that are at high risk for exceeding aquatic and human toxicity levels aids in
watershed management to avoid waterbody impairment. Applying agricultural
chemicals in a region where surface runoff and sedimentation are common leads
to the possibility of waterbody impairment. For a one-day period representing the
worst-case scenario, continuous corn has the potential for extensive stream
impairment (77.8 percent of total stream length impaired), due to the increased

207

pesticide use associated with the large-scale implementation of this intensive row
crop. However, impairment for switchgrass results in only 38 percent of total
stream length exceeding the human toxicity threshold, whereas base impairs 34.5
percent of stream lengths. These findings indicate that converting a large amount
of land into bioenergy crop production may not be suitable in regards to
watershed health.
Based on these results, large environmental cost might be associated with future
large-scale bioenergy crop

expansion. However,

the variability of pesticide

concentrations between different geographic landscapes suggests that these crops may be
suitable if pesticides application managed properly. In addition, due to complexity of
pesticide fate in the environment, water quality monitoring is essential to track the
potential hazardous conditions. In general, intensive bioenergy crops are the least suitable
for large-scale implementation. However, this brings about an important concern: could
continuous corn coupled with widespread conservation practices reduce overland
pesticide contribution to surface waters enough to avoid such far-reaching water quality
impairments? In order to expand our understanding of the effects that large-scale
bioenergy expansion may have on aquatic ecosystems and human health, it is suggested
that additional studies are performed that include multiple aquatic species assessments in
different climatological and physiographic regions to confirm the findings discussed in
this paper. In addition, a more widespread collection of observed data is required to
obtain adequate model calibration for a more detailed and confident validation of
pesticide transport in surface waters. Finally, the authors reinforce the importance of
chemical application timing, in which controlling this is perhaps the most cost-effective

208

and efficient method in directly reducing the amount of pesticides in surface runoff and
thus minimizing waterbody impairments.

Acknowledgements
The authors would like to acknowledge Dr. Lizhu Wang, Mr. Sean Woznicki and
Mr. Mike Prohaska for their added project support, and Dr. Kurt Thelen, Dr. Wesley
Everman, Mr. Dennis Pennington, Dr. Scott Swinton, Ms. Laura James, and Mr. Joseph
Duris for their generous contribution of information.

209

8. CONCLUSIONS
8.1 GENERAL CONCLUSIONS
The research presented here evaluated the potential impacts that large-scale
bioenergy crop expansion could have on water quality, aquatic ecosystems, and human
health. A comprehensive and physically based hydrologic model, the Soil and Water
Assessment Tool (SWAT), was used to predict pollutant loads in a large-scale due to
conversion of current landuse to extensive bioenergy crop production. The study area
mainly consists of agricultural regions of Michigan with highly variable climatic and
physiographic conditions. The model was subject to various landuse scenarios, bioenergy
crop rotations, management practices, and time steps. Model output was evaluated based
on a comparison with the current scenario (current landcover), which for all purposes of
this research, is assumed to represent of future conditions.
In the first section of this research, field studies were conducted in which seven
herbicide treatments were applied to field and silage corn to determine the effects on
biomass accumulation and glucose percentage for cellulosic ethanol. The second section
of this research investigated the effects that management practices have on model output
confidence, and the fluctuation in pollutant loads on a temporal scale to demonstrating
the extreme fluctuation of pollutant loadings throughout the year. In the third section of
this research, a similar approach was taken, only on a much larger scale, which ultimately
helped to develop priority areas for focusing conservation implementation efforts. In the
last section of this research, predicted pesticide concentration was compared to fish and
human toxicity threshold limits to determine the environmental impacts of various

210

bioenergy cropping systems. The results found in these research papers can be concluded
in the following:
Postemergence herbicides can have potentially beneficial biomass and
glucose yield effects when applied to corn that will be used as a feedstock
for cellulosic ethanol production.
Perennial grass species are most suitable for large-scale implementation in
this study area, whereas traditional intensive row crops should be
implemented with caution on such a broad scale as these crops have the
greatest potential to detrimentally affect watershed health.
Only bioenergy crops grown on row crop land experiences significant decreases
in total nitrogen load.
Bioenergy row crops exhibit dramatic pollution load variation caused by
differences in climate and physiographic characteristics throughout the study
area.

Traditional intensive row crops such as canola, corn and sorghum may
negatively impact aquatic life and in most cases affect the availability of
safe drinking water.
Second-generation lignocellulosic bioenergy crops such as switchgrass
show promise in reducing in total impaired stream length that exceeds the
threshold criteria for human consumption.

211

8.2 RECOMMENDATIONS FOR FUTURE RESEARCH
The information presented in this research manuscript is valuable for decision
makers regarding the impacts that landuse management has on environmental quality.
However there are many opportunities to expand on this work and provide more in-depth
research that addresses issues outside of the scope of this project. The following
suggestions are recommended for further research:
Due to climatic and geographic restraints, a number of potential bioenergy
crops (i.e. rice, cassava, sugar cane, jatropha) were not examined based on
poor suitability for our study region. Additional studies in different
climatological and physiographic regions are needed that include such
crops.
In order to expand our understanding of the effects that large-scale
bioenergy expansion may have on aquatic ecosystems and human health,
it is suggested that additional studies are performed that include multiple
specie assessments in different climatological and physiographic regions
to confirm the findings discussed in this paper.
A more widespread collection of observed pesticide data is required to
obtain adequate model calibration for a more detailed and confident
validation of pesticide transport in surface waters.
The version of SWAT used in this project does not adequately model
static waterbodies such as lakes and wetlands. By incorporating these
features in watershed delineation and hydrologic/pollutant transport

212

processes, the predictions of constituent loads would be a better
representation of natural processes.
Bioenergy crops should maximize efficiency and be cost-effective, while
minimizing environmental impact. It would be useful to evaluate many of
the ideas presented in this manuscript based on conservation practices
such as no-till or edge-of-field filter strips.
Implement a mass balance approach to determine the amount of nutrients
leaving the field through the harvesting of biomass. This may help to
demonstrate the effectiveness of bioenergy crops to keep nutrients on site.
Determine the optimal landuse configuration (percentage and location
landuses) to minimize food vs. fuel competition and negative water quality
impacts.

213

REFERENCES

214

REFERENCES
Adler, P.R., Del Grosso, S.J., Parton, W.J., 2007. Life-cycle assessment of net
greenhouse-gas flux for bioenergy cropping systems. Ecological Applications 17,
675-691.
Ajami, N. K., H. Gupta, T. Wagener & S. Sorooshian (2004) Calibration of a semidistributed hydrologic model for streamflow estimation along a river system.
Journal of Hydrology, 298, 112-135.
Al-Abed, N., Al-Sharif, M., 2008. Hydrological modeling of Zarqa River Basin - Jordan
using the Hydrological Simulation Program-FORTRAN (HSPF) model. Water
Resources Management 22, 1203-1220.
Alla, M. M. N. & M. E. Younis (1995) Herbicide effects on phenolic metabolism in
maize (Zea-mays L) and soybean (Glycine max L) seedlings. Journal of
Experimental Botany, 46, 1731-1736.
Alley, R.B., 2004. Abrupt climate change. Scientific American 291, 62-69.
Antizar-Ladislao, B. & J. L. Turrion-Gomez (2008) Second-generation biofuels and local
bioenergy systems. Biofuels Bioproducts & Biorefining-Biofpr, 2, 455-469.
Arnold, J. G., R. Srinivasan, R. S. Muttiah & J. R. Williams. (1998) Large area
hydrologic modeling and assessment part I: model development. Journal of
American Water Resources Association 34(1), 73-89.
Arnold, J.G., Muttiah, R.S., Srinivasan, R., Allen, P.M., 1998. Regional estimation of
base flow and groundwater recharge in the Upper Mississippi River basin. ASAE
Annual International Meeting, Orlando, Florida, USA, 12-16 July, 1998., 14 pp.
Askew, S. D. & J. W. Wilcut (1999) Cost and weed management with herbicide
programs in glyphosate-resistant cotton (Gossypium hirsutum). Weed
Technology, 13, 308-313.
Baker, L.A. (1992) Introduction to nonpoint source pollution in the United States and
prospects for wetland use. Ecological Engineering, v. 1, 1–26.
Balat, M., Balat, H., 2009. Recent trends in global production and utilization of bioethanol fuel. Applied Energy 86, 2273-2282.
Baldwin, D.H., Spromberg, J.A., Collier, T.K., Scholz, N.L. (2009) A Fish of Many
Scales: Extrapolating Sublethal Pesticide Exposures to the Productivity of Wild
Salmon Populations. Ecological Applications. 19(8): 2004-2015.
Balvanera, P., Pfisterer, A.B., Buchmann, N., He, J.S., Nakashizuka, T., Raffaelli, D.,
Schmid, B., 2006. Quantifying the evidence for biodiversity effects on ecosystem
functioning and services. Ecology Letters 9, 1146-1156.
215

Baxter, J., Cummings, S.P., 2008. The degradation of the herbicide bromoxynil and its
impact on bacterial diversity in a top soil. Journal of Applied Microbiology 104,
1605-1616.
Belden, J.B.,Gilliom,R.J., Martin, J.D., Lydy, M.J. (2007) Realtive Toxicity and
Occurrence Patterns of Pesticide Mixtures in Streams Draining Agricultural
Watersheds Dominated by Corn and Soybean Production. Integrated
Environmental Assessment and Management. 3(1): 90 -100.
Berndes, G., 2002. Bioenergy and water - the implications of large-scale bioenergy
production for water use and supply. Global Environmental Change-Human and
Policy Dimensions 12, 253-271.
Berndes, G., Hoogwijk, M., van den Broek, R., 2003. The contribution of biomass in the
future global energy supply: a review of 17 studies. Biomass & Bioenergy 25, 128.
Bicknell, B. R., J. C. Imhoff, J. L. Kittle, Jr., A. S. Donigian, Jr., nd R. C. Johanson.
1997. Hydrological simulation program—Fortran: User’s manual for version 11.
US Environmental rotection Agency, National Exposure Research Laboratory,
Athens, Georgia. EPA/600/R-97/080, 755 pp.
Bicknell, B.R., Imhoff, J.C., Kittle, J.L., Jobes, T.H., Donigian, A.S. (2001) Hydrological
Simulation Program – FORTRAN, HSPF: Version 12. User’s Manual. Reviewed
by EPA. March 2001.
Bingner, R. L., Theurer, F. D., & Yuan, Y. (2007). AnnAGNPS technical processes.
Model version 4.0. USDA-ARS. Oxford, MS: National Sedimentation
Laboratory.
Borah, D.K., Bera, M., 2003. Watershed-scale hydrologic and nonpoint-source pollution
models: Review of mathematical bases. Transactions of the Asae 46, 1553-1566.
Bos, M. M., P. Höhn, S. Saleh, B. Büche, D. Buchori, I. Steffan-Dewenter, and T.
Tscharntke. 2007. Insect diversity responses to forest conversion and agroforestry
management. In T.Tscharntke, C.Leuschner, M.Zeller, E.Guhardja, and A.Bidin
(Eds.). Stability of tropical rain forest margins, linking ecological, economic and
social constraints of land use and conservation, pp. 279–296. Springer, Berlin,
Germany.
Bosch, D., Theurer, F., Bingner, R., Felton, G., Chaubey, I. (1998) Evaluation of the
AnnAGNPS Water Quality Model. Written for Presentation at 1998 ASAE
Annual International Meeting. Paper # 982195.
Carpenter, S.R., Caraco, N.F., Correll, D.L., Howarth, R.W., Sharpley, A.N., Smith,
V.H., 1998. Nonpoint pollution of surface waters with phosphorus and nitrogen.
Ecological Applications 8, 559-568.

216

Carroll, A., Somerville, C., 2009. Cellulosic Biofuels. Annual Review of Plant Biology
60, 165-182.
CDC, Center for Disease Control and Prevention. (2010) Metolachlor, Chamical
information. National Report on Human Exposure to Environmental Chemicals.
CAS No. 93-76-5. Available from:
http://www.cdc.gov/exposurereport/data_tables/Metolachlor_
ChemicalInformation.html
Chan, K.M.A., Shaw, M.R., Cameron, D.R., Underwood, E.C., Daily, G.C., 2006.
Conservation planning for ecosystem services. Plos Biology 4, 2138-2152.
Chapin, F.S., Zavaleta, E.S., Eviner, V.T., Naylor, R.L., Vitousek, P.M., Reynolds, H.L.,
Hooper, D.U., Lavorel, S., Sala, O.E., Hobbie, S.E., Mack, M.C., Diaz, S., 2000.
Consequences of changing biodiversity. Nature 405, 234-242.
Choi, W., Deal, B.M., 2008. Assessing hydrological impact of potential land use change
through hydrological and land use change modeling for the Kishwaukee River
basin (USA). Journal of Environmental Management 88, 1119-1130.
Chum, H.L., Overend, R.P., 2001. Biomass and renewable fuels. Fuel Processing
Technology 71, 187-195.
Coffey, R., Cummins, E., Bhreathnach, N., Flaherty, V.O., Cormican, M., 2010.
Development of a pathogen transport model for Irish catchments using SWAT.
Agricultural Water Management 97.
Cook, M.E., Moore, P.A. (2008) The Effects of the Herbicide Metolachlor on Agonistic
Behavior in the Crayfish, Orconectes rusticus. Arch Environ Contam Toxicol.
55:94-102.
Costanza, R., d'Arge, R., de Groot, R., Farber, S., Grasso, M., Hannon, B., Limburg, K.,
Naeem, S., O'Neill, R.V., Paruelo, J., Raskin, R.G., Sutton, P., van den Belt, M.,
1998. The value of the world's ecosystem services and natural capital (Reprinted
from Nature, vol 387, pg 253, 1997). Ecological Economics 25, 3-15.
Cottingham, C. K. & K. K. Hatzios (1992) Basis of differential tolerance of 2 corn
hybrids (Zea-mays) to metolachlor. Weed Science, 40, 359-363.
Daily, G.C., Polasky, S., Goldstein, J., Kareiva, P.M., Mooney, H.A., Pejchar, L.,
Ricketts, T.H., Salzman, J., Shallenberger, R., 2009. Ecosystem services in
decision making: time to deliver. Frontiers in Ecology and the Environment 7, 2128.
Daily, G.C., Soderqvist, T., Aniyar, S., Arrow, K., Dasgupta, P., Ehrlich, P.R., Folke, C.,
Jansson, A., Jansson, B.O., Kautsky, N., Levin, S., Lubchenco, J., Maler, K.G.,
Simpson, D., Starrett, D., Tilman, D., Walker, B., 2000. Ecology - The value of
nature and the nature of value. Science 289, 395-396.

217

de Fraiture, C., Giordano, M., Liao, Y.S., 2008. Biofuels and implications for agricultural
water use: blue impacts of green energy. Water Policy 10, 67-81.
Domagalski, J., Ator,S., Coupe, R., McCarthy, K., Lampe, D., Sandstrom, M., Baker, N.
(2008) Comparative Study of Transport Processes of Nitrogen, Phosphorus, and
Herbicides to Streams in Five Agricultural Basins, USA. J. Environ. Qual. 37:
1158-1169
Du, B., Saleh, A., Jaynes, D.B., Arnold, J.G., 2006. Evaluation of SWAT in simulating
nitrate nitrogen and atrazine fates in a watershed with tiles and potholes.
Transactions of the Asabe 49, 949-959.
Duke, S.O., Powles, S.B. (2008) Mini Review, Glyphosate: a once-in-a-century
herbicide. Pest Manag. Sci. 64: 319-325.
EXTOXNET (1996) Extension Toxicology Network. Pesticide Information Profiles.
Available from: http://extoxnet.orst.edu/pips.htm
Fan, W.Q., Yanase, T., Morinaga, H., Ondo, S., Okabe, T., Nomura, M., Komatsu, T.,
Morohashi, K.I., Hayes, T.B., Takayanagi, R., Nawata, H., 2007. Atrazineinduced aromatase expression is SF-1 dependent: Implications for endocrine
disruption in wildlife and reproductive cancers in humans. Environmental Health
Perspectives 115, 720-727.
Fargione, J., Hill, J., Tilman, D., Polasky, S., Hawthorne, P., 2008. Land clearing and the
biofuel carbon debt. Science 319, 1235-1238.
Fargione, J.E., Cooper, T.R., Flaspohler, D.J., Hill, J., Lehman, C., McCoy, T., McLeod,
S., Nelson, E.J., Oberhauser, K.S., Tilman, D., 2009. Bioenergy and wildlife:
threats and opportunities for grassland conservation. BioScience 59, 767-777.
Fleeger, J.W., Carman, K.R., Nisbet, R.M. (2003) Indirect Effects of Contaminants in
Aquatic Ecosystems. The Science of the Total Environment. 317(1-3) 207-233.
Foley, J.A., DeFries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, S.R., Chapin,
F.S., Coe, M.T., Daily, G.C., Gibbs, H.K., Helkowski, J.H., Holloway, T.,
Howard, E.A., Kucharik, C.J., Monfreda, C., Patz, J.A., Prentice, I.C.,
Ramankutty, N., Snyder, P.K., 2005. Global consequences of land use. Science
309, 570-574.
Galvan, L., Olias, M., de Villaran, R.F., Santos, J.M.D., Nieto, J.M., Sarmiento, A.M.,
Canovas, C.R., 2009. Application of the SWAT model to an AMD-affected river
(Meca River, SW Spain). Estimation of transported pollutant load. Journal of
Hydrology 377, 445-454.
Garcia-Quijano, J.F., Peters, J., Cockx, L., van Wyk, G., Rosanov, A., Deckmyn, G.,
Ceulemans, R., Ward, S.M., Holden, N.M., Van Orshoven, J., Muys, B., 2007.

218

Carbon sequestration and environmental effects of afforestation with Pinus radiata
D. Don in the Western Cape, South Africa. Climatic Change 83, 323-355.
Gassman, P.W., Reyes, M.R., Green, C.H., Arnold, J.G., 2007. The soil and water
assessment tool: Historical development, applications, and future research
directions. Transactions of the Asabe 50, 1211-1250.
Giller, P.S., Hillebrand, H., Berninger, U.G., Gessner, M.O., Hawkins, S., Inchausti, P.,
Inglis, C., Leslie, H., Malmqvist, B., Monaghan, M.T., Morin, P.J., O'Mullan, G.,
2004. Biodiversity effects on ecosystem functioning: emerging issues and their
experimental test in aquatic environments. Oikos 104, 423-436.
Gillingham, K. T., S. J. Smith & R. D. Sands (2008) Impact of bioenergy crops in a
carbon dioxide constrained world: an application of the MiniCAM energyagriculture and land use model. Mitigation and Adaptation Strategies for Global
Change, 13, 675-701.
Gilliom, R.J. (2007) Pesticides in U.S. Streams and Groundwater. Environmental Science
and Technology. May 15, 2007. 3409-3414
Gleick, P.H., 1998. Water in crisis: Paths to sustainable water use. Ecological
Applications 8, 571-579.
GLIN (Great Lakes Information Network). 2009. Areas of Concern (AOC’s). Great
Lakes Commission, Ann Arbor, Michigan. Available from http://www.greatlakes.net/teach/ pollution/aoc/aoc1.html
Goebel, J. J. 1998. The national resources inventory and its role in U.S. agriculture. In
Agricultural Statistics 2000: An International Conference on Agricultural
Statistics, 181-192. T. E. Holland and M. P. R. Van den Broecke, eds. Voorburg,
The Netherlands: International Statistical Institute.
Goldemberg, J., 2007. Ethanol for a sustainable energy future. Science 315, 808-810.
Goodbred, S.L., Gilliom, R.J., Gross, T.S., Denslow, N.P., Bryant, W.L., and Schoeb,
T.R., 1997, Reconnaissance of 17b-estradiol, 11-ketotestosterone, vitellogenin,
and gonad histopathology in common carp of United States Streams— Potential
for contaminant-induced endocrine disruption: U.S. Geological Survey Open-File
Report 96–627, 47 p.
Green, J. M. & J. F. Ulrich (1993) Responses of corn (Zea-mays L) inbreds and hybrids
to sulfonylurea herbicides. Weed Science, 41, 508-516.
Green, J. M. (1998) Differential tolerance of corn (Zea mays) inbreds to four sulfonylurea
herbicides and bentazon. Weed Technology, 12, 474-477.

219

Gregory, J.M., Mitchell, J.F.B., Brady, A.J., 1997. Summer drought in northern
midlatitudes in a time-dependent CO2 climate experiment. Journal of Climate 10,
662-686.
Groom, M.J., Gray, E.M., Townsend, P.A., 2008. Biofuels and biodiversity: Principles
for creating better policies for biofuel production. Conservation Biology 22, 602609.
Grover, R., Wolt, J.D., Cessna, A.J., Schiefer, H.B. (1997) Environmental Fate of
Trifluralin. Rev Environ Contam Toxicol. 153:1-64
Gupta, H. V., Sorooshian, S., Yapo, P. O., 1999. Status of Automatic Calibration for
Hydrologic Models: Comparison with Multilevel Expert Calibration. Journal of
Hydrologic Engineering, 4(2), 135-143.
Guthrie, G., Kumareswaran, D., 2009. Carbon Subsidies, Taxes and Optimal Forest
Management. Environmental & Resource Economics 43, 275-293.
Hall, M. R., C. J. Swanton & G. W. Anderson (1992) The critical period of weed-control
in grain corn (Zea-mays). Weed Science, 40, 441-447.
Hanlon, P., McCartney, G., 2008. Peak oil: Will it be public health's greatest challenge?
Public Health 122, 647-652.
Harman, W.L., Wang, E., Williams, J.R. (2004) Reducing Atrazine Losses: Water
Quality Implications of Alternative Runoff Control Practices. J. Environ. Qual.
33:7-12.
Harvey, R. G. & C. R. Wagner (1994) Using estimates of weed pressure to establish crop
yield loss equations. Weed Technology, 8, 114-118.
Heathman, G.C., Flanagan, D.C., Larose, M., Zuercher, B.W. (2008) Applications of the
Soil and Water Assessment Tool and Annualized Agricultural Non-Point Source
Models in the St. Joseph River Watershed. Journal of Soil and Water
Conservation; Nov/Dec 2008; 63, 6.
Helfrich, L.A., Weigmann, D.L., Hipkins, P., Stinson, E.R. (2009) Pesticides and
Aquatic Animals: A Guide to Reducing Impacts on Aquatic Systems. Virgina
Tech University
Himmel, M. E., S. Y. Ding, D. K. Johnson, W. S. Adney, M. R. Nimlos, J. W. Brady &
T. D. Foust (2007) Biomass recalcitrance: Engineering plants and enzymes for
biofuels production. Science, 315, 804-807.
Holden, A.V., 1972. Effects of Pesticides on Life in Fresh Waters. Proceedings of the
Royal Society of London Series B-Biological Sciences 180, 383-&.

220

Holshouser, D.L., Coble, H.D., 1990. Compatibility of Sethoxydim with 5
Postemergence Broadleaf Herbicides. Weed Technology 4, 128-133.
Holvoet, K., Griensven A., Gevaert, V., Seuntjens, P., Vanrolleghem, P.A. (2008)
Modifications to the SWAT Code for Modeling Direct Pesticide Losses.
Environmental Modelling and Software. 23(1): 72-81.
Holvoet, K., Griensven, A.v., Seuntjens, P., Vanrolleghem, P.A., 2004. Hydrodynamic
modelling with Soil and Water Assessment Tool (SWAT) for predicting dynamic
behaviour of pesticides. 2nd Young Researchers Conference, University of
Wageningen, Netherlands, 23-24 April 2004, 211-218.
Holvoet, K.M.A., Seuntjens, P., Vanrolleghem, P.A., 2007. Monitoring and modeling
pesticide fate in surface waters at the catchment scale. Ecological Modelling 209,
53-64.
Hoogeveen, J., Faures, J.M., Van de Giessen, N., 2009. Increased Biofuel Production in
the Coming Decade: To What Extent Will it Affect Global Freshwater Resources?
Irrigation and Drainage 58, S148-S160.
Huber, A., Bach, M., Frede, H.G., 1998. Modeling pesticide losses with surface runoff in
Germany. Science of the Total Environment 223, 177-191.
Huber, A., Bach, M., Frede, H.G., 2000. Pollution of surface waters with pesticides in
Germany: modeling non-point source inputs. Agriculture Ecosystems &
Environment 80, 191-204.
Huston, M.A., Marland, G., 2003. Carbon management and biodiversity. Journal of
Environmental Management 67, 77-86.
Hutson, S. S., N. L. Barber, J. F. Kenny, K. S. Linsey, D. S. Lumia, & M. A. Maupin,
2004, Estimated use of water in the United States in 2000: Reston, Va., U.S.
Geological Survey Circular 1268, 46 p.
Jairo Diaz-Ramirez, Zhiyong Duan, William McAnally, James Martin (2005) Sensitivity
of the HSPF Model to Land Use/Land Cover Datasets. Journal of Coastal
Research: Special Issue 52 - Surface Water Modeling [Juang & Chen]: pp. 89-94.
Johnson, R. 2010. "What Is the Farm Bill”. Congressional Research Service Report,
RS22131. Available from http://ncseonline.org/nle/crsreports/10Jun/RS22131.pdf
Kamble, A.K., Tripathi, M.P., Pawar, S.N., Sawant, B.P., 2005. Estimation of surface
runoff from micro watershed using Soil and Water Assessment Tool (SWAT)
model. Indian Journal of Dryland Agricultural Research and Development 20, 4656.
Kannan, N., White, S.M., Worrall, F., Whelan, M.J., 2006. Pesticide modelling for a
small catchment using SWAT-2000. Journal of Environmental Science and

221

Health Part B-Pesticides Food Contaminants and Agricultural Wastes 41, 10491070.
Karickhoff, S.W., Brown, D.S., Scott, T.A., 1979. Sorption of Hydrophobic pollutants on
Natural Sediments. Water Research 13, 241-248.
Karr, J.R., 1991. Bilogical Integrity: A long-neglected Aspect of Water-Resource.
Ecological Applications 1, 66-84.
Khanna, M., Dhungana, B., Clifton-Brown, J., 2008. Costs of producing miscanthus and
switchgrass for bioenergy in Illinois. Biomass & Bioenergy 32, 482-493.
Klein, A.M., Vaissiere, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A.,
Kremen, C., Tscharntke, T., 2007. Importance of pollinators in changing
landscapes for world crops. Proceedings of the Royal Society B-Biological
Sciences 274, 303-313.
Knops, J. M. H. & D. Tilman (2000) Dynamics of soil nitrogen and carbon accumulation
for 61 years after agricultural abandonment. Ecology, 81, 88-98.
Koh, L., Ghazoul, J., 2008. Biofuels, biodiversity, and people: understanding the conflicts
and finding opportunities. Biological Conservation 141, 2450-2460.
Krause P., Boyle, D. P., Bäse, F., 2005. Comparison of Different Efficiency Criteria for
Hydrologic Model Assessment. Advances in Geosciences, 5, 89-97.
Krausz, R. F., G. Kapusta, J. L. Matthews, J. L. Baldwin & J. Maschoff (1999)
Evaluation of glufosinate-resistant corn (Zea mays) and glufosinate: Efficacy on
annual weeds. Weed Technology, 13, 691-696.
Kremen, C., 2005. Managing ecosystem services: what do we need to know about their
ecology? Ecology Letters 8, 468-479.
Kunkel, D. L., R. R. Bellinder & J. C. Steffens (1996) Safeners reduce corn (Zea mays)
chloroacetanilide and dicamba injury under different soil temperatures. Weed
Technology, 10, 115-120.
Lal, R., Pimentel, D., 2007. Biofuels from crop residues. Soil & Tillage Research 93,
237-238.
Landis, D.A., Gardiner, M.M., Werf, W.v.d., Swinton, S.M., 2008. Increasing corn for
biofuel production reduces biocontrol services in agricultural landscapes.
Proceedings of the National Academy of Sciences of the United States of America
105, 20552-20557.
Laroche, A., Gallichand, J., Lagace, Pesant, A. (1996) Simulating Atrazine Transport
with HSPF in an Agricultural Watershed. Journal of Environmental Engineering.
122(7): 622 – 630.

222

Larose, M., Heathman, G.C., Norton, L.D., Engel, B. (2007) Hydrologic and Atrazine
Simulation of the Cedar Creek Watershed Using The SWAT Model. Journal of
Environmental Quality. 36: 521-531.
Laurijssen, J., Faaij, A.P.C., 2009. Trading biomass or GHG emission credits? Climatic
Change 94, 287-317.
Leonard, R.A., W.G. Knisel., and D.A. Still. 1987. GLEAMS: Groundwater Loading
Effects of Agricultural Management Systems. Transactions of ASAE. 30: 14031418.
Lewis, N.S., Nocera, D.G., 2006. Powering the planet: Chemical challenges in solar
energy utilization. Proceedings of the National Academy of Sciences of the
United States of America 103, 15729-15735.
Liebig, M.A., Schmer, M.R., Vogel, K.P., Mitchell, R.B., 2008. Soil Carbon Storage by
Switchgrass Grown for Bioenergy. Bioenergy Research 1, 215-222.
Lin, S. D., Lee, C.C. (2007) Water and Wastewater Calculations Manual. 2nd ed.
McGraw-Hill Companies Inc. page. 333.
Lippke, B., Perez-Garcia, J., 2008. Will either cap and trade or a carbon emissions tax be
effective in monetizing carbon as an ecosystem service. Forest Ecology and
Management 256, 2160-2165.
Lonsdorf, E., Kremen, C., Ricketts, T., Winfree, R., Williams, N., Greenleaf, S., 2009.
Modelling pollination services across agricultural landscapes. Annals of Botany
103, 1589-1600.
Lorenz, A.J., Anex, R.P., Isci, A., Coors, J.G., Leon, N.d., Weimer, P.J., 2009. Forage
quality and composition measurements as predictors of ethanol yield from maize
(Zea mays L.) stover. Biotechnology for Biofuels 2, (9 March 2009).
Love, B.J. and Nejadhashemi, A.P. 2010a. Environmental Impact Analysis of Biofuel
Crops Expansion in the Saginaw River Watershed. Journal of Biobased Materials
and Bioenergy. (Under Review).
Love, B.J. and Nejadhashemi, A.P. 2010b. Water Quality Impact Assessment of large
Scale Biofuel Crops Expansion. Journal of Biomass & Bioenergy (Under
Review).
Luo, Y., Zhang, M. (2009a) A Geo-Refernced Modeling Environment for Ecosystem
Risk Assessment: Organophosphate Pesticides in an Agriculturally Dominated
Watershed. Journal of Environmental Quality. 38: 664-674.
Luo, Y., Zhang, M. (2009b) Management-oriented Sensitivity Analysis for Pesticide
Transport in Watershed-Scale Water Quality Modeling Using SWAT.
Environmental Pollution. 157: 3370-3378.

223

Luo, Y.Z., Zhang, X.Y., Liu, X.M., Ficklin, D., Zhang, M.H., 2008. Dynamic modeling
of organophosphate pesticide load in surface water in the northern San Joaquin
Valley watershed of California. Environmental Pollution 156, 1171-1181.
Lybecker, D. W., E. E. Schweizer & R. P. King (1991) Weed management decisions in
corn based on bioeconomic modeling. Weed Science, 39, 124-129.
Lynd, L.R., Larson, E., Greene, N., Laser, M., Sheehan, J., Dale, B.E., McLaughlin, S.,
Wang, M., 2009. The role of biomass in America's energy future: framing the
analysis. Biofuels, Bioproducts & Biorefining 3, 113-123.
Maler, K.G., Aniyar, S., Jansson, A., 2009. Accounting for ecosystems. Environmental
and Resource Economics 42, 39-51.
Mankin, K.R., Wang, S.H., Koelliker, J.K., Huggins, D.G., deNoyelles, F., 2003.
Watershed-lake water quality modeling: Verification and application. Journal of
Soil and Water Conservation 58, 188-197.
Maski, D., Mankin, K.R., Janssen, K.A., Tuppad, P., Pierzynski, G.M., 2008. Modeling
runoff and sediment yields from combined in-field crop practices using the Soil
and Water Assessment Tool. Journal of Soil and Water Conservation 63, 193-203.
McCuen, R.H., Knight, Z., Cutter, A.G., 2006. Evaluation of the Nash-Sutcliffe
efficiency index. Journal of Hydrologic Engineering 11, 597-602.
McElroy, A. D., S.Y. Chiu, J.W. Nebgen, A. Aleti, and F.W. Bennett. 1976. Loading
functions for assessment of water pollution from nonpoint sources. Environ. Prot.
Tech. Serv., EPA 600/2-76-151.
McLaughlin, S.B., Walsh, M.E., 1998. Evaluating environmental consequences of
producing herbaceous crops for bioenergy. Biomass & Bioenergy 14, 317-324.
MDEP (2010). Massachusettes Department of Environmental Protection (2010) Current
Regulatory Limit: Metolachlor. Water, Wastewater and Wetlands. Available
from: http://www.mass. gov/dep/water/drinking/standards/metolach.htm
Megadi, V.B., Tallur, P.N., Hoskeri, R.S., Mulla, S.I., Ninnekar, H.Z., 2010.
Biodegradation of pendimethalin by Bacillus circulans. Indian Journal of
Biotechnology 9, 173-177.
Menard, S., 2000. Coefficients of determination for multiple logistic regression analysis.
American Statistician 54, 17-24.
Moebius-Clune, B.N., van Es, H.M., Idowu, O.J., Schindelbeck, R.R., Moebius-Clune,
D.J., Wolfe, D.W., Abawi, G.S., Thies, J.E., Gugino, B.K., Lucey, R., 2008.
Long-term effects of harvesting maize stover and tillage on soil quality. Soil
Science Society of America Journal 72, 960-969.

224

Mol, A.P.J., 2007. Boundless biofuels? Between environmental sustainability and
vulnerability. Sociologia Ruralis 47, 297-315.
Moon, H.S., Abercrombie, J.M., Kausch, A.P., Stewart, C.N., 2010. Sustainable Use of
Biotechnology for Bioenergy Feedstocks. Environmental Management 46, 531538.
Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel, R.D., Veith, T.L.,
2007. Model evaluation guidelines for systematic quantification of accuracy in
watershed simulations. Transactions of the Asabe 50, 885-900.
MSU Weed Science. (2010) 2010 Weed Control Guide for Crops. Introduction.
http://msuweeds.com/publications/2010_weed_guide/
MSUEP. Michigan State University Enviroweather project. 2010. MSU, East Lansing,
MI Available from: http://enviroweather.msu.edu
MSUIWR (Michigan State University Institute of Water Research). 2010. Clean Water
Act Section 303(d) List in Michigan. MSUIWR, East Lansing, Michigan.
Available from http://www.gis.iwr.msu.edu/tmdl98/default.htm
Mudgal, A., Baffaut, C., Anderson, S.H., Sadler, E.J., Thompson, A.L. (2008) APEX
Model Assessment of Variable Landscapes on Runoff and Dissolved Herbicides.
Written for Presentation at 2008 ASABE Annual International Meeting. Paper #
084498.
Munro, I.C., Carlo, G.L., Orr, J.C., Sund, K.G., Wilson, R.M., Kennepohl, E., Lynch,
B.S., Jablinske, M. (1992) A Comprehensive, Integrated Review and Evaluation
of the Scientific Evidence Relating to the Safety of the Herbicide 2, 4-D.
International Journal of Toxicology. 11: 559
Nasr, A., Bruen, M., Jordan, P., Moles, R., Kiely, G., Byrne, P., 2007. A comparison of
SWAT, HSPF and SHETRAN/GOPC for modelling phosphorus export from
three catchments in Ireland. Water Research 41, 1065-1073.
NCSU (North Carolina State University). 2005. Organic Field Crop Production and
Marketing in North Carolina. North Carolina State University, Raleigh, NC.
Available from http://www.organicgrains.ncsu.edu/production/croprotation.htm
NCSU, 2005. Organic Field Crop Production and Marketing in North Carolina. North
Carolina State University, Raleigh, NC. Available from
http://www.organicgrains.ncsu.edu/production/ croprotation.htm
Neitsch, S. L., Arnold, J. G., Kiniry, J. R., Williams, J. R., 2005. Soil and Water
Assessment Tool: Theoretical Documentation, Version 2005.
Neitsch, S.L., Arnold, J.G., Srinivasan, R. (2002) Final Report. Pesticides Fate and
Transport Predicted by the Soil and Water Assessment Tool (SWAT): Atrazine,

225

Metolachlor and Trifluralin in the Sugar Creek Watershed. Office of Pesticide
Programs EPA. BRC Publication #2002-03.
Nelson, E., Mendoza, G., Regetz, J., Polasky, S., Tallis, H., Cameron, D.R., Chan,
K.M.A., Daily, G.C., Goldstein, J., Kareiva, P.M., Lonsdorf, E., Naidoo, R.,
Ricketts, T.H., Shaw, M.R., 2009. Modeling multiple ecosystem services,
biodiversity conservation, commodity production, and tradeoffs at landscape
scales. Frontiers in Ecology and the Environment 7, 4-11.
Nelson, E., Polasky, S., Lewis, D.J., Plantingall, A.J., Lonsdorf, E., White, D., Bael, D.,
Lawler, J.J., 2008. Efficiency of incentives to jointly increase carbon
sequestration and species conservation on a landscape. Proceedings of the
National Academy of Sciences of the United States of America 105, 9471-9476.
Ng, T.L., Eheart, J.W., Cai, X.M., Miguez, F., 2010. Modeling Miscanthus in the Soil
and Water Assessment Tool (SWAT) to Simulate Its Water Quality Effects As a
Bioenergy Crop. Environmental Science & Technology 44, 7138-7144.
Nielsen, D. C. & S. E. Hinkle (1996) Field evaluation of basal crop coefficients for corn
based on growing degree days, growth stage, or time. Transactions of the ASAE,
39, 97-103.
Norris, K., 2008. Agriculture and biodiversity conservation: opportunity knocks.
Conservation Letters 1, 2-11.
Nyakatawa, E.Z., Mays, D.A., Tolbert, V.R., Green, T.H., Bingham, L., 2006. Runoff,
sediment, nitrogen, and phosphorus losses from agricultural land converted to
sweetgum and switchgrass bioenergy feedstock production in north Alabama.
Biomass & Bioenergy 30, 655-664.
Oliver, R.J., Finch, J.W., Taylor, G., 2009. Second generation bioenergy crops and
climate change: a review of the effects of elevated atmospheric CO2 and drought
on water use and the implications for yield. Global Change Biology Bioenergy 1,
97-114.
Omalley, M., 1997. Clinical evaluation of pesticide exposure and poisonings. Lancet 349,
1161-1166.
Overmyer, J.P. (2010) Effects of Turf Pesticides on Aquatic Invertebrates. Chapter 3. In
Turf Grass: Pesticide Exposure Assessment and Predictive Modeling Tools; Nett,
M., et al.; ACS Symposium Series; American Chemical Society: Washington,
DC, 2010.
Parker, G.T., Droste, R.L., Kennedy, K.J., 2008. Modeling the effect of agricultural best
management practices on water quality under various climatic scenarios. Journal
of Environmental Engineering and Science 7, 9-19.

226

Parker, R., Arnold, J.G., Barrett, M., Burns, L., Carrubba, L., Neitsch, S.L., Snyder, N.J.,
Srinivasan, R. (2007) Evaluation of Three Watershed-Scale Pesticide
Environmental Transport and Fate Models. Journal of the American Water
Resources Association. 43 (6): 1424-1443.
Pimentel, D. (2005). Environmental and economic costs of the application of pesticides
primarily in the United States, Environment, Development and Sustainability, 7:
229–252.
Pimentel, D., A. Marklein, M. A. Toth, M. N. Karpoff, G. S. Paul, R. McCormack, J.
Kyriazis & T. Krueger (2009) Food Versus Biofuels: Environmental and
Economic Costs. Human Ecology, 37, 1-12.
Pimentel, D., Acquay, H., Biltonen, M., Rice, P., Silva, M., Nelson, J., Lipner, V.,
Giordana, S., Horowitz, A. and D’Amore, M. 1993. Assessment of environmental
and economic impacts of pesticide use. In: Pimentel, D. and H. Lehman (eds),
The Pesticide Question: Environment, Economics and Ethics. Chapman and Hall,
New York, pp. 47–84.
Pimentel, D., P. Hepperly, J. Hanson, D. Douds, & R. Seidel, Bioscience 55, 7 (2005).
Pimentel, D., Patzek, T. W., 2005. Ethanol Production Using Corn, Switchgrass, and
Wood; Biodiesel Production Using Soybean and Sunflower. Natural Resources
Research, 14 (1), 65-76.
Pionke, H.B., Chesters, G., 1973. Pesticide Sediment Water Interactions. Journal of
Environmental Quality 2, 29-45.
Porter, J., Costanza, R., Sandhu, H., Sigsgaard, L., Wratten, S., 2009. The Value of
Producing Food, Energy, and Ecosystem Services within an Agro-Ecosystem.
Ambio 38, 186-193.
Powers, S.E., 2007. Nutrient loads to surface water from row crop production.
International Journal of Life Cycle Assessment 12, 399-407.
Quilbe, R., Rousseau, A.N., Lafrance, P., Leclerc, J., Amrani, M. (2006) Selecting a
Pesticide Fate Model at the Watershed Scale Using a Multi-Criteria Analysis.
Water Qual. Res. J. Canada. 41(3): 283-295.
Raupach, M.R., Marland, G., Ciais, P., Le Quere, C., Canadell, J.G., Klepper, G., Field,
C.B., 2007. Global and regional drivers of accelerating CO2 emissions.
Proceedings of the National Academy of Sciences of the United States of America
104, 10288-10293.
Relyea, R.A. (2009) A Cocktail of Contaminants: How Mixtures of Pesticides at Low
Concentrations Affect Aquatic Communities. Oecologia. 159:363-376.

227

Ribaudo, M.O., Bouzaher, A. (1994) Atrazine: Environmental Characteristics and
Economics of Management. United States Department of Agriculture.
Agricultural Economic Report Number 699.
Rooney, W.L., Blumenthal, J., Bean, B., Mullet, J.E., 2007. Designing sorghum as a
dedicated bioenergy feedstock. Biofuels Bioproducts & Biorefining-Biofpr 1,
147-157.
Rowe, L., E. Rossman & D. Penner (1990) Differential response of corn hybrids and
inbreds to metolachlor. Weed Science, 38, 563-566.
Rowe, R.L., Street, N.R., Taylor, G., 2009. Identifying potential environmental impacts
of large-scale deployment of dedicated bioenergy crops in the UK. Renewable &
Sustainable Energy Reviews 13, 260-279.
Roxburgh, S.H., Shea, K., Wilson, J.B., 2004. The intermediate disturbance hypothesis:
Patch dynamics and mechanisms of species coexistence. Ecology 85, 359-371.
Ruffo, S., Kareiva, P.M., 2009. Using science to assign value to nature. Frontiers in
Ecology and the Environment 7, 3-3.
Runkel, R.L., Crawford, C.G., Cohn, T.A., 2004. Load Estimator (LOADEST)—a
FORTRAN program for estimating constituent loads in streams and rivers, US
Geological Survey Techniques and Methods Book 4, Chap. A5, US Geological
Survey, Reston, VA.
Safe, S., 2004. Endocrine disruptors and human health: is there a problem. Toxicology
205, 3-10.
Sala, O.E., D. Sax, H. Leslie. 2009. Biodiversity consequences of biofuel production.
Pages 127-137 in R.W. Howarth and S. Bringezu (eds) Biofuels: Environmental
Consequences and Interactions with Changing Land Use. Proceedings of the
Scientific Committee on Problems of the Environment (SCOPE) International
Biofuels Project Rapid Assessment, 22-25 September 2008, Gummersbach
Germany. Cornell University, Ithaca NY, USA. Available from
http://cip.cornell.edu/biofuels/
Saleh, A., Arnold, J.G., Gassman, P.W., Hauck, L.M., Rosenthal, W.D., Williams, J.R.,
McFarland, A.M.S., 2000. Application of swat for the Upper North Bosque River
Watershed. Transactions of the Asae 43, 1077-1087.
Saleh, A., Du, B., 2004. Evaluation of SWAT and HSPF within BASINS program for the
Upper North Bosque River watershed in Central Texas. Transactions of the Asae
47, 1039-1049.
Santhi, C., Arnold, J.G., Williams, J.R., Dugas, W.A., Srinivasan, R., Hauck, L.M., 2001.
Validation of the swat model on a large river basin with point and nonpoint
sources. Journal of the American Water Resources Association 37, 1169-1188.

228

Santhi, C., Srinivasan, R., Arnold, J.G., Williams, J.R., 2006. A modeling approach to
evaluate the impacts of water quality management plans implemented in a
watershed in Texas. Environmental Modelling & Software 21, 1141-1157.
Santoro, N., S. L. Cantu, C.-E. Tornqvist, T. G. Falbel, J. L. Bolivar, S. E. Patterson, M.
Pauly & J. D. Walton (2010) A High-Throughput Platform for Screening
Milligram Quantities of Plant Biomass for Lignocellulose Digestibility.
BioEnergy Research, 3, 93-102.
SAS. Statistical Analysis Systems. 2009. SAS/STAT ® 9.2 User’s Guide, Second
Edition. Cary, NC: Statistical Analysis Systems Institute. 1028 p.
Schafer, R. B., Caquet, T., Siimes, K., Mueller, R., Lagadic, L., Liess, M. (2007) Effects
of Pesticides on Community Structure and Ecosystem Functions in Agricultural
Streams of Three Biogeographical regions in Europe. Science of the Total
Environment. 382: 272 – 285.
Schaldach, R., J. A. Priess, J. Alcamo. Simulating the impact of biofuel development on
country-wide land-use change in India. Biomass & Bioenergy (2010),
doi:10.1016/j.biombioe.2010.08.048
Scheffran, J., BenDor, T., 2009. Bioenergy and land use: a spatial-agent dynamic model
of energy crop production in Illinois. International Journal of Environment and
Pollution 39, 4-27.
Schulz, R., Liess, M. (1999) A Field Study of the Effects of Agriculturally Derived
Insecticide Input on Stream Macroinvertebrate Dynamics. Aquatic Toxicology
46: 155-176
Schwartz, M.W., Brigham, C.A., Hoeksema, J.D., Lyons, K.G., Mills, M.H., van
Mantgem, P.J., 2000. Linking biodiversity to ecosystem function: implications for
conservation ecology. Oecologia 122, 297-305.
Scott, G.R., Sloman, K.A. (2004) The Effects of Environmental Pollutants on Complex
Fish Behavior: Integrating Behavioural and Physiological Indicators of Toxicity.
Aquatic Toxicology. 68(4): 369-392.
Searchinger, T., Heimlich, R., Houghton, R.A., Dong, F.X., Elobeid, A., Fabiosa, J.,
Tokgoz, S., Hayes, D., Yu, T.H., 2008. Use of US croplands for biofuels increases
greenhouse gases through emissions from land-use change. Science 319, 12381240.
Serrano, R., Hernandez, F., Pena, J.B., Dosda, V., Canales, J., 1995. Toxicity and
Bioconcentration of Selected Organophosphorus Pesticides in MytilusGallpprovincialis and Venus. Archives of Environmental Contamination and
Toxicology 29, 284-290.

229

Siber, R., Stamm, C., Reichert, P., 2009. Modeling potential herbicide loss to surface
waters on the Swiss plateau. Journal of Environmental Management 91, 290-302.
Solomon, B.D., 2010. Biofuels and sustainability. Annals of the New York Academy of
Sciences 1185, 119-134.
Solomon, K.R., Carr, J.A., Du Preez, L.H., Giesy. J.P., Kendall, R.J., Smith, E.E., Van
Der Kraak, G.J. (2008) Effects of Atrazine on Fish , Amphibians, and Aquatic
Reptiles: A Critical Review. Critical Reviews in Toxicology. 38: 721-772
Spruill, C.A., Workman, S.R., Taraba, J.L., 2000. Simulation of daily and monthly
stream discharge from small watersheds using the SWAT model. Transactions of
the Asae 43, 1431-1439.
StatSoft, Inc. (2010) STATISTICA (data analysis software system), version 9.1. StatSoft,
Inc. Tulsa, OK.
Stone, K.C., Hunt, P.G., Cantrell, K.B., Ro, K.S., 2010. The potential impacts of biomass
feedstock production on water resource availability. Bioresource Technology 101,
2014-2025.
Swift, M.J., Anderson, J.M., 1993. Biodiversity and Ecosystem Function in Agricultural
Systems. Biodiversity and Ecosystem Function 99, 15-41.
Takacs, P., P. A. Martin, and J. Struger. 2002. Pesticides in Ontario: A critical assessment
of potential toxicity of agricultural products to wildlife, with consideration for
endocrine disruption: Volume 2. Triazine herbicides, glyphosate, and metolachlor.
Technical Report Series No. 369. Ottawa, Ontario, Canada: Canadian Wildlife
Service, Environmental Conservation Branch, Ontario Region. Available at:
http://dsp-psd.communication.gc.ca/Collection/CW69-5-369E.pdf
Tallis, H.T., T. Ricketts, E. Nelson, D. Ennaanay, S. Wolny, N. Olwero, K. Vigerstol, D.
Pennington, G. Mendoza, J. Aukema, J. Foster, J. Forrest, D. Cameron, E.
Lonsdorf, C. Kennedy. 2010. InVEST 1.005 beta User’s Guide. The Natural
Capital Project, Stanford.
Tardif, F.J., Holtum, J.A.M., Powles, S.B., 1993. Occurrence of a Herbicide-Resistant
Acetyl-Coenxyme-A Carboxylase Mutant in Annual Ryegrass (Lolium-Rigidum)
Selected by Sethoxydim. Planta 190, 176-181.
Thomas, K. C., S. H. Hynes & W. M. Ingledew (1996) Practical and theoretical
considerations in the production of high concentrations of alcohol by
fermentation. Process Biochemistry, 31, 321-331.
Thomas, M.A., Engel, B.A., Chaubey, I., 2009. Water quality impacts of corn production
to meet biofuel demands. Journal of Environmental Engineering 135, 1123-1135.

230

Thompson, R., Starzomski, B.M., 2007. What does biodiversity actually do? A review for
managers and policy makers. Biodiversity and Conservation 16, 1359-1378.
Tilman, D., Hill, J., Lehman, C., 2006. Carbon-negative biofuels from low-input highdiversity grassland biomass. Science 314, 1598-1600.
Tong, S.T.Y., Naramngam, S., 2007. Modeling the impacts of farming practices on water
quality in the Little Miami River basin. Environmental Management 39, 853-866.
Topp, E., Xun, L.Y., Orser, C.S., 1992. Biodegradation of the Herbicide Bromoxynil
(3,5-bibromo-4-hydroxybenzonitrile) by Purified Pentachlorophenol Hydroxylase
and Whole Cells of Flavobacterium SP Stain ATCC-39723 Is Accompanied by
Cyanogenesis. Applied and Environmental Microbiology 58, 502-506.
Triantafyllidis, V., Hela, D., Salachas, G., Dimopoulos, P., Albanis, T. (2009) Soil
Dissipation and Runoff Losses of the Herbicide Pendimethalin in Tobacco Field.
Water Air Soil Pollution. 201: 253-264.
Tu, M., C. Hurd, and J. M. Randall. 2001. Weed Control Methods Handbook: Tools and
Techniques for Use in Natural Areas. The Nature Conservancy. Available at:
http://www.invasive.org/gist/products/handbook/19.Sethoxydim.pdf
Ullrich, A., Volk, M., 2009. Application of the Soil and Water Assessment Tool (SWAT)
to predict the impact of alternative management practices on water quality and
quantity. Agricultural Water Management 96, 1207-1217.
United States Geological Survey (USGS) The Quality of Our Nation’s Water, Nutrients
and Pesticides. USGS Circular 1225: Reston, VA. 1999.
USDA (U.S. Department of Agriculture) 1997. Glyphosate: Herbicide Information
Profile. United States Department of Agriculture. Forest Service Pacific
Northwest Region. Feb. 1997.
USDA-NRCS (U.S. Department of Agriculture- Natural Resources Conservation
Service). 2009. Conservation Reserve Program. NRCS, Washington, D.C.
Available from http://www. nrcs.usda.gov/programs/crp/
USDOE (U.S. Department of Energy). Biofuels Initiative. Department of Energy, The
Office of Energy Efficiency and Renewable Energy, Washington D.C. Available
from http://www1.eere.energy.gov/biomass/biofuels_initiative.html (2007)
USEIA (U.S. Energy Information Administration). Renewable Energy Consumption and
Electricity Preliminary Statistics 2008. USEIA, Washington, D.C. Available from
http://www.eia.doe.gov/cneaf/alternate/page/renew_energy_
consump/rea_prereport.html (2009)

231

USEPA (U.S. Environmental Protection Agency). 1993. Glyphosate, R.E.D. FACTS.
EPA-738-F-93-011. Available from:
http://www.epa.gov/oppsrrd1/REDs/factsheets/0178fact.pdf
USEPA (U.S. Environmental Protection Agency). 1996. Trifluralin. R.E.D. FACTS.
EPA-738-F-95-035. Available from:
http://www.epa.gov/oppsrrd1/REDs/factsheets/0179fact.pdf
USEPA (U.S. Environmental Protection Agency). 1997. Pendimethalin. R.E.D. Facts.
EPA-738-F-97-007. Available from:
http://www.epa.gov/oppsrrd1/REDs/factsheets/0187fact.pdf
USEPA (U.S. Environmental Protection Agency). 2008. 2, 4-D RED Facts. Pesticides:
Reregistration. June 30, 2005. updated 2008. Available from:
http://www.epa.gov/oppsrrd1/REDs/factsheets/24d_fs.htm
USEPA (U.S. Environmental Protection Agency). 2008. Ecological Risk Assessments.
Pesticides:Environmental Effects. Available from:
http://www.epa.gov/pesticides/ecosystem/ecorisk.htm
USEPA (U.S. Environmental Protection Agency). 2009a. EPA Proposes New
Regulations for the National Renewable Fuel Standard Program for 2010 and
Beyond. USEPA, Washington, D.C. Available from: http://www.epa.gov/oms/
renewablefuels/420f0 9023.htm
USEPA (U.S. Environmental Protection Agency). 2009b. Saginaw River and Bay Area
of Concern. USEPA, Washington, D.C. Available from http://www.epa.gov/
glnpo/aoc/ sagrivr.html
USEPA (U.S. Environmental Protection Agency). 2009c. Great Lakes Areas of Concern.
USEPA, Washington, D.C. Available from http://www.epa.gov/ glnpo/aoc/
index.html
USEPA (U.S. Environmental Protection Agency). 2009d. Office of Pesticide Programs’
Aquatic Life Benchmarks. Pesticides: Environmental Effects. Available from:
http://www.epa.gov/oppefed1/ecorisk_ders/aquatic_life_benchmark.htm
USEPA (U.S. Environmental Protection Agency). 2009e. Pesticide Types. About
Pesticides. Available from: http://www.epa.gov/pesticides/about/types.htm
USEPA (U.S. Environmental Protection Agency). 2010. Michigan 2008 Water Quality
Assessment Report. USEPA, Washington, D.C. Available from
http://iaspub.epa.gov/ waters 10/attains_index.control?p_area=MI
USEPA (U.S. Environmental Protection Agency). 2010b Regulating Public Water
Systems and Contaminates Under the Safe Drinking Water Act. Regulation
Development. Available from: http://water.epa.gov/
lawsregs/rulesregs/regulatingcontaminants/basicinformation.cfm

232

USEPA (U.S. Environmental Protection Agency). 2010c ECOTOX Database. Available
from: http://cfpub.epa.gov/ecotox/
USEPA (U.S. Environmental Protection Agency). 2010d National Primary Drinking
Water Regulations. Available from:
http://water.epa.gov/drink/contaminants/index.cfm
Vanheemst, H. D. J. (1985) The influence of weed competition on crop yield.
Agricultural Systems, 18, 81-93.
Vazquez-Amabile, G., Engel, B.A., Flanagan, D.C., 2006. Modeling and risk analysis of
nonpoint-source pollution caused by atrazine using SWAT. Transactions of the
Asabe 49, 667-678.
Vencill, W. K., K. K. Hatzios & H. P. Wilson (1989) Growth and physiologicalresponses of normal, dwarf, and albino corn (Zea-mays) to clomazone treatments.
Pesticide Biochemistry and Physiology, 35, 81-88.
Walsh, C.J., 2006. Biological indicators of stream health using macroinvertebrate
assemblage composition: a comparison of sensitivity to an urban gradient. Marine
and Freshwater Research 57, 37-47.
Wang, S., Stiles, T., Flynn, T., Stahl, A.J., Gutierrez, J.L., Angelo, R.T., Frees, L., 2009.
A Modeling Approach to Water Quality Management of an Agriculturally
Dominated Watershed, Kansas, USA. Water Air and Soil Pollution 203, 193-206.
Weston, D.P., You, J., Lydy, M.J., 2004. Distribution and toxicity of sediment-associated
pesticides in agriculture-dominated water bodies of California's Central Valley.
Environmental Science & Technology 38, 2752-2759.
Williams, J.R. and R.W. Hann. 1978. Optimal operation of large agricultural watersheds
with water quality contraints. Texas Water Resources Institute, Texas A&M
Univ., Tech. Rept. No. 96.
Williams, J.W., Izaurralde, R.C., Steglich, E.M. (2008) Agricultural Policy /
Environmental Extender Model: Theoretical Documentation. Version 0604.
BREC Report #2008-17.
Winchell, M., Srinivasan, R., Di Luzio, M., Arnold, J., 2009. Soil and Water Assessment
Tool User’s Guide, Version 2005.
Wu, M., Mintz, M., Wang, M., Arora, S., 2009. Water Consumption in the Production of
Ethanol and Petroleum Gasoline. Environmental Management 44, 981-997.
Yuan, Y.P., Bingner, R., Williams, R., Lowrance, R., Bosch, D., Sheridan, J., 2007.
Integration of the models of AnnAGNPS and REMM to assess riparian buffer
system for sediment reduction. International Journal of Sediment Research 22, 6069.

233

Zhang, W., Ricketts, T.H., Kremen, C., Carney, K., Swinton, S.M., 2007. Ecosystem
services and dis-services to agriculture. Ecological Economics 64, 253-260.

234