32 >5

 

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
thesis entitled

ESTABLISHMENT, FERTILITY AND HARVEST
MANAGEMENT FOR OPTIMIZING SWITCHGRASS
YIELD AND QUALITY AS A CELLULOSIC ETHANOL
FEEDSTOCK IN THE GREAT LAKES REGION

presented by

Katherine Kelly Withers

has been accepted towards fulfillment
of the requirements for the

Master of degree in Crop and Soil Science
Science

 

 

M944

 

' Major'Professor’s Signature

lZ AM 5.0/0
d

Date

 

MSU is an Affirmative Action/Equal Opportunity Employer

-4—-DI-.'_-_'_.-_A—LA-n-.-.— -4-.-

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KIProj/AocSPrelelRC/DateDue.indd

 

ESTABLISHMENT, FERTILITY AND HARVEST MANAGEMENT FOR
OPTIMIZING SWITCHGRASS YIELD AND QUALITY As A CELLULOSIC
ETHANOL FEEDSTOCK IN THE GREAT LAKES REGION
By

Katherine Kelly Withers

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Crop and Soil Science

2010

ABSTRACT
ESTABLISHMENT, FERTILITY AND HARVEST MANAGEMENT FOR
OPTIMIZING SWITCHGRASS YIELD AND QUALITY As A CELLULOSIC
ETHANOL FEEDSTOCK IN THE GREAT LAKES REGION
By

Katherine Kelly Withers

Agronomic management Specific to the Great Lakes Region is needed for
producing switchgrass (Panicum virgatum L.), in an economically and
environmentally sustainable manner for the bioconversion platform. Two field
studies were conducted to determine: the best seeding date and planting
methodology for stand establishment; and agronomic N rate and harvest
management system for Optimizing switchgrass yield and quality as a bioenergy
feedstock. Three seeding dates (early, mid, and late- Spring), and three planting
methodologies (no-till, and conventionally tilled and planted with either a grain
drill or double- roller seeder) were assessed. The harvest and fertility study
consisted of three harvest management systems (one- cut fall, one- cut
overwintered, and a two- cut system), and four N rates (0, 39, 78, and 157 kg N
ha“). Both were randomized complete block designs with five replications, the
former arranged as a split-plot. A mid- spring seeding date, with or without
tillage resulted in acceptable stand frequencies and subsequent yields. A Single
harvest made in the fall after a killing frost was found to be the best harvest
management system with a mean yield of 12.3 Mg ha". Biomass yields did not
differ between 78 and 157 kg N ha", indicative of a yield plateau. The yield

response to N was best modeled with a linear response plateau.

ACKNOWLEDGEMENTS

I was very fortunate to be a part of the Bioenergy Cropping Systems Lab.
Thank-you Dr. Kurt D. Thelen for giving me an opportunity to work on the
agronomy project of my dreams, for answering my questions, and for playing
such an integral part in helping me obtain my PhD candidacy. I would not be
where I am today if it weren’t for our field technician Bill Widdicombe, through
instilling lessons on life upon me, and greatly contributing to my professional
development. Thank-you to Lori Williams, Tim Boring, Juan Gao, Leilei Qian and
John Hoben and Dennis Pennington for all their incredible team-work, and to Xin-

Mei Hao for teaching me to love and embrace statistics.

l was also very fortunate to have two supportive committee members.
Thank-you to Dr. Venkatesh Balan for his assistance on the bio-conversion side,
and Dr. Darryl Wamcke for the help in modeling the yield response to nitrogen. I
would like to thank Dr. Wes Everman for the herbicide recommendations, for
teaching me critical thinking skills as my weed science coach, and for mentoring
me through the transition to a Ph.D student. I am very grateful to Cliff Foster for
all of his compositional analysis work and putting up with my never-ending
samples and new ideas. Thank-you to Tim Dietz for his assistance in the lab and
for help with harvesting of the trial, and to John Dahl for the nitrogen content
analysis. Rachel Bethke, thank-you for all the emotional support and help you
gave me. Last but not least, I would like to thank my parents, my sister Julia, and
Laas for their unwavering support. Your timely advice and visits undoubtedly

contributed to my success.

TABLE OF CONTENTS

LIST OF TABLES ................................................................................ iv
LIST OF FIGURES ............................................................................. vii
INTRODUCTION ................................................................................ 1
CHAPTER 1

EARLY-, MlD—, AND LATE-SPRING ESTABLISHMENT OF SWITCHGRASS
UNDER THREE DIFFERENT PLANTING METHODOLOGIES IN THE GREAT
LAKES REGION

INTRODUCTION .............................................................................. 10
MATERIALS AND METHODS ............................................................. 16
Site Description and Experimental Design .............................................. 16
Statistical Analysis ............................................................................ 20
RESULTS AND DISCUSSION ............................................................. 21
CONCLUSION ................................................................................. 34
CHAPTER 2

SWITCHGRASS BIOMASS YIELD RESPONSE TO VARYING RATES OF
NITROGEN FERTILIZER AND HARVEST TIMING IN THE CENTRAL GREAT
LAKES REGION

INTRODUCTION ............................................................................... 37
MATERIALS AND METHODS .............................................................. 46
Site Description and Experimental Design .............................................. 46
Statistical Analysis ............................................................................ 48
RESULTS AND DISCUSSION ............................................................. 50
CONCLUSION .................................................................................. 67
CHAPTER 3

CRYSTALLINE CELLULOSE, LIGNIN CONTENT AND MATRIX
HEMICELLULOSE POLYSACCHARIDE COMPOSITION OF SWITCHGRASS
UNDER VARYING NITROGEN RATES AND HARVEST MANAGEMENT
SYSTEMS '

INTRODUCTION ............................................................................... 68
MATERIALS AND METHODS .............................................................. 80
Site Description and Experimental Design .............................................. 80
Statistical Analysis ............................................................................ 82
RESULTS AND DISCUSSION ............................................................. 83
CONCLUSION ................................................................................. 90
CONCLUSIONS ............................................................................. ..91

LITERATURE CITED ......................................................................... 93

iv

LIST OF TABLES

Table 1. Soil properties obtained from soil samples taken prior to fertilization in
2009 growing year ............................................................................ 17

Table 2. Agronomic operations for both seeding- years over course of field
experiment 2008- 2010 ...................................................................... 18

Table 3. Monthly rainfall totals and average mean monthly air temperatures at
the Michigan State Agronomy Farm, East Lansing, MI, in the 2008 and 2009
growing seasons and the 30-year average (1971-2000) ....... ; ................... 22

Table 4. Harvest Management Systems employed during the study (2008-
2010) with associated harvest dates and plant developmental stages from
Moore et al.(1991) ............................................................................... 48

Table 5. Three functional forms: linear (L), Quadratic Polynomial (OP), and
Linear Response Model (LRP), used to model yield response to N ............. 49

Table 6. Mean monthly rainfall (mm) and temperatures (OC) during the
experiment in comparison to the 30-year mean (1971— 2000) .................... 51

Table‘7. Switchgrass dry biomass yields in response to three different harvest
management systems, and four nitrogen rates in 2008-2010 ........................ 53

Table 8. Switchgrass biomass yields in response to three different harvest
management systems, and four nitrogen rates in 2008 and 2009 ................ 56

Table 9. Switchgrass biomass yields (Mg ha-I) in response to three different
harvest management systems, and four nitrogen rates in 2008 and 2009.....58

Table 10. Dry switchgrass biomass yields (Mg ha-1) responses within the two-
cut system at four different rates of nitrogen fertilizer in 2008 and 2009 ........ 59

Table 11. Regression equations and parameters for the two—cut harvest
system in 2008 ................................................................................ 62

Table 12. Switchgrass biomass yields in response to three different harvest
management systems, and four nitrogen rates in 2008 and 2009 ................ 65

Table 13. Maximum likelihood estimates and coefficients of determination of
regression parameters for quadratic, linear, and linear response plateau (LRP)
equations for the various switchgrass harvest management systems in 2008
and 2009 ....................................................................................... 66

Table 14. Selected lignocellulosic biomass compositions (% dry weight) from
Wyman (1996) ................................................................................ 71

Table 15. Range of Compositions of whole plant ‘Cave-in-Rock’ and leaves
and stems across numerous varieties as a percentage of dry weight from
Wyman (1 996) ................................................................................ 72

Table 16. Chemical compositions of four populations of switchgrass based on
(%) of oven dry weight from Hu et al. 2010 (%) of oven dry weight from Hu et
al. 2010.. ........................................................................................................ 72

Table 17. Comparison of average chemical compositions between three
morphological portions of switchgrass and other published results from Hu et
al. (201 0) ........................................................................................... 73

Table 18. Composition analysis (as a percentage of total dry weight) for the
July and October harvests of Cave-in- Rock switchgrass from Bals et al.
2010 ................................................................................................. 78

v1

Table 19. Changes in composition of ‘Nebraska 28’ switchgrass with crop
Maturity from Lee et al. (2007) ............................................................ 80

Table 20. Cell wall carbohydrates Dien et al. (2006) ............................... 80

Table 21. Matrix polysaccharide composition across varying harvest timings
expressed as molar concentration ....................................................... 89

vii

LIST OF FIGURES

Figure 1. Environmental Conditions during planting in the 2008 Seeding-
year. Weather information was obtained from the Michigan Automated
Weather Network ........................................................................... 23

Figure 2. Stand frequencies obtained at the 2008 seeding-year. Means with
the same letters are not significantly different at a=0.05 .......................... 24

Figure 3. Second-year yields for the 2008 seeding-year. Means with the
same letters are not significantly different at (1:005 level of significance.
Standard error was 0.70 ................................................................... 28

Figure 4. Environmental conditions at planting time for the 2009 seeding
year. Weather information was obtained from the Michigan Automated
Weather Network ............................................................................ 29

Figure 5. Stand frequencies obtained at the 2009 seeding-year, Means with
the same letters are not statistically different at the o=0.05 level of
significance. Stand error was 2.96 ............................................................... 30

Figure 6. First-year yields in the 2009 seeding-year. Means with the same
letters are not significantly different at the o=0.05 level. Standard error was
0176 ........ 31

Figure 7. Stand frequencies across both seeding- years, 2008 and 2009.. ..33

Figure 8. Switchgrass dry biomass yield (Mg ha-1) response to nitrogen
fertilizer 2008-2009. Corresponding regression equations and parameters
can be found in Table 8 ........................................................... . ........ 54

Figure 9. Switchgrass dry biomass yield (Mg ha-1) response to Nitrogen in
2009. Regression equations and parameters can be found in Table 9 ........ 56

viii

Figure 10. Switchgrass biomass yield response to nitrogen under the two-cut
harvest management system in 2008. Corresponding regression equations
can be found in Table 11 .................................................................. 61

Figure 11. Switchgrass biomass yield response to nitrogen under the two-cut
harvest management system in 2009. The vertical lines Show the rate of N
that would optimize the total yield from the two-cut system as determined by
a quadratic (OP) or Linear Response Plateau (LRP) .............................. 63

Figure 12. Crystalline cellulose under varying harvest timings. Means with
the same letter are not statistically different at 0:005. Standard error was
121 ........ 84

Figure 13. Figure 14. Arabinose molar concentration (%), as a propOrtion of
the matrix polysaccharides at four different N rates across harvest timings.
Means with the same letters are not statistically different at a=0.05. Standard
error was 1 .02 ................................................................................ 86

Figure 14. Rhamnose molar concentration (%) , as a proportion of the matrix
polysaccharides at four different nitrogen rates, across harvest timings.
Means with the same letters are not significantly different at a=0.05.
Standard error was 0.06 ................................................................... 86

Figure 15. Galactose molar concentration (%), as a proportion of the matrix
polysaccharides at four different N rates across harvest timings. Means with
the same letters are not statistically different at a=0.05. Standard error was
0.22 ............................................................................................. 87

Figure 16. Lignin content expressed as a weight percentage measurement of
the acetyl bromide soluble lignin as a function of applied nitrogen fertilizer.
Means with the same letters are not statistically significant (a=0.05; scheffe
adjusted means). Standard error was 0.273 .......................................... 89

1x

INTRODUCTION

The quest for a renewable liquid transportation fuel has generated criteria
applicable to the biofuel feedstocks under consideration as well as the resultant
fuel. Scientists, policy makers and society insist that the production of the fuel be
sustainable and that the needs of national security, climate security, and the
environment are met. These important issues have led to extensive research
and the preliminary development of a cellulosic ethanol industry. Cellulosic
ethanol is a viable fossil fuel alternative because it is renewable, can lead to a
reduced dependence on foreign oil, and is compatible with existing automobile
standards (Mitchell et al., 2008). Switchgrass (Panicum virgatum L.), a native,
warm-season, perennial grass, has been selected as a model herbaceous
perennial crop for cellulosic ethanol production by the US Department of

Energy (DOE) (McLaughlin and Wullschleger, 1999).

The Energy Independence and Security Act of 2007 calls for the
Renewable Fuels Standard (RFS) to be increased to 36 billion gallons of
renewable fuels by 2022, of which 21 billion gallons must be derived from
cellulosic feedstocks. Additionally, the new farm bill, the Food, Conservation and
Energy Act of 2008, contains provisions that will provide payments to producers
in an effort to support and expand production of cellulosic ethanol. Michigan is
in a great position to benefit from these policy implementations as it has
approximately 238,120 acres of ideal conservation reserve program (CRP) land

(FSA 2008), and over 4.5 million acres of land that is being neither actively

cropped nor forested according to National Land Cover Data. The three
components of sustainable development: profitability, socio-economic factors,
and the environment, provide a practical framework for assessing why growing

switchgrass as a cellulosic ethanol feedstock is viable option for Michigan.

Profitability

Profitability can be viewed from the perspectives of both the switchgrass
grower and future biorefineries. It is imperative that growers can produce
consistently high-yielding switchgrass in a cost-effective manner and obtain a
reasonable price for their crop (Sokhansanj et al., 2009). It is also reasonable to
assume that biofuel refineries will have constraints on what they can afford to

pay for ethanol feedstocks if they are to be profitable (Sokhansanj et al., 2009).

The ability to use existing farming equipment for stand establishment and
biomass harvest is a great advantage to farmers, as it makes the purchase of
new equipment or capital expenditure unnecessary (Parrish, 1999). Additionally,
methods of adding value to the harvested crop before farm-gate would be
advantageous, rather than simply producing a raw commodity that will likely

command a low price.

Cellulosic ethanol biorefineries need to be profitable over the long-term,
which can only be achieved by purchasing high quality, competitively priced
feedstocks with the option of co- or by-product generation (Sokhansanj et al.,
2009). For a high-quality energy crop, growers Should seek to maximize the

concentration of lignocellulose in the feedstock, minimize N and mineral

concentrations, and limit water concentration (Lewandowski and Kicherer,
1997). It is understood that the profit margins for fuel production are relatively
low and thus the ability to co-produce animal feed or biomaterials, biotextiles, or
biochemical is an important consideration (Laser et al., 2009).The use of waste
streams to power the refineries, such co-firing or combusting the residual lignin,
is also an important aspect because it would lower production costs and improve

production efficiencies (Lynd et al., 2009).

Socio-Economic Factors

Many socio-economic factors on both a national and global scale govern
the transition to second-generation biofuels. Population growth rates and the
need for increased food production in the face of climate change, land
degradation, and desertification, require that arable land be reserved for food
and feed production (Robertson et al., 2008). Increased global demands for
energy at a time when oil reserves are thought to have peaked or at least
reaching peak oil, support the case for acquiring renewable fuels (Farrell et al.,

2006)

A domestically produced, liquid transportation fuel that can replace those
derived from nonrenewable energy sourced from increasingly politically volatile
foreign nations is being sought in an attempt to enhance national security
(Farrell et al., 2006). Stimulating rural economic development is an important
domestic issue (Kim and Dale, 2005). Biorefineries Situated in appropriate
fuelsheds are likely to Significantly improve rural economies and strengthen a

rediversification of agriculture (Sarath et al., 2008b). Localized production and

3

use of biofuels in Michigan can aid in job creation and the generation and
retention of money in rural communities. Michigan's urban centers have
experienced economic woes from the decline of the once prominent
manufacturing sector and stand to benefit from the emerging bioeconomy.

Environmental attributes of cellulosic ethanol

The exploration, production, and combustion of petroleum fuels result in
atmospheric pollution that reduces air quality, and greenhouse gas (GHG)
emissions that contribute to climate change (Robertson et al., 2008). Using a
carbon neutral bioenergy feedstock such as switchgrass to produce cellulosic
ethanol is a superior alternative because the combustion of cellulosic ethanol
results in 94 °/o less GHGS relative to gasoline (Schmer et al., 2008), and the
cultivation of switchgrass can reduce atmospheric carbon through C02
sequestration (Ma et al., 2000). Making ethanol from switchgrass also has
environmental advantages over com-grain ethanol because switchgrass
requires less energy and agronomic inputs during cultivation (Mitchell et al.,
2008). Switchgrass production requires less diesel fuel usage because it is
perennial. Perennials do not have to be established each year and thus have an
economic benefit and require fewer net energy inputs relative to annual crops.
These include a lower need for herbicides, fertilizer and other chemical inputs
(Mitchell et al., 2008). Reduced need for annual cultural inputs, minimizes fossil
fuel use in production, and thereby improves the overall energy balance of the
fuel (Hill, 2007). Additional benefits of the perennial nature of switchgrass
include a large and deep rooting mass (Ma et al. 2000b), which can reduce soil

erosion, enhance carbon sequestration rate, and contribute to soil organic

4

 

 

 

 

lion ti

Nling

EWIC

matter levels and quality overall (Robertson and Swinton 2005). Switchgrass
demonstrates high nitrogen use efficiency (NUE) and remobilizes nutrients to the
root system while undergoing senescence, which means reduced nutrient export
from the field and a lower nitrogen fertilizer requirement (Muir et al., 2001 ).
Nitrogen fertilizer is energy intensive to manufacture and has adverse
environmental impacts when used inefficiently. Nitrogen that is not taken up can
lead to contaminated groundwater or eutrophication of surface bodies of water
when leached to groundwater or lost through lateral runoff as nitrate (N03- )
(Carpenter et al., 1998). Alternatively it can be released to the atmosphere as
nitrous oxide (N20), a greenhouse gas with a high global warming potential
(Mosier et al., 1998). Nitrous oxide has a global warming potential (GWP) of 296
9 eq. C02 (9), based on its capacity to trap heat and its atmospheric residency
time (Mosier et al. 1998).

The need for agro-ecological specific agronomic information

Agro-ecological region-specific information is needed to inform producers
on how to maximize yield and quality of switchgrass grown as cellulosic ethanol
feedstock (McLaughlin et al., 1999). The information needed can be classified
into two categories: stand establishment and post-establishment management.
Successful stand establishment in the first year is critical as switchgrass takes
three-four years to establish (Parrish et al. 1999), and plant populations obtained
in the first year largely determines consequent yield potential and plays a large

role in reducing production costs (Schmer et al., 2006).

 

lair
9C1"?
8318:

ex it

Hit
it

{ill

There is limited information on the influence of tillage and seeding
equipment on plant population and stand yield (Parrish et al. 1999). No-till
establishment is desirable due to reduced soil erosion potential and soil carbon
oxidation, especially if establishment is being considered for fallow land such as
that enrolled in the Conservation Reserve Program (CRP) or land abandoned
due to marginal soil quality (Mitchell et al., 2008). If growers opt to use tillage
they will need to know which type of seeding equipment they Should use to
enhance stand establishment. Producers will also need to be informed of the
best seeding date for their particular region and of viable weed control options.
Ways of overcoming the inherent seed dormancy of switchgrass also need to be

idenfified.

Once switchgrass has been established questions arise as to how much
nitrogen fertilizer should be applied as the nitrogen requirement depends on the
yield potential of the site, productivity of the cultivar, and the management
practices under which it is grown (Vogel et al., 2002). The effects of harvest
frequency on switchgrass productivity and quality is still not clearly understood
(Monti et al., 2008). Discovering the optimal harvest timing and frequency is
critical, as this will govern stand longevity, yield, and cell wall composition. The
proportions of cellulose, hemicellulose and lignin are affected by both nitrogen
fertility and the time at which it is harvested (Vogel et al. 2002, Reynolds et al.

2000, Adler et al. 2006).

Cellulosic ethanol derived from warm-season grasses such as
switchgrass will not replace our entire fossil fuel need but it is a method of
producing a liquid transportation fuel while reducing GHG emissions, improving
air and soil quality, and stimulating rural economic development. Much
agronomic research has been done on switchgrass but more information is
needed on improving stand longevity after multiple years of harvest, optimizing
biomass quality and net energy yield from the feedstock, and achieving cost

effective feedstock production.

Objectives

The major objective of the overall study is to develop agronomic and
economic data and information that will lead to commercially viable and
sustainable systems for establishing and producing switchgrass as a high-quality
cellulosic ethanol feedstock in Michigan and the Great Lakes Region. Agronomic
Information on seeding date, planting methodology, nitrogen fertility, and harvest
system management is much needed in the wake of the emerging cellulosic
ethanol industry. Additionally, we need to understand how harvest regimes and
nitrogen fertility rates influence cell wall composition, which in turn determines

feedstock quality.

Two studies were conducted to assess establishment methods and post-
establishment nitrogen-fertility and harvest management in central Michigan.
The first experiment evaluated the influence of seeding methodology and
Planting date on plant populations and biomass yields at two sites between 2007

and 2010. The second study conducted during the same period on a different

7

Site evaluated the effect on N-fertilization rate and harvest management on

biomass yield and feedstock quality.

The objective of the harvest management and response to nitrogen study
was to determine the nitrogen fertilizer rate and harvest timing for optimizing
yield and quality of switchgrass grown as a cellulosic ethanol feedstock while
minimizing environmental externalities. The second study conducted at two
Sites, assessed three methods of planting switchgrass across three seeding
dates. The objective of the establishment study was to determine the best time
for switchgrass to be planted and whether or not tillage and seeding machinery

play a role in plant populations and first-year biomass yields.

CHAPTER 1

EARLY-, MID-, AND LATE-SPRING ESTABLISHMENT OF SWITCHGRASS
UNDER THREE DIFFERENT PLANTING METHODOLOGIES

Abstract: Economically viable production of switchgrass (Panicum virgatum L.)
as a bioenergy feedstock over the long-term requires successful stand
establishment in the first year. Establishment can be hampered by grassy weed
pressure and poor germination from unsuitable environmental conditions. A field
experiment was conducted with the ‘Cave-in-Rock' switchgrass cultivar at two ,
sites at the East Lansing Field Research Facilities (42.75°N -84.47°W) in
Michigan. The objective of this study was to determine the most suitable seeding
date (early-, mid-, or late-spring) and planting methodology for the Great Lakes
Region, to ensure successful stand establishment and high yields in subsequent
production years. Planting methodologies examined were: no-till drilled (NT);
conventionally tilled and planted with a conventional grain drill (CGD); and
conventionally tilled and planted with a double-roller seedbed conditioning
system (DRS). Planting methodology and seeding date combinations resulted in
stand frequencies that exceeded the desired 40% establishment threshold.
Yields for the 2009 seeding-year were highest when tilled and planted in the
mid-spring, with either the CGD or DRS. Second-year yields at the 2008
seeding-year were highest when tilled and planted in mid- or late-spring with a
CGD. Data suggest that a variety of planting methodologies can be employed for
successful establishment, but a seeding date that is combined with frequent
precipitation and warm air and soil temperatures, as well as proper weed control,

is imperative for obtaining satisfactory first- and second-year yields.

INTRODUCTION

Switchgrass (Panicum virgatum L.) was identified as a biomass feedstock
for energy production based on economic and environmental assessments by
the Oak Ridge National Laboratory’s Biofuels Feedstock Development Program.
Reasons for its selection include: adaptation to a wide geographic region; its
perennial life cycle and associated positive environmental attributes; high-yield
potential on marginal lands; and it can be established via direct seeding (Wright
and Turhollow, 2010). However, like other perennial warm-season grasses,
switchgrass has a long establishment period. Reports on the time it takes
switchgrass to reach full yield potential vary from three years (Heaton et al.,
2004), (McLaughlin and Kszos, 2005), to between three and four years (Sharma
et al., 2003) or more (Christian et al., 2002). A lengthy establishment period
means potential stand encroachment from weeds and delayed profitability for
farmers facing operating costs without a harvestable crop. It is understood that
switchgrass is not generally harvested in the first year because its energy
allocation to below-ground development in the establishment year results in
above-ground biomass yields that are not economical to harvest (McLaughlin et
al. 1999). With proper agronomic practices and normal climatic conditions,
however, year-one biomass yields of 50% of full yield potential can be obtained

(Schmer et al., 2006)

A major research priority is to identify reliable establishment tools and
methods that will ensure productive yields in establishment and post—

establishment years, eliminate the need for re-seeding, and thereby reduce

10

establishment costs (Sanderson et al., 2006). McLaughlin et al. (1999) stated
that this research priority needs to be addressed before farmer adoption and the
scale-up of switchgrass production can be realized. There is limited information
on the influence of tillage and seeding equipment on switchgrass plant
population and stand yield (Parrish et al. 1999). Few studies compare
conventional and no-till methods of establishment without the added factors of
unequal Skill or management and weed control methods (Sanderson et al.
2004). Poor stand establishment can delay biomass production by one or more
years (Schmer et al., 2006). The inability to obtain rapid and consistent
establishment of productive switchgrass stands is a major economic constraint
for bioenergy production (Perrin et al., 2008; Sarath et al., 2008b; Schmer et al.,

2008; Schmer et al., 2006).

Establishing an acceptable stand in the seeding year and obtaining
harvestable biomass early on has significant economic effects on the production
of switchgrass for bioenergy. Growing switchgrass in an economically viable
manner requires establishing a good stand with a stand frequency of 40% or
greater in the establishment year (Schmer et al., 2006). Stand frequency can be
defined as the number of times a species occurs within a given repeated sample
area (Greig-Smith, 1983). Low second-year yields from stands with initial stand
frequencies of 40% or lower can be directly attributed to the initial low-stand
frequency. Low second-year yields from stands that met the establishment

threshold are more likely a result of poor environmental conditions.

11

The ability to use existing farm equipment for stand establishment and
biomass harvest is a great advantage to farmers, as it means the purchase of
new equipment or capital expenditure is unnecessary (Parrish et al. 1999). In a
willingness-to-grow study conducted by Jensen et al. (2007), farmers indicated
fltat they would be more likely to grow switchgrass if they owned hay-making
equipment -- as it is readily adapted to harvesting switchgrass -- which suggests
that data in support of existing forage crop establishment equipment could be a
motivating factor in farmer willingness-to-grow. Furthermore, farmers indicated a
strong willingness to try environmentally friendly technologies and thus the
authors concluded that no-till is positively associated with switchgrass adoption.
The Biomass Crop Assistance Program (BCAP) recently amended by Title IX of
the Food, Conservation, and Energy of 2008 Act (known as the 2008 Farm Bill)
supports farmers in establishing and producing eligible crops for conversion to
bioenergy. The recently proposed changes to BCAP will likely provide incentive
for adopting establishment and management practices that enhance delivery of
ecosystem services such as sustaining biodiversity and soil conservation.

It is hypothesized by economists and fuel vs. food observers that
bioenergy feedstocks will be grown primarily on lands that are classified as
marginal due to poor soil quality or slope. No-till establishment would reduce
potential soil erosion events on converted land (Monti et al., 2001; Wolf, 1989).
Additionally, the lack of a tillage event would maintain existing soil aggregate-
size distribution (Grandy and Robertson, 2006) and minimize soil organic carbon

(SOC) mineralization relative to establishment with tillage, thereby providing a

12

potential to maintain or even improve existing SOM levels. Another ecological
benefit of no-till establishment is keeping Arbuscular Mycorrhizal hyphal
networks intact, with positive implications for AM colonization of roots and future
P uptake potential (G058 and de Varennes, 2002). From an agronomic
standpoint, the soil moisture conservation that a no-till system provides could be
helpful in preventing stand failure when spring rains are not frequent after
planting (Wolf et al. 1989). Switchgrass Should not be planted into soils with
heavy crop or weed residues (Parrish et al 1999). No-till establishment could be
enhanced with a fall burndown with a non-selective herbicide, a spring burn-
down and pre- and post-emergence herbicides (Masters and Nissan, 1998;
Masters et al., 1996).

An economic benefit of no-till establishment is fewer trips across the field,
thereby reducing diesel fuel expenses and other costs involved with
establishment, such as depreciation of tillage machinery. Tillage costs constitute
a large portion of establishment costs. Diesel fuel usage accounted for 29% of
establishment costs in a large multi-location field-scale study by Schmer et al.
(2008), compared to 18% in post-establishment years. Limiting tillage passes
across the field could reduce establishment costs despite the need for additional
herbicides. The concept of no-till switchgrass establishment will be further
developed in this study by verifying its ability to produce stands that meet the
threshold for bioenergy feedstock production determined by Schmer et al.
(2006), and by determining which planting dates will result in the best stands

and post-establishment yields. No-till establishment has many potential

13

ecological, agronomic, and economic benefits, provided that satisfactory yields

are achieved.

There are many ecological challenges governing successful
switchgrass establishment. Soil moisture has implications for seed imbibition,
seedling survival, and root development. The planting date iS an important
consideration regarding soil moisture and weed competition (Smart, 1997)
Studies have found that the soil surface must be moist for three to five days
before adventitious roots will form (Newman and Moser, 1988; Wilson and
Briske, 1979). Research conducted in Texas by Ocumpaugh et al. (2002)
indicates that rainfall as frequent as at least once in a seven to 10 day period is
critical for seedling survival in that climate, or in areas with low soil moisture,
high evaporation or non-ideal pH. The need for regional agro-ecological-specific
agronomic information means that many of the planting dates stated in the
literature will not be directly applicable to all locations. Researchers in eastern
Nebraska found mid-March to be a superior seeding date to late-April or May.
The proposed seeding date of within three weeks of the recommended corn
seeding date (Panclera and Jung, 1984) has become a general rule of thumb
(Parrish et al. 1999, (Mitchell et al., 2008) but has not been widely tested (Moser
et al. 2004 as discussed in Parrish and Fike, 2005). Switchgrass is photoperiod
sensitive (Moser, 1995) with biomass accumulation peaking with the onset of
anthesis and seed development. This implies that seedlings that are slower to
emerge will reach peak biomass at the same time as those seeded earlier, even

though they underwent less vegetative growth. A premature shift to reproductive

14

growth because of a late seeding date could possibly have adverse
consequences for root development. Parrish and Fike (2005) recommend
planting non-donTIant seeds after soils have warmed to at least 15 or 20°C and
while rainfall patterns are favorable.

The Great Lakes Region experiences a shorter growing season, and
cooler-wetter springs than the study sites of previous planting date research.
Our major objective, therefore, was to develop agro-ecological, region-Specific
information needed to inform producers on how to establish switchgrass to
obtain high-plant populations and maximize yield in production years. A field
study involving a Side-by-side comparison of conventional and no-till (NT)
switchgrass establishment was conducted to determine which seeding date and
planting methodology result in the highest second-year yields and that present
the best likelihood for establishment success. The following hypotheses were
formulated based on previous research: (i) switchgrass can be successfully
established as a bioenergy feedstock without tillage; (ii) both a conventional
grain drill (CGD) and a dual-roller seeder (DRS) can be used (with tillage) to
successfully establish switchgrass for energy purposes; and (iii) a mid-spring
seeding would result in the highest first-year stand frequency and, subsequently,

second-year yields in the central Great Lakes Region.

15

MATERIALS AND METHODS

Site Description and Experimental Design

Field experiments were conducted on two sites at the Michigan State
University Agronomy Farm in East Lansing (42°25'48" N, 84°16'48" W), over a
three-year period beginning in the spring of 2008. Soils were a mixture of a
Capac loam (fine-loamy, mixed, mesic Aeric Ochraqualfs) and a Colwood-
Brookston loam (fine-loamy, mixed, active, mesic Typic Endoaquolls and Typic
Haplaquolls). Mean pH levels at the two sites ranged from 6.9-7.9, and K levels
were similar; however the P levels at the 2009 seeding-year location were twice
that of the 2008 seeding-year (Table. 1). The previous crops at the 2008
seeding-year location were com (Zea mays L.) and soybean (Glycine max L.) in
2006 and 2007, respectively, using a conventional chisel plow tillage system.
The previous crop for the second seeding-year was canola (Brassica napus).

A randomized complete block design was arranged as a Split-plot with
five replications at two sites, with three planting methodologies (whole plots) and
three seeding dates (subplots). Planting methodologies were: (NT) drilled;
conventionally tilled and conventionally drilled (CGD); and thirdly, conventionally
tilled and planted with a Brillion Sure Stand grass seeder (DRS) (Brillion Iron
Works, Brillion WI). The Brillion consists of a double-roller seedbed conditioning
system which firms the seedbed with the first roller and presses the seed in

Place with the second.

16

Conventional tillage refers to using a chisel-plow in the fall and a soil
finisher in the Spring. A John Deere conventional grain drill (CGD) (Model No.
450, Westminster, .MD) on a 19cm row spacing was used for both the no-till and
conventional seeding, and a cultipacker (Brillion Iron Works, Brillion WI) followed
the grain drill used in the NT treatment to ensure good seed-to soil contact.
Seeding dates (Table 2) were later in 2009 due to excess precipitation resulting
in soils that were too wet for earlier seeding operations. Switchgrass cultivar
‘Cave-in-Rock' was seeded into plots that were 12m X 3m at the 2008 seeding-
year and 9m X 3m at the 2009 seeding-year, at a seeding rate of 9 kg ha". No
soil amendments were applied in the establishment year. Nitrogen fertilizer was
applied to the 2008 seeding-year the following year on May 8, 2009, as urea
(46-0-0 N-P-K) at a rate of 79 kg N ha",

Table 1. Soil properties obtained from soil samples taken prior to fertilization in
2009 growing year.

 

 

 

 

 

 

 

 

 

 

 

 

Soil Property Seeding- Year

2008 2009
pH 6.9 7.2
p 23.4 48.8
K 111.8 125.6
Ca 1049.4 1617.8
Mg 215.4 280.8
Nitrate 2 3.8
Ammonium 2.9 2.9

 

17

 

Table 2. Agronomic operations for both seeding- years over course of field

experiment 2008- 2010

 

Agronomic Event

Spring Seeding Date

 

2008

2009

 

Early Mid Late

Early Mid Late

 

Tillage & Seeding Event

22- 15- 6-
April May June

 

 

 

12- 26- 15-
May May June

 

 

 

Herbicide Control
Progrgm

 

Pre-plant burndown
(no- till):

 

glyphosate (4.68 L ha'1 +
AMS (12% vlv)

n/a

 

18-May

 

Pre-emergence:

 

atrazine (0.6 kg a.i. ha")
)

+ quinclorac1(0.6kg a.i. ha'

n/a

18- 31 - 22-
May May June

 

 

 

Post-emergence:

 

atrazine (1.12 kg a.i. ha“)

19-May

n/a

 

bromoxynil (1.16 L ha")

25-June + 22-July

n/a

 

Second- year N fertilizer
Application

 

(79 kg N ha" Urea 0-0-46)

 

 

8-May 2009

 

30-April 2010

 

Weed Control

In the 2008 seeding-year no pre-emergence herbicides were used.

Atrazine [6-chloro-N-ethyl-N'-(1-methylethyl)-1,3,5-triazine-2,4-diamine] and

bromoxynil [3,5-Dibromo-4-hydroxybenzonitrile] were used post-emergence to

obtain broadleaf control. Heavy weed pressure from three 0., grasses; fall

panicum (Panicum dichotomiflorum), giant foxtail (Setarl'a faberi), and large

crabgrass (Digitaria sanguinalis) was observed at the site. In 2009, a more

rigorous chemical weed-control program was employed. This included a pre-

18

 

plant glyphosate [N-(phosphonomethyl) glycine] burndown for the NT
treatments, and a pre-emergence application of a quinclorac [3,7-dichloro-8-
quinolinecarboxylic acid] and atrazine tank-mix. A post-emergence application of
the salt of bromoxynil [3,6-dichloro-2-methoxybenzoic acid] was made to the
early-seeded treatments to control an escape of redroot pigweed (Amaranthus

retroflexus).

Biomass Measurements

Stand frequency was measured using a frequency grid as a means of
quantifying establishment success. The grid frequency method was originally
developed by Vogel and Masters (2001), and then used to determine stand
thresholds for growing switchgrass as a commercial biomass energy crop by
Schmer et al. (2006). The grid is comprised of 25 (15- X 150cm) cells, and was
placed randomly within the plot and then turned over four times for a total of 100
cells and each cell is marked for the presence or absence of switchgrass. A
conservative estimate of plant population (plants m'2) can be obtained by

multiplying the stand frequency by 0.4 (Vogel and Masters, 2001).

First-year yields for the 2009 seeding-year were harvested with a flail-
type forage harvester (Carter Mfg Co., Brookston, IN) equipped with an
automated weigh system to determine biomass yield, set at an 18cm cutting
height on October 26, 2009. Second-year cuttings of the 2008 seedings were
made on October 20, 2009 with a commercial forage harvester (Model No. 7650
Hesston Co., Hesston, Kansas), set at an 18cm cutting height that was modified
for plot research with the addition of a weigh bin for yield determination from

19

individual plots. Biomass content for each plot was recorded and biomass
moisture content for each plot was determined via subsamples that were placed
in a forced-air oven at 100°C for 72 hours or until a consistent dry weight was

achieved.

Statistical Analysis

Data were analyzed using PROC MIXED in SAS v9.1 (2003 SAS Institute
Inc., Cary, NV.) Analysis of variance (ANOVA) was conducted to measure the
treatment effects on stand frequency and yield. The analysis was performed as
a RCBD arranged as a Split-plot design. Seeding date and planting methodology
were treated as fixed effects while block, seeding-year, and the interaction
between the whole-plot factor and block were considered to be random effects.
Normality of the residuals and homogeneity of variances were evaluated by
examining normal probability plots and box plots. When variances were found to
be heterogeneous, the REPEATED/GROUP option of PROC MIXED was
implemented. Fisher’s protected least significant difference (LSD) multiple
comparison procedure was used for mean separation when ANOVA was
Significant. Results were reported as statistically significant at 0t = 0.05. Linear
contrast statements were constructed in addition to the multiple comparison
procedure to test the significance Of prior hypothesized effects of the main

factors.

20

RESULTS AND DISCUSSION

Climatological Summary

The 2008 growing season tended to be drier than both the 30-year
average and 2009 (Table 3). April and May were 41% and 46% drier than the
30-year mean and 53% and 75% drier than 2009, respectively. August was also
a very dry month, receiving approximately 82% less precipitation than 2009 and
the 30-year average. Precipitation from April through the month of June was
wetter than average in 2009, and in August precipitation was Similar to the 30-
year mean. Mean air temperatures did not vary considerably between years or
within the 30-year average, although April of 2008 was Slightly warmer than
either 2009 or the climatic norm, indicating that April of 2008 was unusually
warm and dry which had implications for the early-seeded treatment in that year

(Table 3).

Stand Frequency and Yield

Although site -year was considered to be a random effect and found to be
non-significant in the full model, it is useful to consider each year separately in
relation to the environmental conditions and relative weed pressure for further
insight into the risk of potential stand failure. A discussion on Significant findings
and trends for data from both years will follow, with a discussion of the results of

hypothesis testing.

21

Table 3. Monthly rainfall totals and average mean monthly air temperatures at
the Michigan State Agronomy Farm, East Lansing, MI, in the 2008 and 2009
growing seasons and the 30-year average (1971-2000).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Mean Monthly Rainfall Mean Monthly Temperature

Month 2008 2009 30-yr 2008 2009 30-yr
average average
mm °C

April 43.4 147.3 83.1 10.3 8.35 7.6
May 29.5 109.0 69.1 12.8 14.0 14.0
June 112.5 126.3 81.0 20.2 18.9 19.2
July 96.25 60.7 75.7 21.4 18.8 21.4
August 16.51 104.7 87.5 20.5 20.1 20.3
September 206.8 24.1 89.2 17.2 16.8 16.0
April-
septembe' 84.2 95.3 80.9 17.0 16.2 16.4

 

 

 

 

 

 

 

 

 

Results of the 2008 Seeding-year I

For the 2008 seeding-year, soil and air temperature did not start to rise
until the end of May (Fig. 1). There were two significant dips in air temperature
with corresponding decreases in soil temperature within 10 days of the seeding
event, which could potentially have induced seed dormancy. There were no
precipitation events immediately preceding the early-spring (ES) seeding date,
and few precipitation events after the mid-spring (MS) seeding date, which is
necessary for seed imbibition. Soil temperatures were not conducive to rapid
germination for the early-, and mid- dates, however conditions improved greatly

for the late-Spring (LS) seeding event.
22

 

 

 

 

 

 

 

 

 

 

 

 

 

l 30 4 35 i
2 l
25 1 30
J L
f I . 25
A 20 ~ ' A
9 l i e l
E e i— 20 5 i
C
:I o
*5 15 -
3 I I LS i g
a \[V ,I l 15 e-
: O
E. l e
I- 10 ~ IL
f 10
5 l | MS I i 5
O T II. 1 Li ' I II I .— I I i' ‘ t I ll 0
to o to o to o to o to o v t» v
3:. Q! 9‘... £2 ”in t 3: 9‘. Q S2 25 25 S.
<1- v v v to to to to to co
— Precip (mm) --Mean Daily Temp 00
[ ---'Mean Soil Temp (2" Depth)

 

Figure 1. Environmental Conditions during planting in the 2008 Seeding- year.
Weather information was obtained from the Michigan Automated Weather
Network.

23

Stand Frequency

 

100 1*

 

 

g 80

e /

8 60 J /

:7 40 j g IEarly-Spring
L; g I Mid-Spring

g 2: : é aLate-Spring

NT DRS CGD
Planting Methodology

 

 

 

-. M- .. we. “‘W—‘mfl_uv( .2- ,-.,.-__-.7.x “.1.” ”w-WE—w: —-'.z w: —e-..~_,.‘.._..,__._,_._4

Figure 2. Stand frequencies obtained at the 2008 seeding-year. Means with the
same letters are not significantly different at a=0.05. Abbrev: No-till (NT), Dual-
roller Seeder (DRS), Conventional Grain Drill (CGD).

The highest stand frequencies were obtained when tillage was used at
either the mid-, or late-spring seeding date when the CGD was used, and at the
LS date when seeded with the DRS (Fig. 2). The ES seeded DRS treatment
failed to reach the desired 40% establishment threshold determined by Schmer
et al. (2006), and the ES mnventional treatment came very close to failing, and
may have, had this been on a larger scale with increased spatial variation.
Contrast statement estimates indicted that the CGD performed Significantly
better than the DRS at the ES and MS seeding dates but did not differ at the LS
date when moisture was less limiting. The current, un-improved varieties of
switchgrass are known to have issues with the pre-determined length of the

subcoleoptile internode, and the cessation in growth of the coleoptile when it

24

encounters light (Newman and Moser 1988). Adventitious roots form at the
crown node, which can often be very close to the soil surface, especially with
shallow plantings like that afforded by the DRS. If the soil surface is dry, or if
there is a lack of rain, the highly necessary, adventitious roots will abort or
simply cease to grow. Breeding efforts are underway to select for lower crown
node placement (McLaughlin and Kszos, 2005), but in the meantime, drilling
switchgrass to a depth of 1cm is advantageous in dry years (Evers and Butler
2000) and rolling the seed-bed after a shallow seeding event can also be
advantageous (Monti et al., 2008), as discussed in Parrish and Fike (2005).
The results of the linear contrast statements used to test the hypothesis
indicated that NT treatments performed better than tilled treatments at the ES
seeding date, however at the late-seeding date, NT resulted in stand
frequencies that were Significantly lower than either of the tilled treatments.
Stand frequencies in NT plots exceeded the desired establishment threshold,
and did not vary as much as the tilled plots, indicating that they were not
affected by planting date or environmental conditions to the same extent.
Although the soil temperatures under NT were likely cooler (Cox et al., 1990;
Kravchenko and Thelen, 2007), it was not to the detriment of establishment
success. The plots that underwent tillage were observed to have significantly
higher weed pressure from C4 annual grasses than the NT plots at the ES and
MS seeding dates, which reduced the potential establishment success under
tillage that was evident at the later seeding date. The tilled treatments likely had

a competitive advantage at later dates with increased soil temperatures,

25

decreased weed pressure (flushes of annual summer grasses were controlled
with a delayed tillage event), and from improved seed-to-soil contact over the NT
system. Strong competition from three C4 grassy weeds was observed in plots
that were tilled and seeded at the ES and MS seeding date, particularly from
giant foxtail, large crabgrass and fall panicum and by early July these plots had
a closed canopy comprised of weedy biomass, implying competition for light and
water. Surviving seedlings had a spindly appearance. Giant foxtail was
particularly problematic as it was listed as a possible weed species present on
the seed tag, has a shallow rooting system, and was present in the seed bank.
Crabgrass has been previously reported as a concern for switchgrass
establishment due to its similar germination timing and morphological
development (Nelson, 1986). Fall panicum, although an annual, is closely
related to switchgrass, which could have implications for selecting a herbicide

based on physiological differences.

Although winter-annual, biennial and perennial weed species were noted
in the NT plots, the canopy did not close due to the presence of weeds as
observed in the tilled treatments. Interference from crop residue, living weeds
and remnants of annual bluegrass (Poa annua) appear to have reduced seed-to-
soil contact. Species that were present, such as marestail (Conyza canadensis)
and dandelion, could be treated with a fall burndown of a systemic herbicide in
the future. No-till drills could be fitted with row cleaners to prevent potential hair-
pinning of residue, and wave or bubble coulters to improve seed-to-soil contact,

particularly in excessively wet or dry soils.

26

Yield

First-year yields for the 2008 seeding-year were not calculated due to the
presence of weedy biomass and an inability to differentiate the cut switchgrass
from grassy weeds in the post-senescent state. There was a Significant
interaction between seeding date and planting method (p = 0.0026) for second-
year yields of the 2008 seeding-year. Both seeding date (p=0.0004) and planting
method (p=0.0179) were found to be significant (Fig. 3). The highest second-
year yields were achieved when tilled and planted with the CGD at either the
mid-, or late-spring seeding date, similar to the results of stand frequency in the
establishment year (Fig 3). The late-seeded DRS had comparable yields to the
late seeded conventional treatment. According to contrast statements the DRS
and CGD performed similarly, and only differed for the MS seeding date
(P=0.0004). The MS date was only advantageous in second-year yield gains for
the conventional treatment. The early seeded DRS treatment, which failed in the
establishment year, had the lowest second-year yields, as did the early seeded
conventional treatment, which just met the 40% stand frequency establishment
threshold. The NT treatments did not differ significantly from either of the low-
yielding aforementioned treatment combinations despite having met the
establishment threshold. They were also comparable to the yields obtained
under the DRS treatment. The MS and LS conventional treatments yielded
significantly greater than the other two methodologies and it does not appear to

be related to stand frequency, and is possibly due to being planted at an optimal

27

depth with frequent rains that supported adventitious root proliferation allowing

for improved over-wintering capabilities.

1.

 

 

‘iu

.C

E) lEarly-Spring
U)

8 IMid—Spring
co DLate-S rin
a P 9
C)

NT DRS CGD
Planting Methodology

 

 

 

Figure 3. Second-year yields for the 2008 seeding-year. Means with the same
letters are not significantly different at a=0.05 level of Significance. Standard
error was 0.70. Abbrev: No-till (NT), Dual-roller Seeder (DRS),Conventional
Grain Drill (CGD).

Results of the 2009 Seeding-year

The 2009 seeding—year was marked by frequent and plentiful rainfall prior
to, and after, each seeding event (Fig. 4). There were 19, evenly Spaced
precipitation events between mid-May and the end of June that likely facilitated
seed imbibition, high emergence rates and a low incidence of seedling death.
Both primary and adventitious root development would also have benefited from
the frequent rains. Ambient air and soil temperatures rose shortly after the ES
seeding event and soil temperatures remained above 15°C for all seeding
events, climbing to over 20°C at the LS seeding date (Fig. 4). Air temperatures

were favorable for seedling growth; the two dips in air temperature that occurred

28

after both the springs and MS seeding events occurred during their respective

germination and emergence periods and therefore did not affect seedling

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

growth.
50 1
A 40 A i I 1’":
g I ‘ r‘v“ /"\ 'HIII‘,’ ‘
e . t’
3 ‘\ A'"\v’
E ’/ V'I
e
is“
LS
a MS “ 10
ES
- 5
[LT] It'll] rl‘fill‘ [r ‘1‘ r47 0
O I!) O to O U) 0 V O) V O) V O)
QBEEQQQBBEEQQ
V ID ID ID It) to (O (O (D (O
—Precip (mm) “Mean Air Temp (OC) '
----- Mean Soil Temp (2" Depth)

 

Figure 4. Environmental conditions at planting time for the 2009 seeding year.
Weather information was obtained from the Michigan Automated Weather

Network.

29

FSCMC

Stand Frequency

 

90 1'
80 '1
70 '
60 - °
50 1
40 ~
30 ~
20 -
10 e
0 " " “ T

NT DRS CGD

Stand Frequency (%)

 

   

 

 

 

Planting Methodology

 

 

 

 

Figure 5. Stand frequencies obtained at the 2009 seeding-year, Means with the
same letters are not statistically different at the a=0.05 level of significance.
Stand error was 2.96 Abbrev: No-till (NT), Dual-roller Seeder (DRS),
Conventional Grain Drill (CGD).

In 2009, all treatment combinations resulted in stand frequencies above
40%. Although there was an interaction (p=<0.0001) between planting
methodology and seeding date, seeding date itself was not found to be
significant, but planting methodology had a Significant effect (p=<0.0001) (Fig.5).
The highest stand frequency was obtained with the DRS, and the lowest under
NT. The Brillion double-roller seedbed conditioning system appears to have
fared well in this seeding-year, which had adequate precipitation, warm ambient

air and soil temperatures, and minimal weed competition.

30

Yield

s-hmw. ,...‘... , ..L . -..a. -. ... .., .....,-,_ ,, ,, ... ,_,,_ . .7 l

 

 

 

 

 

Tu IEarly-spring
g) 2 a IMid-spring
2 .. b aLate-spring
3 1.5 DC DC DC d Cd
“3 d
E 1 a
.9
co
2 ..
O 0.5

O ..

NT DRS CGD
Planting Methodology

 

 

 

Figure 6. First-year yields in the 2009 seeding-year. Means with the same letters
are not significantly different at the a=0.05 level. The standard error was 0.176
Abbrev: No-till (NT), Dual-roller Seeder (DRS), Conventional Grain Drill (CGD).

There was a significant interaction present in first-year yields in the 2009
seeding-year between planting method and seeding date (p=0.0179), and
although both main effects, planting method (p=0.048), and seeding date
(p=<0.0001) were significant, it is apparent that seeding date was a bigger driver
of yield than planting method (Fig. 6). The highest yields were obtained with the
DRS treatment at the MS seeding date, and the lowest yields were obtained at
the LS seeding date for all planting methods, and when tilled and seeded early

with the conventional CGD (Fig. 6).

31

No-till yield estimates derived from linear contrasts were only significantly
lower than other planting methodologies at the LS date (p=<0.0001). These
results appear to agree with Rehm (1990), who found no yield differences
between conventional and NT establishment in Nebraska. Delayed emergence
and subsequently a shorter vegetative growth period meant that the LS NT
seeding shifted into reproductive growth at approximately the same time as the
other treatment combinations and did not accumulate as much biomass. A MS
seeding date was beneficial when planting with the DRS and CGD, into tilled soil
(Fig. 6). DRS and CGD seedings were comparable except for the MS DRS,
which out-yielded the MS conventional treatment (p=0.0004). Under optimal
establishment conditions like that of the 2009 seeding-year, the DRS appears to
have had rapid emergence and vigorous growth, allowing for ample shoot-
growth. The consequences for second-year yield will be investigated at the

conclusion of the upcoming growing season (2010).

Summary of Establishment Results Across Planting Sites

Pooling stand frequency data across seeding-years was possible
because seeding-year was considered as a random effect, and found to be non-
significant in the full model based on its Pr> fValue from the covariance
parameter table. When stand frequency data was combined across both
seeding-years there was also an interaction between planting method and
seeding date (p=0.008) and both planting method (p=0.0315) and seeding date

(p=0.0091) were significant (Fig. 7).

32

 

 

 

 

90 Early-Sprin g
80 I Mid-Sprln g
A Ilate-Sprlng
§ 70
3’
c 60 ,
3
0' 50
2
E 40
g 4
a 30’
2O 2
10 i
0 :
Plantin g Meth odology

 

 

Figure 7. Stand frequencies across both seeding- years. Means with the same
letters are not statistically different at the a=0.05 level of significant.

When stand frequency data were combined across years it was still
apparent that tillage and either a mid- or late-Spring seeding date resulted in the
highest stand frequencies (Fig. 7). Contrast statements used for hypothesis
testing indicate that there were no differences in final stand frequencies between
MS seeding date and all other dates, and that the results of the CGD and DRS
treatment did not differ. All NT treatments achieved successful stand frequencies
in both years of establishment, and although there was less variation, LS NT
stand frequencies were significantly lower (p= 0.0001) with a trend noted for the
MS date (p=0.06) (Fig 7). This is in agreement with an establishment study of
grass mixtures in Nebraska by Oldfather et al. (1989) that found emergence to

33

be higher when tillage was used rather than direct seeding into undisturbed soil.
It is not in accordance with King et al. (1989), who found a markedly increased
number of seedlings with NT vs. conventional establishment. Although the
planting methodologies that employed tillage achieved higher stand frequencies,
the levels obtained are not necessary for cost-effective establishment, and might
carry a higher inherent risk than NT, where more consistent and less variable

stand frequencies were observed.

CONCLUSION

The data suggest that tillage is not critical for obtaining a stand frequency
that exceeds the 40% establishment threshold, although first- and second-year
yields lagged behind tilled treatments in some instances. However, the use of
one or two pre-plant herbicide burndown treatments, proper residue
management, optimization of NT drills, and seeding into warmed-up soils could
potentially enhance NT stands and yields. Further research is needed to verify
this. Additional growing seasons are needed to determine if the establishment
gap closes among planting methodologies. The data also support the hypothesis
that with tillage, both a CGD and a double-roller seeder can be used to
successfully establish switchgrass for energy purposes. These results suggest
existing equipment can be used to prepare seedbeds, which would mean less
capital expenditure and an increased likelihood of farmer adoption.

The hypothesis that a mid-spring seeding offers the best possibility of

establishment success in the central Great Lakes Region was not fully

34

 

Di

6f

Fer.

Ol'

hi

(I)

supported by the results as there were interactions between planting
methodology and seeding date in some cases, and performance varied between
planting methodologies across seeding dates. The data do not indicate a benefit
from early planting like that found by Smart and Moser (2007) in Nebraska and
Vassey et al. (1985) in central Iowa. However, the wet springs and frequent
precipitation events in Michigan may allow for better establishment success with
a mid-spring seeding than in Nebraska. Planting up to three weeks before the
recommended planting date for corn (Panclera and Jung, 1984) would likely be
on the early side for our region.

However, all planting method and seeding date combinations that ranked
highest in yield for both seeding-years included the mid-spring treatment. A mid-
spring seeding date could potentially offer the best possibility for increasing yield
and profitability in early production years. The seeding date results are in
agreement with Parrish and Fike (2005) and Hsu and Nelson (1986b) who have
conducted similar research in Virginia and Missouri, respecfively, and found that
planting non-dormant seed into warmed soils while precipitation is still frequent
is best for optimizing establishment.

Further research on optimizing no-till establishment systems would be
beneficial for enhancing eco-system services delivery from perennial cropping
systems with the possibility of reducing establishment costs. Research on
chemical control programs for switchgrass establishment and early stand
management would also be of benefit as it would lead to registration of effective

herbicides that could further enhance establishment and cost-effectiveness.

35

CHAPTER 2

SWITCHGRASS BIOMASS YIELD RESPONSE TO VARYING RATES OF
NITROGEN FERTILIZER AND HARVEST TIMING IN THE CENTRAL GREAT
LAKES REGION

Abstract: Switchgrass (Panicum virgatum L.) is a C4, perennial grass that has
been identified as a suitable bioenergy feedstock for marginal land in the United
States. A field experiment was conducted at the MSU Field Research Facilities
in East Lansing (42.75°N- 84.47°W). The objective of this study was to
determine switchgrass biomass yield response to nitrogen fertilizer in the Great
Lakes Region. The cultivar ‘Cave-in-Rock’ was established in a RCBD and
uniformly harvested in 2007. Four different nitrogen rates (0, 38, 79, and 157 kg
ha'” were applied to the plots that were then harvested either under a two-cut
system (anthesis and post-senescence), late fall or left to overwinter and then
harvested in the spring. Agronomic recommendations on rate of N application
and harvest timing for this agroecoregion will be derived from these findings.
Yields across all N rates and harvest systems ranged from 401025 Mg ha-1 in

2008, and 9.0—12.25 Mg ha-1 in 2009.

36

INTRODUCTION

The Energy Independence and Security Act of 2007 (EISA) mandates the
production of 36 billion gallons of renewable fuel by 2022, of which 21 billion
gallons must be from advanced fuels such as cellulosic ethanol or biobutanol. A
number of possible feedstock sources are under consideration for cellulosic
ethanol production, including residues from annual crops, herbaceous perennial
grasses and woody perennials. Financial assistance to establish and produce
this material iS available to producers through a provision of the Farm Services
Bill 2008, the Biomass Crop Assistance Program (BCAP). The government will
match each dollar paid out by a Biomass Conversion Facility (BCF), up to a
maximum of $45 ton'1 over two years of production. It is expected that the
emerging cellulosic ethanol industry will consist of regional BCFs that will
acquire biomass feedstock from the land surrounding the facility in a 48-mile
radius, known as a fuelshed (Sarath et al., 2008b). The boundaries of the
fuelshed are determined by the distance biomass feedstock can be transported
without the costs of hauling the material becoming uneconomical (Sarath et al.,
2008b). Three years of field trial data conducted by the ORNL Bioenergy
Feedstock Development Program (BFDP) between 1993 and 1996 emphasized
the need for regional specificity of optimum harvest strategies (McLaughlin et al.

1999)

Switchgrass populations can be divided into two ecotypes: lowland and
upland (Porter 1966). This was confirmed by Hultquist et al. (1996), who found

that upland ecotypes can be either hexaploids (2n=6x=54) or octaploids

37

(2n=8x=72), and that lowland ecotypes are tetraploids (2n=4x=36). Lowland
ecotypes were developed in lower-lying, ‘hydric’ sites, and tend to be more
sensitive to moisture stress and have recently been regarded as a facultative
wetland species (Barney et al., 2009). Upland ecotypes were developed on
higher, ‘mesic’ Sites and are known to be more drought resistant. Switchgrass
also demonstrates adaptation to geographic areas and performs better within its
latitude of origin (Casler et al., 2004). From a morphological standpoint, the
upland ecotypes tend to be Shorter, and have thinner stems than lowland
ecotypes. Northern upland populations, for instance, would be more productive
in the northeast. Upland cultivars have a greater potential for yield response with
two-cut management due to phonological differences - uplands have a faster
rate of development, maintain higher rates of photosynthesis, and a greater
capacity to recover from stress (6.9. higher drought tolerance) (Wullschleger et

aL,1996)

Obtaining high biomass yields, inter alia, is a top priority for switchgrass
bioenergy production, regardless of the conversion platform or the desired end
use. The quantity of dry matter produced by a biomass species per unit area of
production determines the potential energy production capacity or yield of the
available land area (McKendry, 2002). High yields equates to better land- and
water- use efficiency, and higher profit per acre, the latter depending on the ratio
of input costs to feedstock value. Management factors, such as fertility
programs and harvest frequency, determine stand survival, productivity,

feedstock quality, and economic returns (Fike et al., 2006). N fertilizer

38

application rate and harvest management systems must be optimized to
minimize dry matter loss and maximize biofuel quality (Lemus et al., 2008).
Nitrogen fertilizer is a costly input from both a monetary and energy standpoint,
and is a significant source of global warming potential (GWP) in cropping
systems fertilized with synthetic nitrogen (Robertson, 2004). Furthermore, the
profitability of switchgrass as a biomass crop would be enhanced if acceptable
yields were produced with a minimum amount of applied N (Madakadze et al.,
1999)

Harvest Management Systems

Seasonal time of harvest affects switchgrass yield (Adler et al. 2006,
Madakadze et al., 1999; Sanderson et al., 1999; Vogel et al., 2002; Casler and
Boe, 2003). The effect of harvest frequency on switchgrass productivity and
quality as a bioenergy crop is still not clearly understood (Vogel et al. 2002,
Monti et al. 2008). Producing switchgrass as a biofuel feedstock has different
management objectives than that of a forage crop, or production as a feedstock
for energy production via combustion. Vogel et al. (2002) report that the optimal
time to harvest switchgrass for maximizing biomass yields in the Midwest is at
the R3 to R5 stages of maturity, which means that the cut would be made
between panicle emergence and postanthesis (Moore et al., 1991 ). They also
state that with the right growing conditions, sufficient regrowth could be obtained
in time for a killing frost. However, long-terrn yield depression is not precipitation
related, below ground root reserves are depleted from not being replenished by
mineral nutrients during natural crop senescence (Sanderson et al., 1996).

Decreased productivity over the long-tenn has been reported with multiple

39

harvests (Casler and Boe, 2003; Cuomo and Anderson, 1996; Mulkey et al.,
2008). Optimal harvest timing is not based on maximizing yield in the Short-term,
but on obtaining high yields while maintaining stand longevity, and economic
viability in terms of reducing N-P-K inputs.
One-cut harvest system Described

A harvest management system consisting of a Single cut made after
switchgrass has naturally senesced and at least two weeks after a killing frost is
generally believed to be the most suitable from a biogeochemical standpoint.
This system facilitates the mobilization of carbohydrates and nutrients into the
below-ground, which means less nutrient mining of the soil with harvest, and,

consequently, lower fertilizer requirements.

Adler et al. (2006) found that yield decreased when harvest was delayed
until spring by 20-24% and 32-43% at the plot and field scale respectively.
These yield losses were attributed to biomass not picked up by the baler, and a

decrease in standing tiller weight, which accounted for <10% of the loss.

Two-cut Harvest System Described

The feasibility of a two-cut harvest system is contingent on quality (N and
ash content), seasonal yield, long-term regrowth capacity, economic lifespan of
the stand, and cost of multiple harvests (Monti et al. 2008). Cutting switchgrass
twice per year may increase biomass production but the impact on fuel quality
and economic return needs to be investigated to determine if the additional
yields from the two-cut system are warranted (Fike et al., 2006). ‘Cave-in-Rock’,

a northern upland ecotype could potentially provide higher yields with a two-cut

40

system; however, this yield increment could be countered by the need for
additional N fertilization; harvest operations and associated variable costs; and
reduced fuel quality due to high N concentrations. It is well understood that a
two—cut harvest system takes up and exports more nitrogen than a single-cut
system does (Reynolds et al., 2000). In northern regions that have shorter
growing seasons, the timing of the first-cut in a two-cut system is critical as it
affects regrowth potential. The earlier the first harvest occurs the higher the
regrowth will be; provided that the switchgrass is harvested above the growing

points and that there is adequate precipitation.

Decreased productivity over the long-tenn has been reported with
multiple harvests (Cuomo et al.1996, Casler and Boe 2003, Mulkey et al. 2007).
The two-cut system leads to increased seasonal biomass yield in the two years
following establishment (+15% and +38%, respectively) but third-year yields
were significantly lower (28% compared to the one-cut system) (Monti et al.,
2008). Reynolds et al. (2002) found that in Tennessee, the two-cut system only
out-yielded the Single-cut system when summer precipitation was not limiting.
The two-cut system performed poorly at the Texas research site. This is
believed by the authors to have been caused by the effect of the cutting on root
persistence, and the consequent impediment on late-season water uptake under
dry growing conditions (McLaughlin et al. 1999). Stand losses can occur if there
is not a six-week span between the last harvest and a killing frost (Moser and

Vogel 1995).

41

Yll

Yield Response to Applied Nitrogen Fertilizer
Achieving high Nitrogen Use Efficiency (NUE) in switchgrass is important if

this biofuel feedstock is to be grown in an environmentally sustainable manner
and with producer profitability in mind. Important considerations in selecting a
nitrogen rate in, addition to increasing yield, include minimizing nitrous oxide
emissions to the atmosphere, and the loss of nitrate-N to groundwater through
leaching. Minimizing excessive N fertilizer in a cropping system translates into
lower nitrous oxide (N20) emissions, which are 310 more potent than 002 on an
equivalent basis as a GHG (Fargione et al., 2008).

Switchgrass production is optimized and quality stands are maintained with
good fertility management (Sarath et al., 2008b). While the calculation of the
most economic rate of nitrogen is important for cost-effective production, a more
pressing research need is estimating the optimal nitrogen rate for maximizing
yield: yield potential data are lacking in the region; no commercial cellulosic
ethanol production currently exists; the expected feedstock value is not known;
and the price of nitrogen will fluctuate over time and from region to region.

The main fertilizer requirement of switchgrass is nitrogen (N) (Vogel et al.
2002). Economically viable production of perennial grass monocultures coupled
with substantial quantities of annual biomass removal would be expected to
require inputs of N fertilizer (Heaton et al., 2004). The N requirement of
switchgrass is a function of a number of factors ,including: harvest timing and
frequency, location, rainfall and length of growing season, yield potential of the
site, productivity of the cultivar, and other agronomic management practices

(McLaughlin and Kszos, 2005; Vogel et al., 2002). Nitrogen recommendations
42

should be Site-specific and take into account soil type, and the crop’s use of N
and system NUE (Brejda et al., 1998).

Soil moisture and water-holding capacity can be important factors in yield
(Fike et al., 2006). These factors interact with the above stated environmental
considerations and thus information Specific to the particular agro-ecological
region, coupled with local trials, are needed to determine how these
management factors affect yield and to fine-tune N-fertilization levels to local
conditions (McLaughlin and Kszos 2005). A major objective of the fertility
regime, in addition to economic and environmental concerns, Should be to avoid
application of N-fertilizer that could cause stand lodging, which has implications
for harvesting and yield.

Previous research on the estimation of switchgrass biomass yield response
to N has been summarized by McLaughlin and Kszos (2005) and more recently
in Haque et al. (2009). Considerable differences were found across location,
switchgrass varieties, and harvest timings; further, genotype environment
interactions play a large role in switchgrass production (Fike et al., 2006).
Therefore this review will focus on studies that are applicable to our
agroecological region, and that used upland varieties (e.g. Cave-in-Rock) under
similar harvest regimes.

In a study conducted at Ames, Iowa, and Mead, Nebraska, that examined
switchgrass yield response to nitrogen fertilizer and harvest frequency, Vogel et
al. (2002) found that the response to increased nitrogen was linear for both first-

cut (late-boot to early anthesis), and total biomass yields at Mead, and

43

curvilinear at Ames. They reported a yield range of10.5-12.6 Mg ha'1 and an N
level of 120 kg ha'1 associated with maximum yield. A yield increase of s 0.6 Mg
ha‘1 was gained with N rates above 60 kg ha'1 for the first-cut harvests made at
Mead, and that yield plateau was present at Ames, with no increase in yield with
N rates above 180 kg ha". Further, their regrth yields made from successive
harvests of the plots that were cut at late-boot at two-week intervals, increased
linearly with increasing N at Mead but did not respond to nitrogen at the Ames

site.

In southern Iowa, Lemus et al. (2008) obtained a yield of 8.5 Mg ha'1
when switchgrass was harvested after a killing frost (single-harvest) and found
the N level associated with maximizing yield was 112 kg N ha“. The authors
noted a non-linear response to nitrogen, with diminishing yields occurring at
higher N levels and calculated the most efficient rate of fertilizer to be 56 kg N

ha'1 from a fertilizer use efficiency (FUE) standpoint.

Haque et al. (2009) compared two harvest systems in Stillwater,
Oklahoma: one harvest per year in October, and a two-harvest system (July and
October) in which the nitrogen rates were split and applied at the beginning of
the season and again after the first harvest. They found that the one-cut system
out-yielded the two-cut system, and that there was no response to nitrogen

above their applied rate of 67 kg N ha", indicative of a yield plateau.

Much of the previous work used quadratic polynomials or quadratic

response models to model the yield response of switchgrass to nitrogen. A

44

recent paper by Haque et al. (2009) and the associated work by
(Aravindhakshan, 2008) examined the nitrogen response of four perennial grass
species that are bioenergy candidates. The researchers found that the linear
response plateau (LRP) model was the most appropriate model for switchgrass
in their case, based on the likelihood dominance criterion of Pollack and Whales
(1999), being 12 times more likely than the linear model and five times more

likely than the quadratic model (Aravindhakshan et al. 2008).

Information is needed on growing and harvesting switchgrass (Panicum
virgatum L.) as a bioenergy feedstock in the Central Great Lakes region. In
order for rapid scale-up of commercialized cellulosic ethanol production, large
quantities of biomass feedstock will be required. Farmers require agro-ecological
Specific information for economic production of switchgrass. A combination of
low N inputs and a single harvest taken at the end of the season may be an
Optimal and sustainable strategy for managing switchgrass for biomass for

bioenergy production.

Objectives

The objectives of this study were to determine the optimum nitrogen rate
and harvest timing and frequency to maximize switchgrass yield on a dry matter,
per hectare basis in the Great Lakes Region. Three harvest management
systems will be compared: one-cut fall, one-cut overwintered, and a two-cut
System in which a cut is made in July (two-cut summer) and then re-harvested
after a killing frost in the fall (two-cut fall). Further analysis was conducted on the

components of the two-cut system alone over the two growing seasons. It was

45

tn

hypothesized that yield would increase with increasing N until a plateau was
reached due to moisture, fertility or other limitations and that N-fertilizer rate
would not significantly impact crop quality. Harvest timing is hypothesized to
significantly impact yield. Regrowth in the two-cut system was not anticipated to
contribute enough yield to justify the two-cut system due to our Short growing

season .

MATERIALS AND METHODS

Site Description and Experimental Design

Field experiments were conducted on the Michigan State
University Agronomy Farm in East Lansing (42°25'48" N, 84°16'48" W), Ml,
over a three-year period beginning in the fall of 2007. Soils at East Lansing are
classified as a mixture of Aubbeenaubbee-Capac sandy loams (fine-loamy,
mixed, active, mesic Aeric Epiaqualfs) and Colwood-Brookston loams (fine-
loamy, mixed, active, mesic Typic Endoaquolls and Typic Haplaquolls). The
upland switchgrass cultivar ‘Cave-in-Rock’ was seeded at a rate of 9 kg ha‘1 on
May 24, 2007, into 4.6m X 12.2 m plots. The previous crop was soybean
(Glycine max (L.) Merr.) Switchgrass was seeded into relatively weed-free plots
that were chisel-plowed in the fall of 2006 and cultivated again in the spring of
2007. The switchgrass was seeded using a dual roller seeder to a depth of
0.50m. A double roller Seeder (Brillion, Brillion Iron Works, Wl) was used to
plant the seeds to a ~1.25 cm depth, while simultaneously acting as a

cultipacker, thus firming the seedbed. Plots were fertilized in the spring at

46

green-up in the years after establishment year with (0, 38, 79, 157 kg N ha")
applied as granular urea (46-0-0 N-P-K).

Weeds were controlled with a tank-mix of S-metalochlor (2-chloro-N-(2-
ethyl-6-methylphenyl)-N-[(1S)-2-methoxy-1-methylethyl] acetamide) and
atrazine (2-chloro-4-(ethylamino)-6-(isopropylamino)-s-triazine) at rates of 1.3
and 1.8 kg a.i. ha", respectively“

The experimental design was a randomized complete block design with
five replications. Soil testing was done in April 2007 to measure pH, P, and K
levels and re-tested again in April of 2008 and 2009 more intensively to measure
soil nitrate levels. The field has two different soil types and consequently two
distinct pH ranges; the eastern two thirds of the field has a mean pH of 6.4 while
the sandier, western third of the field had a mean pH of 5.6. The entire field has
a lime index of 68-71. Initial P and K levels were within a close range but lower

than desirable for initial nutrient levels of these two important macronutrients.

Biomass Measurements

The various harvest management systems were imposed in 2008 (Table
4), the first year of production after establishment. Cuttings were made with a
commercial forage harvester (Model NO. 7650 Hesston Co., Hesston, Kansas),
set at an 18cm cutting height that was modified for plot research with the
addition of a weigh bin for yield determination from individual plots. Total fresh
weight and biomass content for each plot was recorded and biomass moisture
content for each plot was determined via subsamples that were placed in a

forced air oven at 100°C for 72 hr or until a consistent dry weight was achieved.

47

The total dry matter (DM) per plot was then converted to Mg DM ha".

Table 4. Harvest Management Systems employed during the study (2008-2010)
with associated harvest dates and plant developmental stages from Moore et al.
(1991)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Harvest Harvest Date Plant Developmental Stage
Treatment (Moore et al. 1991)
2008-2009 2009-2010 2008-2009 2009-2010
One-Cut Harvest Managgnent Systems
One-Cut Fall Nov. 5 Oct. 26 Post Post
Senescent Senescent
One-Cut March March 19 Post Post
Overvvintered 22 Senescent Senescent
Two-Cut Harvest management Systems
Two-Cut July 21 July 17 3.7 (Anther 3.0 (Boot
Summer emergence/and stage) -3.1
or anthesis) (Inflorescence
emergence)
Two-Cut Fall Nov. 5 Oct. 26 Post Post
Senescent Senescent

 

 

 

 

 

 

 

Statistical Analysis

Data were analyzed using PROC MIXED in SAS v9.1 (2003 SAS Institute
Inc., Cary, NV.) Analysis of variance (ANOVA) was conducted to measure the
treatment effects on biomass yield. Year, nitrogen rate and harvest timing were
treated as fixed effects while blocks were considered to be random effects.
Repeated measures analysis was not employed at this time because there have
not been enough yields taken over time to warrant its use. Fisher's protected
least significant difference (LSD) multiple comparison procedure was used for
mean separation when ANOVA was significant (Saxton, 1998). Single degree of
freedom contrast statements were used in PROC MIXED to assess the effects
of harvest timing on yield, how the harvest timings responded in each year, and

specifically where yield plateaus occurred within a given harvest timing.

48

Yield response to nitrogen was modeled using the NLMIXED procedure
to fit three functional forms: quadratic polynomial (QP), linear (L), and linear
response plateau (LRP). N fertilizer was listed as a continuous variable,
whereas harvest timing was considered a categorical variable and listed in the
class statement. Model appropriateness was based on comparisons of the -2
log likelihood value and Akaike's information criterion (AIC). The log likelihood
value was used to compute coefficients of determination (R2) values using the
method from (Nagelkerke, 1991). In the LRP model, the plateau of the
regression estimates the maximum yield, and the join point of the regression
estimates the N fertilizer rate at which the maximum biomass yield was
achieved. In the quadratic model, the level of N that Optimizes yield was
calculated by taking the derivative of the Slope and setting it to zero and solving
for the independent variable (N) (Bowley, 2008). Response fit was visually
inspected against observed yield to further ensure the best model was chosen.

Table 5. Three functional forms: linear (L), Quadratic Polynomial (QP), and
Linear Response Model (LRP), used to model yield response to N.

 

 

 

 

Linear Model Y= mX + b

Quadratic Model Y = a + bX — cXY

Linear Response Model Y = a + bX if X < C
Y: P if X2 C

 

 

Where Y is the yield of biomass (Mg ha“) and X is the rate of N
application (kg ha"), b is the intercept, m, is the linear coefficient and C is the

critical rate of N which occurs at the intersection of the linear response and the

49

 

occurrence of the plateau (Cerrato and Blackmer 1990). In the LRP model, P is
the join point or the N fertilizer rate at which the plateau begins, and a, b, c, and
plateau are model coefficients. Coefficients of determination (R2) values were

calculated as follows:

R2 = 1 — [1.(0) -:— L(6)]2/"

Where 0 and 6 represent the log likelihood values for the full and the null

(intercept only) model, respectively.

Normality of the residuals and homogeneity of variances were evaluated
by examining normal probability plots and box plots. When variances were found

to be heterogeneous, the REPEATED/GROUP option of PROC MIXED was

implemented. Results were reported as statistically Significant at a = 0.05.

RESULTS AND DISCUSSION

Climatological Summary

Mean air temperatures did not vary considerably between years or
against the 30-yr average (Table 6). The median date for first frost (28°) at East
Lansing was 18 October with temperatures falling to (28°) by November 4, 90%

of the time over the last 30 years.

50

Table 6. Mean monthly rainfall (mm) and temperatures (°C) during the
experiment in comparison to the 30-year mean (1971—2000)

 

 

 

 

 

 

 

 

 

 

 

 

 

Mean Monthly Rainfall Mean Monthly Temperature

Month 2008 2009 30 yr-avg. 2008 2009 30 yr-avg.
April 4.8 10.4 8.3 10.3 8.4 7.6
May 3.7 15.0 6.9 12.8 14.0 14.1
June 9.2 10.7 8.1 20.0 18.9 19.2
July 9.5 5.9 7.6 21.4 18.8 21.4
Aggist 1.5 8.7 8.7 20.5 20.1 20.3
September 15.7 3.5 9.2 17.2 16.8 16.0
April- 7.4 9.0 8.1 17.0 16.2 16.4
September

 

 

 

 

 

 

The 2008 growing season tended to be drier than both the 30-yr average
and 2009 April and May were 41% and 46% drier than the 30-year mean and
53% and 75% drier than 2009, respectively (Table 6). August 2008 was also a
very dry month, receiving approximately 82% less precipitation than 2009 and
the 30-year average, which had negative consequences for regrowth in the two-
cut system (Table 6). Precipitation from April through the month of June in 2009
was wetter than average, and in August 2009 precipitation was similar to the 30-
year mean and hence more conducive for regrowth of switchgrass under the
two-cut system.

Three Harvest Management Scenarios Compared

A three-way interaction was not present between year, N and harvest
timing, nor was there a two-way interaction between year and nitrogen in the
ANOVA results. Harvest timing interacted with nitrogen (p=0.0161) and year
(p=0.0001). The fixed effects: harvest timing (p=<0.0001), N (p=0.0003), and

year (p=<0.001) were Significant. Overall yields in 2009 were significantly higher

51

than in 2008 (p=<0.0001). Yields across all N rates and harvest systems ranged
from 4.0-10.25 Mg ha-1 in 2008, and 9.0-12.25 Mg ha-1 in 2009. The stand was
at two-thirds and possibly full yield potential in 2008 and 2009, respectively,
which mirrors the concept put forth by McLaughlin and Kszos, 2005) in their
summary of the 10-year assessment of switchgrass in the ORNL BFDP
program, and (Madakadze et al., 1998) in the Short growing season of eastern
Ontario.
Year By Harvest Interaction

Yields in the second production year (2009) were higher due to either
precipitation or the fact that the stand in 2008 was not at full-yield potential. In
assessing the effect of harvest timing on yield in 2008, we found that the two-cut
system out-perfon'ned the one-cut system (p=-0.001), however the combined
system yielded less than the one-cut fall harvest system in 2009 (Fig. 2). In both
years the one-cut fall yields were significantly higher than one-cut overwintered
yields (p=0.001). This is in agreement with Adler et al. (2006), who found that
yields generally decreased by more than 20% when harvest was delayed from
fall to spring. The researchers estimated that leaf and panicle weights and tiller
weights (-7%) decreased, while stem weights increased. Mechanical harvest
losses were also incurred during their research, with harvest equipment not
picking up 21% and 45% of the harvested biomass in the fall and following
spring, respectively. Our use of a commercial forage harvester with a direct -cut
header was efficient at collecting harvestable biomass. Yield losses on a plot
scale varied between (51 - s 9%) in the spring of 2009 and 2010 (data not

shown).
52

Nitrogen By Harvest Interaction

There were no significant differences among N rates in either the one-cut
fall or one-cut overwintered system (Table 7). This was not due to a lack of
response to nitrogen, but rather a loss of biomass over time due to translocation
of carbohydrates and nutrients, leaf and seed drop, and further physical losses
due to environmental conditions (e.g. peduncles being lost to wind). The results
of the combined yield from the two-cut system demonstrated different responses
to N when tested with linear contrast statements. A significant difference was
found between the 39 and 78 kg N ha'1 rate (p=0.0351), but not between 78 and

157 kg N ha", indicative of a yield plateau.

Table 7. Switchgrass dry biomass yields (Mg ha") in response to three different
harvest management systems, and four nitrogen rates in 2008-2010

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Nitrogen Fertilizer Application Rate
kg N ha”
0 39 78 157
Harvest Management 2008
System Mg ha'1
One-Cut Fall 6.31 6.77 7.18 6.99
One-Cut 4.06 4.34 3.88 3.64
Overwintered
Two-Cut Combined 8.34" 10.52 10.91 1 1.24
(Summer + Fall)
2009
Mgha‘1
One-Cut Fall 11.82 12.02 12.36 12.80"
One-Cut 8.56 8.95 9.31 9.34“
Overwintered
Two-Cut Combined 9.54" 10.30 12.20 12.28
Summer + Fall)
** Yields differed from other means within the same row at 0 =0.05
LSD values for N*Harvest is 0.91

 

 

In 2008, one-cut overwintered yields did not respond positively to applied
nitrogen fertilizer (Fig. 9). The one-cut fall harvest management system was best
modeled with a quadratic polynomial that reaches a plateau at 7.07 Mg ha'1 with
a corresponding N rate of 60 kg N ha". The log likelihood values for the two-cut
combined system were 83.2 and 83.5 for the LRP and QP functions,
respectively, and corresponding AIC values were similar (Table 13). The
outcomes for critical agronomic N values and plateaus varied. The LRP model
predicts a plateau in yield at 11.07 Mg ha'1, while the QP predicts a plateau at
11.56 Mg ha". The former occurs at 111 kg N ha", and the QP model predicts

the plateau to occur at 106.9 kg N ha".

 

 

 

 

 

 

 

, ° 43-- One-Cut
A 12 “7 ____-____- Overwintered
0-45.8. .. - - - - ---e
at 10 “ ...,a’f? I -X—One-Cut Fall
a ;-r
'3 8
O \ X X
S 26" A - 6- Two-Cut
a 6 (Summer+Fall)
N P
E Q
.2 4 [19.3.8.2 3.9-2.1.!!!SAQLII II can... ...-0B0. o row I “0‘63" 51.”.- -;-._;.,_.._._; ... ;... ;.............. U.
m ' -' - Two-Cut
E" 2 (Summer +
0 Fall) LRP
0 -.-_,-....-_._-.-_--...-.--._-._.--_T_.............-..h-..-.-......-\_..T ---.-_.--. “.-.-.T... .-....-_.-c....--...._..-__.1
0 40 80 120 160
Nitrogen Fertilizer Rate kg N ha'1

 

 

Figure 8. Switchgrass dry biomass yield (Mg ha") response to nitrogen fertilizer
2008-2009. Corresponding regression equations and parameters can be found
in Table 8.

54

 

 

A42 x 880.3

US$53.90

 

 

 

 

 

 

 

 

 

 

 

mod on 94 a: do + Az x 358.8 + m3 u> 50.26
m2 x 5833
one «.8 «.8 a: do + 2 x 80.8 + 9% u >
sen.
... BEE—.3
and new new we? 65 me. x Bee + m; u> 30.65
A z x BBB
and EN in new do .2 x 2o. 0 + 88 u > =6". 26.25
28:68 E ..2 z 9. spasm
0005.95 .EeEe.._:uem Eon. EeEommcos.
Nm me. u. 6.4. peaszem .2285". .86: amaze...

 

 

meow c_ mew. comet:
So.— ucm .mEmugm EoEommcmE «mote; EoLmEu 095 2 mmcoamm: c_ 3.03 39:05 mmm..mco=3m .w cam...

55

 

 

 

 

 

 

, Z; , _____ >6 "*3“ One—Cut
12 W ‘;-.;-:av‘mw--‘ ‘ Overwintered
‘3 c5 +One-Cut Fall
D: l . ......... El .......... El ........................ E]
2 til
I 8 {.-_ ,_,__-,_ _ _ __
3 '9‘ Two-Cut
g 6 - ----- (Summer+Fall)
is
a 4
e
!
i
0 40 80 120 160
Nitrogen Fertilizer Rate kg N ha:1 _ _

 

 

Figure 9. Switchgrass dry biomass yield (Mg ha") response to Nitrogen in 2009.
Regression equations and parameters can be found in Table 9.

In 2009, one-cut overwintered switchgrass yields did notvary with applied
nitrogen, yield reached a plateau at 9.1 Mg ha“. Both the one-cut fall system
and the combined two-cut system were best modeled with the LRP functional
form with a plateau occurring at 12.80 Mg ha'1 and 12.28 Mg ha", respectively.
These yields would be obtained with a nitrogen rate of 134 kg N ha'1 for the fall-

one cut system, and 123 kg N ha‘1 for the latter system.

56

Table 9. Switchgrass biomass yields (Mg ha") in response to three different
harvest management systems, and four nitrogen rates in 2009.

 

 

 

 

Harvest Model Estimated N AIC -2 log R2
Management Requirement, Likelihood

System kg N ha" Fit

Statistic
One-Cut Fall Y = 11.84 + 134 58.4 48.4 0.16
(0.0069 X N)

Two-Cut Y= 9.83 + 123 79.0 69.0 0.51
(Summer + (0.020 X N)
Fall)
One-Cut Y = 8.71 + 72.5 54.7 0.22
Overwintered (0.0053 X N )

 

 

 

 

 

 

 

 

Break Down of the Two-Cut Harvest Management System

The statistical analysis of the two-cut system did not result in a three-way
interaction between N, yield and harvest, nor were there tvvo-way interactions
between year and harvest, or year and nitrogen. There was, however, a
significant interaction between harvest timing and nitrogen (p=0.0024). Fixed
effects, nitrogen rate and harvest timing were significant (p=<0.0001), and yields
were higher in 2009 than in 2008 (p=0.0352). Despite the lack of Year X H and
Year X N interactions, data are reported and discussed for each year individually
to improve our understanding of what occurred. This was done because 2008
was the first production year and thus the stand was not at full yield potential
and there may have been a decreased response to applied N fertilizer due to the

residual N in the system from the previous soybean crop in 2006.

57

Table 10. Dry switchgrass biomass yields (Mg ha") responses within the two-cut
system at four different rates of nitrogen fertilizer in 2008 and 2009.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Nitrogen Fertilizer Application Rate
kg N ha"
0 39 78 157
Harvest Management 2008
System Mg ha"
Two-Cut Summer 76“ 9.5 9.9 10.2
Two-Cut Fall 0.7 1.0 1.0 1.0
Two-Cut Combined 8.3“ 10.5 10.9 11.2
(Summer + Fall)
2009
Mg ha’1
Two-Cut Summer 8.2" 9.0 10.4 10.4
Two-Cut Fall 1.3 1.3 1.5 1.9"
Two-Cut Combined 9.5“ 10.3 12.2 12.3
(Summer + Fall)
** Yields differed from other means within the same row at a=0.05
LSD values for N*H is 1.14

 

 

Harvest

Two-cut summer and the combined yields (summer and fall) from the two-
out system did not differ between 2008 and 2009. The regrowth in the two-cut
system was higher in 2009 than in 2008 (p=<0.0001) because there was less
precipitation during the latter half of the 2008 growing season (Table 1.) and due
to the timing of the first-cut. The first cut made in the two-cut system varied with
calendar date and plant growth stage and so yield differences cannot be directly
attributed to year effect. In 2008 switchgrass was harvested during anthesis
(Moore et al. 1991 staging 3.0-3.3), which meant there was less of a growing
season (Shorter photoperiod) to accumulate yield prior to senescence that was
further confounded by low, late-summer precipitation (Table 1). In 2009,

switchgrass was harvested at the late-boot stage, which meant the crop likely

58

had higher carbohydrate and nutrient reserves available and there was also
ample precipitation to facilitate the regrowth (Table 1).

Harvesting the switchgrass earlier in the season in 2009 and obtaining
more regrowth resulted in the combined and one-cut system having smaller yield
gaps than in 2008. The one-cut fall system resulted in higher yields than the two-
cut combined system at the 39 kg ha'1 rate, and had comparable yields at 79
and 157 kg N ha". The stand was in its second production year, with more than
adequate precipitation for optimal growth, and the two-cut system accumulated a
large amount of biomass prior to the first-cut and had appreciable regrowth
yields (1.57 Mg ha“), 39.02% more than regrowth yields in 2008.

In 2009, the two-cut system did not perform better than the one-cut
system, this is in agreement with findings by Casler and Boe (2003), Monti et al.
(2008) and elsewhere reported in with Sanderson et al. (1999), and in the review
by McLaughlin and Kszos (2005). Haque et al. (2009) found that switchgrass did
not respond well to a harvest in July followed by a harvest in October. Our two-
cut system is not directly comparable to (Haque et al. 2009) because we applied
the full rate of N at the beginning of the season, with no additional input after the
first harvest. Hence our yield for the second cut was a function of residual soil

nitrate and not from an application of additional fertilizer.

59

Nitrogen
In 2008 there was no difference in regrowth yields of the two-cut system
between the 39, 78 and 157 kg N ha'1 rates, and in 2009, there was no

significant yield response to N fertilizer applied at the beginning of the season.

Single-degree of freedom contrasts applied to the pooled data from 2008
and 2009 show that yields from the two-cut summer treatment did increase with
increased nitrogen, but there was no difference in yield between the 39 and 78
kg N ha‘1 rates, indicative of a yield plateau. For yields from the second cut
made in the two-cut system, a response was noted between the 0 and 157 kg N
ha'1 rates, with a trend (p=0.0580) noted for the difference in yield between the 0
and 78 kg N ha‘1 rate. Total yields from the two-cut system were higher with
applications of 78 and 157 kg N ha", but not 39 kg N ha“, and there were no

differences in yield between either the 39 and 78, or 78 and 157 kg N ha'1 rates.

60

 

.— -' "" "I‘ - ~ — 'Two-Cut Total j
. — C 0 do
9 a ; . ~0’ ° 9 (Summer+

N Fall) LRP

"GI" Two-Cut Fall

 

 

i
i
l
5

Dry Biomass (Mg ha")
0)

l —)(—Two-Cut ‘
2 . — .-. - - -- -€- Two-CutTotaI

E ooooooooooo D oooooooooooo D 0000000000000000000000000 U (summer+

ctr Fall) ‘
o 1_ """""‘"‘“'“"T"““" - --~-——~——-r-~------ ‘ 'I ' ' "" “—1

0 4O 80 120 160

 

Nitrogen Fertilizer Rate kg N ha'1

 

 

Figure 10. Switchgrass biomass yield response to nitrogen under the two-cut
harvest management system in 2008. Corresponding regression equations can
be found in Table 11.

Table 11. Regression equations and parameters for the two-cut harvest system
in 2008.

 

 

 

 

Harvest Model Estimated N AIC -2 log R2
System Required Likelihood
kg N ha" Fit Statistic

Two-Cut Y= 7.72+ 148 69.7 59.7 0.28
Summer (0.051 X N)+

( -0.00024 X

N2)
Two-Cut Y= 0.84 + 41 17.6 7.9 0.23
Fall 0.002 X N)
(Re-
growth) Plateau =

11.07 Mg ha"
Two-Cut Y = 8.47+ 112 83.5 93.5 0.30
(Summer (0.057 X N)+
«I Fall) (-0.000027 X

N2)

 

 

 

 

 

 

 

 

61

In 2008, the two-cut summer treatment was best modeled by a quadratic
polynomial, and the regrowth yields from that system was best modeled by a
LRP function, although the regression parameters for the second-cut were very
small because there were no significant differences between N rates (Fig. 11).
The summer cut of the two-cut system reached a plateau at 10.03 Mg ha", '
obtained with an N rate of 148 kg N ha". The log likelihood values from the
combined yields from the two-cut system were very similar for both the QP and
LRP functional forms as previously discussed. The QP model predicted the N
rate that maximizes biomass to be 112 kg N ha'1 at a yield of 11.09 Mg ha",
whereas the LRP model predicts an agronomic optimal N rate of 109 kg N ha'1

at a join point yield of11.07 Mg ha“ (Fig 11).

62

 

 

 

 

 

 

12 i .- ., WW.-. “_...-. O . L - ,a!’"---.-..'G
it: i ..-—1”" "El" Two-Cut
-: i X Fall
2 f
'o .
E: 8 X- ~- -
5. l +Two-Cut
3] Summer
in 6 f m - -
E
.9 i
In 4
a. ‘9' Two-Cut
D j Total
2 “ :0" ........ ';.i.__‘,;',“,';;""t:l (Summer +
l? ........... 13 Fall)
0 I 7' '— I'” "I" ”'TW "I
0 40 80 120 160
Nitrogen Fertilizer Rate kg N ha'1

 

Figure 11. Switchgrass biomass yield response to nitrogen under the two-cut
harvest management system in 2009. Corresponding regression equations are

found in Table 12.

In 2009, yield responses to nitrogen were best modeled with the LRP

functional form. The critical value at which the linear portion joins the plateau

region of the non-linear response was 120 kg N ha'1 for the combined system,

where yield was maximized at 12.28 Mg ha". The yield from the summer cut

reached a plateau at 10.41 Mg ha“, and required 118 kg N ha“. The response

and consequent N requirement for regrowth was neg

harvest.

63

ligible for the two-cut fall

 

 

 

 

 

 

 

 

 

 

 

...eu +

E... «.8 «.8 rm... 2 9. as .2589 + 8s u > 555$ 50.65
2:65

B... e... e... 6.: 2.9.8.9 +e«.. u> .65 _.en. 5065
B... we ewe re; 2 9. m: .2855. + new u> .6555 50.23
6.8.55 P...... z 9. 526.6

«E 305.3...— JceEogaaom EoEommcas.

km me. «. o.< z 8838 .28: .826...

 

.mcow .._ GEE comet:
58 new @8996 EoEmmmcmE «meta: EEOEU we... 9 omcoamoc c. 3.03 89:05 mmflmzogsm .9 632.

64

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.3. 0.00 000 0.00 0.0.4 .00 0005.9... 003.

FN.0 l m 0F.0 I w. 0.0.0 I m 00.0 I m. 00.0 i m 00.0 i w. .000

l« i« i« l« i« I« .0.Eo:>_0n.

. . . . . 0.2 600.0030
«2 ...0000 0 «2 F000 0 «2 F F000 0 «2 00000 0 «2 0500 0 0N000 0
- 20F0.0+ - 2080+ - 2080+ - ZNF000.0+ -2F00.0+ - 2000.0+
00.0 u> FF.FF u> 34.0 u> 00.0 u> 00.0 n> ..F0 u>

. . . 0 : . . oo _ 0 _ mo

0 3. F 00 0 F0 \ N 0.0 ... 00 0 F... e... .N

00.0 n«m 0F.0 ”N”. F00 "mm 0000 "N”. 00.0 "NW. F00 0%. Ex... 3090....

2050.0 09.0000”.

.2: S... 208... 28.0 .268... -0 + .08...
- 00.0 n> - 00.NFn> l 0N.NFu> vm...u> -K.0Fn> 000. F Fn>

0.0.4 0.00 0.00 0.00 0.00 0.00 0005.9... 00.0

0F.0 u «m 0F.0 n «2 F00 u «m F00 n «m 00.0 n «m Fwd u «m. A... .00:.._

2300.0 2FF0.0 2000.0 22.000 20000. - 20F0.0 .

+FR.0u> + 00. F Fn> +00.0u> + 00.0u> 0+N000.0Fn> + 00.0n>

05.00 .000h 05.00 .20... :00030m
300:0 =0". 500:0 50-0.: 500:0 =0". 500:0 50.02....

0000 000a

 

 

 

 

.88 0:0 08« a. peace
.:0E000:0E .0030: 0005:0530 m:0..0> 0.... .0. 0:000:00 Em... 300.03 00:800. .00:._ 0:0 ..00:._ 6.00.0000
.0. 0.0.0:.000 5.00900. .0 .«mv :o_.0:.E.0.00 .o 3:00.008 0:0 00.0...000 0005.9... E:E.x0.>. .0F 0.00...

65

CONCLUSION

Data from the formative years of this switchgrass stand indicate that yield
response to nitrogen is best modeled with a linear response plateau. Quadratic
and LRP functional forms were very similar in some instances and had minor
implications for the predicted optimal N within the range tested. Yields across all
N rates and harvest systems ranged from 401025 Mg ha-1 in 2008, and 9.0- .
(12.25 Mg ha-1 in 2009, which is similar to reported yields in the mid- west, and
north- east regions of the US. The efficacy of deriving agronomic and optimal N
rates for switchgrass biomass production from either an LRP or QP model needs
to be ascertained beyond the formative years and across all years of a mature
stand. In 2009, the fall one-cut system outperformed the two-cut system.
However, data representative of a mature stand is required before we can
conclude that a Single cut-system is superior to the two-cut system in the Great
Lakes Region. Delaying harvest until two weeks after a killing frost or the
following spring appears to make any gains in yield from nitrogen fertilizer
application negligible during the early years of the stand. Further research is
needed on how nitrogen contributes to increased biomass yield and why those
gains diminish with delaying harvest. In both years there were sizeable yield
losses when switchgrass was overwintered, (41.5% and 26.3% in 2008 and
2009, respectively). It does not appear to be an ideal harvest management

system for the bioconversion platform.

66

CHAPTER 3

CRYSTALLINE CELLULOSE, LIGNIN CONTENT AND MATRIX
HEMICELLULOSE POLYSACCHARIDE COMPOSITION OF SWITCHGRASS
UNDER VARYING NITROGEN RATES AND HARVEST MANAGEMENT
SYSTEMS

Abstract: Understanding the chemical composition of switchgrass is an important
issue for future utilization of switchgrass for biofuel production by- and 00- product
generation. The objective of this research was to determine the effects of nitrogen
fertilizer and harvest timing on switchgrass biomass composition. Four different
nitrogen rates (0, 38, 79, and 157 kg ha") were applied to the plots that were then
harvested either under a two-cut system (anthesis and post-senescence), late fall or
left to overwinter and then harvested in the spring. Crystalline cellulose content
increased with maturity but did not increase with increased nitrogen fertilizer. Acetyl
bromide soluble lignin content did not differ between the fall and spring, but was
higher with 79 kg and 157 kg N ha'1 compared to the two lower application rates of N.
Hemicellulose monosaccharide composition varied with harvest timing, and mannose,
rhamnose and galactose concentrations were significantly reduced at the 79 and 157

kg N ha‘1 rate of fertilizer.

Key Words: lignocellulosic, ethanol, cellulose, hemicellulose, lignin, xylose, harvest,

nitrogen

67

INTRODUCTION

Cellulosic ethanol is being pursued as a domestically produced,
renewable liquid transportation fuel in the United States as part of a mandate
stating that 21 billion gallons of renewable, advanced fuels must be produced
annually by 2022. More specifically, the 2005 Energy Act mandates that 250
million gallons of ethanol must be produced from cellulose materials by 2012.
The positive attributes of domestic production of cellulosic ethanol from perennial

grasses has been discussed in previous chapters.

Understanding the chemical compositions of switchgrass is an important
issue for future utilization of switchgrass for biofuel production (Hu et al., 2010).
With a thorough understanding of the composition of feedstocks, we are better
equipped to maximize cellulosic ethanol production from various feedstocks, as
well as derive co- and by-products. Utilizing all major components in the plant
improves the cost effectiveness of the production system (Cherney et al., 1988;
Pauly and Keegstra, 2008). Feedstock quality is the primary determining factor in
conversion efficiency. Increasing cell wall carbohydrate (principally as cellulose)
content in biomass can be expected to increase the conversion efficiency and
yield of liquid fuel (Weimer et al., 2005; Dien et al., 2006, (Sarath et al., 2008b).

The processing steps for depolymerization by both chemical or enzymatic
processes and the subsequent fermentation of the various sugars will need to be
optimized to accommodate different feedstocks. Within a given species,
information on the ease of digestibility due to lignin content and composition,

crystalline cellulose content, and the composition of the hemicelluloses and other

68

matrix polysaccharides will also be important for feedstock processing
optimization (Foster et al, 2010) It follows that information regarding specific
feedstock characteristics form feedstock grown under varying levels of nitrogen
fertilizer and harvest frequencies/timing will be needed for planning harvest and

delivery logistics.

.lf Bioconversion Facilities (BCF) desire specific quality parameters for
improved conversion efficiency, growers could be compensated accordingly (e.g.
increased cellulose, low lignin content). Economic incentives will have to be
provided for producers who choose to grow cultivars with quality characteristics
that have been improved through genetic manipulation (Vogel and Jung, 2001).
Vogel and Jung (2000) speculated on a scenario in which future bioconversion
facilities would probe and collect feedstock samples at delivery and be able to
evaluate feedstock quality value in s 30 min. In the absence of improved cultivars
for conversion to cellulosic ethanol, agronomic management serves as the

primary method of increasing yield and optimizing quality traits.

Biomass Cell Wall Constituents

Lignocellulosic biomass is a mixture of carbohydrate polymers (cellulose,
hemicellulose and pectin in varying ratios) and the non-carbohydrate polymer
lignin (Doran-Peterson et al., 2008). The breakdown of crystalline cellulose yields
glucose, while hemicellulose yields mostly pentoses (xylose) and some hexoses
(Sarath et al., 20083). Grass cell walls consist of a network of cellulose fibers

surrounded by a matrix of non-cellulosic polysaccharides (Vogel 2008).

69

Switchgrass, being a monocot and a member of the Poales family, has a
Type II primary cell wall. It is comprised of cellulose fibers encased in
glucuronoarabinoxylans (GAX), high levels of hydroxycinnamates, quantities of
mixed linkage glucans to varying degrees, and levels of pectin and structural
proteins (Harris and Smith, 2006). Once the cell walls have ceased elongation,
secondary cell wall deposition initiates and cell walls lignify, replacing water with
lignin for improved protection from pathogens, viruses and mechanical strength.
Secondary cell walls are largely composed of cellulose, GAX, and lignin (~20%)
and are prominent features of xylem, fibers and sclerenchyma (Vogel, 2008).
They account for the dominant fractions of lignin, hemicellulose and cellulose in
the biomass of mature plants. Tables 14-17 provide information on switchgrass

chemical composition from the literature.

Table 14. Selected lignocellulosic biomass compositions (% dry weight) from
Wyman (1996) and Li et al. (2001)

 

 

 

 

 

 

 

 

 

 

Cell Wall Component Wyman (1996) Li et al., (2001)
% D.M.

Cellulose 32.0 37.3
Hemicellulose 25.2 28.5
Xylan SC 21.1 22.8
Arabinan 5C I 2.8 3.1
Mannan 60 0.3 0.3
Galactan 60 1.0 1.4
Lignin 18.1 19.1

 

 

 

 

 

70

 

Table 15. Range of Compositions of whole plant ‘Cave-in-Rock’ and leaves and
stems across numerous varieties as a percentage of dry weight from Wyman
(1996)

 

 

 

 

 

Feed- Glucan Xylan Galactan Arabinan Lignin Mannan

stock

Whole- 31 .4-35.0 20.2- 0.9-1.6 2.7-2.9 11.4- 02-05

plant 23.8 16.5

Stems 33.5-37.5 21 .8- 0.6-1.2 2.2-2.7 13.6- 02-04
25.0 17.2

Leaves 26.8-30.9 17.6- 1.2-2.6 3.2-3.7 5.5- 0.206
21.8 15.4

 

 

 

 

 

 

 

 

Table 16. Chemical compositions of four populations of switchgrass based on
(%) of oven dry weight from Hu et al. 2010 (%) of oven dry weight from Hu et al.
2010

 

Variety/Populations Arabinose Galactose Glucose Xylose Lignin
(Whole plant)

 

 

 

Alamo 3.8 1.2 38.8 23.1 21.2
Kanlow 3.4 1 .3 37.4 23.1 22.6
GA993 3.7 1.3 37.8 23.1 22.4

 

 

 

 

 

 

 

 

In a study by Sarath et al. (2007) that looked at lignification of successive
internodes in switchgrass, significant secondary cell wall deposition was noted in
the cortical fibers, the fiber sheaths surrounding the vascular bundles, and in the
cortical parenchyma. Grabber et al. (1991) also showed increased amounts of

lignin in older parenchyma cells of switchgrass.

71

Table 17. Comparison of average chemical compositions between three
morphological portions of switchgrass and other published results from Hu et al.
(2010)

 

Ara Gal Gluc Xyl Lignin Reference

 

3.7 1.3 38.0 22.8 22.1 Hu et al.2010

 

WhP'e-P'am 3.2 1.1 34.3 20.9 17.5* Agblevoretal.1994
Swrtchgrass

 

3.6 2.1 34.8 23.4 21.4* Thammasouk etal.
1997

 

 

 

 

 

 

 

 

 

*Klason lignin content

Cellulose

Cellulose is the most abundant organic polymer on earth, and an important
trait for biofuels being used to produce ethanol via fermentation (Lemus et al.,
2002). Cellulose is made up of long microfibrils containing repeating units of
beta-linked (8 1,4) dimmers of glucose molecules known as cellobiose (Vogel
and Jung 200, Doran-Peterson et al. 2008). Cellulose from 1 kg of dry straw
yields 0.111 kg of ethanol, it has an enzymatic conversion efficiency of 0.76 and
a fermentation efficiency of 0.75 (Badger, 2002). The susceptibility of cellulose to
hydrolysis is restricted due to lignin matrix and hemicellulose protection
surrounding these cellulose microfibrils (Alizadeh et al., 2005).

Hemicellulose

Hemicellulose is the second most renewable biomass polymer next to
cellulose (Saha, 2003). Hemicellulose is a branched polysaccharide composed of
both 6-carbon and 5-carbon sugars. Hemicelluloses are heterogeneous polymers
of pentoses (xylose, arabinose), hexoses (mannose, glucose, galactose), and

sugar acids (Saha, 2003). It can be used in the production of xylitol, 2,3-

72

 

butanediol, and other value-added fermentation products in addition to fuel
ethanol (Saha, 2003). Although more susceptible to enzymatic hydrolysis than
lignin, it is not as readily fermented by the industrial yeast Saccharomyces
cerevisiae. Further research is needed to identify microorganisms that can
ferment sugars derived from hemicellulose in an economic and efficient manner
(Saha, 2003). Improvements in hemicellulosic conversion will improve the

economic efficiency of the conversion process (Pauly and Keegstra, 2008).

Lignin

Lignin is a biologically resistant, non—carbohydrate polymer that surrounds
cellulose and hemicelluloses. Lignin improves water conductance through the
xylem, provides mechanical strength to the plant and prevents the infiltration and
spread of pathogens in cell walls (Iiyama et al., 1994). As plants mature, the
efficiency of enzymatic degradability of cell walls is decreased due to the
accumulation and progressive lignification of vascular and sclerenchyma tissue
cell walls (Grabber, 2005). The primary cell walls of parenchyma and epidermal
tissues undergo Iignification in stem, and to a lesser extent, leaf cell walls (Wilson
and Hatfield, 1997). Lignin content negatively affects hydrolysis of structural
carbohydrates because it restricts access to enzymes, which can limit the

conversion of plant material into liquid fuels (Cherney et al. 1988).

Bioconversion of biomass to ethanol is hindered by the structure and
content of lignin in the cell wall. Many researchers have proposed modifying
lignin biosynthetic pathways to either reduce the recalcitrance of the lignin or

reduce its quantity. Hu et al. (1999) found that repressing lignin promotes

73

cellulose. However, this has implications for plant defense mechanisms, vascular

integrity, plant fitness, and lodging (Anterola and Lewis, 2002).

The remaining lignaceous biomass after saccharification and fermentation
can be used as a combustion fuel source for the distillation process and to power
the cellulosic ethanol biorefinery with the possibility of selling surpluses to the

electrical grid (Farrell et al., 2006; Lynd et al., 2009).

The Bioconversion Platform and the Bio-refinery Concept

Most conversion technology research efforts in the United States are
focused on the sugar or bioconversion approach due to the strong interest in the
development of alternative liquid transportation fuel (Sanderson et al., 2006). In
the biological conversion of feedstock carbohydrate fractions cellulose,
hemicellulose and pectins are chemically separated from lignin and split into
hexose and pentose monomers that are then fermented to produce ethanol
(Patzek 2008). The bioconversion platform is made up of an array of
pretreatment, hydrolysis or saccharification, fermentation and distillation
processes. In some forms of the platform, processes can be combined, including
simultaneous saccharification and fermentation (SSF), a process in which
cellulose hydrolysis and glucose fermentation as well as C6 hemicellulose sugars
(galactose and mannose) occur in the same vessel (Chang et al., 2001). SSF is
thought to reduce inhibition caused by glucose fermentation and thereby
increase the rate of saccharification (Chang et al., 2001). Consolidated
bioprocessing (CBP) is a single step, mature technology (not yet developed)

process, that combines cellulase production, cellulose hydrolysis, and hexose

74

 

and pentose fermentation, thereby eliminating the need for separate enzyme

production (Lynd et al., 2005).

Pretreatment

The susceptibility of cellulose to hydrolysis is restricted due to lignin matrix
and hemicellulose protection surrounding the cellulose microfibrils (Bals et al.,
2010a; Wyman, 1996). Effective pretreatment is required to increase the surface
accessibility for hydrolysis, promote cellulose decrystallization, partially
depolymerize hemicelluloses, and reduce the lignin recalcitrance in the treated
biomass (Alizadeh et al., 2005). An ideal pretreatment must reduce
lignocellulosic recalcitrance while minimizing the formation of degradation
products that inhibit subsequent hydrolysis and fermentation. Pretreatment
processes can be physical or chemical. Examples of available pretreatment
techniques include acid hydrolysis, steam explosion, ammonia fiber expansion,
alkaline wet oxidation, and hot water pretreatment. AFEX is an important
pretreatment technology that utilizes both physical (high temperature and
pressure) and chemical (ammonia) processes to achieve effective pretreatment
(Alizadeh et al., 2005; Bals et al., 2010b). AF EX offers nearly complete recovery
of the pretreatment chemical (ammonia) and does not require a washing step,
which facilitates high solid load hydrolysis. The residual ammonia on the

pretreated biomass serves as a nutrient addition for microbial growth.

75

Hydrolysis

The disrupted biomass then undergoes hydrolysis (saccharification),
which liberates glucose, xylose and other monomer sugars which are then
fermented and distilled to produce bioethanol (Yuan et al., 2008). There are four
major structural characteristics of lignocellulosic biomass that hinder enzymatic
hydrolysis: cellulose fibre crystallinity, hemicelluloses acetylation, inaccessible
surface area and lignin content. Glucose can be fermented by industrial yeast
strains (Saccharomyces). Hemicellulosic pentoses, however, are difficult to

ferment with currently available yeast strains (Saha, 2003).

Effect of nitrogen fertilization and harvest timing on cell wall composition

It is generally understood that the proportions of cellulose, hemicellulose,
and lignin components vary from plant to plant, but less is known about the
extent to which these proportions differ within in a given species, or how these
proportions can be manipulated with harvest timing and frequency, or nitrogen

fertilizer.

Nitrogen fertilization and its effect on chemical composition of biomass is
complex. A survey discussed in Chemey et al. (1998) in which Wilson (1982)
looked at studies that examined the effects of nitrogen on composition and
degradability found that a third of them stated there was no effect, another third
said there was a positive effect and the remainder stated that there was a
negative effect. A previous study by Lemus et al. (2008) found that cellulose and

lignin increased with increased N rate.

76

 

The Effect of Nitrogen Fertilizer and Harvest Timing on Feedstock Quality

The different responses to harvest date and location/ecotype have large
implications for biomass refining (Table 18, Bals et al., 2010a). Harvest
management systems must be tailored to local needs of the BCF in order to
maximize ethanol production (Bals et al., 2010). The authors found that
harvesting in July provided lower costs for pretreatment and higher potential
ethanol yields, which may help offset the cost of a second harvest, with the
possibility for additional revenue from co-product generation. However,
harvesting prior to senescence and a killing frost has implications for stand

longevity as discussed in the previous chapter.

Table 18. Composition analysis (as a percentage of total dry weight) for the July
and October harvests of Cave-in-Rock (CIR) switchgrass from Bals et al. 2010

 

 

 

 

 

 

 

 

 

Cell-Wall Constituent CIR: July CIR: October
(% of total dry weight)

Glucan 30.6 :I: 0.2 33.6 :I: 0.5

Sucrose* 5.1 :l: 0.3 2.4 t 0.2

Xylan 19.4 :I: 0.3 25.3 :I: 0.2

Arabinan 2.0 :t 0.1 2.0 t 0.1

Lignin 10.4 :I: 0.4 16.7 :I: 0.5

 

 

 

*Soluble carbohydrate

77

 

Cell wall component differences also exist between switchgrass harvested
in the fall and the spring. Adler et al. (2006) found that switchgrass that was over-
wintered and harvested in the spring had higher concentrations of cell wall
glucose and non-glucose sugars than switchgrass harvested in the fall but that
lignin increased over the same period. They attribute the higher cell wall
carbohydrates values in spring to the leaching of soluble components such as
sugars, protein, and organic acids when overwintered (Adler et al., 2006).

Maturity is the single most important determinate of cell-wall
concentration and composition (Vogel and Jung 2000). As plants mature, they
undergo many processes that affect the relative concentrations of cellulose,
hemicellulose, lignin, pectin, proteins, ash, and other cell wall constituents. Some
of the processes include: cell elongation; a shift in leafzstem ratio; secondary wall
deposition; Iignification; carbohydrate translocation; and senescence. The stem
component increases as switchgrass matures and leaf components decrease as
a percent of total biomass, primarily because of stem internode (Table 19, Lee et
al., 2007) elongation (Sanderson, 1992). The proportion of leaves to stems is an
important determinant in biomass yields and potentially for conversion. On a
component basis, leaves are generally lower in lignin, higher in solubles,
cellulose and hemicellulose, whereas switchgrass stems will contain greater
proportions of lignin and cellulose (Sarath et al., 2008a). Table 20 provides
information on changes in cell-all carbohydrates with maturity from Dien et al.

(2006)

78

 

Table 19. Changes in composition of ‘Nebraska 28’ switchgrass with crop
Maturity from Lee et al. (2007)

 

 

 

 

 

 

 

 

Anthesis | Post Physiological Maturity j Overwintered
% of dry matter
Cellulose 33.0 35.2 41.8
Hemicellulose 31.0 31.1 33.7
Acid 3.8 4.6 6.2
Detergent
Lignin

 

 

 

 

 

Table 20. Cell wall carbohydrates (9 kg“) from Dien et al. (2006)

 

 

 

 

 

Switchgrass Gluc* Xyl Ara Gal Man Rha Fuc
Pre—boot 273 179 31 13 5 2 1
Anthesis 283 195 27 10 4 1 1
Post-frost 322 223 30 12 5 2 1

 

 

 

 

 

 

 

 

 

 

*Abbrev: Glucose (Gluc); Xylose (Xyl); Arabinose (Ara); Galactose (Gal);
Mannose (Man); Rhamnose (Rha); Fucose (Fuc).

Objectives

Cost effective lignocellulosic ethanol production from switchgrass requires
high biomass yields and a high concentration of cellulose. The objective of this
research was to quantify the crystalline cellulose and polyphonol lignin content of
switchgrass cell walls, and assess the composition of matrix polysaccharide
content mainly from hemicelluloses at a whole-plant level under three harvest
management systems and four nitrogen fertilizer rates. Research on
pretreatment optimization, hydrolysis/saccharification, SSF, BCP, enzyme and

catalyst design could benefit from this feedstock quality data.

79

MATERIALS AND METHODS

Site Description and Experimental Design

Field experiments were conducted at the Michigan State University
Agronomy Farm, in East Lansing (42°25'48" N, 84°16'48" W), MI over a 3-yr
period beginning in the spring of 2007. The experimental design was a
randomized complete block design with five replications and was conducted in a
field with predominantly Capac loam (Fine-loamy, mixed, mesic Aeric
Ochraqualfs) soils. Plots were fertilized (0, 38, 79, 157 kg N ha“) in the spring
applied as granular urea (46-0-0 N-P-K). Plots were then harvested under a one,
or two- cut system. The plots under the one-cut system were out either in the fall
approximately two weeks after a killing frost (one-cut fall), or left to overwinter,
and harvested in the spring (one-cut ovenivintered). The two cut-system
consisted of a harvest during the late-boot to early anthesis stage (two-cut
summer), and again in the late fall (two-cut fall) after a killing frost (<28° C for a
sustained period of time). Total fresh weight per plot and biomass content for
each plot was recorded, and biomass moisture content for each plot was
determined via subsamples that were placed in a forced air oven at 100°C for 72
hours or until a consistent dry weight was achieved. The total dry matter (DM)
per plot was then converted to Mg DM ha". Samples were finely ground with a
Christy-Norris mill (Christy and Norris Limited, Chelmsford, England) fitted with a

1-mm screen for further analysis.

80

 

Compositional Analysis

Crystalline cellulose content and the analysis of the monosaccharide
constituents Of cell wall hemicellulose were based on five field replicates and all
four nitrogen rates and harvest timings, while lignin content analysis was based
on three field replicates, two harvest timings (one-cut fall and one-cut
overwintered) and three N rates (0, 79, 157 kg ha"). Prior to compositional
analysis, the dried biomass was ball milled to a fine powder (<1 mm). From the
powder, the alcohol insoluble residue (AIR) was prepared and thereafter,
underwent amylase treatment to remove the residual starch resulting in purified

lignocellulosic material (York et al., 1986).

Cellulose and Hemicellulose

For hemicellulose compositional analysis, the destarched AIR was used
for the weak acid hydrolysis using [2M] trifluoroacetic acid. The hydrolysate was
separated from the insoluble material and the monosaccharides were derivatized
into their corresponding alditol acetates and quantitated using a GC-MS system
(Albersheim, 1967). The remaining insoluble material from the hydrolysis was
stripped further of hemicelluloses and amorphous glucan by washing with the
Updegraff reagent, an acetic acid and nitric acid mixture (Updegraff 1969). The
remaining material, primarily crystalline cellulose, was fully hydrolyzed with 72%
sulphuric acid and the glucose in the resulting hydrolysate was quantitated using
the colorimetric Anthrone assay (Seaman hydrolysis; (Selvendran and Oneill,

1987)

81

 

Lignin

Samples collected from three replications, and not including the 38 kg N
ha‘1 rate, or the two- cut system treatment, were used for lignin content analysis.
Starting from the purified lignocellulosic material collected from the sugar
compositional analysis, the lignin was quantified with the acetyl bromide soluble
lignin assay. The material was treated with a 1:4 acetyl bromide/acetic acid (v/v)
solution to render the lignin acetic acid soluble. After a volumetric dilution with
glacial acetic acid, the solublized lignin was quantitated using a UV
spectrophotometer set at 280 nm wavelength. The lignin content data was
expressed as percent acetyl bromide soluble lignin from the dry weight of the

isolated cell wall material (Fukushima et al., 1991).

Statistical Analysis

All compositional analyses were done in triplicate, and data were
corrected to a 100% dry matter (DM) basis. Data was analyzed using PROC
MIXED is SAS v9.1 (2003 SAS Institute Inc., Cary, NV.) Nitrogen fertilization
level, harvest timing, and year were considered as fixed effects, and replication
was considered to be a random effect. Unequal variances were analyzed using
the REPEATED/Group option. Comparisons between individual factor level or
treatments were conducted using multiple t-tests with the post-hoc scheffe
adjustment to generate post-specified contrasts and estimate statements. When
nitrogen fertilizer was found to be significant in the ANOVA linear contrast
statements were used to examine the nitrogen effect. Contrast statements were

also used to determine if concentrations of the various hemicellulose sugars

82

varied with harvest timing. Matrix polysaccharide composition data were
transformed with the arcsine square-root transformation prior to analysis and
year was treated as a random effect for this analysis. Results were reported as

statistically significant at 01 = 0.05.

RESULTS AND DISCUSSION

Crystalline Cellulose Content

There was no interaction between nitrogen fertilizer and harvest timing,
and nitrogen was not significant at a=0.05. but a trend is evident (P=0.0667) (Fig.
12). Harvest timing was a Significant factor (p=<0.0001), plots that were
harvested under the two-cut harvest system contained ~12.8% less crystalline
cellulose (pg mg") than those that reached full maturity and underwent natural

senescence (Fig. 12).

Studies of lowland ecotypes in Alabama found that cellulose contents
decreased by 12% between the first and second harvests within a two-cut
system (Sladden et al., 1991). Although the second-cut of the two-cut system
was made at the same time as the one-cut fall, plants experiencing regrowth may
have had an accelerated vegetative growth and therefore a reduced period of
secondary cell wall deposition; and may have ceased growth as a result of the
killing-frost as opposed to reaching full maturity, and undergoing natural
senescence. The resulting tissue may have higher proportions of soluble

carbohydrates at harvest time.

83

 

400 b a a
350 b
300
250
200
150
100
50
0

Two-Cut Two-Cut Fall One-Cut Fall One-Cut
Summer Over Wintered

Crsytalline Cellulose (pg/mg)

 

 

Figure 12. Crystalline cellulose under varying harvest timings. Means with the
same letter are not statistically different at a=0.05. Standard error was 12.1

The lack of a change in cellulose content with increased fertilizer present
in our data is not in agreement with previous research. Bilal and Sindhu (2001)
found that cellulose content increased with nitrogen fertilizer level and advancing
growth stage with the maximum concentrations occurring in the stem portion.
Lemus et al. (2008) conducted a study that looked at quality as affected by
nitrogen rate, but had a single harvest treatment in which switchgrass was
harvested between September 27 and November 27 over a five-year period.
They found that cell wall constituents differed between years, with strong linear
and quadratic responses for all traits and nitrogen content of earlier cut
switchgrass to differential leaching severity among their harvest dates. They do
not mention the possible change in stemzleaf ratio or potential for enhanced
lignification of the secondary cell wall with the delayed onset of frost in some

years.

84

Matrix polysaccharides composition

The molar concentrations of seven cell wall constituents that comprise or
are associated with hemicellulose were quantified and their molar concentrations
were compared among harvest timings, across both years (Table 8). The
monosaccharides were derivatized into their corresponding alditol acetates and
thus we are not able to discuss the results as they pertain to specific
hemicellulose types, or other measures of quantification such as NDF and ADF.
A measure of total change of hemicellulose over time is not reported. All
monosaccharide’s molar concentrations were Significantly affected by harvest
timing (p=<0.0001) and nitrogen was found to be a significant effect in the
derivation of three of the sugars, namely arabinose (p=0.0155) (Fig 14),
rhamnose (p=0.0002) (Fig 15) and galactose (p=0.0002) (Fig 16) although there

was no interaction between the main effects.

85

40 -~

 

 

a
E) 38 "
0)
3 36 ~-
0
U)
a 34
(‘3
E 32‘
O 39 78 157

Nitrogen Fertilizer Rate kg N ha-1

Figure 13. Figure 14. Arabinose molar concentration (%), as a proportion of the
matrix polysaccharides at four different N rates across harvest timings. Means
with the same letters are not statistically different at a=0.05. Standard error was
1.02

Rhamnose (uglmg)
O
(I)

 

 

 

O 39
Nitrogen fertilizerrate kg N ha-1

Figure 14. Rhamnose molar concentration (%) , as a proportion of the matrix
polysaccharides at four different nitrogen rates, across harvest timings. Means
with the same letters are not significantly different at a=0.05. Standard error was
0.06.

86

Galactose (uglmg)

 

 

0 35 70 140
Nitrogen Fertilizer Rate kg N ha-1

 

 

 

y“-.. ,, . -.-_-_FF.-r-« .7 , FF2~er_.;:—Fe.—m--’. . .-.

Figure 15. Galactose molar concentration (%), as a proportion of the matrix
polysaccharides at four different N rates across harvest timings. Means with the
same letters are not statistically different at a=0.05. Standard error was 0.22

Our data is not in agreement with Chemey et al. (1998) who found that
arabinose concentrations, among other cell wall monosaccharides, were not

influenced by environmental conditions or N fertilization (Cherney et al., 1988).

The two-cut summer biomass had high levels of xylose (80.1% molar
concentration) and the lowest levels of all other monosaccharides. Of the two
harvests made in the fall, there were significant differences in all
monosaccharides between harvest systems except for fucose and glucose. The
regrowth from the two-cut system had the highest proportions of all
monosaccharides except in the case of xylose, which was relatively higher in the
one-cut fall harvested biomass. The one-cut system had Similar levels of xylose,

rhamnose and galactose across harvest systems. However, glucose, mannose

87

and arabinose were all higher in proportion in the biomass that was over-

wintered. The highest glucose concentration was present in the one-cut

overwintered biomass.

Table 21. Matrix polysaccharide composition across varying harvest timings
expressed as molar concentration

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Monosaccharide Two-Cut Two-Cut One-Cut One-Cut Standard
Constituents Summer Fall Fall Overwintered Error
Molar Concentration (%)
Xylose 80.10°' 68.93° 74.70° 6965" 0.002
Arabinose 9.81 d 13.48° 11 .20° 12.25° 0.0031
Mannose+ 017° 135° 0.39c 0.84° 0.0050
Fucose 0.0? 0.12° 0.14° 0.09° 0.0023
Rhamnose+ 0.29c 0.39a 0.72b 0.45b 0.0022
Galactose+ 1 .94° 3.59° 232° 245° 0.0037
Glucose 758° 1 1 .82° 10.88° 14.28°

 

 

 

 

 

 

*Means with the same letters within the same row are not different at a=0.05
+ N was a significant factor but no interaction was present

88

 

Lignin Content

27 ab

Lignin Content (pg/mg)

   

 

 

Nitrogen Fertilizer Rate kg N ha'1

Figure 16. Lignin content expressed as a weight percentage measurement of
the acetyl bromide soluble lignin as a function of applied nitrogen fertilizer.
Means with the same letters are not statistically significant (a=0.05; scheffe
adjusted means). Standard error was 0.273.

The lignin content was determined as a weight percentage measurement
of the acetyl bromide soluble lignin based off the dry weight of the isolated cell
walls. Lignin content of the biomass did not differ when harvested in the fall or
delayed until the spring and there were no significant interactions between any
of the fixed effects. Analyses on the summer one- and two-cut samples were
not performed. Both year (p=<0.001) and nitrogen (p=0.0099) were affected
although caution is exercised in the interpretation of the year results as
samples were analyzed at separate times. Switchgrass that was fertilized with

157 kg N ha'1 had a lignin content that was 4.5% greater than unfertilized

89

switchgrass, whereas switchgrass fertilized at the 79 kg N ha‘1 rate was not
statistically different from either the unfertilized biomass, or the 157 kg N ha'1
rate (Fig. 9). Lignin content was increased by 16% when fertilized with 200 kg
N ha'1 (Cherney et al., 1988).

CONCLUSION

The data do not show a correlation between nitrogen fertilizer and
crystalline cellulose content during the initial years of the switchgrass stand.
Hemicellulose composition did change with harvest timing. The greatest
amount of xylose was present in the two-cut summer harvest. Nitrogen
fertilizer negatively affected the concentrations of arabinose, galactose and
rhamnose. Further research is needed to examine why nitrogen might affect
these specific hemicelluloses monosaccharides. Lignin content did not differ
between the fall and spring but did increase with increased rates of applied
nitrogen fertilizer. Increased sampling in terms of harvest timing and nitrogen
rates is needed to further validate this finding. Farmers should harvest once

annually, once the crop has senesced to enhance feedstock quality.

90

CONCLUSIONS

The data from three sites over two years demonstrate that switchgrass
can be successfully established in the Great Lakes Region. The data suggest
that tillage is not critical for obtaining a stand frequency that exceeds the 40%
establishment threshold. The data also support the hypothesis that with tillage,
both a CGD and a double-roller seeder can be used to successfully establish
switchgrass for energy purposes. These results suggest that existing equipment
can be used to prepare seedbeds, which would mean less capital expenditure
and an increased likelihood of farmer adoption. A mid-spring seeding date (mid-
may to early June) could potentially offer the best possibility for increasing yield

and profitability in early production years.

A yield of 12.0 Mg ha'1 per year is a reasonable estimate of yield potential
for the fall one- cut harvest system, Yield losses from delaying harvest until the
spring is of concern if switchgrass is being grown for the bioconversion platform.
Nitrogen rate was a significant determinant of yield for the two- out summer, and
two- out combined yields and was best fir with a LRP functional form. Less
differentiation by level of nitrogen fertilizer was evident when harvest was

delayed until the fall or following spring.

We have shown that lignin content was increased with increasing N and
that crystalline cellulose was not. Crystalline cellulose increased with harvests
made after maturity had been reached. Hemicellulose composition was affected

by harvest timing, which subsequently affects the concentration of hemicellulosic

91

monosaccharides. Arabinose, rhamnose, and galactose concentrations were

decreased with higher rates of N,

Further research is needed to optimize the no- till establishment systems,
This includes the use of two pre-plant herbicide burndown treatments, proper
residue management, optimization of NT drills, and seeding into warmed-up soils
could potentially enhance NT yields. Research on chemical control programs for
switchgrass establishment and early stand management would also be of benefit
as it would lead to registration of effective herbicides that could further enhance
establishment and cost-effectiveness, The efficacy of deriving agronomic and
optimal N rates for switchgrass biomass production from wither an LRP or QP
model needs to be ascertained, Lastly, research is needed on how nitrogen
contributes to increased biomass yield and why those gains diminish with

delaying harvest.

92

LITERATURE CITED

Adler, P.R., M.A. Sanderson, A.A. Boateng, P.l. Weimer, and H.J.G. Jung. 2006.
Biomass yield and biofuel quality of switchgrass harvested in fall or spring.
Agron. J. 98:1518-1525.

Albersheim, P., et al. 1967. "A method for the analysis of sugars in plant cell wall
polysaccharides by gas-liquid chromatography,". Carbohydr. Res. 5.
(ed.) 2005.

Anterola, AM, and N.G. Lewis. 2002. Trends in lignin modification: a
comprehensive analysis of the effects of genetic manipulations/mutations
on lignification and vascular integrity. Phytochemistry 61 :221-294.

Aravindhakshan, S.C., Epplin, F. and C. Taliaferro. 2008. Biomass Yield to
Nitrogen Response Functions for Four Candidate Biorefinery Feedstock
Perennial Grass Species American Agricultural Economics Association
Annual Meeting, Orlando, FL.

Badger, P.C., (ed.) 2002. Ethanol from cellulose: A general review., pp. 1-17-21.
ASHS Press, Alexandria, VA.

Bals, B., H. Murnen, M. Allen, and B. Dale. 2010a. Ammonia fiber expansion
(AFEX) treatment of eleven different forages: Improvements to fiber
digestibility in vitro. Animal Feed Science and Technology 155:147-155.

Bals, B., C. Rogers, M.J. Jin, V. Balan, and B. Dale. 2010b. Evaluation of
ammonia fibre expansion (AF EX) pretreatment for enzymatic hydrolysis of

switchgrass harvested in different seasons and locations. Biotechnology
for Biofuels 3.

Barney, J.N., J.J. Mann, G.B. Kyser, E. Blumwald, A. Van Deynze, and J.M.
DiTomaso. 2009. Tolerance of switchgrass to extreme soil moisture
stress: Ecological implications. Plant Science 177:724-732.

Bilal, M.Q., and AA. Sindhu. 2001. Hemi-cellulose and Cellulose Contents of
Mott Grass as Affected by Nitrogen Fertilizer and Stage of Maturity. Online
Journal of Biological Sciences 12436-437.

Bowley, SR. 2008. A Hitchhiker’s Guide to Statistics in Plant Biology Any Old
Subject Books, Guelph, ON.

Brejda, J.J., L.E. Moser, and KP. Vogel. 1998. Evaluation of switchgrass
rhizosphere microflora for enhancing seedling yield and nutrient uptake.
Agronomy Journal 90:753-758.

93

Carpenter, S.R., N.F. Caraco, D.L. Correll, R.W. Howarth, A.N. Sharpley, and
V.H. Smith. 1998. Nonpoint pollution of surface waters with phosphorus
and nitrogen. Ecological Applications 82559-568.

Casler, M.D., and AR. Boe. 2003. Cultivar X environment interactions in
switchgrass. Crop Science 43:2226-2233.

Casler, M.D., K.P. Vogel, C.M. Taliaferro. and R.L. Wynia. 2004. Latitudinal
adaptation of switchgrass populations. Crop Science 44:293-303.

Chang, V.S., W.E. Kaar, B. Burr, and MT. Holtzapple. 2001. Simultaneous
saccharification and fermentation of lime-treated biomass. Biotechnology
Letters 23:1327-1333.

Cherney, J.H., K.D. Johnson, J.J. Volenec, and KS. Anliker. 1988. Chemical
Composition of Herbaceous Grass and Legume Species Grown for
Maximum Biomass Production. Biomass 17:215-238.

Christian, D.G., A.B. Riche, and NE. Yates. 2002. The yield and composition of
switchgrass and coastal panic grass grown as a biofuel in Southern
England. Bioresource Technology 83:115-124.

Cox, W.J., R.W. Zobel, H.M. Vanes, and DJ. Otis. 1990. Tillage Effects On
Some Soil Physical And Corn Physiological-Characteristics. Agronomy
Journal 82:806-812.

Cuomo, G.J., and BE. Anderson. 1996. Nitrogen fertilization and burning effects
on rumen protein degradation and nutritive value of native grasses.
Agronomy Journal 88:439-442.

Dien, B.S., H.J.G. Jung, K.P. Vogel, M.D. Casler, J.F.S. Lamb, L. lten, R.B.
Mitchell, and G. Sarath. 2006. Chemical composition and response to
dilute-acid pretreatment and enzymatic saccharification of alfalfa, reed
canarygrass, and switchgrass. Biomass & Bioenergy 30:880-891.

Doran-Peterson, J., D.M. Cook, and SK. Brandon. 2008. Microbial conversion of
sugars from plant biomass to lactic acid or ethanol. Plant Journal 54:582-
592.

Fargione, J., J. Hill, D. Tilman, S. Polasky, and P. Hawthorne. 2008. Biofuels:
Putting current practices in perspective - Response. Science 320:1420-
1422.

Farrell, A.E., R.J. Plevin, B.T. Turner, A.D. Jones, M. O'Hare, and OM. Kammen.

2006. Ethanol can contribute to energy and environmental goals. Science
31 1 :506-508.

94

Fike, J.H., D.J. Parrish, D.D. Wolf, J.A. Balasko, J.T. Green, M. Rasnake, and
J.H. Reynolds. 2006. Switchgrass production for the upper southeastern
USA: Influence of cultivar and cutting frequency on biomass yields.
Biomass & Bioenergy 30:207-213.

Foster, C.E., Martin, TM. and M. Pauly J Vis Exp. Comprehensive compositional
analysis of plant cell walls (lignocellulosic biomass) part II: carbohydrates.
12.

Fukushima, R.S., B.A. Dehority, and SC. Loerch. 1991. Modification Of A
Colorimetric Analysis For Lignin And Its Use In Studying The Inhibitory
Effects Of Lignin On Forage Digestion By Ruminal Microorganisms.
Journal of Animal Science 69:295-304.

Goss, M.J., and A. de Varennes. 2002. Soil disturbance reduces the efficacy of
mycorrhizal associations for early soybean growth and N-2 fixation. Soil
Biology & Biochemistry 34:1167-1173.

Grabber, J.H. 2005. How do lignin composition, structure, and cross-linking affect
degradability? A review of cell wall model studies. Crop Science 45:820-
831.

Grabber, J.H., and GA. Jung. 1991. Isolation Of Parenchyma And Sclerenchyma
Cell-Types From The Plant-Parts Of Grasses. Crop Science 31 :838-842.

Grandy, AS, and GP. Robertson. 2006. Aggregation and organic matter
protection following tillage of a previously uncultivated soil. Soil Science
Society of America Journal 70:1398-1406.

Greig-Smith, P. 1983. Quantitative plant ecology Blackwell Scientific Publishers,
London.

Haque, M., F.M. Epplin, and CM. Taliaferro. 2009. Nitrogen and Harvest
Frequency Effect on Yield and Cost for Four Perennial Grasses.
Agronomy Journal 101 :1463-1 469.

Harris, P.J., and B.G. Smith. 2006. Plant cell walls and cell-wall polysaccharides:
structures, properties and uses in food products. lntemational Journal of
Food Science and Technology 41 :129-143.

Heaton, E., T. Voigt, and SP. Long. 2004. A quantitative review comparing the

yields of two candidate C-4 perennial biomass crops in relation to
nitrogen, temperature and water. Biomass & Bioenergy 27:21-30.

95

Hill, J. 2007. Environmental costs and benefits of transportation biofuel
production from food- and lignocellulose-based energy crops. A review.
Agronomy for Sustainable Development 2721-12.

Hu, W.J., S.A. Harding, J. Lung, J.L. Popko, J. Ralph, D.D. Stokke, C.J. Tsai,
and V.L. Chiang. 1999. Repression of lignin biosynthesis promotes
cellulose accumulation and growth in transgenic trees. Nature
Biotechnology 17:808-812.

Hu, Z.J., R. Sykes, M.F. Davis, E.C. Brummer, and A.J. Ragauskas. 2010.
Chemical profiles of switchgrass. Bioresource Technology 101:3253-3257.

Hultquist, S.J., K.P. Vogel, D.J. Lee, K. Arumuganathan, and S. Kaeppler. 1996.
Chloroplast DNA and nuclear DNA content variations among cultivars of
switchgrass, Panicum virgatum L. Crop Science 36:1049-1052.

liyama, K., T.B.T. Lam, and BA. Stone. 1994. Covalent Cross-Links In The Cell-
Wall. Plant Physiology 104:315-320.

Jensen, K., C.D. Clark, P. Ellis, B. English, J. Menard, M. Walsh, and D. Ugarte.
2007. Farmer willingness to grow switchgrass for energy production.
Biomass & Bioenergy 31:773-781.

Kim, S., and BE. Dale. 2005. Life cycle assessment of various cropping systems
utilized for producing biofuels: Bioethanol and biodiesel. Biomass &
Bioenergy 29:426-439.

Kravchenko, A.G., and KD. Thelen. 2007. Effect of winter wheat crop residue on
no-till corn growth and development. Agronomy Journal 99:549-555.

Laser, M., H.M. Jin, K. Jayawardhana, and LR. Lynd. 2009. Coproduction of
ethanol and power from switchgrass. Biofuels Bioproducts & Biorefining-
Biofpr 32195-218.

Lee, D.K., V.N. Owens, C. Boehmel, and P. Jeranyama. 2007. Compostion of
Herbaceous Biomass Feedstocks, pp. 16. North Central Sun Grant
Center, Brookings, SD.

Lemus, R., E.C. Brummer, K.J. Moore, N.E. Molstad, C.L. Burras, and M.F.
Barker. 2002. Biomass yield and quality of 20 switchgrass populations in
southern Iowa, USA. Biomass & Bioenergy 23:433-442.

Lemus, R., E.C. Brummer, C.L. Burras, K.J. Moore, M.F. Barker, and NE.
Molstad. 2008. Effects of nitrogen fertilization on biomass yield and quality

in large fields of established switchgrass in southern Iowa, USA. Biomass
& Bioenergy 32:1187-1194.

96

Lewandowski, l., and A. Kicherer. 1997. Combustion quality of biomass: Practical
relevance and experiments to modify the biomass quality of Miscanthus x
giganteus. European Journal of Agronomy 62163-177.

Lynd, L.R., W.H. van Zyl, J.E. McBride, and M. Laser. 2005. Consolidated
bioprocessing of cellulosic biomass: an update. Current Opinion in
Biotechnology 16:577-583.

Lynd, L.R., E. Larson, N. Greene, M. Laser, J. Sheehan, B.E. Dale, 8.
McLaughlin, and M. Wang. 2009. The role of biomass in America's energy
future: framing the analysis. Biofuels Bioproducts & Biorefining-Biofpr
3:1 13-123.

Ma, 2., CW. Wood, and DJ. Bransby. 2000. Soil management impacts on soil
carbon sequestration by switchgrass. Biomass & Bioenergy 18:469-477.

Madakadze, l.C., B.E. Coulman, A.R. McElroy, K.A. Stewart, and D.L. Smith.
1998. Evaluation of selected warm-season grasses for biomass

production in areas with a short growing season. Bioresource Technology
65:1-12.

Madakadze, l.C., K.A. Stewart, P.R. Peterson, B.E. Coulman, and D.L. Smith.
1999. Cutting frequency and nitrogen fertilization effects on yield and
nitrogen concentration of switchgrass in a short season area. Crop
Science 39:552-557.

Masters, RA, and SJ. Nissen. 1998. Revegetating leafy spurge (Euphorbia
esula)—infested rangeland with native tallgrasses. Weed Technology
12:381-390.

Masters, R.A., S.J. Nissen. R.E. Gaussoin, D.D. Beran, and RN. Stougaard.
1996. lmidazolinone herbicides improve restoration of Great Plains
grasslands. Weed Technology 10:392-403.

McKendry, P. 2002. Energy production from biomass (part 2): conversion
technologies. Bioresource Technology 83:47-54.

McLaughlin, S., J. Bouton, D. Bransby, B. Conger, W.R. Ocumpaugh, D.J.
Parrish, C. Taliaferro, K.P. Vogel, and SD. Wullschleger. 1999.
Developing Switchgrass as a Bioenergy Crop ASHS Press, Alexandria,
VA.

McLaughlin, S., Bouton, J., Bransby, D., Conger, B., Ocumpaugh, W., Parrish,
D., Taliaferro, C., Vogel, K. and, and S. Wullschleger, (eds.) 1999.
Developing Switchgrass as a Bioenergy Crop. ASHS Press, Alexandria,
VA.

97

McLaughlin, SB, and LA. Kszos. 2005. Development of switchgrass (Panicum
virgatum) as a bioenergy feedstock in the United States. Biomass &
Bioenergy 28:515-535.

Mitchell, R., K.P. Vogel, and G. Sarath. 2008. Managing and enhancing
switchgrass as a bioenergy feedstock. Biofuels Bioproducts & Biorefining-
Biofpr 22530-539.

Monti, A., P. Venturl, and H.W. Elbersen. 2001. Evaluation of the establishment
of lowland and upland switchgrass (Panicum virgatum L.) varieties under
different tillage and seedbed conditions in northern Italy. Soil & Tillage
Research 63:75-83.

Monti, A., G. Bezzi, G. Pritoni, and G. Venturi. 2008. Long-term productivity of
lowland and upland switchgrass cytotypes as affected by cutting
frequency. Bioresource Technology 99:7425-7432.

Moore, K.J., L.E. Moser, K.P. Vogel, S.S. Waller, B.E. Johnson, and J.F.
Pedersen. 1991. Describing And Quantifying Growth-Stages Of Perennial
Forage Grasses. Agronomy Journal 83:1073-1077.

Moser, LE, and KP. Vogel. 1995. Switchgrass, big bluestem, and indiangrass.
5th ed. Iowa State Univ. Press, Ames.

Mosier, A.R., J.M. Duxbury, J.R. Freney, O. Heinemeyer, and K. Minami. 1998.
Assessing and mitigating N20 emissions from agricultural soils. Climatic
Change 4027-38.

Muir, J.P., M.A. Sanderson, W.R. Ocumpaugh, R.M. Jones, and R.L. Reed.
2001. Biomass production of 'Alamo' switchgrass in response to nitrogen,
phosphorus, and row spacing. Agronomy Journal 93:896-901.

Mulkey, V.R., V.N. Owens, and OK Lee. 2008. Management of warm-season
grass mixtures for biomass production in South Dakota USA. Bioresource
Technology 99:609-617.

Nagelkerke, N.J.D. 1991. A Note On A General Definition Of The Coefficient Of
Determination. Biometrika 782691-692.

Nelson, F.H.H.a.C.J. 1986. Planting Date Effects on Seedling Development of
Perennial Wann-Season Forage Grasses. l. Field Emergence Agron J
78:33—38.

Newman, PR, and LE. Moser. 1988. Grass Seedling Emergence, Morphology,

And Establishment As Affected By Planting Depth. Agronomy Journal
80:383-387.

98

Panciera, M.T., and GA. Jung. 1984. Switchgrass Establishment By
Conservation Tillage - Planting Date Responses Of 2 Varieties. Journal of
Soil and Water Conservation 39:68-70.

Parrish, D.J., and J.H. Fike. 2005. The biology and agronomy of switchgrass for
biofuels. Critical Reviews in Plant Sciences 24:423-459.

Parrish, D.J., D.D. Wolf, P.R. Peterson and W.L. Daniels. 1999. Switchgrass as a
biofuels crop for the upper southeast: Variety trials and cultural
improvement. Bioenergy Feedstock Development Program, Oak Ridge
National Laboratory, Oak Ridge, Tennessee.

Pauly, M., and K. Keegstra. 2008. Cell-wall carbohydrates and their modification
as a resource for biofuels. Plant Journal 54:559-568.

Perrin, R., K. Vogel, M. Schmer, and R. Mitchell. 2008. Farm-Scale Production
Cost of Switchgrass for Biomass. Bioenergy Research 1:91-97.

Pollack, RA, and R.J. Whales. 1991. Likelihood Dominance Criterion. Journal of
Econometrics:227-242.

Reynolds, J.H., C.L. Walker, and M.J. Kirchner. 2000. Nitrogen removal in
switchgrass biomass under two harvest systems. Biomass & Bioenergy
19:281-286.

Robertson, G.P., V.H. Dale, O.C. Doering, S.P. Hamburg, J.M. Melillo, M.M.
Wander, W.J. Parton, P.R. Adler, J.N. Barney, R.M. Cruse, C.S. Duke,
P.M. Fearnside, R.F. Follett, H.K. Gibbs, J. Goldemberg, D.J. Mladenoff,
D. Ojima, M.W. Palmer, A. Sharpley, L. Wallace, K.C. Weathers, J.A.
Wiens, and WW. Wilhelm. 2008. Agriculture - Sustainable biofuels Redux.
Science 322:49-50.

Robertson, G.P.a.P.R.G. 2004. Greenhouse Gas Fluxes in Tropical and
Temperate Agriculture: The need for a Full-Cost accounting of Global

Warming Potentials Environment, Development and Sustainability 6:51-
63.

Saha, BC. 2003. Hemicellulose bioconversion. Journal of Industrial Microbiology
& Biotechnology 30:279-291.

Sanderson, MA. 1992. Morphological Development Of Switchgrass And
Kleingrass. Agronomy Journal 84:415-419.

Sarath, G., L.M. Baird, K.P. Vogel, and RB. Mitchell. 2007. lnternode structure

and cell wall composition in maturing tillers of switchgrass (Panicum
virgatum. L). Bioresource Technology 98:2985-2992.

99

Sarath, G., D.E. Akin, R.B. Mitchell, and KP. Vogel. 2008a. Cell-wall composition
and accessibility to hydrolytic enzymes is differentially altered in
divergently bred switchgrass (Panicum virgatum L.) genotypes. Applied
Biochemistry and Biotechnology 15021-14.

Sarath, G., R.B. Mitchell, S.E. Sattler, D. Funnell, J.F. Pedersen, R.A.
Graybosch, and KP. Vogel. 2008b. Opportunities and roadblocks in
utilizing forages and small grains for liquid fuels. Journal of Industrial
Microbiology & Biotechnology 35:343-354.

SAS, 2003. SAS Procedures Guide. Version 9. SAS Inc., Cary, NC.

Schmer, M.R., K.P. Vogel, R.B. Mitchell, and R.K. Perrin. 2008. Net energy of
cellulosic ethanol from switchgrass. Proceedings of the National Academy
of Sciences of the United States of America 105:464-469.

Schmer, M.R., K.P. Vogel, R.B. Mitchell, L.E. Moser, K.M. Eskridge, and R.K.
Perrin. 2006. Establishment stand thresholds for switchgrass grown as a
bioenergy crop. Crop Science 46:157-161.

Selvendran, RR, and MA. Oneill. 1987. Isolation And Analysis Of Cell-Walls
From Plant-Material. Methods of Biochemical Analysis 32:25-153.

Sharma, N., I. Piscioneri, and V. Pignatelli. 2003. An evaluation of biomass yield
stability of switchgrass (Panicum virgatum L.) cultivars. Energy
Conversion and Management 44:2953-2958.

Sladden, S.E., D.l. Bransby, and GE. Aiken. 1991. Biomass Yield, Composition
And Production Costs For 8 Switchgrass Varieties In Alabama. Biomass &
Bioenergy 121 19-122.

Smart, A.J.a.L.E.M. 1997. Morphological development of swicthgrass as affected
by planting date. Agron. J. 89.

Sokhansanj, 8., S. Mani, A. Turhollow, A. Kumar, D. Bransby, L. Lynd, and M.
Laser. 2009. Large-scale production, harvest and logistics of switchgrass
(Panicum virgatum L.) - current technology and envisioning a mature
technology. Biofuels Bioproducts & Biorefining-Biofpr 3:124-141.

Vassey, T.L., J.R. George, and RE. Mullen. 1985. Early-Spring, Mid-Spring, And
Late-Spring Establishment Of Switchgrass At Several Seeding Rates.
Agronomy Journal 77:253-257.

Vogel, J. 2008. Unique aspects of the grass cell wall. Current Opinion in Plant
Biology 11:301-307.

100

Vogel, K.P., and RA Masters. 2001. Frequency grid - a simple tool for
measuring grassland establishment. Journal of Range Management
54:653-655.

Vogel, K.P., and H.J.G. Jung. 2001. Genetic modification of herbaceous plants
for feed and fuel. Critical Reviews in Plant Sciences 20:15-49.

Vogel, K.P., J.J. Brejda, D.T. Walters, and DR. Buxton. 2002. Switchgrass
biomass production in the Midwest USA: Harvest and nitrogen
management. Agronomy Journal 94:413-420.

Wilson, AM, and DD. Briske. 1979. Seminal And Adventitious Root-Growth Of
Blue Grama Seedlings On The Central Plains. Journal of Range
Management 32:209-213.

Wilson, J.R., and RD. Hatfield. 1997. Structural and chemical changes of cell
wall types during stern development: Consequences for fibre degradation
by rumen microflora. Australian Journal of Agricultural Research 48:165-
180.

Wolf, D.D., Parrish, D. J., Daniels, W. L. and J. R. McKenna. 1989. No-till
establishment of perennial, warm-season grasses for biomass production
Biomass 20:209-217.

Wright, L., and A. Turhollow. 2010. Switchgrass selection as a "model" bioenergy
crop: A history of the process. Biomass & Bioenergy 34:851-868.

Wullschleger, S.D., M.A. Sanderson, S.B. McLaughlin, D.P. Biradar, and AL.
Rayburn. 1996. Photosynthetic rates and ploidy levels among populations
of switchgrass. Crop Science 36:306-312.

Wyman, C.E., (ed.) 1996. Hanbook on Bioethanol: Production and Utilization.
Taylor and Francis, Washington, DC.

York, W.S., A.G. Darvill, M. McNeil, T.T. Stevenson, and P. Albersheim. 1986.
Isolation And Characterization Of Plant-Cell Walls And Cell-Wall
Components. Methods in Enzymology 118:3-40.

Yuan, J.S., K.H. Tiller, H. AI-Ahmad, N.R. Stewart, and CM Stewart. 2008.

Plants to power: bioenergy to fuel the future. Trends in Plant Science
13:421-429.

101

   

“'illillllllliilliillllllll2