THE CARBON SEQUESTRATION AND SOIL RESPIRATION
AFTER LAND USE CONVERSION IN BIOFUEL CROPPING ECOSYSTEMS
By
Yahn-Jauh Su

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
GeographyâDoctor of Philosophy
2017

ABSTRACT
THE CARBON SEQUESTRATION AND SOIL RESPIRATION AFTER LAND USE
CONVERSION IN BIOFUEL CROPPING ECOSYSTEMS
By
Yahn-Jauh Su
Global climate change alters Earthâs carbon, hydrological and energy cycles from local to
global scales, changing our climate patterns and impacting our lifestyles and prosperity. The
development of bioenergy may partially mitigate the release of carbon dioxide during the
combustion of fossil fuel. However, the carbon emissions from the bioenergy-induced land use
change have long been debated and it is not certain whether they really represent a reduction of
carbon emission. In this study, I monitored the components of the net ecosystem exchange
(NEE) of CO2, including gross primary production (GPP), ecosystem respiration (Reco), total soil
respiration (Rs), autotrophic soil respiration (Ra) and heterotrophic soil respiration (Rh), to
understand their responses to climate variability and in particular a severe drought event. I
studied three major bioenergy crops (continuous corn, switchgrass and restored multicultural
prairie) on fields with two different land use histories (conventional corn-soybean rotation and
Conservation Reserve Program brome grass fields). I found that the amplitude, the duration and
the seasonality of microclimatic variables (temperature and precipitation) were important for the
carbon dynamics in the bioenergy cropping systems. The soil water content affected the annual
NEE, GPP and Reco although it did not have strong correlations with these components of carbon
fluxes at short-term scale. The short-term (1-2 week) normal summer water deficit may affect
annual NEE while long-term (spring-summer) drought may change the community structure and
affect the carbon cycling processes in the following years. The temperature sensitivities of soil
respiration were shifted within and between years. In addition, crop types and land use histories

affect the responses of ecosystem to climate events. The different phenology between annual and
perennial crops and the establishment of dense root systems in perennial crops can change the
ratio of the components of NEE and change the direction and the amounts of net ecosystem
carbon flux. Annual and perennial crops have different strategies responding to different climate
scenarios and their combinations. The monitoring of climate patterns at intra-annual scale is
required to understand how the ecosystem respond to climate change.

Dedicated to my parents and my wife,
who have been always supportive and encouraging

iv

ACKNOWLEDGEMENTS

I would like to thank my advisor, Dr. Jiquan Chen, for his support and guidance
throughout the past five year. He gave me a chance to start my PhD study in environmental
sciences and ecosystem ecology with the wide exploration of geography, biogeochemistry and
mathematic modeling, as well as pushing me for academic journal writing. I learned independent
learning, problem solving in field, and developing new methods to resolve present knowledge
gaps. I also deeply thank my committees. Dr. G. Philip Robertson for teaching me to look at
academic questions from a big picture and encouraging me to think about fundamental questions
and how research findings can be implemented in our society. Dr. Stephen K. Hamilton
stimulates my thinking from details of study results and encourage me to try any new ideas. He
and Suzanne Sippel give me the warmest help when I faced big challenges during my PhD life.
Dr. Jeffrey A. Andresen assisted me for the basic knowledge of agricultural meteorology and
gave me support in Geography.
I like to thank all members in LEES lab, Robertson lab and Hamilton lab and professors,
researchers and students in the W.K. Kellogg Biological Station (KBS). I cannot finish the heavy
field work, data analysis, method development and dissertation writing without their support. I
also thank Michael Abraha, who worked with me closely in the field and in the lab. The
discussions with him, both regarding research and life, always helped me mitigate my stress. I
am grateful to Ilya Gelfand, Stecey VanderWulp, Kevin Kahmark, Sven Bohm, Changliang
Shao, Housen Chu, Zutao Ouyang, Sarah Roley, Gabriela Shirkey, Ranjeet John, Terenzio
Zenone, Wei Shan, Yi Fan, and Cheyenne Lei. I also thank the support from the Department of
Geography, Environment, and Spatial Sciences. Dr. Ashton Shortridge and Ms. Sharon Ruggles

v

gives me help when I have any question in Geography and assist me when I need departmentâs
support. I specially thank my wife, Hsun-Yi Hsieh, and my parents. Your support at any time is
the most important in enhancing my progress.
My research was funded by the US DOE Great Lakes Bioenergy Research Center (DOE
Office of Science, DE-FC02-07ER64494 & DOE Office of Energy Efficiency and Renewable
Energy, DE-AC05-76RL01830).

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. xi
LIST OF FIGURES .......................................................................................................... xii
KEY TO ABBREVIATIONS ............................................................................................xv
CHAPTER 1: DISSERTATION INTRODUCTION ..........................................................1
1.1 OVERVIEW ......................................................................................................1
1.1.1 Global carbon cycle and climate changes ...........................................1
1.1.2 The development of biofuel and its impacts on global carbon
balance ...............................................................................................4
1.1.3 The drivers of carbon budget in biofuel cropland ecosystem after
land use change .................................................................................5
1.1.4 Major carbon processes and pools of ecosystems...............................7
1.1.5 The importance of grassland and cropland ecosystems ......................9
1.2 THE CONCEPTUAL FRAMEWAORK ........................................................10
1.3 OBJECTIVES ..................................................................................................13
1.3.1 Objectives of study ...........................................................................13
1.3.2 Hypotheses ........................................................................................15
1.4 STUDY AREA ................................................................................................17
1.5 EXPERIMENTAL DESIGN AND SCHEDULE ............................................19
LITERATURE CITED ..........................................................................................22
CHAPTER 2: INTERANNUAL VARIATIONS OF SOIL RESPIRATION AND ITS
RESPONSE TO MICROCLIMATE .................................................................................28
ABSTRACT ...........................................................................................................28
2.1 INTRODUCTION ...........................................................................................29
2.2 METHODS ......................................................................................................31
2.2.1 Study area..........................................................................................31
2.2.2 Experimental design and schedule ....................................................31
2.2.3 The climatic and microclimatic measurements and growth season
identification ....................................................................................32
2.2.4 The total and heterotrophic soil respiration measurements ..............34
2.2.5 Vegetation Index data .......................................................................36
2.2.6 The temperature-water-vegetation (TWV) models & Ra:Rh ratio ....36
2.2.7 Data analysis .....................................................................................37
2.3 RESULTS ........................................................................................................38
2.3.1 Climatic and microclimatic variables and the growing seasons .......38
2.3.2 The Vegetation Indices: NDVI and EVI ...........................................42
2.3.3 CROP and LUH effects on total (Rs) and heterotrophic soil
respiration (Rh) ................................................................................42

vii

2.3.3.1 The effects of CROP, LUH and their interactions on Rs
and Rh .................................................................................42
2.3.3.2 The effects of CROP in different LUH ...............................43
2.3.3.3 The effects of LUH in different CROP ...............................44
2.3.4 Soil respiration regulators .................................................................49
2.3.5 Temperature sensitivity (Q10) of soil respiration ..............................54
2.3.6 Temperature-Water-Vegetation (TWV) multiple variable models
for soil respiration ...........................................................................54
2.3.7 Ra:Rh ratio and the root contribution of soil respiration ...................55
2.4 DISCUSSION ..................................................................................................58
2.4.1 The major biophysical drivers of soil respiration .............................58
2.4.2 The effects of crop type (CROP) and land use history (LUH) .........63
2.4.3 The root contribution (RC) to total soil respiration across years ......64
2.4.4 The responses of Ra and Rh to future climate scenarios ...................65
2.5 CONCLUSIONS..............................................................................................67
APPENDIX ............................................................................................................69
LITERATURE CITED ..........................................................................................75
CHAPTER 3: SEASONAL PATTERNS OF SOIL RESPIRATION AND ITS
RESPONSE TO MICROCLIMATE .................................................................................79
ABSTRACT ...........................................................................................................79
3.1 INTRODUCTION ...........................................................................................80
3.2 METHODS ......................................................................................................82
3.2.1 Study area..........................................................................................82
3.2.2 Experimental design and schedule ....................................................82
3.2.3 The climatic and microclimatic measurements and growth season
identification ....................................................................................84
3.2.4 Total and heterotrophic soil respiration measurements ....................85
3.2.5 Vegetation index data .......................................................................86
3.2.6 The temperature-water-vegetation (TWV) models ...........................87
3.2.2 Seasonality of root contribution (RC) to total soil respiration ..........87
3.3 RESULTS ........................................................................................................88
3.3.1 Temporal variations of biophysical factors.......................................88
3.3.1.1 Climate and microclimate ...................................................88
3.3.1.2 Vegetation indices ...............................................................92
3.3.1.3 Soil respiration ....................................................................93
3.3.2 The seasonality of the root contribution (RC) to total soil
respiration ........................................................................................95
3.4 DISCUSSION ..................................................................................................97
3.4.1 How seasonal patterns of biophysical drivers affect the seasonal
patterns of soil respiration ...............................................................97
3.4.2 The seasonality of Ra and Rh in the dry and hot year .....................100
3.4.3 The seasonality of Ra and Rh in humid spring and cool summers ..102
3.4.4 Seasonal variations in RC ...............................................................101
3.5 CONCLUSIONS............................................................................................106
LITERATURE CITED ........................................................................................108

viii

CHAPTER 4: ECOSYSTEM RESISTANCE AND RESILIENCE TO SEVERE
DROUGHTS: A BAYESIAN MODELING OF SOIL RESPIRATION ........................112
ABSTRACT .........................................................................................................112
4.1 INTRODUCTION .........................................................................................113
4.2 METHODS ....................................................................................................116
4.2.1 Study area........................................................................................116
4.2.2 Experimental design and schedule ..................................................116
4.2.3 The climatic and microclimatic measurements and growth season
identification ..................................................................................117
4.2.4 The total and heterotrophic soil respiration measurements ............119
4.2.5 The measurements of Ta, PRCP, Ts, VWC, NDVI, EVI, Ra and Rh119
4.2.6 The concept of Bayesian modeling .................................................120
4.2.7 The estimates of prior models .........................................................121
4.2.8 The equation of likelihood ..............................................................122
4.2.9 Calculation of estimations of posterior models ..............................122
4.2.10 Estimations of Î˛ and LR20 of Ra....................................................123
4.3 RESULTS ......................................................................................................124
4.3.1 Bayesian models for the R-Ts relationship .....................................124
4.3.2 The differences of Î˛ and LR20 between CROP and LUH ...............126
4.3.3 The distribution of Î˛ and LR20 of Ra and Rh and their trajectories
across years....................................................................................128
4.4 DISCUSSION ................................................................................................132
4.4.1 The distribution of Î˛ and LR20 across CROP and LUH
combinations .................................................................................132
4.4.2 The stability of Î˛ and LR20 in the severe spring-summer drought in
the Ref field ...................................................................................132
4.4.3 Different responses of Î˛-LR20 on Rs and Rh after severe springsummer drought .............................................................................133
4.4.3.1 The resistance of Rs-Ts and Rh-Ts relationships to the
drought ............................................................................133
4.4.3.2 The changes of Rs-Ts and Rh-Ts relationships in the
recovery years ..................................................................135
4.4.3.3 The systematic responses of the R-T relationship to the
severe drought ..................................................................136
4.5 CONCLUSIONS............................................................................................137
APPENDIX ..........................................................................................................139
LITERATURE CITED ........................................................................................141
CHAPTER 5: HOW THE NET ECOSYSTEM EXCHANGE (NEE) OF CARBON
DIOXIDE AND ITS PARTITIONING RESPOND TO THE BIOPHYSICAL
VARIABLES DURING GROWING SEASON ..............................................................146
ABSTRACT .........................................................................................................146
5.1 INTRODUCTION .........................................................................................147
5.2 METHODS ....................................................................................................149
5.2.1 Study area........................................................................................149
5.2.2 Experimental design and schedule ..................................................149

ix

5.2.3 The calculations of NEE, GPP and Reco by eddy-covariance
approach ........................................................................................150
5.2.3.1 The measurements of eddy-covariance and microclimatic
data ...................................................................................150
5.2.3.2 Data processing and gap-filling .........................................152
5.2.3.3 Estimation of NEE, GPP and Reco .....................................153
5.2.4 Estimations of Reco and Rs ..............................................................153
5.2.4.1 The TWV model and its variables .....................................153
5.2.4.2 Data processing and gap filling .........................................154
5.2.4.3 The components of ecosystem respiration and their
calculation ........................................................................155
5.3 RESULTS ......................................................................................................156
5.3.1 The seasonal variations of NEE, GPP, Reco and Rs.........................156
5.3.2 The mid-growing season NEE, GPP, Reco, Rabove and Rs among
CROP and LUH.............................................................................163
5.3.3 The mid-growing season Reco:GPP, Rs:GPP and Rs:Reco ratios
among CROP and LUH .................................................................163
5.3.4 The shift of mid-growing season Reco:GPP and Rs:Reco ratios across
years...............................................................................................165
5.3.5 The shift of annual NEE, GPP, Reco and Rs among CROP and LUH
across years....................................................................................169
5.4 DISCUSSION ................................................................................................173
5.4.1 The seasonality of NEE, GPP & Reco between annual and perennial
crops ..............................................................................................173
5.4.2 The mid-growing season Reco:GPP and Rs:Reco ratios ....................174
5.4.3 The interannul trend of mid-GS Rs:Reco ratios ................................175
5.4.4 The limitation of the TWV model ..................................................176
5.5 CONCLUSIONS............................................................................................177
LITERATURE CITED ........................................................................................179
CHAPTER 6: CONCLUSIONS AND IMPLEMENTATIONS .....................................183
LITERATURE CITED ........................................................................................190

x

LIST OF TABLES

Table 1.1. Soil texture and soil properties of the seven experiment sites before the land
use conversion ...............................................................................................19
Table 2.1. The root contribution (RC) to total soil respiration ..........................................56
Table S.2.1. Monthly Climate and microclimate ...............................................................70
Table S.2.2. The Pearsonâs correlation between soil respiration (R or log10R) and its
controls (Ts, VWC, NDVI and EVI ................................................................71
Table S.2.3. The exponential model of total soil respiration (Rs)/heterotrophic soil
respiration (Rh) and soil temperature (Ts) and the temperature sensitivity
(Q10) during 2011 and 2014 ..........................................................................72
Table S.2.4. The two-level interactive TWV models for Rs and Rh ..................................73
Table S.2.5. The effects of CROP and LUH on total (Rs) and heterotrophic soil
respiration (Rh) ..............................................................................................74
Table 4.1. The mean and standard deviation of log10 soil respiration by year ................124
Table 4.2. The values of LR20 and Î˛ in Rs-Ts and Rh-Ts relationships ............................127
Table S.4.1. The root contribution of soil respiration (RC) for each site in each year ....140
Table 5.1. The minimum NEE, maximum GPP and maximum Reco among sites across
years ............................................................................................................158
Table 5.2. The Reco:GPP, Rs:GPP and Rs:Reco ratios among CROP and LUH over four
years ............................................................................................................167

xi

LIST OF FIGURES

Figure 1.1. The major anthropogenic CO2 emissions and the partitioning among three
major carbon reservoirs...................................................................................2
Figure 1.2. A simplified schematic of the global carbon reservoirs and the annual fluxes .3
Figure 1.3. The conceptual framework of the major carbon processes between
ecosystem and the atmosphere, their biophysical drivers and how human
management affect them ...............................................................................11
Figure 1.4. The theoretical framework on the impact of LUH and CROP on the net
ecosystem exchange (NEE) of CO2 and soil respiration (Rs) .......................12
Figure 1.5. The location of research sites and experimental design ..................................18
Figure 1.6. The experimental design and schedule of agricultural management...............21
Figure 2.1. The measurements of total (Rs) and heterotrophic soil respiration (Rh) .........35
Figure 2.2. Temporal changes in nine-day moving averages of daily mean air
temperature (Ta) ............................................................................................40
Figure 2.3. Annual precipitation by water year from 1989 to 2016 ..................................41
Figure 2.4. The seasonal patterns of NDVI and EVI .........................................................42
Figure 2.5. The CROP effect on total (Rs) and heterotrophic soil respiration (Rh) in
fields with different LUH..............................................................................46
Figure 2.6. The LUH effect of total (Rs) and heterotrophic soil respiration (Rh) in
different crop types .......................................................................................47
Figure 2.7. Linear regression of log10 soil respiration (log10R) on soil temperature (Ts) ..50
Figure 2.8. Linear regression of soil respiration (R) on soil moisture (VWC) ..................51
Figure 2.9. Linear regression of soil respiration (R) on MODIS vegetation indices
(NDVI and EVI) ...........................................................................................52
Figure 2.10. Residuals of the two-level interaction TWV models ....................................55

xii

Figure 2.11. The root contribution (RC) to total soil respiration across CROP (crop
types) .............................................................................................................56
Figure 2.12. The root contribution to total soil respiration across LUH (land use
histories)........................................................................................................57
Figure 2.13. The root contribution to total soil respiration over 2011-2014 .....................58
Figure 2.14. Schematic depiction of the impacts of three major forces on soil
respiration at different temporal scales .........................................................61
Figure 3.1. Temporal changes of soil respiration and environmental variables across
sites and years ...............................................................................................91
Figure 3.2. Temporal changes of total (Rs) and heterotrophic soil respiration (Rh) ..........94
Figure 3.3. Temporal changes of root contribution (RC) to total soil respiration and the
correlation between RC and VWC ...............................................................96
Figure 3.4. The seasonality of soil respiration and their drivers in 2011...........................98
Figure 3.5. The seasonality of Rs and Rh across four years ...............................................99
Figure 3.6. The responses of Ra and Rh to VWC .............................................................105
Figure 3.7. The changes of C3:C4 biomass ratio and ecosystem water-use efficiency
(eWUE) .......................................................................................................106
Figure 4.1. The probabilities of log10 soil respiration in each year..................................125
Figure 4.2. The distributions of log10 soil respirations temperature sensitivities (Î˛) and
the logarithm soil respiration rates at 20 Â°C (LR20) ....................................127
Figure 4.3. The distributions of the log10 soil respiration temperature sensitivities (Î˛)
and the log10 soil respiration rates at 20 Â°C (LR20) across years .................129
Figure 4.4. The inter-annual shifts of the log10 soil respiration temperature sensitivities
(Î˛) and the log10 soil respiration rates at 20 Â°C (LR20) of Ra and Rh for
each site across all years .............................................................................130
Figure 4.5. The inter-annual shifts of the log10 soil respiration temperature sensitivities
(Î˛) and the log10 soil respiration rates at 20 Â°C (LR20) of Ra and Rh for
different crops and land use histories across all years ................................131
Figure 5.1. Eddy-covariance tower with instruments for carbon, water and energy cycle
monitoring ...................................................................................................152

xiii

Figure 5.2. The seasonal and interannual variations of daily NEE, GPP, Reco, Rabove, and
Rs among sites during 2011-2014 ...............................................................159
Figure 5.3. The mid-growing season (21 May â 20 August) average NEE, GPP, Reco,
Rabove, and Rs across for each of four years .................................................164
Figure 5.4. The Reco:GPP, Rs:GPP and Rs:Reco ratios among CROP and LUH over four
years ............................................................................................................166
Figure 5.5. The mid-growing season Reco:NPP and Rs:Reco ratios across years in
different sites ...............................................................................................168
Figure 5.6. The cumulative annual NEE, GPP, Reco and mid-growing season Rs across
years in different sites .................................................................................171
Figure 5.7. The linear regression models of Reco and mid-growing season Rs to annual
GPP .............................................................................................................172
Figure 6.1. The framework of coupled human and nature systems (CHANS) on the
impacts that bioenergy development on managed ecosystems ...................184

xiv

KEY TO ABBREVIATIONS

9d-Ta

Nine day moving average of daily mean air temperature

AIC

Akaike Information Criterion

AGR

Corn-Soybean Rotational Conventional Agricultural Field

Î˛ Ra

The temperature sensitivity of logarithm autotrophic soil respiration

Î˛ Rh

The temperature sensitivity of logarithm heterotrophic soil respiration

Î˛ Rs

The temperature sensitivity of logarithm total soil respiration

C

Corn

CO2e

Carbon dioxide equivalent, a standard unit for measuring carbon footprints

CROP

Crop types

CRP

Conservation Reserve Program

DLUC

Direct land use change

EC

Eddy-covariance

ECE

Extreme climatic event

ET

Evapotranspiration

EVI

Enhanced Vegetation Index

eWUE

Ecosystem water-use efficiency, calculated by GPP/ET

F:B ratio

Fungi to bacteria ratio

GHG

Greenhouse gas

GPP

Gross primary production

GWI

Global warming impact

GWP

Global warming potential

xv

ILUC

Indirect land use change

LR20

Log10 soil respiration rate at 20 Â°C

LR 20 R

a

LR 20 R

h

LR 20 R

s

The logarithm autotrophic soil respiration rate at 20Â°C
The logarithm heterotrophic soil respiration rate at 20Â°C
The logarithm total soil respiration rate at 20Â°C

LUC

Land use conversion/change

LUH

Land use history

MODIS

Moderate Resolution Imaging Spectroradiometer

NDVI

Normalized Difference Vegetation Index

NEE

Net ecosystem exchange

nir

Near infrared red

PAR

Photosynthetically active radiation

PN

Photosynthesis

R

Respiration, including different types of respiration. i.e., Rs and Ra in Ch.1 and 2
and Ra, Reco, Rh, and Rs in Ch. 3.

Ra

Autotrophic soil respiration

RC

Root contribution to total soil respiration (Ra/Rs ratio)

Reco

Ecosystem respiration

Ref

Reference site

Rh

Heterotrophic soil respiration

PAR

Photosynthetically active radiation

Pr

Prairie mixture

PRCP

Precipitation

Rs

Total soil respiration
xvi

SOC

Soil organic carbon

Sw

Switchgrass

Ts

Soil temperature

uz

Vertical wind speed

VWC

Volumetric soil water content

VI

Vegetation indices, including NDVI and EVI

xvii

CHAPTER 1
DISSERTATION INTRODUCTION

1.1 OVERVIEW

1.1.1 Global carbon cycle and climatic changes
Carbon, the element constructing the chemical frame of most living things on the Earth,
plays an important role in biogeochemical cycles and global climate change. Carbonâs
distribution, form and presence in reservoirs, the water cycle, energy flow and other cycles is
modified by physical, chemical and biological processesâall of which affect climate, from the
local to the global scale. Atmospheric greenhouse gases (GHGs), such as CO2, N2O and CH4,
can absorb radiation and trap heat when radiation penetrates the atmosphere. Atmospheric GHG
concentrations can thus affect the surface energy balance, change the surface temperature and
water distribution and can alter climate. The anthropogenic carbon release from fossil fuels and
land use change (LUC) has been recognized as the most important cause of atmospheric CO2
increase (Fig. 1.1, IPCC, 2013). The increase of atmospheric CO2 concentration is believed to
majorly increase total radiative forcing and induce global warming (Fig. 1.2, IPCC, 2013). The
carbon fluxes between an ecosystem and atmosphere are crucial for estimating the dynamics of
the global atmospheric CO2 concentration. However, there is still much uncertainty in how
ecosystem carbon budgets respond to climate. We need more research to understand the
processes of carbon exchange in different ecosystem-atmosphere interfaces under different
climate patterns and future climate changes, and how ecosystems and the atmosphere exchange
carbon after LUC at different spatial and temporal scales.

1

Figure 1.1. The major anthropogenic CO2 emissions and the partitioning among three
major carbon reservoirs. The top panel shows the dynamics of anthropogenic CO2 emissions
from fossil fuel combustion (gray) and land use change (khaki) after 1750. Part of the carbon
sinks into land (green) or oceans (dark blue) and part remains in the atmosphere (light blue),
which increases the atmospheric CO2 concentration. The emission values were estimated from
four models and data during 1750-1850 while the partitioning data is estimated from the
combination of multiple-source atmospheric CO2 concentration data, ice cores and models
(source: IPCC, 2013).

2

Figure 1.2. A simplified schematic of the global carbon reservoirs and annual fluxes. The
units of the carbon stocks in major carbon reservoirs (PgC) and the annual fluxes between
reservoirs (PgC yr-1). The red arrows and numbers denote the annual anthropogenic carbon
fluxes and accumulated changes over the industrial period (1750-2011) while the black arrows
and numbers show annual natural fluxes and the stocks of the natural carbon reservoirs before
Industrial Era (after 1750) and assumed stable. The largest two fluxes between atmosphere and
other reservoirs are the land flux and the ocean flux that is stable before 1750. The annual
contribution of anthropogenic carbon to the atmosphere is 7.8 and 1.1 PgC yr-1 by fossil fuel and
cement production and land use change, respectively. Part of them were neutralized by ocean and
land sink. The net annual increase of atmospheric carbon is 4 PgC yr-1 (source: IPCC, 2013).

3

1.1.2 The development of biofuel and its impacts on global carbon balance
Bioenergy crops have the potential to mitigate GHG emissions, as their net climate
forcing is lower than the equivalent energy derived from fossil fuels. This reduction in the
amount of GHG emissions, compared with those from fossil fuels, is denoted as âfeedstock
carbon uptake creditâ or âfossil fuel offset creditâ (Searchinger et al., 2008; Gelfand et al.,
2013). However, there remain many uncertainties and ongoing debates on the âclimate
neutralityâ of biofuel crops (Hansen, 1993; Roberston et al., 2008; Searchinger et al., 2008;
Robertson et al., 2011). Land use change from previously uncultivated fields, food-based farms
or marginal lands into biofuel plantations, for example, can result in large CO2 emissions during
or after the conversion that increase carbon debt (Davidson and Ackerman, 1993; Guo and
Gifford, 2002; Searchinger et al., 2008; Kim et al., 2009; Gelfand et al., 2011; Zona et al., 2013)
and may offset the GHG benefit of biofuels. Several studies concluded that the net GHG
emissions of biofuel croplands are positive after the direct LUC (DLUC) and indirect LUC
(ILUC) (Liu, 2015) are counted. However, some research has demonstrated that the management
practices also play a major role. Robertson et al. (2000) reported the significant difference in
GHG emission between till and no-till croplands and between different successional stages of
forest regeneration, while Gelfand et al. (2011) showed that the tillage practices (68 Mg
CO2eďha-1) tripled the carbon dioxide equivalent (CO2e) of no-till practice (222) after land use
conversion if the alterations in C stocks were counted, which changed N2O and CH4 fluxes and
the fossil fuel offset credit. Ruan et al. (2013) also made evident that soil CO2 emissions from
conventional tillage fields were 1.2 times those in non-tillage fields and 3.1 times those in nonconversion fields. The global warming impact (GWI) of converted soybean fields with nontillage and conventional tillage fields were 15 and 6.37 Mg CO2eďha-1 higher than that at non-

4

converted fields. Thus, management practices directly affect soil carbon and nitrogen cycles and
change the GHG emissions under different climate patterns at short-term and long-term scales.
Further understanding how the major carbon processes in biofuel agricultural ecosystems
respond to climate change and variability can help us evaluate the effects of biofuel-induced
LUC. More so, it is required in order to improve the estimate of the consequences of biofuelinduced LUC based on the Earth System Models (ESMs).

1.1.3 The drivers of carbon budget in biofuel cropland ecosystem after land use change
The amount of GHG emission due to LUC, also known as carbon debt, will diminish
over time before reaching a new equilibrium in cropping systems (Fargione et al., 2008; Mitchell
et al., 2012). The duration that an established biofuel crop system needs to be in production in
order to âpaybackâ the carbon debt after LUC can range from years to over a century. Some
studies estimated the payback time using mathematical models that contained multiple variables
such as land use history, crop type, and management practices (Gelfand et al., 2011; Mitchell et
al., 2012). However, the corresponding mechanisms to understand the responses of GHG
emissions are complex and uncertain (Robertson et al., 2011; Anderson-Teixeira et al., 2013),
and hence we need more evidence from field studies to understand these mechanisms.
Land cover directly or indirectly affects soil erosion, water drainage and soil carbon and
nutrient processes, which are important factors driving the soil carbon cycle (Montgomery, 2007;
Love and Nejadhashemi, 2011). The carbon stocks prior to LUC and the carbon dynamics during
and after LUC determine how the soil acts as either a carbon source or a sink after LUC (Guo
and Gifford, 2002; Searchinger et al., 2008). Land cover and land use type, including vegetation
type and prior (land use history, LUH) and after (crop type, CROP) conversion, therefore, alter
the carbon stocks and dynamics in soils and vegetation. However, LUH and CROP may alter

5

ecosystem carbon dynamics through different biogeochemical processes. While the processes
and potential of carbon accumulation in reforestation from abandoned agricultural lands are well
known (Compton et al., 1998; Post and Kwon, 2000; Silver et al., 2000; Paul et al., 2002), the
climate implications that result from converting agricultural or conservation land into biofuel
croplands are not well understood. We need more studies to understand how the expansion or
intensification of bioenergy cropping systems around the world impact GHG emissions.
Current and historical land use can modify soil carbon, nutrient and soil pools, as well as
microbial community composition, affecting plant growth and soil microbial activities and
contributing to soil biogeochemical processes. The LUH, therefore, affects the carbon fluxes
among soil carbon pools and the surrounding atmosphere (Lugo and Brown, 1993; Compton et
al., 1998; Post and Kwon, 2000; Silver et al., 2000; Paul et al., 2002; Kasel and Bennett, 2007).
Relatively nutrient-rich soil (e.g., a grassland in the Conservation Reserve Program, CRP) can
respond very differently from nutrient-impoverished soil (e.g., fields that have been subject to
long-term cultivation of crops, AGR). In the United States, the repatriation of CRP, the
conversion of food-based farms (AGR) and marginal lands to biofuel crops, has been a proposed
solution to the growing demand for liquid biofuel (Robertson et al., 2011; Duke et al., 2013).
The reduction of soil carbon and nitrogen content after AGR on the Great Plains has long been
reported in rangeland and prairie soil (Campbell and Souster, 1982; Tiessen et al., 1982; Aguilar
et al., 1988; Bowman et al., 1990). Comparatively, converting farmland into CRP grassland
generally recovers soil carbon and nitrogen (Burke et al., 1995; Reeder et al., 1998). The
understanding of the response of ecosystem carbon flux to climate patterns between CRP and
AGR LUH after LUC can help the trajectory of GHG emissions associated with land conversion
for biofuel production, and assist the policy making of bioenergy development.

6

Crop types (CROP) of a biofuel agricultural system, after LUC, may regulate the amount
and timing of carbon gain/loss of the ecosystem via photosynthesis, carbon allocation strategy
and soil respiration (Lugo and Brown, 1993; Raich and Tufekcioglu, 2000; Paul et al., 2002; Zak
et al., 2003; Dias et al., 2010; Barron-Gafford et al., 2011; Han et al., 2014). Grain-based biofuel
crops and cellulosic biofuel crops are two major liquid biofuel crops that produce a significant
amount of ethanol nowadays and in the near future, respectively. Grain-based corn biofuel is the
major type of biofuel in the USA nowadays because corn grains contain mostly starch and can be
easily converted to ethanol. However, the potential environmental impacts, such as soil erosion
and nutrient runoff, of the intensive grain-based crops have been noted as a drawback of corn
biofuel (Robertson et al., 2011). Perennial cellulosic biofuels, such as switchgrass and mixed
prairie grasses, are considered alternative biofuel feedstock because they provide better
ecosystem services (Farrell et al., 2006), such as reducing nitrogen leakage and fertilizer
subsidies (Duke et al., 2013). Perennial cellulosic biofuel crops can recover the soil organic
carbon (SOC) by establishing perennial vegetation without tillage and can further mitigate the
soil erosion and increase SOC stocks in roots (West and Marland, 2003). These alternatives are
currently considered as target biofuel crops by the European Union (Roberston et al., 2008).

1.1.4 Major carbon processes and pools of ecosystems
Many studies explored the carbon fluxes and storage in ecosystems (Chen et al., 2014).
The dynamics of carbon in an ecosystem are mainly determined by carbon gain (GPP) from the
photosynthesis (PN) and carbon loss from ecosystem respiration (Reco) (Eq. 1-1). Carbon, energy
and water participate in the reactions. The crop captures CO2 and solar energy and transpires
water to atmosphere by photosynthesis. The fixed carbon and energy are allocated to different
requirements, such as basal metabolism, growth and development of stems, roots, leaves, and the

7

production of secondary compounds and seeds, depending on the phenology, the signal of
environmental condition they receive and their survival strategies. Some carbon is allocated to
belowground tissues or chemicals, such as coarse and fine roots, root exudates or root symbionts
by the plant itself, whereas some eventually enter soil carbon pools after senescence, such as
plant residues, litter, dead roots and root exudates. Soil microbial respiration is fueled by organic
carbon, which ultimately comes from photosynthesis.
Organisms in agricultural ecosystems, whether they are producers, consumers or
decomposers, get energy by oxidizing organic matter and releasing CO2 when they respire. The
relative rates of photosynthesis and respiration determine the sink or source balance of carbon in
an ecosystem and, thereby, net flux of carbon dioxide from the ecosystem to the atmosphere.
This is denoted as the net ecosystem exchange (NEE) of CO2:
đđ¸đ¸ = đđđđ â đşđđ

Eq. 1-1

where gross primary production (GPP) is the amount of carbon fixed from the atmosphere into
the ecosystem via photosynthesis (PN). Ecosystem respiration (Reco) is the sum of carbon loss
from the ecosystem/soil to the atmosphere, and NEE is the difference between Reco and GPP.
Some ecologists use net ecosystem production (NEP), which is the negative value of NEE, more
often than NEE. This is because NEP determines the carbon flux from the atmosphere to the
ecosystem, which better represents the ecosystem-centric perspective. One can present the NEP
as the following equation.
đđ¸đ = đşđđ â đđđđ

Eq. 1-2

The total ecosystem respiration can be partitioned into aboveground respiration (Rabove)
and belowground respiration (a.k.a. soil respiration, Rs) (Eq. 1-3).
đđđđ = đđđđđŁđ + đđ 

Eq. 1-3

8

Belowground respiration can also be partitioned into autotrophic and heterotrophic
respiration. In my dissertation, I focus on soil respiration. The autotrophic soil respiration (i.e.,
by plants) is denoted as Ra, while heterotrophic soil respiration (mainly due to microorganisms)
is denoted as Rh.
đđ  = đđ + đâ

Eq. 1-4

Soil respiration is an important component in the global carbon cycle. First, soil stores
1500-2400 Pg C globally, which is twice as much as the atmosphere (829 ďą 10 Pg C) and three
times as much as terrestrial vegetation (420-630 Pg C) (Fig. 1.2, IPCC, 2013). A small change of
soil carbon efflux due to widespread drivers, such as climate change and land use conversion,
can alter global carbon budgets, especially the atmospheric carbon dioxide concentration.
Second, soil respiration can be 60-90% of total ecosystem respiration (Goulden et al., 1996). The
variation of soil respiration can change the direction and the amount of carbon flux at local
scales. Third, soil carbon efflux responds sensitively and quickly to the climate pattern shifts,
climate-derived vegetation changes and agricultural management practices. Soil temperature, soil
water and vegetation condition are major drivers of soil respiration. The temperature and rainfall
patterns affect the temporal variation of soil respiration directly or indirectly via the vegetationderived carbon subsidy.

1.1.5 The importance of grassland and cropland ecosystems
Grassland ecosystems may contain different types of grasses and forbs, with some woody
plants and shrubs, and occupy 52.5 million square kilometers, or 40.5 percent of the global
terrestrial area excluding Greenland and Antarctica (Suttie et al., 2005). Grassland plays an
important role in the global carbon cycle due to its large extent and its large carbon storage
capacity. However, our understanding of grassland response to climate is still very limited due to

9

the highly variable vegetation types and compositions, temperature and precipitation patterns,
soil properties and disturbances, and the worldwide distribution of these ecosystems. Many
studies have shown that grasslands can be carbon sinks, at least during the growing season
(Belelli et al., 2007; Flanagan et al., 2002). In contrast, some studies demonstrated that grassland
ecosystems can vary between a sink or source of carbon inter-annually (Dugas et al., 1998),
depending on precipitation amounts and patterns. Even in the same ecosystem in the same year,
the NEE may be highly variable due to the seasonal and intra-annual temperature and rainfall
patterns or extreme climate events (Kim, 1992). The different responses between gross primary
production (GPP) and total ecosystem respiration (Reco) to the climate patterns and extremes may
vary and, sometimes, shift the carbon balance from sink to source.
Agricultural ecosystems are similar to grasslands and include variable vegetation types
that are distributed worldwide in different climate regimes and soils, store carbon in soil, and
experience frequent disturbances that keep the ecosystem in an early successional stage. They
exchange all three major GHGs with the atmosphere and are crucial for the mitigation of global
warming (Robertson et al., 2000). Cropping ecosystems are intensively managed by humans.
Agricultural management practices, such as crop type, the rotation of crops, cover crops,
fertilization, tillage, agricultural chemical applications, harvest, and the successional stage, not
only change the dominant vegetation, but also affect the physical and chemical status of soil and
the soil microbial community, which impacts the biogeochemical cycles and the global warming
potential (GWP) (Robertson et al., 2000).

1.2 THE CONCEPTUAL FRAMEWORK
Many studies explored the carbon stocks in different carbon pools and the fluxes among
the pools. The carbon fluxes are controlled by biophysical variables underlying different

10

mechanisms. The different responses of the carbon fluxes due to different mechanisms result in
the seasonal and annual fluctuations of carbon dioxide NEE and the relative contributions of
different carbon fluxes components (i.e., GPP:Reco, Ra: Rs ratios). The conceptual framework of
how the land use conversion affects carbon processes was presented in Fig. 1.3. I hypothesized
that the abiotic and biotic variables (i.e., temperature, water and nutrient availability, vegetation
types, leaf area index, phenology and soil microbial activities) determine the major carbon
processes (i.e., photosynthesis, ecosystem respiration, autotrophic and heterotrophic soil
respiration).

Figure 1.3. The conceptual framework of the major carbon processes between ecosystem
and the atmosphere, their biophysical drivers and how human management affect them.
GPP and Reco determine the net ecosystem exchange (NEE) of carbon dioxide between the
ecosystem and the atmosphere. Biotic and abiotic drivers affect photosynthesis and respiration,
indirectly altering NEE. Human management, such as land use change, changes the crop types
and soil properties and, therefore affect the biophysical variables and carbon processes. BGC:
biogeochemical cycle; GPP: gross primary production; Reco: ecosystem respiration; LAI: leaf
area index; LUH: land use history; CROP: crop types.

More specifically, managed land use conversion from different historical land use types
may affect the current crop types and soil nutrient content. Different crop types may have
different length of phenology stages and carbon allocation strategies, different water tolerance
abilities and, water use efficiency and, therefore, they may respond to climate differently. The

11

different soil carbon and nitrogen alter the available soil nitrogen to plants and soil microbes,
and, thus, affect plant growth and photosynthesis and carbon emission from soil (Fig. 1.4). In this
dissertation, I discuss how soil respiration responds to specific climate patterns and severe
droughts across different LUH and CROP at inter-annual (Ch. 2) and intra-annual (Ch. 3) scales.
I also develop Bayesian models to understand the R-Ts relationship change resulting from a
severe drought (Ch. 4). In Chapter 5, I explored the different responses of GPP, Reco and Rs to
the climate pattern and the severe drought in a particular year.

Figure 1.4. The theoretical framework on the impact of LUH and CROP on the net
ecosystem exchange (NEE) of CO2 and soil respiration (Rs). PN: photosynthesis; Ra:
autotrophic soil respiration; Rh: heterotrophic soil respiration; CRP: Conservation Reserve
Program brome grassland; AGR: conventional corn-soybean rotation agricultural farm; SOC:
soil organic carbon. CROP affects the biological reaction in the green square while LUH affects
the soil microbiological reactions in the orange square. Meteorological factors influence all
reactions.

12

1.3 OBJECTIVES
My dissertation aims to understand the temporal changes of net ecosystem exchange
(NEE) and soil respirations (R) in an annual grain (continuous corn, C), a monoculture perennial
grass (switchgrass, Sw), and a multicultural perennial grassland crop (prairie mixture, Pr), which
are all managed as biofuel crops on land converted either from the Conservation Reserve
Program (CRP) grassland or conventional agricultural farmland (AGR). The ecosystems with
different CROP and LUH may respond differently to climate patterns within and across years in
their gross primary production (GPP), ecosystem respiration (Reco), and the partitioning of
ecosystem respiration into total soil respiration (Rs), autotrophic respiration (Ra) and
heterotrophic soil respiration (Rh). The responses of respiration to long-term climate patterns,
short-term-regular climate events, and the resistance and the resilience to short-term events
reveals how carbon processes in these ecosystems respond to climate after land use conversion.

1.3.1 Objectives of Study
1. How do biophysical variables affect soil respiration (R) across years after land use
change:
(A) Examine the correlation between soil respiration and biophysical variables.
(B) Develop regression models of the effects of soil temperature, soil moisture and
vegetation index on soil respiration for each study site to predict autotrophic soil
respiration (Ra) and the root contribution (RC) to total soil respiration (Ra:Rs ratio).
(C) Study the interannual changes of growing-season Ra:Rs ratio after land use changes
and a severe drought in fields with different crop types (CROP) and land use histories
(LUH).

13

2. How does seasonal variation in biophysical variables affect soil respiration (R):
(A) Quantity how the seasonal patterns of temperature and soil moisture relate to the
seasonality of soil respiration (Rs and Rh)
(B) Study the shift of the relative importance of biophysical variables through the
growing season.
(C) Compare the seasonality of the Ra: Rh ratio in different CROP and LUH.
3. How does the soil respiration (R) â temperature (Ts) relationship change after the severe
drought?
(A) Develop an analysis method by the Bayesian approach to elucidate the relationship
between soil respiration (R) and its major driver (Ts). The temperature sensitivity (ď˘) of
log R and the log R at 20 oC (LR20) will be estimated based on the models.
(B) Understand how the R-Ts relationship shifts after a severe spring-summer drought in
different CROP and LUH combinations at the inter- and intra-annual scales.
(C) Recognize the immediate and prolonged effects of severe drought on the R-Ts
relationship in different CROP and LUH.
4. How do the various carbon fluxes respond to the changes of the major drivers?
(A) Explore the predominant regulators of the major carbon fluxes.
(B) Examine how the seasonal patterns of GPP:Reco, Rabove:Rs, and Ra:Rh ratios are
affected by different microclimate patterns, LUH and CROP.
(C) Study the response of important ecological processes, such as GPP, Reco, Rabove, Rs,
Ra and Rh to different climatic events (e.g. spring-summer drought), and how these
processes alter the NEE of CO2.

14

1.3.2 Hypotheses
The estimate of carbon stocks and fluxes were complex and multifaceted, since different
drivers alter carbon flow differently, underlying different mechanisms. Biophysical variables on
soil respiration, ecosystem respiration and GPP vary across different biomes and climate regimes
(Chen et al., 2014). The following is my hypotheses.
1. How do biophysical variables affect soil respiration (R) across years after land use
change:
(A) Soil respiration exponentially correlates to soil temperature and vegetation indices,
but is not significantly correlated to soil water content.
(B) Total and heterotrophic soil respiration have different parameters to biophysical
variables in different combinations of CROP and LUH fields.
(C) Perennial crops have increasing root contributions to total soil respiration, while
annual crop does not. The CRP fields have lower RC than AGR fields, but the RC values
in CRP fields increase year by year.
2. How does seasonal variation in biophysical variables affect soil respiration (R):
(A) The seasonality of soil temperature and vegetation index determines the seasonality
of Rs and Rh.
(B) Soil water content may be crucial for soil respiration during summer dry period.
(C) The RC may be high in early spring and mid-summer due to the high growth and
activities of roots while soil microbial activities respond to climate and photosynthesisderived carbon input may be delayed and gentle. Perennial and annual crop have different
RC seasonality due to different phenology among the crops.

15

3. How does the soil respiration (R) to temperature (Ts) relationship change with the severe
drought?
(A) The probability distributions of Rs and Rh are located differently across the ď˘-LR20
biplots. The Reference (Ref) field has high ď˘ and LR20 while corn fields have low LR20.
Other perennial crops have intermediate LR20 values.
(B) Both ď˘ and LR20 of Ra and Rh in perennial crops decrease in the drought year (2012)
and then rebound in the following years (2013 and 2014). The prolonged effects of
drought on the R-T relationship in perennial crop fields were remarked. Both ď˘ and LR20
of Ra and Rh in annual crops decrease in the drought year (2012). However, the prolonged
effect of Ra disappears.
4. How do the various carbon fluxes respond to changing drivers?
(A) GPP correlated strongly to photosynthetically active radiation (PAR) while Reco is
dominated by temperature.
(B) The higher GPP:Reco, Rabove:Rs, and Ra:Rh ratios occurred in early spring and midsummer since production was stronger than respiration, aboveground parts were quicker
than belowground parts and because roots were quicker to respond than soil microbes,
respectively. The different seasonality of these ratios between perennial and annual crops
are due to difference in phenology among the crop types.
(C) NEE decreases in order: Ref > annual crop > perennial crops. Ref, the late succession
grassland has weak carbon sequestration capability. NEE in annual crop fields are larger
than those in perennial crops fields due to shorter growing seasons (GS).
(D) The impacts of severe drought on GPP are stronger than it is on Reco, resulting in an
increase of NEE. The increasing drought events in the future may decrease the ability to

16

sequester carbon in cropping ecosystems, especially for the annual crop.
(E) The Rs:Reco ratio in perennial crop fields increase due to the development of dense
root systems while those in annual crop fields do not.

1.4 STUDY AREA
My experimental sites are located at the Great Lakes Bioenergy Research Center scale-up
fields of the Kellogg Biological Station (KBS, 42Â°40âN, 85Â°40âW), established in association
with the KBS Long-Term Ecological Research (LTER) site. The sites are located in the south
Michigan, USA (Fig. 1.2). with a humid continental (warm summer) climate (Dfa) (Peel et al.,
2007). The mean annual air temperature and mean annual precipitation at KBS are 10.1 Â°C and
1005 mm yr-1 (1981-2010), respectively (Robertson & Hamilton, 2015).

17

(A)

(B)
AGR

CRP

KBS

KBS

(C)

Figure 1.5. The location of research sites and experimental design. The research fields are
located in two areas that have different land use histories at Kellogg Biological Station (KBS)
scale-up sites in southwest Michigan (A & B). The two land use histories include CRP and AGR
(C). CRP: Conservation Reserve Program brome grassland at Marshall Farms; AGR:
conventional corn-soybean rotation agricultural farms at Lux Arbor Reserve. AGR converted to
three crop types, switchgrass, prairie mixture and corn. CRP changed to the same crops in AGR,
adding a reference site (Ref). The Ref were maintained as CRP brome grass without agricultural
practices.

18

1.5 EXPERIMENTAL DESIGN AND SCHEDULE
Seven experimental plots were divided between two locations, each with its own land use
history (LUH): (1) Conservation Reserve Program (CRP) grasslands at Marshall Farm; and (2)
corn-soybean rotation agricultural fields (AGR) at Lux Arbor Reserve. The CRP sites have been
in a monoculture of smooth brome grass (Bromus inermis Leyss) since 1987, while the AGR
fields have been under conventional corn-soybean rotation cultivation for several decades (Fig.
1.6). The soil texture at all sites is sandy loam, except for a sandy clay loam at one field.
However, soil carbon and nitrogen contents at the CRP sites were significantly higher than those
at the AGR sites before the land conversions (Table 1.1).
Table 1.1. Soil texture and soil properties of the seven experiment sites before the land use
conversion. In the land use history (LUH) column, CRP sites are grasslands of the Conservation
Reserve Program and AGR sites are conventional corn-soybean agricultural farms. Different
superscripts denote significant difference of nitrogen or carbon content using the t-test (p<0.05).
See text and Fig. 1.6 for treatment codes (from KBS LTER datatables;
https://lter.kbs.msu.edu/datatables)
Treatment

Area

LUH

CRP-C

(ha)
17

CRP

CRP-Sw

13

CRP

CRP-Pr

11

AGR-C

11

Sandy loam

(g kg-1 soil)
664Âą29 257Âą24
80Âą8

6.0Âą0.1

Bulk
density
(g cm-3)
1.58

Sandy loam

688Âą25

265Âą23

48Âą6

5.7Âą0.1

1.66

1.6d

18.5c

CRP

Sandy loam

697Âą53

245Âą42

58Âą12

6.3Âą0.2

1.59

1.7d

19.5c

AGR

Sandy loam

577Âą26

337Âą23

86Âą12

6.0Âą0.1

1.54

1.2a

12.2a

a

10.8b

AGR-Sw

14

AGR

AGR-Pr

13

AGR

Ref

9

CRP

Soil texture

Sandy loam
Sandy clay
loam
Sandy loam

Sand

Silt

Clay

Soil
pH

N

C

(g kg-1 soil)
2.0c 21.2c

651Âą31

271Âą24

79Âą12

6.1Âą0.1

1.79

1.1

495Âą32

360Âą31

146Âą10

5.8Âą0.1

1.69

1.4b

13.5a

583Âą28

342Âą25

75Âą8

6.2Âą0.1

1.56

1.9c

20.9c

The experiment was conducted at seven scale-up fields ranging in size from 9 to 17
hectares. Four fields are located in Marshall Farm (i.e., CRP sites) and three in Lux Arbor
Reserve (i.e., AGR sites). All sites, except the reference site (Ref), were sprayed with herbicide
at the end of 2008 to prepare the lands for soybean planting in 2009. The CRP and AGR sites

19

were then cultivated with either continuous corn (Zea mays, Dekalb DK-52), switchgrass
(Panicum virgatum) or a mixture of native prairie grasses that had been dominated by Canada
wild rye (Elymus Canadensis), little bluestem (Schizachyrium scoparium), Indian grass
(Sorghastrum nutans), big bluestem (Andropogon gerardii) and switchgrass (Panicum virgatum)
since 2010. One CRP grassland, dominated by brome grass, was not disturbed and retained as the
historical reference site (Gelfand et al., 2011; Zenone et al., 2011; Deal et al., 2013; Zenone et
al., 2013). I used an experimental code for these sites by abbreviating them as âLUH-CROPâ.
CRP-C, CRP-Sw and CRP-Pr represent the CRP sites that were converted to corn, switchgrass
and prairie mixture respectively, while AGR-C, AGR-Sw and AGR-Pr are AGR farms converted
to corn, switchgrass and prairie mixture. In 2010, when the crops were established, the perennial
crops (switchgrass and prairie) were accompanied by oats as nurse crop with and without
fertilization, respectively (Fig. 1.6). Switchgrass fields were applied with 55 kg N ha-1 (28%
liquid urea ammonium nitrate) on DOY 172. No other management practices were applied
beyond harvesting at the end of each growing season. No-till continuous corn was seeded in midMay with a one-time herbicide mix (Lumax, Atrazine 4L, Honcho Plus and (NH4)2SO4).
Phosphorus, potassium (P2O5 + K2O, 168.5 kg ha-1 on DOY 95, 2010) and nitrogen fertilizers
(112 kg N ha-1 on DOY 165 & 160 in AGR-C and CRP-C, respectively) were applied in 2010.
Phosphorus, potassium (P2O5 + K2O, 294 kg ha-1 on DOY 104, 2011) and nitrogen fertilizers
(liquid nitrogen168 kg N ha-1 on DOY 172, 2011) were applied (Bhardwaj et al., 2011; Zenone
et al., 2013).

20

Figure 1.6. The experimental design and schedule of agricultural management. Land use
before 2008 (land use history, LUH) included brome grass fields that were enrolled in the
Conservation Reserve Program (CRP) for 22 years, as well as conventional corn-soybean
rotation cultivations in the other fields (AGR). Research sites, except Ref, were planted in
soybean in 2009 to facilitate conversion. We planted three bioenergy cropsâcontinuous corn
(C), a monoculture of switchgrass (Sw) and a polycultural prairie mixture (Pr).

21

LITERATURE CITED

22

LITERATURE CITED

Aguilar, R., Kelly, E.F., Heil, R.D., 1988. Effects of cultivation on soils in northern Great Plains
rangeland. Soil Science Society of America Journal, 52, 1081-1085.
Anderson-Teixeira, K.J., Masters, M.D., Black, C.K., Zeri, M., Hussain, M.Z., Bernacchi, C.J.,
DeLucia, E.H., 2013. Altered belowground carbon cycling following land-use change to
perennial bioenergy crops. Ecosystems, 16, 508-520.
Barron-Gafford, G.A., Scott, R.L., Jenerette, G.D., Huxman, T.E., 2011. The relative controls of
temperature, soil moisture, and plant functional group on soil CO2 efflux at diel, seasonal,
and annual scales. Journal of Geophysical Research, 116, G01023.
Belelli Marchesini, L., Papale, D., Reichstein, M., Vuichard, N., Tchebakova, N., Valenini, R.,
2007. Carbon balance assessment of a natural steppe of southern Siberia by multiple
constaint approach. Biogeosciences, 4, 581-595.
Bowman, R.A., Reeder, J.D., Lober, R.W., 1990. Changes in soil properties in a central plains
rangeland soil after 3, 20, and 60 years of cultivation. Soil Science, 150, 851-857.
Burke, I.C., Lauenroth, W.K., Coffin, D.P., 1995. Soil organic matter recovery in semiarid
grasslands: Implications for the Conservation Reserve Program. Ecological Applications,
5, 793-801.
Campbell, C.A., Souster, W., 1982. Loss of organic matter and potentially mineralizable nitrogen
from Saskatchewan soils due to cropping. Canadian Journal of Soil Science, 62, 651-656.
Chen, J., John, R., Sun, G., McNulty, S., Noormets, A., Xiao, J., Turner, M.G., Franklin, J.F.,
2014. In: Azevedo, J.C., Perera A.H., Pinto, M.A. (Eds.). Forest Landscapes and Global
Change: Challenges for Research and Management. DOI 10.1007/978-1-4939-0953-7_6,
ÂŠ Springer Science+Business Media New York.
Compton, J.E., Boone, R.D., Motzkin, G., Foster, D.R., 1998. Soil carbon and nitrogen in a pineoak sand plain in central Massachusetts: Role of vegetation and land-use history.
Oecologia, 116, 536-542.
Davidson, E.A., Ackerman, I.L., 1993. Changes in soil carbon inventories following cultivation
of previously untilled soils. Biogeochemistry, 20, 161-193.
Deal, M.W., Xu, J., John, R., Zenone, T., Chen, J., Chu, H., Jasrotia, P., Kahmark, K.,
Bossenbroek, J., Mayer, C., 2013. Net primary production in three bioenergy crop systems
following land conversion. Journal of Plant Ecology, 6, 1-10.
Dias, A.T.C., van Ruijven, J., Berendse, F., 2010. Plant species richness regulates soil respiration
through changes in productivity. Oecologia, 163, 805-813.

23

Dugas, W.A., Heuer, M.L., Mayeus, H.S., 1999. Carbon dioxide fluxes over bermudagrass,
native prairie, and sorghum. Agricultural and Forest Meteorology, 93, 121-139.
Duke, C.S., Pouyat, R.V., Robertson, G.P., Parton, W.J., 2013. Ecological dimensions of
biofuels. Issues in Ecology, 17, 1-17.
Fargione, J., Hill, J., Tilman, D., Polasky, S., Hawthorne, P., 2008. Land clearing and the biofuel
carbon debt. Science, 319, 1235-1238.
Farrell, A.E., Plevin, R.J., Turner, B.T., Jones, A.D., O'Hare, M., Kammen, D.M., 2006. Ethanol
can contribute to energy and environmental goals. Science, 311, 506-508.
Flanagan, L.B., Wever, L.A., Carlson, P.J., 2002. Seasonal and interannual variation in carbon
dioxide exchange and carbon balance in a northern temperate grassland. Global Change
Biology, 8, 599-615.
Gelfand, I., Sahajpal, R., Zhang, X., Izaurralde, R.C., Gross, K.L., Robertson, G.P., 2013.
Sustainable bioenergy production from marginal lands in the US Midwest. Nature, 493,
514-517.
Gelfand, I., Zenone, T., Jasrotia, P., Chen, J., S.K.Hamilton, Robertson, G.P., 2011. Carbon debt
of conservation reserve program (CRP) grasslands converted to bioenergy production.
PNAS, 108, 13864-13869.
Goulden, M.L., Munger, J.W., Fan, S.M., Daube, B.C., Wofsy, S.C., 1996. Exchange of carbon
dioxide by a deciduous forest: Response to interannual climate variability. Science, 271:
1576-1578.
Guo, L.B., Gifford, R.M., 2002. Soil carbon stocks and land use change: a meta analysis. Global
Change Biology, 8, 345-360.
Han, G., Xing, Q., Luo, Y., Rafique, R., Yu, J., Mikle, N., 2014. Vegetation types alter soil
respiration and its temperature sensitivity at the field scale in an estuary wetland. PLoS
One, 9, e91182.
Hansen, E.A., 1993. Soil carbon sequestration beneath hybrid poplar plantations in the north
central Uniter States. Biomass and Bioenergy, 5, 431-436.
Houghton, R.A., House, J.I., Pongratz, J., van der Werf, G.R., DeFries, R.S., Hansen, M.C., Le
QuĂŠrĂŠ, C., Ramankutty, N., 2012. Carbon emissions from land use and land-cover change.
Biogeosciences, 9, 5125-5142.
IPCC, 2013: Climate Change 2013: The Physical Science Basis. Contribution of Working Group
I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. In:
Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y.
Xia, V. Bex and P.M. Midgley (Eds.), Cambridge University Press, Cambridge, United
Kingdom and New York, NY, USA, 1535 pp, doi: 10.1017/CBO9781107415324.

24

Kasel, S., Bennett, L.T., 2007. Land-use history, forest conversion, and soil organic carbon in
pine plantations and native forests of south eastern Australia. Geoderma, 137, 401-413.
Kim, H., Kim, S., Dale, B.C., 2009. Biofuels, land use change and greenhouse gas emissions:
some unexplored variables. Environmental Science and Technology, 43, 961-967.
Kim, J., Verma S.B., Clement, R.J., 1992. Carbon dioxide budget in a temperate grassland
ecosystem. Journal of Geophysical Research, 97 (D5), 6057-6063.
Liu, J., Mooney, H., Hull, V., Davis, S.J., Gaskell, J., Hertel, T., Lubchenco, J., Seto, K.C.,
Gleick, P., Kremen, C., Li, S., 2015. Systems integration for global sustainability. Science,
347(6225), 1258832-8.
Love, B.J., Nejadhashemi, A.P., 2011. Water quality impact assessment of large-scale biofuel
crops expansion in agricultural regions of Michigan. Biomass and Bioenergy, 35, 22002216.
Lugo, A.E., Brown, S., 1993. Management of tropical soils as sinks or sources of atmospheric
carbon. Plant and Soil, 149, 27-41.
Mitchell, S.R., Harmon, M.E., O'Connell, K.E.B., 2012. Carbon debt and carbon sequestration
parity in forest bioenergy production. Global Change Biology Bioenergy, 4, 818-827.
Montgomery, D.R., 2007. Soil erosion and agricultural sustainability. Proceedings of the
National Academic of Sciences of the USA, 104, 13268-13272.
Paul, K.I., Polglase, P.J., Nyakuengama, J.G., Khanna, P.K., 2002. Change in soil carbon
following afforestation. Forest Ecology and Management, 168, 241-257.
Peel, M.C., Finlayson, B.L., McMahon, T.A., 2007. Updated world map of the Ko Ěppen-Geiger
climate classification. Hydrology and Earth System Sciences Discussions, 4, 439-473.
Post, W.M., Kwon, K.C., 2000. Soil carbon sequestration and land-use change: Processes and
potential. Global Change Biology, 6, 317-328.
Raich, J.W., Tufekcioglu, A., 2000. Vegetation and soil respiration: Correlations and controls.
Biogeochemistry, 48, 71-90.
Reeder, J.D., Schuman, G.E., Bowman, R.A., 1998. Soil C and N changes on conservation
reserve program lands in the Central Great Plains. Soil and Tillage Research, 47, 339-349.
Robertson, G.P., Paul, R.R., Harwood, R.R., 2000. Greenhouse gases in intensive agriculture:
Contributions of individual gases to the radiative forcing of the atmosphere. Science, 289:
1922â1925.
Roberston, G.P., Dale, V.H., C., D.O., Hamburg, S.P., Melillo, J.M., Wander, M.M., Parton,
W.J., Adler, P.R., Barney, J.N., Cruse, R.M., Duke, C.S., Fearnside, P.M., Follett, R.F.,
Gibbs, H.K., Jose, G., Mladenoff, D.J., Ojima, D., Palmer, M.W., Sharpley, A., Wallace,

25

L., Weathers, K.C., Wiens, J.A., Wilhelm, W.W., 2008. Sustainable biofuels redux.
Science, 322, 49-50.
Robertson, G.P., Hamilton, S.K., Del Grosso, S.J., Parton, W.J., 2011. The biogeochemistry of
bioenergy landscapes: carbon, nitrogen, and water considerations. Ecological Applications,
21, 1055-1067.
Robertson, G.P., Hamilton, S.K., 2015. Long-term ecological research at the Kellogg Biological
Station LTER site: conceptual and experimental framework. In: Hamilton, S.K., Doll, J.E.,
Robertson, G.P. (Eds.), The Ecology of Agricultural Landscapes. Oxford University Press,
New York, pp. 1-32.
Ruan, L., Robertson, G.P., 2013. Initial nitrous oxide, carbon dioxide, and methane costs of
converting conservation reserve program grassland to row crops under no-till vs.
conventional tillage. Global Change Biology, 19, 2478-2489.
Searchinger, T., Heimlich, R., Houghton, R.A., Dong, F., Elobeid, A., Fabiosa, J., Tokgoz, S.,
Hayes, D., Yu, T.-H., 2008. Use of U.S. croplands for biofuels increases greenhouse gases
through emissions from land use change. Science, 319, 1238-1240.
Silver, W.L., Ostertag, R., Lugo, A.E., 2000. The potential for carbon sequestration through
reforestation of abandoned tropical agricultural and pasture lands. Restoration Ecology, 8,
394-407.
Suttie, J.M., Reynolds, S.G., Batello, C., 2005. Grasslands of the World. Food and Agriculture
Organization of the United Nations. Plant production and protection series, 0259-2025; No.
34. Rome. ISBN: 9251053375.
Tiessen, H., Stewart, J.W.B., Bettany, J.R., 1982. Cultivation effects on the amounts and
concentration of carbon, nitrogen, and phosphorus in grassland soils. Aaron Journal, 74,
831-835.
West, T.O., Marland, G., 2003. Net carbon flux from agriculture: Carbon emissions, carbon
sequestration, crop yield, and land-use change. Biogeochemistry, 63, 73-83.
Zak, D.R., Holmes, W.E., White, D.C., Peacock, A.D., Tilman, D., 2003. Plant diversity, soil
microbial communities, and ecosystem function: are there any links?. Ecology, 84, 20422050.
Zenone, T., Chen, J., Deal, M.W., Wilske, B., Jasrotia, P., Xu, J., Bhardwaj, A.K., Hamilton,
S.K., Robertson, P.G., 2011. CO2 fluxes of transitional bioenergy crops: effect of land
conversion during the first year of cultivation. GCB Bioenergy, 3, 401-412.
Zenone, T., Gelfand, I., Chen, J., Hamilton, S.K., Robertson, G.P., 2013. From set-aside
grassland to annual and perennial cellulosic biofuel crops: Effects of land use change on
carbon balance. Agricultural and Forest Meteorology, 182-183, 1-12.

26

Zona, D., Janssens, I.A., Aubinet, M., Gioli, B., Vicca, S., Fichot, R., Ceulemans, R., 2013.
Fluxes of the greenhouse gases (CO2, CH4 and N2O) above a short-rotation poplar
plantation after conversion from agricultural land. Agricultural and Forest Meteorology,
169, 100-110.

27

CHAPTER 2
INTERANNUAL VARIATIONS OF SOIL RESPIRATION AND ITS RESPONSE TO
MICROCLIMATE

ABSTRACT
In this study, I discovered that soil temperature and vegetation type affect soil respiration,
whereas soil moisture has weak or no effect. Soil respiration declined in 2012 as the springsummer drought depressed plant growth and soil microbial activity. The drought lasted for just a
couple of months but had an impact on soil respiration that lasted for more than one year. The
root contribution to total soil respiration deceased in AGR-Sw, CRP-Pr and AGR-Pr fields,
implying the drought affected autotrophic respiration more than heterotrophic respiration. The
reference field, which did not experience land use change, had the highest soil respiration
compared to all experimental treatments except for the CRP-Pr site in 2014, suggesting that the
multicultural prairie ecosystem with high soil carbon and nitrogen content had established their
extensive root systems within five years after its establishment. The annual crop (corn) had the
lowest soil respiration without an obvious trend, considering the drought impact. Perennial crops
(switchgrass and prairie mixture) had higher soil respiration than corn. In contrast to corn,
perennial cropsâ total soil respiration increased over the years after land use change. Pr and CRP
had higher rates of increase in Rs than Sw and AGR, respectively. The Rs and Rh in the CRP-Pr
site were higher or equal to those in Ref, demonstrating that well established root systems have
an association with increasing soil microbial activities. The CRP sites had higher soil respiration
than the AGR sites in 2011 due to high soil carbon and nitrogen content. I did not find a clear
decline in heterotrophic soil respiration, perhaps due to the legacy effect of the 2012 drought or
because the experimental period was too short to detect the change. My study clarified how

28

autotrophic and heterotrophic soil respiration responds to the impacts of a particular yearâs
climate pattern and severe drought, as well as the effects caused by land use conversion (i.e.,
different land use histories and crop types at different temporal scales).

2.1 INTRODUCTION

Soil respiration (Rs) plays an important role in the global carbon cycle, as soil contains a
huge amount of carbon (Post et al., 1982). Soil respiration is usually the major component of
total ecosystem respiration (Goulden et al., 1996; Longdoz et al., 2000) and is an important flux
to the atmosphere (Raich et al., 2002; Raich & Schlesinger, 1992; Ryan & Law, 2005;
Schlesinger & Andrews, 2000). It is also sensitive to climate and susceptible to climate change,
anthropogenic management and agricultural practices. Soil respiration responses to climate- and
human-induced environmental changes are crucial components in the overall ecosystem response
to global climate change.

Many previous studies explored how major biophysical drivers affect soil respiration and
how to better model these processes. Temperature, water availability, and carbon substrate
supply from canopy photosynthesis (Arrhenius, 1889; Curiel Yuste et al., 2003; Reichstein et al.,
2003; Lloyd & Taylor, 1994; vanât Hoff, 1885) affect biogeochemical or ecological processes
from individual to ecosystem levels (Hopkins et al., 2013). These drivers can be distinguished as
either climate-driven or microclimate-driven, where microclimate is defined here as the
conditions at a particular field site (e.g., due to soils, vegetation, and management). However,
there are still many gaps in our understanding of how soil respiration responds to the changing
climate and human activity, as the net balance of soil carbon is influenced by multiple abiotic
(e.g., temperature, moisture and nutrient availability) and biotic (e.g., crop types and soil

29

microbial composition) factors and their interactions. The relative importance of the biophysical
drivers that regulate soil respiration vary with climate regime and biome (Hanson et al., 2000).
The different components of soil respiration (i.e., Ra and Rh) may be confounding when they
have different magnitudes, timing and response durations to the drivers, and when the major
physiological and biogeochemical processes are different (Carbone et al., 2011). The situation
can be even more difficult to understand when anthropogenic influences, such as land use change
and agricultural practices (e.g., due to biofuel crop production), are considered.

In this chapter, I discuss how total soil respiration (Rs) and its components (Ra and Rh)
respond to variability in climate and microclimate at interannual scale. I examined the effects of
land use histories (LUH) and crop types (CROP) on autotrophic and heterotrophic soil
respiration. I hypothesized that annual crop had lower annual soil respiration compared to
perennial crops due to the shorter growing season. I also hypothesized that conventional
agricultural fields (AGR), which have low soil carbon and nitrogen content, have lower soil
respiration than CRP fields. I clarify the correlations between soil respiration (Rs and Rh) and the
major biophysical drivers (Ta, VWC, NDVI and EVI) and establish the temperature-watervegetation (TWV) models for Rs and Rh based on the correlations. The adjusted Rh and Ra were
calculated and the root contribution (RC) of soil respiration was estimated for each year. I
assumed that RC would be lower in severely dry years, where water deficit depresses the growth
of plants during growing seasons and, thus, limits the activities of roots more than the soil
microbial community.

2.2 METHODS

2.2.1 Study area

30

My experimental sites are located at the Great Lakes Bioenergy Research Center scale-up
fields at the Kellogg Biological Station (KBS, 42Â°40âN, 85Â°40âW), established in association
with the KBS Long-Term Ecological Research (LTER) site. The sites are located in the
southwest of Michigan, USA (Fig. 1.2), with a humid continental (warm summer) climate (Dfa)
(Peel et al., 2007). The mean annual air temperature and mean annual precipitation at KBS are
10.1 Â°C and 1005 mm yr-1 (1981-2010), respectively (Robertson & Hamilton, 2015).

2.2.2 Experimental design and schedule
Seven experimental plots were located at two locations, each with its own land use
histories (LUHs): (1) Conservation Reserve Program (CRP) grasslands at Marshall Farm, and (2)
corn-soybean rotation agricultural fields (AGR) at Lux Arbor Reserve. The CRP sites have been
a monoculture of smooth brome grass (Bromus inermis Leyss) since 1987, while the AGR fields
have been under conventional corn-soybean rotation cultivation for several decades (Fig. 1.6).
The soil texture at all sites is sandy loam, except for one field, which has sandy clay loam.
However, soil carbon and nitrogen contents at the CRP sites were significantly higher than those
at the AGR sites before land conversion (Table 1.1).
The experiment was conducted at seven scale-up fields ranging from 9-17 hectares. Four
fields are located in Marshall Farm (i.e., CRP sites) and three in Lux Arbor Reserve (i.e., AGR
sites). All sites, except the reference site (Ref), were sprayed with herbicide at the end of 2008 to
prepare the lands for soybean planting in 2009. The CRP and AGR sites were then cultivated
with either continuous corn (Zea mays, Dekalb DK-52), switchgrass (Panicum virgatum), or a
mixture of native prairie grasses dominated by Canada wild rye (Elymus Canadensis), little
bluestem (Schizachyrium scoparium), Indian grass (Sorghastrum nutans), big bluestem
(Andropogon gerardii), and switchgrass (Panicum virgatum) since 2010. One CRP grassland,

31

dominated by brome grass, was not disturbed and retained as the reference site (Gelfand et al.,
2011; Zenone et al., 2011; Deal et al., 2013; Zenone et al., 2013). I used an experimental code
for these sites by abbreviating them as âLUH-CROPâ. CRP-C, CRP-Sw and CRP-Pr represent
the CRP sites that were converted to corn, switchgrass and prairie mixture respectively, while
AGR-C, AGR-Sw and AGR-Pr are AGR farms converted to corn, switchgrass and prairie
mixture. In 2010, when the crops were established, the perennial crops (switchgrass and prairie)
were accompanied by oats as a nurse crop with and without fertilization, respectively (Fig. 1.6).
No other management practices were applied beyond harvesting at the end of each growing
season. No-till continuous corn was seeded in mid-May with a one-time herbicide mix (Lumax,
Atrazine 4L, Honcho Plus and (NH4)2SO4). Phosphorus, potassium and nitrogen fertilizers were
applied during April and June in each year (Bhardwaj et al., 2011; Zenone et al., 2013).

2.2.3 The climatic and microclimatic measurements and growth season identification
Climatic factors, such as air temperature (Ta) and daily precipitation (PRCP), and
microclimate, such as soil temperature and soil water content, were recorded as the independent
variables. They were recorded at different spatial and temporal scales based on the requirement
of research and the limitation of equipment and labor.
The daily mean Ta and daily accumulative PRCP were collected and calculated from the
KBS LTER Weather station (42Â°24â47.1â N, 85Â°22â15.3â W; lter.kbs.msu.edu). The air
temperature was recorded hourly by a thermometer at 3 m height. The mean daily air
temperature was calculated from the hourly data. Precipitation was measured by a NOAH IV
total precipitation gauge (ETi Instrument Systems Inc., Fort Collins, CO).
Soil temperature (Ts) and volumetric soil water content (VWC) were measured
simultaneously with soil respiration rate measurements biweekly during late April to September

32

and monthly during other times. Soil temperature was measured at 10 cm depth using a Taylor
8940N digital thermometer (Taylor Precision Products, Las Cruces, NM, USA) while VWC was
monitored from the ground surface to 10 cm depth by a HydroSense II with a CS659 sensor
(Campbell Scientific, Inc. (CSI), Logan, UT).
The length of the growing season is an important index for climate change and
agriculture. There are several identifications that are based on climate and crop growth. Here, I
used temperature-based identification, which is the most common in the North America (a.k.a.
frost-free season), instead of the crop-based growing season, since soil respiration includes the
soil microbe-derived carbon efflux, not only plant-derived soil carbon efflux. The onset and the
end of the climatic growing season (GS) in each year was calculated based on the 9-day moving
average of daily mean air temperature (9d-Ta) to filter short-term warm or cold events, which
may not induce the start of plant growth and development in spring. The onset of the GS was
identified as the first day of exceeding 0ď°C 9d-Ta in spring while the end of the GS was
recognized as the latest day of exceeding 0ď°C 9d-Ta in fall. The lengths of the GS were the
number of dates between the onset and the end of the GS in each year.
To compare the soil respiration rates, climate and microclimate, and the vegetation
indices, I separated the data into different years based on the USGS âwater yearâ concept using
October 1st to September 30th since it is more meaningful than the calendar year in agriculture
research. I identified the water year in my dissertation from the end of the previous GS until the
end of this GS because the post-GS precipitation may subsidize the soil water pool and will
support the consumption for plants and soil microbes in the next GS. The length of the water
year may be different due to the different seasonal patterns of air temperature in fall and winter.

2.2.4 The total and heterotrophic soil respiration measurements

33

Four plots at each site were randomly selected to install 1 x 1 m root exclusion plots for
heterotrophic soil respiration (Rh). We dug a 1 m deep trench at each edge of the square (Fig. 2.1
(B)). The trenches were lined with root-barrier sheets before refilling the soil back according to
its soil profile (Fig. 2.1 (C); Tang et al., 2005). The plants and roots inside the square were
manually removed and/or killed by herbicide (Monsanto RoundupÂŽ). Two 10 cm inner diameter
PVC collars were installed inside the square to accommodate flux chambers to measure Rh.
Another two collars were installed surrounding the square to measure total soil respiration (Rs)
(Fig. 2.1(D)). A total of 112 measurements (including Rh and Rs) were taken between 10 am to 7
pm when soil respiration is at its diel maximum.
Soil respiration was measured biweekly during the growing season and monthly during
the non-growing season when snow cover was minimal to allow for measurements from October
to December. The chamber-based infrared gas analyzer (IRGA) approach for soil respiration
measurement was done with an LI-6400 portable photosynthesis system with a 10 cm diameter
6400-09 soil chamber or an LI-8100 with a 10 cm diameter 8100-102 soil chamber (LI-COR
Biosciences, Lincoln, NE).

34

(A)
1m
1m

Root exclusion area

Control

10 cm

1m
depth

(B)

(C)

(D)

Figure. 2.1. The measurements of total (Rs) and heterotrophic soil respiration (Rh). A 1m x 1m x 1m cube was entrenched in the
soil and isolated for heterotrophic soil respiration measurement. The total soil respiration plots were placed near the heterotrophic soil
respiration plots (A). The four vertical sides of the cube were entrenched (B) and material-exchangeable sheets were embedded in the
trenches (C) to prevent the invasion of roots. The crops were killed by Monsanto RoundupÂŽ. Soil respiration rate was measured by an
infrared gas analysis (IRGA) of CO2 changes in the chamber headspace (D). Two 10 cm diameter collars were installed at least one
day before respiration measurement.

35

2.2.5 Vegetation Index data
Vegetation indices, including normalized difference vegetation index (NDVI) and
enhanced vegetation index (EVI), were calculated based on satellite remote sensing using band 1
(nir), band 2 (red), and band 3 (blue) of the Moderate Resolution Imaging Spectroradiometer
(MODIS). The NDVI was determined by the ratio of the difference between near-infrared
reflectance and red reflectance to their sum (Tucker et al., 2005), while EVI was further
modified with a soil adjustment factor, L, and two atmospheric aerosol scattering coefficientsâ
C1 and C2â(Huete et al., 1999) and had a higher sensitivity at the âgreenâ portion of the
spectrum when vegetation was very dense. These data were developed from MODIS daily
products by selecting an optimal day during every 16-day interval (i.e., low cloud cover).
Missing data may appear when clouds obscured the site over an entire 16-day period. NDVI and
EVI data were downloaded from the Oak Ridge National Laboratory Distributed Active Archive
Center for Biogeochemical Dynamics (ORNL DAAC,
http://daac.ornl.gov/MODIS/modis.shtml). I used a single pixel from each siteâs MODIS image
at the finest spatial resolution of 250 m found mostly within each experimental site. The dates of
NDVI and EVI values were paired with the nearest dates of soil respiration measurement.

2.2.6 The temperature-water-vegetation (TWV) models & Ra:Rh ratio
Multiple factor models that contain the effects of Ts, VWC, EVI and their interactions
were developed and tested. The experiment treatments (seven sites: 2 LUH * 3 CROP + Ref)
were set as a categorical factor and the following function was tested:
đ = đ(đđ  ) â đ(đđđś) â â(đ¸đđź) â đ(đśđđđ, đżđđť)

Eq. 2-1

where R is total (Rs) or heterotrophic (Rh) soil respiration, and f(Ts), g(VWC) and h(EVI) are the

36

effects of soil temperature, soil moisture and vegetation condition on R, respectively.
The functions of f(Ts), g(VWC) and h(EVI) were assumed as log-linear, quadratic, or
linear based on observed relationships between R and Ts, VWC, and EVI, from previously
published research, and the best correlations between R-Ts, R-VWC and R-EVI. The equations
containing two-level interactions and the variables themselves were tested. The final models
would include the equation with first and secondary orders in terms (Supplement 2-4). The
variables and their parameters were modeled by actual data. The best models for Rs and Rh were
separately selected based on Akaike Information Criterion (AIC).
The level of Rh was adjusted to the same Ts and VWC condition of Rs for calculation of
the autotrophic soil respiration (Ra). The root contribution (RC) to soil respiration (Ra: Rs ratio)
was calculated for all measurement dates in the growing season.

2.2.7 Data analysis
The temperature sensitivity (Q10) of R was computed based on the exponential
relationship between soil temperature at 10 cm depth and soil respiration at each site by year.
The coefficients and Q10 were estimated by the function, đ = đź â đđĽđđ˝đđ  , where đ10 = đđĽđ10đ˝ .
To compare across treatments, Q10 were then analyzed using a one-way analysis of variance
(ANOVA) among sites for each year. The Pearson correlation coefficients were calculated
between soil respiration (Rh and Rs) and their potential controlling factors (Ts, VWC, NDVI, and
EVI) by site and year.
The LUH effects (i.e., AGR, CRP and Ref) on Rs and Rh were compared in different
crops (i.e., C, Sw, and Pr) in each year by one-way ANOVA and Tukeyâs HSD test. Similar
comparison of CROP effects on Rs and Rh were analyzed in different LUH in each year. The Ra:
Rs ratios among years were compared by ANOVA and Tukeyâs HSD test as well. All statistical

37

analyses and graphs were conducted in R (R Development Core Team, 2010).

2.3 RESULTS

2.3.1 Climatic and microclimatic variables and the growing seasons
The onsets of the GS occurred each year in late March (DOY=90, 85, and 87) while the
ends of the GS were in mid-November to early-December (DOY= 341, 326, and 318) in 2011,
2013, and 2014, respectively, based on the 9-day moving average air temperature (9d-Ta). The
lengths of the GS were 251, 214, and 231 days in 2011, 2013, and 2014, respectively. However,
2012 had a very long GS (299 days) because of its early onset (DOY=57) and late end
(DOY=356) (Fig. 2.2). Air temperature in March 2012 was extremely high (ca. 10ď°C higher than
other years), although it fell back to normal levels in April (Table S.2.1). The maximum 9d-Ta
was 26.51, 27.91, 25.84, and 22.28ď°C for 2011â2014, respectively. The maximum air
temperature in 2014 was lower than other years. Reviewing the Ta data in more detail, 2012 had
the hottest summer with a 30.67 ď°C mean daily Ta on July 6th. Years 2011 and 2013 also had
high mean daily Ta, reaching 29.88 and 29.25 ď°C, respectively, while 2014 had a mean daily Ta
of 24.81 ď°C. July Ta means in 2011 and 2012 were significantly higher than those in 2013 and
2014 (Supplement 2-1). Soil temperature showed the same patterns (Supplement 2-1) although
the data were only collected biweekly during the daytime. Generally speaking, 2011 and 2012
had hotter growing seasons while 2013 and 2014 had cooler, especially in July. The year 2012
had very hot March temperatures that advanced the onset of growing season.
Annual precipitation (PRCP) ranged from 636 to 1238 mm between 1989 and 2016.
Precipitation patterns seem to have shifted over those years. The precipitation was lower with
smaller inter-annual variations before 1999, and was very low in 1999. Precipitation increased

38

and had a higher variation after 1999. During my experimental period 2011â2014, the PRCPs
were 1023, 796, 1139, and 971 mm, respectively. The difference between actual PRCP and the
5-year averages were 69, -177, 119, and -33 mm in 2011 to 2014, respectively. The year 2012
had a very low PRCP while the year 2013 was very wet (Fig. 2.3). Changes in VWC were highly
synchronized with precipitation. The VWCs in 2012 were very low during April and June under
canopy and during April and July across bare land (Table S.2.1).

39

Figure 2.2. Temporal changes in nine-day moving averages of daily mean air temperature (Ta). Air temperature was measured at
the KBS LTER Weather station. The green, red, blue and purple lines present daily mean Ta in 2011â2014, respectively. The Ta was
calculated by an average nine-day (ďą 4 days) moving filter in order to more clearly display the seasonal trend. The colored arrows and
the colored numbers indicate the DOY of the onset, the maximum Ta and the end of the growing season in spring, summer and fallwinter, respectively. The numbers in brackets show the maximum 9-day average of daily mean Ta (ď°C) in each year.

40

Figure 2.3. Annual precipitation by water year from 1989 to 2016. Five-year moving averages were calculated to better show the
long-term trend. The 95% confidence intervals represent the variation in five-year average. The dots show the annual precipitation of
each water year. Water year is defined from 1 October to 30 September from USGS. The residual is the difference between the actual
annual precipitation and the five-year moving average. Green, red, blue and purple dots were the experimental years 2011, 2012, 2013
and 2014, respectively. The data were collected by the KBS LTER weather station.

41

2.3.2 The Vegetation Indices: NDVI and EVI
NDVI and EVI showed a hump shape, reflecting the phenology of crops. However, both
NDVI and EVI dropped in June in 2012, revealing the slow growth (or even wilting) of crops
due to the drought stress in that year (Fig. 2.4). The values of EVI rebounded in August, but the
maximum values were lower than those in other years. The vegetation indices coincided with my
field observations that crops wilted in June.

Figure 2.4. The seasonal patterns of NDVI and EVI. The NDVI and EVI values were
calculated from MODIS images. Each site had one pixel. The dates of data were fit to the nearest
dates of soil respiration measurements. There were dates with two missing data due to
continuously cloudy days in late summer 2013.

2.3.3 CROP and LUH effects on total (Rs) and heterotrophic soil respiration (Rh)

2.3.3.1 The effects of CROP, LUH and their interactions on Rs and Rh
Both CROP and LUH had significant effects on Rs and Rh. The Ref site always had the
highest means and variances of Rs and Rh in all years. In comparison to the other crop types, Ref
had the highest Rs, followed sequentially by Pr, Sw and C. Also, LUH had the highest effect on
Ref, followed by CRP and AGR (Supplement 2-5).
At the annual scale (2011â2014), corn had the lowest R although it was sometimes close
42

to Sw (Rs in 2011; Rh in 2012 and 2014). In first three years (2011â2013), R in Sw and Pr were
similar and smaller than R in Ref and larger than R in corn. However, over time, the R in Pr
increased and became very close to R in Ref in 2014. A similar pattern was apparent in Rs: Pr
was similar to Ref, but not Sw.
The effects of CROP and LUH in 2012 were different from the other years. Both Rs and
Rh in this year were lower than other years, especially for Rh. The effects of LUH and CROP
were smaller since the values of Rs and Rh were very low in 2012. For CROP effects, the Rh in
C-Sw-Pr and in Sw-Pr-Ref showed no significant difference while AGR-CRP showed the similar
R (Supplement 2-5). The Ref field still had highest mean soil respiration but its variance was
high.

2.3.3.2 The effects of CROP in different LUH
Comparing the CROP effects between the two different LUHs separately, I found that the
Ref field usually had the highest Rs and Rh in all years, except CRP-Pr in 2014; Corn always had
the lowest; while Sw and Pr had intermediate rates that were lower than Ref but higher than corn.
The chronological comparison showed the highest soil respiration rates in 2011, which was a
warm and wet year with higher than usual crop yields in the region. Both Rs and Rh were lower
in 2012 when it was warm and dry and rebounded in the subsequent cooler and wetter years,
2013 and 2014. Thus, the crops had different magnitudes of response that reflected their
respective LUH.
In CRP sites, both Rs and Rh were separated into two groups, annual and perennial crops,
based on Tukeyâs HSD test. All Rh in 2012 was reduced to the similar level and the difference
disappeared while the Rs in Ref was still significantly higher than Rs in Sw, Pr and C. After the
dry year, Rh in Sw and Ref increased, although they did not become statistically different. The Rs

43

of perennial crops in 2013 and 2014 were higher than Rh. This was especially true for Pr. The Rs
in Pr was higher than those in Sw in 2014 (Fig. 2.5 (A)).
In AGR sites, the relative differences of R among crops and years were similar to those in
CRP sites. The Ref field always had highest Rs and Rh, followed by Pr, Sw, and C. However, the
three crops usually had larger differences relative to Ref, compared to those at CRP sites. Soil
respiration in Pr did not rebound strongly like those in CRP sites (Fig. 2.5 (B)).

2.3.3.3 The effects of LUH in different CROP
Comparing the two LUH (i.e., CRP vs. AGR) and Ref, AGR had the lowest soil
respiration rates while Ref usually had the largest rates. The changes of Rs and Rh among years
were similar to those in the description in CROP effects: they were highest in 2011, depressed in
2012 and then rebounded in 2013 and 2014. However, the speed and the magnitude of the
changes of soil respiration were different between crops in fields of different LUH.
In corn fields, Rh was different between AGR and CRP in 2011, 2012 and 2014, while Rs
was similar. The Ref field always had significantly higher Rs and Rh. No obvious chronological
trends were found for either Rs or Rh (Fig. 2.6 (A)).
Comparing the interannual patterns in Rs and Rh between AGR- and CRP-Sw fields for
only the second year after land use change (2011), CRP had a higher Rh and Rs than AGR. After
2011, however, both Rh and Rs in AGR and CRP were similar; yet, Ref always had a higher Rs
than AGR. The Rs in CRP was similar to that of Ref in 2011, but reduced and then became
similar to AGR and Ref in 2013 and 2014 (Fig. 2.6 (B)).
The Rh of CRP-Pr was closer to Ref and was higher than that in AGR-Pr in 2011. All Rh
in three sites were not different when they were depressed in 2012, like all other crops. However,
Rh and Rs in CRP-Pr increased a lot in 2013 and 2014, reaching even higher values than those in

44

Ref although there was no significant difference (Fig. 2.6 (C)).

45

(A)

Rs

Rh

Rs

Rh

(A)

(B)

(B)

Figure 2.5. The CROP effect on total (Rs) and heterotrophic soil respiration (Rh) in fields
with different LUH. CRP (A) and AGR (B) refer to the previous land use. The letters presented
the statistical results from Tukeyâs TSD test between crops in the same LUH and in the same
year. The letters display the means of R values in each treatment from small to large
alphabetically.

46

(A)

Rs

Rh

Rs

Rh

(B)

Figure 2.6. The LUH effect of total (Rs) and heterotrophic soil respiration (Rh) in different
crop types. Corn (A), switchgrass (B), and prairie mixture (C) are shown with total (Rs) and
heterotrophic soil respiration (Rh) in different panels. The letters presented the statistical results
from Tukeyâs HSD test between crops in the same crop types and in the same year. The letters
display the means of R values in each treatment from small to large alphabetically.

47

Figure 2.6. (contâd)
(C)

Rs

Rh

48

2.3.4 Soil respiration regulators
Soil respiration had different relationships with three major drivers: temperature, soil
water content, and vegetation. Both Rs and Rh had similar correlations with these drivers. Soil
respiration was positively, exponentially related to Ts (Tables S.2.1, 2.2 and Fig. 2.7) and
positively linearly correlated to VIs (Fig. 2.9 (A) & (B)), but was weakly negative or not
correlated to VWC (Fig. 2.8).
The percent variation in log10R explained by Ts ranged from 47% to 90% in all data,
mostly between 60% and 80%. The results revealed the strong exponential relationship between
R and Ts. The regression slopes for log10Rs-Ts were greater than the slopes for log10Rh-Ts (Fig.
2.7), suggesting that the Rs had higher temperature sensitivity than Rh, although the difference
was not significant because of high within-group variation (Q10, Table S.2.3).
Soil VWC showed no correlation (p > 0.05) or a weak, negative linear correlation (0.05 >
p > 0.01) with Rs and Rh (Table S.2.2). A quadratic relationship fit better. However, the
importance of VWC to soil respiration appeared different among years. VWC was a weak driver
in wet years, such as 2013 and 2014. For example, R and VWC at all sites except AGR-Sw were
not correlated (p > 0.05). On the contrary, R in most sites was correlated to VWC in 2011 and
2012 when the spring and the summer were dry (Fig. 2.8; Table S.2.2).
NDVI and EVI had strong, positive linear correlations with log10R (Figs. 2.9(A) & (B);
Table S.2.2). The slopes of Rs-NDVI/EVI and Rh-NDVI/EVI are close, but most Rs-NDVI/EVI
regressions have larger intercepts. The regression slopes in 2012 are visibly higher than those in
2011, 2013, and 2014 (Figs. 2.9(A) & (B)), suggesting that the relationship of VI on Rs was
stronger in 2012 (Table S.2.2).

49

Figure 2.7. Linear regressions of soil respiration (log10R) on soil temperature (Ts). The soil respiration values were the average of
measurements from four plots in each field. The solid circles and solid lines were Rs. The empty circles and dashed lines show Rh.

50

Figure 2.8. Linear regressions of soil respiration (R) on soil moisture (VWC). The solid circles show the average Rs from four
plots in each research field, and the empty circles show the mean Rh. The solid lines are linear regressions for Rs-VWC while the
empty circles and dashed lines are the linear regressions of Rh-VWC.

51

(A)

Figure 2.9. Linear regression of soil respiration (log10R) on MODIS vegetation indices (NDVI and EVI). The soil respiration
values were the average of measurements from four plots in each research site. The solid circles and solid lines were Rs. The empty
circles and dash lines were Rh.

52

Figure 2.9. (contâd)
(B)

53

2.3.5 Temperature sensitivity (Q10) of soil respiration
The Q10 values for soil respiration ranged from 1.20 to 1.86, although the differences are
not significant due to high within-group variation. The ranges of Q10 were smallest in 2011
(1.31-1.57) followed by 2012 (1.20-1.60), 2014 (1.30-1.86), and 2013 (1.26-1.81 Îźmol CO2 m-2s1

). The range of Q10 was only 0.26 in 2011 and 0.56 and 0.55 in 2013 and 2014, respectively.

The smallest Q10 (1.20) occurred in the dry summer of 2012, and the largest Q10 (1.86) in the
wettest year (2013) (Supplement 2-3). It is not clear whether the high variations of Q10 are due to
the 2013 wet spring or the dry-rewet recovery in 2012-2013.

2.3.6 Temperature-Water-Vegetation (TWV) multiple variable models for soil respiration
After testing many regression models including three biophysical variables: Ts, VWC,
EVI, separately and their interactions, the AIC values shows that the models that contain all three
factors (i.e., Ts, VWC, and EVI) with their relationships (exponential, exponential and linear for
Ts, VWC, and EVI, respectively) and some of their second-order interactions, explain Rs and Rh
well. I do not include higher-order interactions between biophysical factors since it is difficult to
explain the ecological meaning. The parameters of each term in the best-fit biophysical R model
are listed in Supplement 2-4.
The Rs and Rh models were similar but the interactions of log10VWC:EVI and
Ts:CROP/LUH terms were removed from the Rs model. The adjusted R2 are 0.62 and 0.53 for Rs
and Rh, respectively. The residuals of the TWV models of log10Rs and log10Rh seem acceptable
(Fig. 2.9).

54

Figure 2.10. Residuals of the two-level interaction TWV models. The x-axis is the predicted
value of log10Rs (left) and log10Rh (right) from the TWV model, while the y-axis is the residuals
(i.e., difference between measured data and modeling estimates).
2.3.7 Ra:Rh ratio and the root contribution to soil respiration
The estimates of Rs and Rh were calculated based on TWV models with actual Ts, VWC
and EVI values at Rs plots (canopy-covered) during the GS. The Ra was calculated as the
difference between the paired measurements of Rs and Rh. The values of RC ranged from 0.31 in
CRP-C field to 0.57 in CRP-Sw field. Generally speaking, CRP-C had the smallest RC values
(0.31 ~ 0.33) while the Ref site always had highest RC values (0.54 ~ 0.56). The RC values in
other sites were between CRP-C and Ref (Table. 2.1). RCs in CRP fields increased in the order C
< Pr < Sw < Ref. In AGR fields, Sw significantly had lower RC than other three crop fields (Fig.
2.11). Between the two LUH, CRP fields usually had higher RC than AGR, except switchgrass
fields. The RCs in the AGR-Pr field were the same as those in the Ref field (Figure 2.12). The
RC values in most fields decreased in 2012, except for CRP-Sw (Table 2.1 & Fig. 2.13). Tukeyâs
HSD tests showed that the Ra:Rs ratios in Ref and in AGR-Pr field were all the same over the
four years. The AGR-Sw and CRP-Pr had lower RC in 2012 compared to other years while
CRP-Sw had high RC in 2012 (Fig. 2.13).
55

Table. 2.1. The root contribution (RC) to total soil respiration. RCs (a.k.a. Ra:Rs ratios) were
calculated from the estimates of TWV models. The values were mean Âą 1 s.d. CROP: crop type;
LUH: land use history; C: corn; Sw: switchgrass; Pr: prairie; Ref: reference; CRP: Conservation
Reserve Program; AGR: corn-soybean rotation agriculture.
2011
2012
2013
2014
CROP LUH
0.32 Âą 0.05
0.32 Âą 0.04
0.31 Âą 0.05
0.33 Âą 0.06
C
CRP
0.46 Âą 0.10
0.43 Âą 0.06
0.49 Âą 0.12
0.51 Âą 0.13
AGR
0.47 Âą 0.08
0.57 Âą 0.17
0.46 Âą 0.08
0.47 Âą 0.10
Sw
CRP
0.45 Âą 0.14
0.35 Âą 0.24
0.43 Âą 0.14
0.44 Âą 0.17
AGR
0.45 Âą 0.06
0.29 Âą 0.18
0.39 Âą 0.12
0.41 Âą 0.09
Pr
CRP
0.51 Âą 0.03
0.51 Âą 0.04
0.52 Âą 0.05
0.53 Âą 0.04
AGR
0.56 Âą 0.04
0.56 Âą 0.04
0.54 Âą 0.06
0.54 Âą 0.06
Ref

Figure 2.11. The root contribution (RC) to total soil respiration across CROP (crop types).
The CROP effects on Ra/Rs ratio in CRP fields (A) and in AGR fields (B) are shown, where Rs is
estimated from the TWV model and Ra is calculated as the difference between Rs and Rh. The
letters indicate the results of Tukeyâs HSD test for different sites. The letters display the means
from small to large alphabetically. CRP: Conservation Reserve Program; AGR: corn-soybean
rotation agriculture; C: corn; Sw: switchgrass; Pr: prairie; Ref: reference.

56

Figure 2.12. The root contribution to total soil respiration across LUH (land use histories). The LUH effects on Ra/Rs ratio in
corn (A), switchgrass (B) and prairie fields (C) are shown, where Rs is estimated from the TWV model and Ra is calculated as the
difference between Rs and Rh. The letters indicate the results of Tukeyâs HSD test for different sites. The letters display the means
from small to large alphabetically. LUH: land use history; CRP: Conservation Reserve Program; AGR: corn-soybean rotation
agriculture; C: corn; Sw: switchgrass; Pr: prairie; Ref: reference

57

Figure 2.13. The root contribution to total soil respiration over 2011â2014. The Rs was
estimated from the TWV model and Ra was calculated as the difference between Rs and Rh. The
yearly variations of Ra/Rs ratios are indicated by their CROP (between panels) and LUH (within
panel). Solid dots with solid lines show CRP while empty circles with dashed lines show AGR
sites. The error bars were one standard deviation. The letters presented the results of Tukeyâs
HSD test for different sites. The capitals present the statistical results of CRP and the lowercase
letters represent the results of AGR fields. The letters display the means from small to large
alphabetically.

2.4 DISCUSSION

2.4.1 The major biophysical drivers of soil respiration
Many recent studies show that the major biophysical drivers for soil respiration are
typically temperature, soil water content, substrate supply from plants, and their interactions
(Luo et al., 2001; Curiel Yuste et al., 2004). These drivers may affect soil respiration via
different mechanisms and have different correlations to soil respiration. The relative importance
of different biophysical factors may change depending on the biogeochemical processes and their
feedback in different climate regimes and ecosystems within and among years. In my
ecosystems, soil temperature and vegetation indices were positively and exponentially correlated

58

with soil respiration. Soil water content had weak or no correlation with soil carbon dioxide
efflux and was more likely a limiting factor for soil respiration only when it was low (Figs. 2.7,
2.8 & 2.9).
The inter-annual and intra-annual variations and the pulses of biophysical factors affect
the temporal variations of soil respiration during the growing season. Legacy effects play
important roles as well. The different biophysical factors may control soil respiration by different
ways and at different temporal scales. Long-term climate patterns (e.g., the temporal variations
of temperature, photoperiods and light quality), short-term regular climate events (e.g., periodic
flooding or drought), and short-term occasional events (e.g., heat waves, rain pulses) can alter
ecosystem functions. Some of the influences of the abiotic factors explicitly change soil
respiration when it is happening while others may not present effects immediately. The âcarryover effectsâ may alter community structure, change species richness and evenness, and affect
important ecological processes within or among years (Flanagan, 2009). For example, the
cumulative effect of temperature (e.g., growing degree days) affect the plant phenology and
biomass production, which influence both carbon uptake and water consumption by the plants,
and further alter autotrophic and heterotrophic soil respiration via the root growth and soil
substrate supply and water availability. On the contrary, the soil water availability, which
depends predominately on precipitation inputs and evapotranspiration outputs, can limit the
activities of plant and soil microbes and affect soil respiration in the growing season when water
availability is limited.
In the studied ecosystems, I hypothesized that three major forces control the soil
respiration at different temporal scales and with different consequences (Fig. 2.12). I made the
schematic chart for soil respiration across years based on the correlation between R and

59

biophysical factors and the actual values of the drivers in each year (Fig. 2.12 (A)). If I add the
impacts of severe drought and assume that the sites require more than one year to recover, the
values of R fit better for our data (Fig. 2.12 (B)). The effects of CROP and LUH after LUC may
be different on Ra and Rh. The perennial crops may develop their extensive root systems after
land use change and increase the autotrophic soil respiration while annual crop does not have
such an extensive root system (Fig. 2.12 (C) CRPpere and AGRpere). Annual crops should have
consistent Ra from year to year since they will die and be replanted every year (Fig. 2.12 (C)
CRPann and AGRann). The CRP sites should have relatively higher root system establishment
due to the more fertile soil. The year-to-year trends of Rh will be opposite to those of Ra at CRP
sites since plant and soil microbes will consume soil nutrients, especially nitrogen. The changes
of soil respiration may or may not be evident depending on the relative strength of different
drivers and their interactions and feedback. The crop may adapt to its environment and diminish
the magnitude of changes.

60

(A)

(B)

(C)

Ts

Ra

Ra or Rh

CRPpere
AGRpere
CRPann
AGRann

2011

VWC

Ra

LUC

drought

2012

2013

2014

LUC

Rh
CRP

Rh

year

Ra or Rh

AGR
2010

2011

vegetation

2012

2013

2014

year

(D)
LUC

climate disturbance

annual biophysical factors

annual biophysical factors

2011

2012

2013

2014

0

1

year

10

100

year

Figure 2.14. Schematic depiction of the impacts of three major forces on soil respiration at
different temporal scales. (A) The interannual variation in R is driven by biophysical factors
and the actual biophysical conditions in each year. The interannual variation in biophysical
factors (Ts, VWC and vegetation) affect annual R. (B) The 2012 severe drought depressed soil
respiration, which took more than one year to recover. (C) The Ra was affected by the
establishment of root systems. Perennial crops (CRPpere & AGRpere) will have increasing Ra
while the annual corn crop (CRPann & AGRann) will have consistent Ra across years after land
use change. The rate of root establishment will be higher in CRP sites owing to its more fertile
soil. CRP sites will have decreasing Rh chronologically that reflects the gradual depletion of soil
nutrients while AGR sites will have relatively consistent Rh. The LUC occurred in 2009. (D) The
temporal scales of biophysical effects on R may range from less than a year to decades in the
case of LUC.
During my study period, both Rs and Rh had similar year-to-year trends, although the
magnitudes were different in different CROP or LUH treatments. In 2011, soil respiration rates
were highest when both temperature and precipitation were high. In 2012, the hot and springsummer dry year, soil respiration was depressed in all sites. Soil respiration rates in 2013 and

61

2014 were higher than those in 2012, although it had lower summer temperatures (Figs. 2.5 &
2.6).
The combination of biophysical variables (TWV models) explained 62% and 53% of
variation in Rs and Rh for all sites over all four years. The individual variables in each
experimental site may explain up to 89% of the variation (i.e., the effect of Ts on Rs in AGR-Pr
in 2013; Table S.2.2). However, the changes of soil respiration were confounded between
responses to the biophysical variables in a particular year, the recovery after extreme climate
events (i.e., severe drought), and the long-term trends after LUC. The former two were climateinduced effects while the latter one was due to human management. The Ref field is the only one
that was driven by climate variability as opposed to LUH and management changes.
The Rs and Rh in Ref were highest in 2011, dropped in 2012 and then rebounded in 2013
and 2014 (Fig. 2.5). The warm temperatures with enough water in 2011 supported a good
environment for plants and soil microbes, thus explaining the high soil respiration. In 2012, the
hottest and longest growing season resulted in lower soil respiration. The severe spring-summer
drought suppressed both plant growth and soil microbial activities, resulting in the lowest soil
respiration observed over the four years. The sudden drop of NDVI and EVI in June 2012 (Fig.
3.1 (C)) confirms the drought effect on above ground plant biomass, supporting the hypothesis
that the spring-summer drought affected both Ra and Rh directly or indirectly via reduced plant
growth and soil carbon substrate supply. The consequences fit our hypothesis that soil water
content plays a crucial role at times when water is limiting, which also tend to be the times with
warmest temperatures.
Comparing the post-drought years, I found that soil respiration was higher in 2014 than in
2013, which was the reverse of the summer soil temperature levels. It means that the roots and

62

soil microbes had a lower metabolism in 2013 than in 2014, even though the summer
temperature was higher in 2013 and when soil water was not a limitation. The drought-recovery
hypothesis was a better fit (Fig. 2.12 (B)) than the biophysical hypothesis alone (Fig. 2.12 (A)).
The results imply that soil respiration requires at least two years to recover from the damage of a
severe spring-summer drought. The reasonable hypothesis is that the carbon-related
biogeochemical processes, the community structure of plant and soil microbes, and/or their
physiological traits changed during the drought and it spent longer than one year to reach a new
equilibrium. However, there is a difference in the temperature sensitivity of soil respiration
across the years, which had high variations. I developed a method to understand the change in
the R-T relationship and find its trajectory across years, which is discussed in Chapter 4.

2.4.2 The effects of crop type (CROP) and land use history (LUH)
The six experimental sites had similar year-to-year trends to the Ref site: the highest soil
respiration in 2011, lowest R in 2012, and the rebound in 2013 and 2014. However, the different
crops and land use histories induced different magnitudes of impacts and different recovery
speeds of R.
Comparing the inter-annual trend of Rs between the three cropping systems, the annual
crop had relatively constant values from year to year while perennial crops increased
chronologically. The 2014 Rs and Rh at CRP-Pr had even exceeded those in Ref, although there
was no significant difference (Fig. 2.5 (A)). It implied that the extensive root system in CRP-Pr
had become well-established and that the soil microbial community associated with the dense
root systems had high heterotrophic respiration. In contrast to the CRP fields, the Rs and Rh in
AGR-Pr fields did not quickly increase but always had similar values with AGR-Sw (Fig. 2.5
(B)). The results agree with my field observations that the AGR-Pr site changed its plant

63

composition and became more dominated by switchgrass after the dry year (Fig. 3.7; Abraha et
al., 2016). The highly similar plant compositions between AGR-Pr and AGR-Sw had similar
responses to annual climate regime and to severe drought.
In 2011, most Rs and Rh rates were different between CRP and AGR. This should be the
LUH effect from high carbon and nitrogen in the former CRP field soils, which stimulated the
growth of plants and soil microbes. Most Rs and Rh were the same in 2012 due to the drought
depression of plant and microbial activity. In 2013 and 2014, most corn and switchgrass lands
had the similar Ra and Rh, but CRP-Pr had significantly higher Ra and Rh than AGR-Pr. The
initial high soil carbon and nitrogen content may stimulate the establishment of root-microbe
association underneath the multicultural prairie.

2.4.3 The root contribution (RC) to total soil respiration across years
Other studies have estimated that the root contribution (RC) to total soil respiration
ranges 10â90% in different ecosystems, but lower RC contributions are common in non-forest
ecosystems. The mean annual RC of grassland and cropping ecosystems from 14 studies was
60.4% (Hanson et al., 2000). My annual growing-season RCs ranged from 0.31 to 0.57 (Table
2.1). The CRP-C had very low means and variations of RCs (0.31 ~ 0.33, Table 2.1 & Fig. 2.13)
because high soil carbon and nitrogen content supported the high soil microbial activities, adding
the low root respiration of annual crop. The RC values in CRP-C are in the range of maize fields
in Hanson et al. (2000) review (0.35 ~ 0.40 in growing season). However, RC values in AGR-C
are high (0.43 ~ 0.51). This may be due to the long-term (decades) cultivation that reduced soil
fertility in the AGR field, which limited soil microbial activities and thus reduced Rh. The Ref
site had high RC values with small within- and between-year variations implying a highly
developed root system and a very stable carbon allocation to root and soil organic carbon pool.

64

The resistance of RC to severe drought is high. CRP-Pr and AGR-Sw had lower RC in 2012,
revealing that the impact of drought on root was stronger than that on soil microbial community.
My field observation, that most plants in CRP-Pr were withered in 2012 summer, which
seemingly proves this.
The dry year (2012) had lower RC with higher variation, implying that the 2012 springsummer drought depressed Ra more than Rh and the effect lasted until the second year (2013).
The responses of RC in different crop types and land use histories were different: Ref and AGRPr fields were stable; CRP-Pr and AGR-Sw decreased significantly; CRP-Sw increased; CRP-C
and AGR-C did not change or changed slightly. In current studies, RC may change seasonally or
be stimulated by climate pulses (e.g., huge precipitation or temperature fluctuation) since Rh and
Ra may respond to climate differently in timing and magnitude (Carbone et al., 2011; GomezCasanovas et al., 2012). I will discuss the seasonal changes of RC and examine how the
seasonality of the responses of Ra and Rh affect the annual RC in Chapter 3.

2.4.4 The responses of Ra and Rh to future climate scenarios
The trend of increasing air temperature (IPCC, 2013) and increasing extreme
precipitation events (Andresen et al., 2012) were expected in the Midwestern USA in the future.
Grassland and cropping ecosystems are sensitive to drought events and reduce both GPP and Reco
in response (Hoover et al., 2014; Reichstein et al., 2013; Ciais et al., 2005; Baldocchi, 2008). In
a grassland ecosystem, Rs is the second largest carbon process and can return 50-90% of annual
GPP back to the atmosphere (Bahn et al., 2008). The relative strength of the reduction of GPP
and Rs under drought stress may change a carbon sink to a carbon source or the amount of net
CO2 it can sequestrate from the atmosphere (Nagy et al., 2007) depending on different responses
of respiration and the assimilation. The two major components of soil respiration, Ra and Rh, thus

65

play crucial roles on the carbon dynamics of cropping and grassland ecosystems. The Ra and Rh
have their own mechanisms and are controlled by different biophysical variables.
Photosynthetic carbon supply is one of the crucial variables for both Ra and Rh. The
changes of photosynthetic rates and the carbon allocation strategies responding to drought,
therefore, affect Ra and Rh severely. Vegetation adapts to environment conditions by modifying
its ecophysiological habits. Different crop types have their own strategies and capability to adapt
to different magnitudes, durations and frequencies of drought under different environmental
conditions (van der Molen et al., 2011). My study explains the different response of Ra and Rh in
different CROP and LUH. The Ref field, the late successional grassland, always had high RC
each of the four years during my study due to its highly developed root systems. The highly
developed root system may connect root and soil microbial community tightly and thus the
responses of Ra and Rh to severe drought were similar, resulting the constant RC even in the
drought year. The RC in the AGR-Pr fields responded to the severe drought similar to Ref field.
In CRP-Pr and AGR-Sw fields, RC decreased to adapt to drought and returned back to the initial
values of 2011 soon after the drought year. The water stress of the 2012 drought year caused an
early senescence of crops in mid-summer in the CRP-Pr. The AGR-Pr maintained the same RC
in the drought year, but the composition of the C4 plant biomass increased after 2012 (Fig. 3.7;
from Abraha et al., 2016). The changes of the composition of vegetation and the ecosystem
water use efficiency (eWUE) in CRP-Sw, AGR-Pr and AGR-Sw fields revealed how the
ecophysiological adaptated to severe drought. The increases of C3 vegetation composition can be
expected if the occasional severe drought increases in the future. However, the mechanisms and
the effect of the adaptation on the ecosystem carbon cycling and atmospheric CO2 dynamics
require more research to explore.

66

2.5 CONCLUSIONS
The regulatory roles of biophysical factors on soil respiration are complex. Their
relationships are always non-linear, interconnected, and need to include the consideration of their
threshold and relative strength of force. The dominant factor may change under different biomes
and climate regimes. The adaptation of vegetation and soil microbial community increases the
complexity of their responses to climate patterns and occasional climate events and human
management.
In this chapter, I found that soil respiration positively and exponentially correlated to both
Ts and EVI. Soil moisture had weak or no correlation to soil respiration, but instead decoupled
the R-Ts or R-EVI correlations when it was low. I also explored the different responses of Rs and
Rh under different LUH and CROP after land use conversion at the interannual scale. The mature
grassland ecosystem had high Ra and Rh and with high resistance to drought. The CRP-Pr field,
which was a multicultural prairie with carbon and nitrogen enriched soil, established a root
system 5 years after land use change, despite a severe drought in the third year, and had high Rs
and Rh in 2014. Corn, a annual crop, had a lower RC comparing to perennial crops, implying the
high importance of heterotrophic soil respiration response to climate change.
I found evidence for the different effects of three forces on soil respiration: the particular
year biophysical condition; the severe 2012 drought; and the land use conversion at different
timescales. Particular year biophysical variables (annual temperature, water availability and
vegetation index) support the potential values of soil respiration in each year. The severe drought
in 2012 decreased both Ra and Rh in the same year but its effect prolonged to the following years.
The effect of LUC was not very clear other than the CRP fields, which have high soil carbon and
nitrogen content, had higher Ra and Rh than AGR fields. Perennial crops had increasing Rs and

67

Rh while annual did not, revealing the different capacity and ecophysiological adaptation
between annual and perennial crops.
The spring-summer severe drought in 2012 depressed Ra and Rh in all fields and induced
different responses of Ra:Rs ratios. The RC values in most of fields were decreased or maintained
in 2012 but returned to the levels in 2011 although most Rs and Ra were still lower than those in
2011. The changes of the C3:C4 ratio of vegetation biomass and eWUE revealed the
ecophysiological adaptations expressed in vegetation to severe drought. It will happen more
frequently if the amplitude and frequency of extreme drought events increase in the future.

68

APPENDIX

69

Table S.2.1. Monthly climate and microclimate. The monthly air temperature (Ta) and the monthly precipitation (PRCP) were
calculated from daily mean air temperature and daily precipitation measured at the KBS LTER weather station. The monthly soil
temperature (Ts), and the monthly soil volumetric water content (VWC) were measured biweekly (growing season) or monthly (nongrowing season) when soil respiration was measured. The values were mean Âą standard deviation. The Ta, Ts, and VWC among four
years were compared in a one-way analysis of variance (ANOVA). Different letters after the values denote significant (p < 0.05)
differences among years by Tukeyâs HSD (honest significant difference) test. The Ts and VWC for total soil respiration (Rs) and
heterotrophic soil respiration (Rh) experiments were measured and analyzed separately.
R

year

Ta
(â°C)

2011
2012
2013
2014

Jan
-6.3Âą3.6ac
-1.6Âą4.8b
-2.6Âą3.68ab
-8.7Âą6.4c

PRCP
(mm)

2011
2012
2013
2014

19.2
68.3
178.8
73.1

Ts
(â°C)

Rs

Rh

VWC
(%)

Rs

Rh

2011
2012
2013
2014
2011
2012
2013
2014

Month
Jun
Jul
20.2Âą3.5a
24.1Âą2.5a
21.0Âą4.3a
25.3Âą2.9a
19.4Âą3.3a
21.7Âą3.9b
20.4Âą2.7a
19.2Âą2.4c

Feb
-3.5Âą5.77ab
-0.4Âą3.04a
-3.9Âą3.88b
-8.0Âą5.02c

Mar
1.0Âą4.0a
10.0Âą8.0b
-0.6Âą3.9a
-2.6Âą5.9a

Apr
7.6Âą4.3a
8.8Âą3.4a
6.5Âą4.9a
8.1Âą4.4a

May
15.1Âą5.0a
17.2Âą4.3a
16.7Âą5.3a
15.0Âą5.6a

75.7
52.3
107.4
60.5

66.8
78.0
29.5
52.1

146.0
109.0
170.4
67.3

142.4
30.2
118.6
80.5

47.5
22.9
107.9
148.1

10.6Âą1.9a
10.8Âą3.4a

15.3Âą3.4a

2.1Âą2.5

18.2Âą3.2b
11.8Âą3.1c

2.6Âą3.6

10.5Âą2.3a
11.3Âą4.4a

15.0Âą3.2a
21.0Âą3.0b
13.4Âą3.8c

Aug
20.8Âą2.4a
20.7Âą3.0a
20.1Âą2.9a
20.3Âą2.1a

Sep
15.6Âą4.8a
16.5Âą4.3a
16.7Âą4.6a
15.6Âą4.1a

Oct
10.6Âą4.7a
9.8Âą5.0a
11.0Âą5.2a
9.9Âą3.9a

Nov
6.2Âą4.2a
3.4Âą4.3ab
2.9Âą5.2ab
1.2Âą5.7b

Dec
1.0Âą3.1a
1.4Âą4.6a
-3.7Âą5.1b
-0.1Âą4.1a

186.7
45.5
82.6
102.4

96.3
70.1
117.1
74.2

82.8
58.3
19.3
67.6

89.9
143.0
106.2
98.0

109.0
12.7
73.2
75.7

62.5
52.1
66.3
33.3

21.3Âą2.5ab
20.1Âą3.4a
22.0Âą2.6b
21.8Âą2.3b

27.5Âą1.4a
28.3Âą4.4a
21.6Âą2.5b
20.2Âą1.5c

22.0Âą2.4a
22.2Âą3.1a
23.2Âą2.2a
20.4Âą2.1b

14.7Âą1.9a
15.7Âą3.2a
18.5Âą1.8b
16.1Âą2.8a

11.8Âą1.0a
10.8Âą1.9b
17.3Âą2.3c

6.9Âą1.5a

3.7Âą1.4

21.3Âą2.4a
23.2Âą5.3b
24.8Âą3.3c
25.6Âą3.2c

29.3Âą2.3a
30.3Âą5.4a
25.1Âą4.0b
23.3Âą2.8c

23.5Âą2.7a
22.8Âą4.2a
25.7Âą2.3b
22.8Âą3.4a

15.3Âą2.5a
15.7Âą4.7a
19.4Âą2.0b
17.9Âą2.6c

12.0Âą1.7a
11.1Âą3.3a
18.8Âą2.3b

9.5Âą0.3b

7.1Âą2.3a

3.3Âą1.7

9.2Âą0.3b

2011
2012
2013
2014

23.57Âą5.14a 21.89Âą5.70a
28.71Âą5.74 20.96Âą4.11b
19.44Âą5.99b
24.14Âą5.50a

16.49Âą4.91a
8.20Âą4.53b
22.32Âą4.74c
21.02Âą5.66c

7.43Âą2.06a 18.77Âą5.14a 13.61Âą3.13a 23.11Âą4.20a
8.84Âą7.24a 15.90Âą6.43b 11.89Âą6.26a 21.27Âą4.38a
9.94Âą3.52a 16.57Âą5.24ab 12.45Âą3.83a 10.82Âą4.49b
15.18Âą6.80b 12.47Âą7.49c 23.59Âą6.23b

26.43Âą3.96a 31.29Âą5.14

2011
2012
2013
2014

23.22Âą5.33a 21.25Âą5.61a
27.12Âą6.26 18.59Âą3.87b
17.33Âą4.77b
20.91Âą4.98a

16.68Âą5.14a
10.84Âą3.85b
20.05Âą4.42c
20.74Âą4.43c

14.79Âą5.24a
10.19Âą5.65b
14.36Âą4.09a
15.23Âą5.23a

25.46Âą4.32a 30.19Âą5.11

70

20.68Âą5.05a
15.26Âą4.91b
18.45Âą4.64a
15.20Âą5.61b

19.00Âą4.91a 22.96Âą4.58a
13.04Âą4.17b 19.35Âą5.82b
16.31Âą4.18a 13.99Âą4.70c
22.28Âą5.15c

25.88Âą3.99a

23.59Âą3.77b

Table S.2.2. The Pearsonâs correlation between soil respiration (R or log10R) and its potential controls (Ts, VWC, NDVI and
EVI).
2011
CROP LUH

R

C

Rs
Rh
Rs
Rh
Rs
Rh
Rs
Rh
Rs
Rh
Rs
Rh
Rs
Rh

CRP
AGR

Sw

CRP
AGR

Pr

CRP
AGR

Br

CRP

CROP LUH

R

C

Rs
Rh
Rs
Rh
Rs
Rh
Rs
Rh
Rs
Rh
Rs
Rh
Rs
Rh

CRP
AGR

Sw

CRP
AGR

Pr

CRP
AGR

Br

CRP

log R ~ Ts
r
p
0.83 2.22E-16
0.79 7.68E-14
0.68 9.64E-09
0.60 1.21E-06
0.73 3.16E-10
0.64 1.34E-07
0.64 5.81E-08
0.70 2.14E-09
0.65 3.99E-08
0.69 3.71E-09
0.80 7.59E-14
0.66 3.27E-08
0.77 7.27E-12
0.65 1.06E-07
log R ~ Ts
r
p
0.78 < 2.2E-16
0.78 < 2.2E-16
0.66
2.24E-13
0.64
2.08E-12
0.76
4.44E-16
0.78 < 2.2E-16
0.90 < 2.2E-16
0.84 < 2.2E-16
0.73
1.95E-14
0.80 < 2.2E-16
0.89 < 2.2E-16
0.84 < 2.2E-16
0.60
3.22E-09
0.71
2.03E-13

R ~ VWC
r
p
-0.42 9.46E-04
-0.25 5.71E-02
-0.29 3.37E-02
0.04 7.94E-01
-0.37 4.35E-03
-0.28 3.40E-02
-0.28 3.03E-02
-0.31 1.79E-02
-0.33 1.18E-02
-0.43 5.79E-04
-0.32 1.27E-02
-0.31 1.59E-02
-0.35 6.06E-03
-0.57 3.11E-06
2013
R ~ VWC
r
p
-0.09 3.36E-01
-0.16 9.99E-02
0.07
4.86E-01
0.01
9.08E-01
0.07
5.02E-01
-0.10 2.92E-01
-0.18 7.02E-02
-0.27 5.10E-03
-0.05 6.40E-01
-0.04 6.86E-01
-0.22 2.19E-02
-0.05 6.25E-01
-0.24 1.44E-02
-0.05 6.31E-01

log R ~ NDVI
r
p
0.67 7.09E-09
0.70 8.52E-10
0.53 3.57E-05
0.07 6.15E-01
0.63 9.91E-08
0.52 2.27E-05
0.62 1.66E-07
0.48 1.16E-04
0.47 2.02E-04
0.40 1.78E-03
0.68 3.96E-09
0.49 7.20E-05
0.63 7.13E-08
0.67 8.47E-09
log R ~ NDVI
r
p
0.38 2.09E-04
0.35 9.45E-04
0.44 7.57E-06
0.33 1.12E-03
0.69 2.09E-12
0.72 3.75E-14
0.50 7.91E-07
0.54 9.07E-08
0.69 1.43E-12
0.78 < 2.2Ee-16
0.64 2.55E-11
0.62 1.80E-10
0.60 2.37E-09
0.71 1.17E-13

log R ~ EVI
r
p
0.75 9.92E-12
0.77 1.81E-12
0.57 4.74E-06
0.14 3.25E-01
0.59 7.12E-07
0.51 3.44E-05
0.59 9.23E-07
0.43 7.79E-04
0.42 9.54E-04
0.40 1.51E-03
0.72 1.15E-10
0.47 1.80E-04
0.62 1.55E-07
0.75 8.89E-12
log R ~ EVI
r
p
0.56 8.31E-09
0.51 5.52E-07
0.57 1.73E-09
0.43 8.81E-06
0.74 2.89E-15
0.72 6.80E-14
0.55 2.2E-08
0.55 3.98E-08
0.74 4.22E-15
0.78 < 2.2e-16
0.65 1.16E-11
0.62 1.66E-10
0.69 6.97E-13
0.77 < 2.2e-16

log R ~ Ts
r
p
0.71 1.02E-14
0.80 < 2.2E-16
0.72 2.40E-14
0.72 4.89E-15
0.50 3.66E-06
0.67 1.25E-12
0.73 1.33E-15
0.79 < 2.2e-16
0.58 2.38E-09
0.60 9.70E-10
0.80 < 2.2E-16
0.82 < 2.2E-16
0.61 2.01E-09
0.71 6.44E-15
log R ~ Ts
r
p
0.72 1.67E-12
0.63 3.43E-09
0.70 1.01E-11
0.48 1.77E-05
0.73 3.09E-13
0.54 8.68E-07
0.72 6.61E-13
0.82 < 2.2E-16
0.78 4.44E-16
0.80 < 2.2E-16
0.79 2.22E-16
0.47 2.69E-05
0.53 1.49E-06
0.69 1.40E-11

2012
R ~ VWC
log R ~ NDVI
r
p
r
p
-0.24 2.25E-02
0.73 4.44E-16
-0.23 2.89E-02
0.72 1.78E-15
-0.24 3.10E-02
0.72 1.02E-14
-0.35 8.16E-04
0.70 3.73E-14
-0.42 1.17E-04
0.81 < 2.2E-16
-0.40 1.18E-04
0.67 1.86E-12
-0.39 1.66E-04
0.75 < 2.2E-16
-0.58 2.27E-09
0.78 < 2.2E-16
-0.20 5.98E-02
0.75 < 2.2E-16
-0.40 1.22E-04
0.74 4.44E-16
-0.54 7.26E-08
0.72 7.11E-15
-0.63 7.34E-11
0.76 < 2.2E-16
-0.37 6.42E-04
0.55 1.52E-07
-0.48 1.92E-06
0.44 1.49E-05
2014
R ~ VWC
log R ~ NDVI
r
p
r
p
-0.14 2.39E-01
0.33 4.31E-03
-0.02 8.95E-01
0.39 6.35E-04
-0.01 9.56E-01
0.64 1.48E-09
-0.03 8.10E-01
0.47 3.01E-05
-0.48 2.10E-05
0.72 1.15E-12
-0.08 4.81E-01
0.51 4.63E-06
-0.36 1.77E-03
0.60 3.19E-08
-0.35 2.90E-03
0.65 5.22E-10
-0.22 5.86E-02
0.79 4.44E-16
-0.14 2.40E-01
0.71 1.76E-12
-0.28 1.73E-02
0.70 7.65E-12
-0.01 9.05E-01
0.56 3.28E-07
-0.07 5.64E-01
0.65 5.90E-10
0.05 6.71E-01
0.75 3.29E-14

* CROP (crop type): C: corn; Sw: switchgrass; Pr: prairie mixture; Br: brome grass in reference
* LUH (land use history): CRP: Conservation Reserve Program; AGR: corn-soybean rotation conventional agriculture
* R (soil respiration): Rs: total soil respiration; Rh: heterotrophic soil respiration
* r: Pearsonâs correlation; p: p-value

71

log R ~ EVI
r
p
0.73 4.44E-16
0.66 1.74E-12
0.73 5.33E-15
0.70 5.82E-14
0.68 1.62E-11
0.56 2.17E-08
0.65 1.15E-11
0.71 2.18E-14
0.60 4.46E-10
0.56 1.19E-08
0.71 3.40E-14
0.70 9.77E-14
0.56 7.53E-08
0.34 1.02E-03
log R ~ EVI
r
p
0.46 5.01E-05
0.33 5.27E-03
0.64 2.06E-09
0.35 2.75E-03
0.65 6.10E-10
0.50 9.73E-06
0.57 1.61E-07
0.65 7.57E-10
0.67 9.31E-11
0.65 8.55E-10
0.76 1.78E-14
0.54 8.63E-07
0.64 1.37E-09
0.66 3.15E-10

Table S.2.3. The exponential model of total soil respiration (Rs)/heterotrophic soil
respiration (Rh) and soil temperature (Ts) and the temperature sensitivity (Q10) during
2011 and 2014. The soil coefficients a and b were calculated from the equation: R = a * ebTs,
temperature sensitivity of soil respiration (Q10) were calculated by e10b.
2011
CROP

C

LUH R
CRP

AGR

a

b

2012

Q10

r2

n

p

a

b

Q10

r2

n

p

Rs 0.73Âą1.67 0.04Âą0.03

1.57Âą1.35

0.69

59

2.73E-16

0.74Âą1.76 0.03Âą0.03 1.34Âą1.34 0.51 88

1.03E-14

Rh 0.75Âą1.72 0.04Âą0.03

1.46Âą1.34

0.63

59

7.67E-14

0.59Âą1.83 0.04Âą0.03 1.46Âą1.34 0.64 88

<2E-16

Rs 0.72Âą2.14 0.04Âą0.04

1.45Âą1.50

0.47

55

9.64E-09

1.48Âą2.51 0.04Âą0.04 1.49Âą1.54 0.46 88

3.83E-13

Rh 0.70Âą2.05 0.03Âą0.04

1.31Âą1.45

0.36

55

1.21E-06

0.40Âą2.32 0.04Âą0.04 1.48Âą1.46 0.52 88

2.28E-15

Rs 1.13Âą1.67 0.03Âą0.03

1.36Âą1.35

0.53

59

3.16E-10

1.04Âą2.01 0.02Âą0.03 1.20Âą1.40 0.25 88

3.55E-06

Rh 1.05Âą1.92 0.03Âą0.04

1.34Âą1.45

0.41

59

1.34E-07

0.63Âą2.13 0.03Âą0.03 1.35Âą1.40 0.45 88

1.25E-12

Rs 0.84Âą2.14 0.03Âą0.04

1.41Âą1.53

0.41

59

5.80E-05

0.56Âą2.45 0.05Âą0.04 1.58Âą1.55 0.53 88 1.45E--15

Rh 0.63Âą2.24 0.04Âą0.04

1.51Âą1.56

0.47

59

3.16E-09

0.44Âą2.25 0.05Âą0.04 1.59Âą1.44 0.63 88

<2E-16

Rs 1.10Âą1.92 0.03Âą0.04

1.38Âą1.48

0.42

59

3.99E-08

0.84Âą2.31 0.03Âą0.05 1.39Âą1.59 0.34 88

2.38E-09

Rh 0.97Âą1.97 0.04Âą0.04

1.42Âą1.47

0.47

59

3.71E-09

0.73Âą2.13 0.03Âą0.04 1.31Âą1.44 0.36 88

9.70E-10

Rs 0.82Âą1.74 0.04Âą0.03

1.51Âą1.37

0.65

59

7.59E-14

0.62Âą1.98 0.05Âą0.03 1.59Âą1.42 0.65 85

<2E-16

Rh 0.68Âą2.39 0.04Âą0.05

1.52Âą1.64

0.43

59

3.27E-08

0.43Âą2.02 0.05Âą0.03 1.58Âą1.38 0.68 85

<2E-16

Rs 1.18Âą1.62 0.04Âą0.03

1.43Âą1.37

0.59

59

7.27E-12

1.03Âą2.10 0.03Âą0.03 1.42Âą1.62 0.37 88

2.01E-09

(Ref)

Rh

0.62Âą2.18 0.05Âą0.05 1.60Âą1.60 0.51 88
2014

6.43E-15

CROP

LUH R

Sw CRP

AGR

Pr

CRP

AGR

Br

C

CRP

CRP

AGR

Sw CRP

AGR

Pr

CRP

AGR

Br
(Ref)

CRP

1.06Âą1.9

0.03Âą0.04

1.38Âą1.49
2013

0.42

59

1.06E-07

a

b

Q10

r2

n

p

Rs 0.55Âą2.38 0.05Âą0.04

1.65Âą1.55

0.58

107

<2E-16

0.83Âą1.90 0.03Âą0.03 1.40Âą1.40 0.51 72

1.67E-12

Rh 0.56Âą3.46 0.03Âą0.06

1.40Âą1.75

0.27

106

1.01E-08

0.72Âą2.37 0.03Âą0.04 1.38Âą1.49 0.39 72

3.43E-09

Rs 0.79Âą2.39 0.03Âą0.04

1.30Âą1.52

0.29

112

1.21E-09

0.80Âą1.94 0.03Âą0.03 1.38Âą1.40 0.49 72

1.01E-11

Rh 0.43Âą3.24 0.04Âą0.05

1.44Âą1.68

0.33

113

2.29E-11

0.71Âą2.72 0.02Âą0.04 1.27Âą1.55 0.23 72

1.77E-05

Rs 0.61Âą2.89 0.06Âą0.06

1.81Âą1.86

0.48

104

2.56E-16

0.67Âą2.69 0.06Âą0.06 1.81Âą1.75 0.53 72

3.09E-13

Rh 0.44Âą2.45 0.06Âą0.05

1.74Âą1.62

0.57

104

<2E-16

0.88Âą2.26 0.03Âą0.04 1.31Âą1.53 0.29 72

8.68E-07
6.61E-13

a

b

Q10

r2

n

p

Rs 0.71Âą2.16 0.04Âą0.04

1.55Âą1.51

0.54

104

<2E-16

0.71Âą2.48 0.05Âą0.05 162Âą1.60 0.52 72

Rh 0.38Âą2.52 0.06Âą0.05

1.77Âą1.58

0.62

105

<2E-16

0.47Âą2.27 0.05Âą0.04 1.71Âą1.47 0.66 72 < 2.2E-16

Rs 0.57Âą2.67 0.06Âą0.05

1.86Âą1.71

0.58

104

<2E-16

1.03Âą1.85 0.04Âą0.03 1.51Âą1.40 0.61 72

Rh 0.43Âą2.16 0.06Âą0.04

1.76Âą1.44

0.71

104

<2E-16

0.74Âą1.87 0.04Âą0.03 1.47Âą1.33 0.65 72 < 2.2E-16

Rs

0.562.17

5.72E-16

0.05Âą0.04

1.72Âą1.50

0.64

104

<2E-16

0.77Âą2.10 0.05Âą0.04 1.63Âą1.47 0.62 72 < 2.2E-16

Rh 0.37Âą1.86 0.05Âą0.03

1.72Âą1.34

0.78

106

<2E-16

0.86Âą2.65 0.02Âą0.04 1.26Âą1.55 0.22 72

2.69E-05

Rs 0.75Âą2.32 0.06Âą0.05

1.76Âą1.66

0.56

107

<2E-16

1.29Âą2.14 0.03Âą0.04 1.31Âą1.55 0.28 72

1.49E-06

Rh 0.47Âą2.20 0.05Âą0.04

1.67Âą1.47

0.65

104

<2E-16

0.71Âą2.32 0.04Âą0.04 1.45Âą1.48 0.48 72

1.40E-11

* CROP (crop type): C: corn; Sw: switchgrass; Pr: prairie mixture; Br: brome grass in CRP
reference site.
* LUH (land use history): CRP: Conservation Reserve Program; AGR: corn-soybean rotation
conventional agriculture.
* R (soil respiration): Rs: total soil respiration; Rh: heterotrophic soil respiration

72

Table S.2.4. The two-level interactive TWV models for Rs and Rh. The parameters of the two-level interaction multiple variable
soil respiration regression models (TWE models) from soil temperature (Ts), soil water content (VWC), enhanced vegetation index
(EVI) and the research sites with different land use histories and crop types. The function is:
logR = a + b*Ts + c*logVWC + d*EVI+ e1*(AGR-Pr) + e2*(AGR-Sw) + e3*(CRP-C) + e4*(CRP-Pr) + e5*(CRP-Ref) + e6*(CRPSw) + f*(Ts:logVWC) + g*(Ts:EVI) + h*(logVWC:EVI) + i1*(Ts:AGR-Pr) + i2*(Ts:AGR-Sw) + i3*(Ts:CRP-C) + i4*(Ts:CRP-Pr) +
i5*(Ts:CRP-Ref) + i6*(Ts:CRP-Sw) + j1*(logVWC:AGR-Pr) + j2*(logVWC:AGR-Sw) + j3*(logVWC:CRP-C) + j4*(logVWC:CRPPr) + j5*(logVWC:CRP-Ref) + j6*(logVWC:CRP-Sw) + k1*(EVI:AGR-Pr) + k2*(EVI:AGR-Sw) + k3*(EVI:CRP-C) +
k4*(EVI:CRP-Pr) + k5*(EVI:CRP-Ref) + k6*(EVI:CRP-Sw).
The blank cells were the terms removed from best-fit model via AIC. The Rs and Rh models were established separately. Both use
actual data in 2011-2014.
a

b

c

d

e1

e2

e3

e4

e5

e6

f

Rs

-2.137

0.045

0.609

4.658

0.446

0.223

0.390

0.307

0.990

0.576

0.017

-0.110 -1.154

Rh

-1.085

0.037

-0.150

1.913

0.116

0.595

0.061

0.342

0.369

-0.077

0.024

-0.095

j1

j2

j3

j4

j5

j6

k1

k2

k3

k4

k5

k6

Rs

-0.226

-0.106

-0.166

-0.003

-0.304

-0.193

-0.391

-0.378

-0.391

-0.425

-0.727

-0.705

Rh

-0.110

-0.569

-0.052

-0.215

-0.172

0.193

0.585

0.842

0.649

0.816

0.598

0.359

73

g

h

i1

i2

i3

i4

i5

i6

0.012

0.013

0.005

0.014

0.007

0.013

Table S.2.5. The effects of CROP and LUH on total (Rs) and heterotrophic soil respiration
(Rh). (A) Effect of crop type (CROP and land use history (LUH) and their interaction on total
soil respiration (Rs) and heterotrophic soil respiration (Rh). The p-value was calculated from the
two-way analysis of variance (ANOVA) by R language for each year. (B) Pair comparisons
between different CROP (C: corn, Sw: switch grass, Pr: prairie mixture and ref: brome grass in
reference site) and different LUH (CRP: Conservation Reserve Program, AGR: corn-soybean
rotation agriculture and Ref: reference). The p-value from Tukey HSD post-hoc test represented
the difference of Rs (right-up corner of table) and Rh (left-down corner of table) between
different crop types and between different LUHs.
(A)
Rs

CROP
LUH

2011
1.93e-08 ***
3.74e-10 ***

2012
1.99e-13 ***
0.00693 **

2013
<2e-16 ***
4.77e-07 ***

2014
<2e-16 ***
1.34e-05 ***

Rh

CROP
LUH

6.6e-09 ***
<2e-16 ***

9.94e-10 ***
0.00795 **

2.11e-14 ***
2.48e-06 ***

1.28e-06 ***
5.67e-05 ***

***: p-value < 0.001; **: 0.001 < p < 0.01; *: 0.01 < p < 0.05; . : 0.05 < p < 0.1
(B)
CROP
year effect
2011 Rh Rs
C
Sw
Pr
Ref
C
0.2111714 0.0092386***
0***
Sw
0.0383932*
0.6094264 0.0000172***
Pr
0.0000677*** 0.2911074
0.0010832***
Ref
0***
0.0000297*** 0.0072699***

2012

2013

2014

LUH
Rh Rs
CRP
AGR
Ref

CRP
0***
0.0131907*

AGR
Ref
0.0081931** 0.0000962***
0***
0***

CROP
Rh Rs
C
Sw
Pr
Ref
C
0.0019099** 0.0000154***
0***
Sw
0.4263716
0.8407322 0.0000026***
Pr
0.4261766
1
0.0002107***
Ref
0.0028281*** 0.1072257
0.1072911

LUH
Rh Rs
CRP
AGR
Ref

CROP
Rh Rs
C
Sw
C
0.0000006***
Sw
0.0286266*
Pr
0.0000187*** 0.2306837
Ref
0.0000105*** 0.0484249*

LUH
Rh Rs
CRP
AGR
Ref
CRP
0.0000011*** 0.0001504***
AGR 0.0002140***
0***
Ref
0.2538962 0.000214***

CROP
Rh Rs
C
C
Sw
0.0566624.
Pr
0.0001247***
Ref
0.000114***

Sw
0***
0.292746
0.1214564

Pr
0***
0.8765797

Ref
0***
0.0000859***
0.0010844**

CRP
0.7236371
0.0285623*

AGR
0.4526613

Ref
0.0000001***
0***

0.0057242**

0.7113536

Pr
Ref
0***
0***
0.0000812*** 0.0023843**
0.9999831
0.8778098

LUH
Rh Rs
CRP
AGR
Ref
CRP
0.0043459** 0.0141944*
AGR 0.0011549***
0.0000017***
Ref
0.421979 0.0005237***

***: p-value < 0.001; **: 0.001 < p < 0.01; *: 0.01 < p < 0.05; . : 0.05 < p < 0.1

74

LITERATURE CITED

75

LITERATURE CITED

Abraha, M., Gelfand, I., Hamilton, S.K., Shao, C., Su, Y.J., Robertson, G.P., Chen, J., 2016.
Ecosystem water-use efficiency of annual corn and perennial grasslands: contributions
from land-use history and species composition. Ecosystems, 19, 1001-1012.
Andresen, J., S. Hilberg, K. Kunkel, 2012. Historical climate and climate trends in the
Midwestern USA. In: U.S. National Climate Assessment Midwest Technical Input Report.
J. Winkler, J. Andresen, J. Hatfield, D. Bidwell, and D. Brown, coordinators. Available
from the Great Lakes Integrated Sciences and Assessments (GLISA) Center,
http://glisa.msu.edu/docs/NCA/MTIT_Historical.pdf.
Bahn, M., Rodeghiero, M., Anderson-Dunn, M., Dore, S., Gi- meno, C., DrĂśsler, M., Williams,
M., Ammann, C., Berninger, F., Flechard, C., Jones, S., Balzarolo, M., Kumar, S.,
Newesely, C., Priwitzer, T., Raschi, A., Siegwolf, R., Susiluoto, S., Tenhunen, J.,
Wohlfahrt, G., Cernusca, A., 2008. Soil respiration in European grasslands in relation to
climate and assimilate supply, Ecosystems, 11, 1352â1367.
Baldocchi, D., 2008. Breathing of the terrestrial biosphere: lessons learned from a global
network of carbon dioxide flux measurement systems. Australian Journal of Botany, 56
(1), 1â26.
Carbone, M.S., Still, C.J., Ambrose, A.R., Dawson, T.E., Williams, A.P., Boot, C.M., Schaeffer,
S.M., Schimel, J.P., 2011. Seasonal and episodic moisture controls on plant and microbial
contributions to soil respiration. Oecologia, 167, 265-278.
Ciais, P., Reichstein, M., Viovy, N., Granier, A., Ogee, J., Allard, V., Aubinet, M., Buchmann,
N., Bernhofer, C., Carrara, A., Chevallier, F., De Noblet, N., Friend, A.D., Friedlingstein,
P., GrĂźnwald, T., Heinesch, B., Keronen, P., Knohl, A., Krinner, G., Loustau, D., Manca,
G., Matteucci, G., Miglietta, F., Ourcival, J.M., Papale, D., Pile- gaard, K., Rambal, S.,
Seufert, G., Soussana, J.F., Sanz, M.J., Schulze, E.D., Vesala, T., Valentini, R., 2005.
Europe-wide reduction in primary productivity caused by the heat and drought in 2003.
Nature, 437 (7058), 529â533.
Curiel Yuste, Y., Janssens, I.A., Carrara, A., Meiresonne, L., Ceulemans, R., 2003. Interactive
effects of temperature and precipitation on soil respiration in a temperate maritime pine
forest. Tree Physiology, 23, 1263-1270.
Flanagan, L.B., Johnson, B.G., 2005. Interacting effects of temperature, soil moisture and plant
biomass production on ecosystem respiration in a northern temperate grassland.
Agricultural and Forest Meteorology, 130, 237-253.
Goulden, M.L., Munger, J.W., Fan, S.M., Daube, B.C., Wofsy, S.C., 1996. Exchange of carbon
dioxide by a deciduous forest: Response to interannual climate variability. Science, 271,
1576-1578.

76

Gomez-Casanovas, N., Matamala, R., Cook, D.R., Gonzalez-Meler, M.A., 2012. Net ecosystem
exchange modifies the relationship between the autotrophic and heterotrophic components
of soil respiration with abiotic factors in prairie grasslands. Global Change Biology, 18,
2532-2545.
Hanson, P.J., Edwards, N.T., Garten, C.T., Andrews, J.A., 2000. Separating root and soil
microbial contributions to soil respiration: A review of methods and observations.
Biogeochemistry, 48, 115-146.
Hoover, D., Knapp, A., Smith, M., 2014. Resistance and resilience of a grassland ecosystem to
climate extremes. Ecology, 95, 2646â 2656, doi:10.1890/13-2186.1.
Hopkins, F., Gonzales-Meler, M.A., Flower, C.E., Lynch, D.J., Czimczik, C., Tang, J., Subke,
J.A., 2013. Ecosystem-level controls on root-rhizosphere respiration. New Phytologist,
199, 339-351.
Huete, A., Justice, C., van Leeuwen, W., 1999. MODIS vegetation index (MOD 13) algorithm
theoretical basis document, Center, N.G.S.F. (Eds.), p. 129.
IPCC, 2013: Climate Change 2013: The Physical Science Basis. Contribution of Working Group
I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
[Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y.
Xia, V. Bex and P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, United
Kingdom and New York, NY, USA, 1535 pp, doi: 10.1017/CBO9781107415324.
Lloyd, J., Taylor, J.A., 1994. On the temperature dependence of soil respiration. Functional
Ecology, 8, 315-323.
Longdoz, B., Yernaux, M., Aubinet, M., 2000. Soil CO2 efflux measurements in a mixed forest:
impact of chamber distances, spatial variability and seasonal evolution. Global Change
Biology, 6, 907-917.
Luo, Y., Wan, S., Hui, D., Wallace, L.L., 2001. Acclimatization of soil respiration to warming in
a tall grass prairie. Nature, 413, 622-625.
Nagy, Z., PintĂŠr, K., CzĂłbel, S., Balogh, J., HorvĂĄth, L., FĂłti, S., Barcza, Z., Weidinger, T.,
Csintalan, Z., Dinh, N.Q., Grosz, B., Tuba, Z., 2007. The carbon budget of semi-arid
grassland in a wet and a dry year in Hungary. Agriculture, Ecosystems & Environment,
121, 21-29.
Post, W.M., Emanuel, W.R., Zinke, P.J., Stangenberger, A.G., 1982. Soil carbon pools and world
life zones. Science, 298, 156-159.
R Development Core Team, 2010. R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria.
Raich, J.W., Schlesinger, W. H., 1992. The global carbon dioxide flux in soil respiration and its
relationship to vegetation and climate. Tellus, 448, 81-99.

77

Raich, J.W., Tufekcioglu, A., 2000. Vegetation and soil respiration: Correlations and controls.
Biogeochemistry, 48, 71-90.
Raich, J.W., Potter, C.S., Bhagawati, D.B., 2002. Interannual variability in global soil
respiration, 1980-94. Global Change Biology, 8, 800-812.
Reichstein, M., Rey, A., Freibauer, A., Tenhunen, J., Valentini, R., Banza, J., Casals, P., Cheng,
Y., GrĂźnzweig, J. M., Irvine, J., Joffre, R., Law, B. E., Loustau, D., Miglietta, F., Oechel,
W., Ourcival, J. M., Pereira, J. S., Peressotti, A., Ponti, F., Qi, Y., Rambal, S., Rayment,
M., Romanya, J., Rossi, F., Tedeschi, V., Tirone, G., Xu, M., and Yakir, D., 2003.
Modeling temporal and large-scale spatial variability of soil respiration from soil water
availability, temperature and vegetation productivity indices. Global Biogeochemical
Cycles, 17(4), 1104.
Reichstein, M., Bahn, M., Ciais, P., Grank, D., Mahecha, M.D., Seneviratne, S.I., Zscheischler,
J., Beer, C., Buchmann, N., Frank, D.C., Papale, D., Rammig, A., Smith, P., Thonicke, K.,
van der Velde, M., Vicca, S., Walz, A., Wattenbach, M., 2013. Climate extremes and the
carbon cycle. Nature, 500, 287-296.
Ryan, M.G., Law, B.E., 2005. Interpreting, measuring, and modeling soil respiration.
Biogeochemistry, 73, 3-27.
Schlesinger, W.H. & Andrews, J.A., 2000. Soil respiration and the global carbon cycle.
Biogeochemistry, 48, 7-20.
Tang, J., Misson, L., Gershenson, A., Cheng, W., Goldstein, A.H., 2005. Continuous
measurements of soil respiration with and without roots in a ponderosa pine plantation in
the Sierra Nevada Mountains. Agricultural and Forest Meteorology, 132, 212-227.
Tucker, C., Pinzon, J., Brown, M., Slayback, D., Pak, E., Mahoney, R., Vermote, E., El Saleous,
N., 2005. An extended AVHRR 8-km NDVI dataset compatible with MODIS and SPOT
vegetation NDVI data. International Journal of Remote Sensing, 26, 4485-4498.
van der Molen, M.K., Dolman, A.J., Ciais, P., Eglin, T., Gobron, N., Law, B.E., Meir, P., Peters,
W., Phillips, O.L., Reich- stein, M., Chen, T., Dekker, S.C., DoubkovĂĄ, M., Friedl, M.A.,
Jung, M., van den Hurk, B.J.J.M., de Jeu, R.A.M., Kruijt, B., Ohta, T., Rebel, K.T.,
Plummer, S., Seneviratne, S.I., Sitch, S., Teuling, A.J., van der Werf, G.R., and Wang, G.,
2011. Drought and ecosystem carbon cycling. Agricultural and Forest Meteorology, 151,
765-773.
vanât Hoff, J. H., 1884. Ătudes de Dynamique Chimique (Studies of Chemical Dynamics).
Frederik Muller and Co., Amsterdam, the Netherlands.

78

CHAPTER 3
SEASONAL PATTERNS OF SOIL RESPIRATION AND ITS RESPONSE TO
MICROCLIMATE
ABSTRACT
Water deficits vary in timing, duration and magnitude, and may affect plant growth and
soil microbial activities differently. A short-term, normal seasonal water deficit and a long-term,
occasional drought may bring very different consequences to ecosystems. For example, a midsummer short-term (days to weeks) drought is common in humid ecosystems during the growing
season, and affects the soil respiration during the dry period. In the experimental treatments
studied here, soil respiration decreased when water deficit occurred, but recovered soon after
once rainfall had replenished the soil water. Many cropping systems had the highest soil
respiration peak in August following the drought in each year. I observed that a short-term water
deficit decoupled the R-Ts relationship, which quickly recovered and possibly altered the annual
soil carbon efflux depending on the timing of summer drought. In contrast to the short-term
drought, the 2012 May-July long-term (months) drought, the abnormally warm March, and the
consequent early onset of the growing season, depleted available soil water in June and induced
the obvious decrease of vegetation indices and lowered soil respiration for the whole growing
season. The result of that spring-summer 2012 drought was the extremely low Ra in the CRP-Pr,
CRP-Sw and AGR-Sw fields and very low Rh in CRP-Sw. The dramatic decrease of Ra when
VWC was low suggests that root respiration diminished greatly. The change of C3:C4 biomass
composition and ecosystem water-use efficiency implies that the community structure and the
carbon- and water-related ecological processes had altered to adapt to the drought.
I concluded that the occasional long-term spring-summer drought had very different
effects on soil respiration compared to the normal short-term summer dry period. The spring-

79

summer drought impacted the soil respiration through the whole growing season and may have
extended to the subsequent years. In the prairie and switchgrass treatments, the decreased soil
respiration in subsequent years may reflect the shift of plant species composition toward greater
dominance by C4 grasses, which may be better adapted to drought disturbance. However, we
need more research to understand how these systems respond to climate variability.

3.1 INTRODUCTION
Soil respiration, as one of the largest fluxes of carbon between ecosystems and the
atmosphere, is important for the atmospheric carbon budget and represents a major feedback
mechanism that may affect future climate changes (Goulden et al., 1996; Post et al., 1982; Raich
et al., 2002; Raich & Schlesinger, 1992; Ryan & Law, 2005; Schlesinger & Andrews, 2000). The
strong seasonality of soil respiration based on the intra-annual variations of soil temperature, soil
moisture and plant productivity has been widely reported (Luo and Zhou, 2006; Tang and
Baldocchi, 2005). However, the components of soil respiration, including autotrophic (Ra) and
heterotrophic soil respiration (Rh), have different mechanisms and feedbacks in response to
climate and human management. The net balance of carbon is determined by the sum of the
processes of Ra and Rh and the drivers that affect them. Thus, the partitioning of soil respiration
becomes essential for explaining the behavior of soil respiration and its effects on soil carbon
pools.
Autotrophic respiration includes respiration from roots and rhizosphere microbes that
generally respire photosynthetically produced organic carbon after a short lag time after
photosynthesis. In some climate regimes and biomes, Ra may be more strongly related to
daytime net ecosystem exchange (NEE) of CO2 rather than temperature. The lags of carbon
allocation from photosynthesis in leaves to root respiration was only 4-6 days (Gomez-

80

Casanovas et al., 2012). Heterotrophic soil respiration is due to the free-living soil microbial
community and is known to be controlled by soil temperature, soil moisture and daytime NEE.
The major carbon source to support Rh is from photosynthesis but it may be delayed 36-42 days
from photosynthesis (Gomez-Casanovas et al., 2012). Climate dynamics or disturbance may
impact Ra and Rh at different times intra-annually due to their distinct magnitude, the length of
lags âthe legacy effects.
Many studies have partitioned soil respiration into these two components to understand
the relative contribution of them in different ecosystems across climate regimes with diverse
results. However, studies focusing on grassland and cropping systems are very limited,
especially for those with history of land use conversion. My study is important because of the
increased land use conversion from grassland and other land use types to bioenergy crops in the
past decade (Lark et al., 2015) and the prospect for more conversion in the future. Understanding
the responses of biofuel cropping systems to climate regimes and events in the Midwest US will
help us evaluate the contribution of biofuel crops to net carbon balances in the US.
In Chapter 2, I explored how autotrophic (Ra) and heterotrophic soil respiration (Rh)
responds to climate variability, including a severe drought at inter-annual scales. However, the
timing of drought events within a growing season can alter the annual soil respiration if they
depress the growth and development of plants in certain phenological stages (Curiel Yuste et al.,
2004; Hanson et al., 2000) and can have a crucial effect on soil carbon efflux. For example,
Savage and Davidson (2001) suggested that the spring and summer droughts and the advanced
onset of spring had a stronger impact on annual soil respiration than those that happened during
the later growing season. In this chapter, I explored the effect of that seasonality of biophysical
variables have on the seasonal patterns of Ra and Rh. I hypothesized that soil temperature and the

81

photosynthesis-derived carbon control Ra and Rh during most of the growing season, while low
soil water in dry periods diminished soil respirationâeven if Ta and EVI are high. I also
examined the seasonal variations of root contribution to soil respiration (RC) in each year under
different climate patterns. I hypothesized that the normal climate regime drives the seasonal
patterns of soil respiration within most years, while a severe drought disturbance may change
ecosystem properties and the biogeochemical processes (i.e., soil respiration), as well as impact
the ecosystem well after the drought ends.

3.2 METHODS

3.2.1 Study area
My experimental sites are located at the Great Lakes Bioenergy Research Center scale-up
fields of the Kellogg Biological Station (KBS, 42Â°40âN, 85Â°40âW), established in association
with the KBS Long-Term Ecological Research (LTER) site. The sites are located in the
southwest of Michigan, USA (Fig. 1.2.). The climate is humid continental (warm summer)
climate (Dfa) (Peel et al., 2007). The mean annual air temperature and mean annual precipitation
at KBS are 10.1 Â°C and 1005 mm yr-1 (1981-2010), respectively (Robertson & Hamilton, 2015).

3.2.2 Experimental design and schedule
Seven experimental plots were chosen at two locations, each with its own land use
history (LUH): (1) Conservation Reserve Program (CRP) grasslands at Marshall Farm, and (2)
corn-soybean rotation agricultural fields (AGR) at Lux Arbor Reserve. The CRP sites have been
in a monoculture of smooth brome grass (Bromus inermis Leyss) since 1987, while the AGR
fields have been under a conventional corn-soybean rotation cultivation for several decades (Fig.
1.6). The soil texture at all sites is sandy loam, except for a sandy clay loam at one field.

82

However, soil carbon and nitrogen contents at the CRP sites were significantly higher than those
at the AGR sites before land conversion (Table 1.1).
The experiment was conducted at seven scale-up fields ranging from 9 to 17 hectares.
Four fields are located in Marshall Farm (i.e., CRP sites) and three in Lux Arbor Reserve (i.e.,
AGR sites). All sites, except the reference site (Ref), were sprayed with herbicide at the end of
2008 to prepare the lands for soybean planting in 2009. The CRP and AGR sites have been
cultivated since 2010 with either continuous corn (Zea mays, Dekalb DK-52), switchgrass
(Panicum virgatum), or a mixture of native prairie grasses, which include: Canada wild rye
(Elymus Canadensis), little bluestem (Schizachyrium scoparium), Indian grass (Sorghastrum
nutans), big bluestem (Andropogon gerardii), and switchgrass (Panicum virgatum). One CRP
grassland, which is dominated by brome grass, has not been disturbed, but instead retained as a
reference site (Ref) (Gelfand et al., 2011; Zenone et al., 2011; Deal et al., 2013; Zenone et al.,
2013). I used an experimental code for these sites by abbreviating them as âLUH-CROPâ. CRPC, CRP-Sw and CRP-Pr represent the CRP sites that were converted to corn, switchgrass and
prairie mixture respectively, while AGR-C, AGR-Sw and AGR-Pr are AGR farms converted to
corn, switchgrass and prairie mixture, respectively. In 2010 when the crops were established, the
perennial crops switchgrass and prairie were accompanied by oats as a nurse crop, with and
without fertilization, respectively (Fig. 1.6). No other management practices were applied
beyond harvesting at the end of each growing season. No-till continuous corn was seeded in midMay with a one-time herbicide mix (Lumax, Atrazine 4L, Honcho Plus and (NH4)2SO4).
Phosphorus, potassium and nitrogen fertilizers were applied during April and June each year
(Zenone et al., 2013).

83

3.2.3 The climatic and microclimatic measurements and growth season identification
The climate, such as air temperature (Ta) and daily precipitation (PRCP), and
microclimate, such as soil temperature and soil water content, were recorded as independent
variables. They were recorded at different spatial and temporal scales based on the requirement
of research and the limitation of equipment and labor.
The daily mean Ta and daily accumulative PRCP were collected and calculated from the
KBS LTER Weather station (42Â°24â47.1â N, 85Â°22â15.3â W; lter.kbs.msu.edu). Air temperature
was recorded hourly by a thermometer at 3 m height and mean daily air temperature was
calculated from the hourly data. Precipitation was measured by a NOAH IV total precipitation
gauge (ETi Instrument Systems Inc., Fort Collins, CO).
Soil temperature (Ts) and volumetric soil water content (VWC) were measured
simultaneously with biweekly soil respiration rate measurements late April to September and,
during other times, monthly. Soil temperature was measured at 10 cm depth using a Taylor
8940N digital thermometer (Taylor Precision Products, Las Cruces, NM, USA) while VWC was
monitored from the ground surface to 10 cm depth by a HydroSense II with a CS659 sensor
(Campbell Scientific, Inc. (CSI), Logan, UT).
The length of the growing season is an important index for climate change and
agriculture. There are several identifications based on climate and crop growth. Here, I use the
temperature-based identification, which is the most commonly used in North America (a.k.a.
frost-free season), instead of the crop-based growing season since soil respiration includes the
soil microbe-derived carbon effluxânot only plant-derived soil carbon efflux. The onset and the
end of the climatic growing season (GS) each year was calculated based on the nine-day moving
average of daily mean air temperature (9d-Ta). This filters out short-term warm or cold events

84

that may not induce the start of plant growth and development in spring. The onset of the GS was
identified as the first day exceeding 0ď°C 9d-Ta in spring while the end of the GS was recognized
as the last day exceeding 0ď°C 9d-Ta in fall. The length of the GS ranges between the onset and
the end of the GS in each year.
To compare soil respiration rates, climate and microclimate, and the vegetation indices, I
separated the data into different years based on the USGS âwater yearâ concept. I began my
yearly analysis with October 1st and ended at September 30th, as this method is more meaningful
for agriculture research. I identified the water year in my dissertation as the end of the previous
GS until the end of the current GS, as post-GS precipitation may subsidize soil water pools and
support the plants and soil microbes consumption in the next GS. The length of the water year
may be different due to the different seasonal patterns of air temperature in fall and winter.

3.2.4 Total and heterotrophic soil respiration measurements
Four plots at each site were randomly selected to install 1x1 m root exclusion plots for
measuring heterotrophic soil respiration (Rh). We dug a 1 m deep trench around the edges of
each square (Fig. 2.1 (B)). The trenches were lined with root-barrier sheets before refilling the
soil according to its original soil profile (Fig. 2.1 (C); Tang et al., 2005). The plants and roots
inside the square were manually removed and/or killed by herbicide (Monsanto RoundupÂŽ).
Two 10 cm inner diameter PVC collars were installed inside the square to accommodate flux
chambers to measure Rh. Another two collars were installed surrounding the square to measure
total soil respiration (Rs) (Fig. 2.1(D)). A total of 112 measurements (including Rh and Rs) were
taken between 10 am and 7 pm when soil respiration is at diel maximum.
Soil respiration was measured biweekly during the growing season and monthly during
the non-growing season to allow the measurements when snow cover is minimal (from October

85

to December). The chamber-based infrared gas analyzer (IRGA) approach for soil respiration
measurement used an LI-6400 portable photosynthesis system with a 10 cm diameter 6400-09
soil chamber or an LI-8100 with a 10 cm diameter 8100-102 soil chamber (LI-COR Biosciences,
Lincoln, NE).

3.2.5 Vegetation index data
Vegetation indices, including normalized difference vegetation index (NDVI) and
enhanced vegetation index (EVI), were calculated based on satellite remote sensing using band 1
(nir), band 2 (red), and band 3 (blue) of the Moderate Resolution Imaging Spectroradiometer
(MODIS). The NDVI was determined by the ratio of the difference between near-infrared
reflectance and red reflectance to their sum (Tucker et al., 2005), while EVI was further
modified with soil adjustment factor L and two atmospheric aerosol scattering coefficients C1
and C2 (Huete et al., 1999). The EVI has a higher sensitivity at the âgreenâ portion of the
spectrum when vegetation is very dense. These data were developed from MODIS daily products
by selecting the optimal day from each 16-day interval (i.e., low cloud cover). Missing data may
appear when clouds obscured the site over an entire 16-day period. The NDVI and EVI data
were downloaded from the Oak Ridge National Laboratory Distributed Active Archive Center
for Biogeochemical Dynamics (ORNL DAAC, http://daac.ornl.gov/MODIS/modis.shtml). I used
a single pixel from the MODIS image for each site with a spatial resolution of 250 m. The dates
of NDVI and EVI values were paired with the nearest dates of soil respiration measurement.

86

3.2.6 The temperature-water-vegetation (TWV) models
Multiple factor models containing the effects of Ts, VWC, EVI and their interactions
were developed and tested. The experiment treatments (seven sites: 2 LUH * 3 CROP + Ref)
were set as a categorical factor and the following function was tested:
đ = đ(đđ  ) â đ(đđđś) â â(đ¸đđź) â đ(đśđđđ, đżđđť)

Eq. 2-1

where R is total (Rs) or heterotrophic (Rh) soil respiration, and f(Ts), g(VWC) and h(EVI) are the
effects of soil temperature, soil moisture and vegetation condition on R, respectively.
The functions of f(Ts), g(VWC) and h(EVI) were assumed as log-linear, quadratic, or
linear based on observed relationships between R and Ts, VWC, and EVI, from previously
published research, and the best correlations between R-Ts, R-VWC and R-EVI. The equations
containing two-level interactions and the variables themselves were tested. The final models
would include the equation with first and secondary orders in terms (Supplement 2-4). The
variables and their parameters were modeled by actual data. The best models for Rs and Rh were
independently selected based on Akaike Information Criterion (AIC).

3.2.7 Seasonality of root contribution (RC) to total soil respiration
The Ra, Rh and Rs data were calculated based on the TWV models (Supplement 2-4) and
the equation (5) in Hanson et al. (2000). The root contribution (RC) of total soil respiration was
calculated based on the Hanson et al. (2000) equation (6). The RCs were plotted against DOY
and compared between different treatments.

87

3.3 RESULTS

3.3.1 Temporal variations of biophysical factors

3.3.1.1 Climate and microclimate
As mentioned in Chapter 2, the growing season (GS), defined by the non-frost season
described in Chapter 2, lasts from late-March to late November or early December in 2011, 2013
and 2014. However, it warmed up early in March 2012 and the onset of the GS advanced (Fig.
2.2).
After the onset of the growing season, the temperature gradually increased until its peak
in July and then decreased until the end of the GS (Fig. 2.2). The maximum daily mean air
temperatures were 29.88, 30.67 and 29.25 Â°C in 2011, 2012 and 2013, respectively, whereas it
was only 24.39Â°C in 2014 (data not shown).
The length of the GS, highest temperature and hottest date differed across years. The GS
in 2012 started very early (DOY=57) and finished late (DOY=356). The length of the GS in
2012 lasted almost 300 days with high degree-day (data not shown). The growing seasons in
other years started from late March (DOY=57~90) and finished in mid-November to early
December (DOY=318~341). Two of the years (2013 and 2014) had lower July temperatures
(Fig. 2.2, Table S.2.1) and lower degree-days than in 2011 and 2012 (data not shown).
In my sites, precipitation was distributed evenly through the whole year although there
were occasional very large rain events. However, VWC revealed distinct seasonal patterns in the
GS. The pre-season (October-April) recharged the soil water with precipitation, which kept soil
moisture high throughout winter and early spring. The VWC started to decline after the increase
of temperature and crop growth, sometimes also because of the lack of rainfall. The timing,

88

duration and magnitude of low VWC periods varied inter-annually. For example, the major low
VWC occurred during 3 June and 19 July in 2011, while in 2012 it was 6 June to 30 July in
2012. In 2013 and 2014, these were shortened and delayed and lasted only during 13-25 July and
26 Julyâ16 August, respectively (Fig. 3.1 (B)). The most severe water deficit always happened at
the end of the dry period, around late July in 2011-2013 and early August in 2014. In 2013 and
2014, large amounts of spring rainfall replenished soil water pools and delayed, shortened and
weakened the water deficit (Fig. 3.1 (B), Table S.2.1). A second low VWC period happened at
the end of the GS (late August to late September) when rainfall was low. However, that may not
be important for the growth of plants.
When comparing the climate patterns across years, 2012 was an outlier. First, the months
of February and March in 2012 had a mean Ta that was significantly higher than the other years.
Historically, the March mean Ta tends to be close to or below 0Â°C. However, in March 2012 the
mean Ta was 9.96Â°C while those in other years it ranged between 0.96 and -2.57Â°C (Table S.2.1).
The nine-day moving average of daily mean air temperature (9d-Ta) revealed the early onset of
warmth in the 2012 GS (Fig. 2.1). The onset of the GS based on the 9d-Ta was very early (25
February) in 2012 but later in 2011, 2013 and 2014, falling on 30, 26, and 28 March,
respectively. Secondly, the low May-August precipitation in 2012 induced the low springsummer VWC, which greatly impacted the growth of both annual and perennial crops. The
cumulative precipitation from May-July was 377, 99, 309 and 331 mm from 2011 to 2014,
respectively. The 2012 precipitation was less than one third of other years. Thus, the VWC in
June was significantly lower than in the other years (Supplement 2-1). Third, the maximum air
(Ta) and soil temperatures (Ts) were higher in 2012 than those in the other years. In contrast, the

89

coolest summer was in 2014 (Fig. 2.2). In addition, the variations of growing season Ts among
the research sites in 2012 were higher than other years, especial in July (Fig. 2.2 (A)).
The low rainfall in 2012 induced a severe water deficit in summer and VWC was very
low at all sites. However, the perennial crops in CRP sites (CRP-Sw, CRP-Pr and Ref)
experienced the most acute drought, as their June VWC levels dropped to near zero. Surface soil
in AGR perennial crop sites reached around 5% VWC, while the two corn sites fell to around
10% VWC (Fig. 3.1 (B)). The differences of VWC may be due to site-to-site variability or the
phenology of crops.

90

(A)

2011

2012

3
3
0
0

2013

3
0

2014

3
0

30
3
0

2
0

2
2
0
0

2
0

2
0

20

10

1
1
0
0

1
0

1
0

1
0

0

0
0

0
0

0
0

0

â
1
â
0
1
0

â
1
â
0
1
0

1
0

0

-10

â
1
0

â
1
â
0
1
0

(B)

CRP-C
CRP-Sw
CRP-Pr
CRP-Ref

AGR-C
AGR-Sw
AGR-Pr

precipita on

Ta

2
0

10
0

-10
â
1
0

/

/

80

8
0
3
0

8
0
3
0

8
0

6
0

6
0

6
0

6
0

2
0

2
0

2
0

4
0

4
0

4
0

4
0

1
0
2
0

1
0
2
0

1
0
2
0

2
0

0
.8

0
.8

0
.8

0.6

0
.6

0
.6

0
.6

0
.6

0.4

0
.4

0
.4

0
.4

0
.4

0.2

0
.2

0
.2

0
.2

8
0
3
0

30

VWC (%)

3
0

20

2
0

10

1
0

Ta (Â°C)

Ts (Â°C)

3
0

20

60
40
20

Precipita on
(mm)

30

/

0.8

NDVI

0
.8

(C)

0
.2

/

//

/

0
.8

0.8

0
.8

0
.8

0
.8

0.6

0
.6

0
.6

0
.6

0
.6

0.4

0
.4

0
.4

0
.4

0
.4

0.2

0
.2

0
.2

0
.2

EVI

/

Rs
Rh
(Îźmol CO2 m-2 s-1) (Îźmol CO2 m-2 s-1)

0
.2

/

/

(D)
10

1
0

1
0

1
0

5

5

5

5

1
0

5

0

0

0

0
5
0

1
0
0

1
5
0

2
0
0

2
5
0

3
0
0

5
0

1
0
0

1
5
0

2
0
0

2
5
0

3
0
0

/

0
5
0

1
0
0

1
5
0

2
0
0

2
5
0

3
0
0

10

1
0

1
0

1
0

1
0

5

5

5

5

5

0

0

0

0

50
5
0

100
1
0
0

150
1
5
0

200
2
0
0

250
2
5
0

300
3
0
0

50
5
0

100
1
0
0

150
1
5
0

200
2
0
0

250
2
5
0

300
3
0
0

/

5
0

1
0
0

1
5
0

2
0
0

2
5
0

3
0
0

50

100

150

200

250

300

0

50
5
0

100
1
0
0

150
1
5
0

200
2
0
0

250
2
5
0

300
3
0
0

5
0

1
0
0

1
5
0

2
0
0

2
5
0

3
0
0

DOY

Figure 3.1. Temporal changes of soil respiration and environmental variables across sites and years. (A) Daily air
temperature (Ta) and biweekly mean soil temperature (Ts) at 10 cm depth from each site; (B) daily precipitation from KBS LTER
weather station and soil volumetric water content (VWC) at 0-10 cm depth; (C) normalized difference vegetation index (NDVI)
and enhanced vegetation index (EVI) from MODIS; (D) biweekly total soil respiration rate (Rs); and heterotrophic soil respiration
(Rh). The soil temperature and soil moisture shown in (A) and (B) were measured simultaneously with Rs measurements.
91

3.3.1.2 Vegetation indices
Seasonal changes in NDVI and EVI were distinctly different between the annual and
perennial crops. The annual crop had only one peak, whereas the perennials had a rise-and-fall
plateau June-September and longer growing seasons. At corn sites, VIs showed a small peak
with weed growth before the corn was planted, and dropped after herbicide application in early
May. In 2011, 2013, and 2014, corn growth began immediately after planting, reached its
maximum in early July and stayed at a high biomass until late July or early August. The
maximum NDVI and EVI in 2011, 2013, and 2014 occurred between mid-July and mid-August
(solid and empty circles in Fig. 3.1 (C)). These results demonstrate how corn rapidly grew after
seeding, and how its growth coupled with a high chlorophyll density at maturation. The VIs
maintained their high values until September, decreased slightly because of the senescence and
the drying of leaves before it eventually decreased sharply after the harvesting. The growth rate
of corn seemed slower in 2012 compared to other years. The maximum VI delayed to late
September. The average height of corn was only 1.5 m, which is around 1 m shorter than other
years (data not shown). My VI data, field observations of crop conditions and eventual crop yield
measurements confirm that crop growth in 2012 was lower than other years.
In contrast to annual crop, the VIs of the perennial crops during the growing season had
three distinct stages: a rapid increase in the spring (March or April), stable high levels in summer
(June-September) and a decrease after harvest in the fall (September-October). Perennial crops
always emerge a couple of weeks after the snow melts, which is far earlier than corn is planted
(128-129 DOY). However, the temporal dynamics of crop growth in 2012 were different. The
plants emerged early in 2012 in response to the warmth in March. The March and April VIs were
higher than that in other years, as there was a higher temperature and less limitation on water in

92

their beginning weeks. However, later in July the VIs dropped, accompanied by the very low
VWC. In mid- to late August, a second peak appeared since new rainfall mitigated the water
deficit. However, the second EVI peak was lower than EVI in other years. The maximum values
of NDVI and EVI in 2012 were 0.79 and 0.64, respectively, whereas those were 0.84 and 0.62 in
2011, 0.88 and 0.73 in 2013, and 0.87 and 0.74 in 2014 (triangles and squares in Fig. 3.1 (C)).

3.3.1.3 Soil respiration
Briefly speaking, both the seasonal patterns of the total (Rs) and heterotrophic soil
respiration (Rh) were hump-shaped during the GS. The Rs and Rh had similar seasonal changes,
although Rh were less than Rs during GS, especially in their peaks (Fig. 3.2). Like the VI, the
seasonal patterns of soil respiration were different between perennial grasses and annual crop.
The R in perennial crops increased in mid- to late April when Ts increased and crops germinated.
R had several peaks during June-September and then decreased in September when Ts declined
and the leaves withered up. It sharply dropped in October after harvest. The soil respiration in
corn rose slightly in April. The values of Rs and Rh were very close. The Rs obviously increased
after mid-May when the corn was cultivated. The Rs values were higher than Rh later in the GS.

93

Figure 3.2. Temporal changes of total (Rs) and heterotrophic soil respiration (Rh). The soil respiration values are the averaged
measurements from four plots in each research field. Soil respiration was measured biweekly during the growing season (GS) and
monthly in the non-GS. The solid circles and solid lines are for Rs. The empty circles and dashed lines are for Rh.

94

3.3.2 The seasonality of the root contribution (RC) to total soil respiration
The mean annual RC fractional contributions to total soil respiration were 0.460, 0.427,
0.448, and 0.462 in 2011, 2012, 2013, and 2014, respectively. The RC was significantly lower in
2012 than that in other years (data not shown). The CRP-Pr and AGR-Sw had significantly
decreased in 2011 across all treatments (Fig. 2.13). Seasonal variations of RC in AGR-Pr and
Ref were small during the GS, while, in summer, AGR-C, CRP-C, CRP-Sw and AGR-Pr had
higher RC. The AGR-Sw and CRP-Pr had low RC in the mid- to late GS (Fig. 2.3 (A)). The
correlation between RC and VWC presented very low RCs when VWC close to zero in AGR-Sw
and CRP-Pr in 2012. In contrast to AGR-Sw and CRP-Pr, CRP-Sw had a reverse trend with low
VWC values at with high RC. Other sites did not have a distinct pattern between RC and VWC
(Fig. 2.3 (B)).

95

(A)

(B)

Figure 3.3. Temporal changes of the ratio of root contribution (RC) to total soil respiration
and the correlation between RC and VWC. (A) The temporal variations of RC (Ra:Rs ratio)
through growing seasons. The colors of circles present different years: Green, red, blue and
purple represent 2011, 2012, 2013 and 2014, respectively. (B) The correlation between RC and
VWC. The negative values of the ratio occurred when the model estimated Rh is larger than
estimated Rs.

96

3.4 DISCUSSION

3.4.1 How seasonal patterns of biophysical drivers affect the seasonal patterns of soil
respiration
The major biophysical drivers affecting soil respiration are known to be temperature, soil
water content, the substrate supply from plants, and their interactions (Luo et al., 2001; Curiel
Yuste et al., 2004). However, the influences of these biophysical drivers on soil respiration may
be different in different climate regimes and in different biomes due to the complex mechanisms
of different components of soil respiration and their feedbacks. In arid and semi-arid ecosystems,
water availability plays crucial roles in soil respiration, since it is the limiting factor (Davidson et
al., 2000; Luo et al., 2001). The dynamics of soil moisture determines the seasonal patterns of
soil respiration. In the Mediterranean climate, the soil respiration is limited by temperature in
winter, while it is depressed by soil moisture in summer. The major control during the growing
season often is soil moisture (Carbone et al., 2011). In mesic ecosystems, temperature and
radiation are always the most important drivers of seasonal soil respiration, since water supply is
less of limiting (Billing et al., 1998; Epron et al., 2001; Borken et al., 2002).
My results demonstrate that without irrigation soil respiration in these managed crop
ecosystems, which often experience hot and water-deficient summers and cold winters, is mainly
driven by temperature and that water deficit also plays an important role in the hottest and driest
periods. My experimental treatments are in a humid continental climate and soil temperature and
vegetation regulate soil respiration most of the time during the growing season. However, the
soil respiration rate in mid-July, the hottest period, was low in 2011 and 2012 (Fig. 3.2 and 3.4).
The wet springs in 2013 and 2014 diminished the water stress in the growing seasons but the soil
respiration still decreased later in late July or August, following the seasonal decrease of soil

97

moisture (Figs. 3.1 and 3.5). The findings imply that the projected increases in precipitation
variation and extreme precipitation events in the Midwest US (Andresen et al., 2012) may alter
ecosystem processes and carbon dynamics and seriously impact the ecosystem. The impacts of
drought on the ecosystem processes, such as photosynthesis or carbon dioxide losses from the
soil, in agricultural ecosystems can become more severe if the precipitation variability
exacerbates the summer water stress.

10

1
0

10

1
0

0

0

0

-10

â
1
0

CRP-C
CRP-Sw
CRP-Pr
CRP-Ref

AGR-C
AGR-Sw
AGR-Pr

precipita on

Ta

M

Rs
Rh
-1
-2
(Îźmol CO2 m s ) (Îźmol CO2 m-2 s-1)

20

20

2
0

0

VWC (%)

30

3
0

2
0

-10

â
1
0

80

8
0

30

3
0

60

6
0

20

2
0

40

4
0

10

1
0

20

2
0

Precipita on (mm)

Ts (Â°C)

3
0

Ta (Â°C)

2011

30

0.8

0
.8

0.6

EVI

0
.6

A

M

J

J

A

S

N

O

10

1
0

5

5

0

0
5
0

1
0
0

1
5
0

2
0
0

2
5
0

3
0
0

50

100

150

200

250

300

10

1
0

5

5

0

0
5
0

1
0
0

1
5
0

2
0
0

2
5
0

3
0
0

DOY

0.4

0
.4

0.2

0
.2

50

100

150

200

250

300

DOY
Figure 3.4. The seasonality of soil respiration and their drivers in 2011. (A) The seasonality
of biophysical factors (Ts, VWC and VI) were shown by different colors. (B) The seasonality of
Rs and Rh were predominantly controlled by different drivers at different times in the growing
season. The color band at upper-right corner shows their period of control. DOY means the day
of year.

98

(A)

AGRâC Rs

AGRâSw Rs

AGR-C

AGRâPr Rs

AGR-Pr

AGR-Sw

10

1
0

5

5

0

0

2011
2012
2013
2014

1
0

n
a
e
m
R
S

n
a
e
m
R
S

n
a
e
m
R
S

1
0

5

5

1
5
0

2
0
0

2
5
0

3
0
0

3
5
0

0

CRPâSw Rs

1
0
0

1
5
0

D
O
Y

2
0
0

2
5
0

3
0
0

CRP-C
1
0

50

100

150

2
0
0

200

D
O
Y

2
5
0

250

3
0
0

300

3
0
0

3
5
0

350

3
5
0

Ref
1
0

5

0

0

CRPâRef Rs

2
5
0

n
a
e
m
R
S
5

5

2
0
0

1
0

n
a
e
m
R
S

5

1
5
0

1
5
0

CRP-Pr

1
0

1
0
0

CRPâPr Rs

1
0
0

D
O
Y

CRP-Sw

10

0

3
5
0

D
O
Y

n
a
e
m
R
S

1
0
0

n
a
e
m
R
S

Rs (Îźmol CO2 m-2 s-1)

0
0

CRPâC Rs

5

0

1
0
0

100

1
5
0

150

2
0
0

200

D
O
Y

2
5
0

250

3
0
0

300

3
5
0

350

0

1
0
0

100

1
5
0

150

2
0
0

200

D
O
Y

2
5
0

250

3
0
0

300

3
5
0

350

1
0
0

100

1
5
0

150

2
0
0

200

D
O
Y

2
5
0

250

3
0
0

300

3
5
0

350

DOY

(B)

AGRâC Rh

AGRâSw Rh

AGR-C

AGR-Sw

10

AGR-Pr

n
a
e
m
R
S
5

5

0

0

0

CRPâC Rh

1
0
0

1
5
0

2
0
0

5

2
5
0

3
0
0

3
5
0

0

CRPâSw Rh

1
0
0

1
5
0

D
O
Y

2
0
0

2
5
0

3
0
0

CRP-C

0

1
5
0

150

2
0
0

200

D
O
Y

2
5
0

250

3
0
0

300

3
0
0

3
5
0

350

3
5
0

Ref
1
0

5

0

1
0
0

100

CRPâRef Rh

2
5
0

n
a
e
m
R
S
5

50

2
0
0

1
0

n
a
e
m
R
S

0

1
5
0

CRP-Pr

1
0

n
a
e
m
R
S

1
0

5

CRPâPr Rh

1
0
0

D
O
Y

CRP-Sw

10

5

3
5
0

D
O
Y

n
a
e
m
R
S

5

2011
2012
2013
2014

1
0

n
a
e
m
R
S

1
0

n
a
e
m
R
S

1
0

Rh (Îźmol CO2 m-2 s-1)

AGRâPr Rh

5

0

1
0
0

100

1
5
0

150

2
0
0

200

D
O
Y

2
5
0

250

3
0
0

300

3
5
0

350

0

1
0
0

100

1
5
0

150

2
0
0

200

D
O
Y

2
5
0

250

3
0
0

300

3
5
0

350

1
0
0

100

1
5
0

150

2
0
0

200

D
O
Y

2
5
0

250

3
0
0

300

3
5
0

350

DOY

Figure 3.5. The seasonality of Rs and Rh across four years. (A) Rs and (B) Rh were plotted
with DOY, which means the day of year. The measurement of Rs and Rh were biweekly during
late-April and September and monthly in October to December. The colors and different types of
lines display different years.

The summer decrease of R with maximum soil temperature displays the decoupling of
soil respiration and soil temperature that happened commonly in all experimental years. There

99

are two main hypotheses for the decoupled respiration-temperature relationship. The first is heat
suppression of biochemical reactions. High temperatures may break down the protoplasm system
and depress the metabolism of roots and soil microbes. The second is limitation of water deficit
to plant roots and microorganisms. High temperatures increase both transpiration and
evaporation rates with high photosynthetic and metabolic rates and high diffusion rates of water
from the surface of the ground to the atmosphere, respectively. Maximum temperature always
occurred simultaneously with minimum soil moisture in late July. The first hypothesis was not
supported by my data since the thermal threshold for the suppression of biochemical reactions is
generally higher than 35ď°C and, although it may vary among different plant species and soil
microbes (pp. 86-87, Luo & Zhou, 2006), soil temperatures in my data were always less than
35ď°C. The second hypothesis is more likely because of R-T decoupled since the July VWC fell
to below 10%. The low VWC (for weeks) in late summer depressed the metabolism of roots and
soil microbes, and thus decreased soil respiration.

3.4.2 The seasonality of Ra and Rh in the dry and hot year
The impacts of drought on ecosystems can vary depending on the magnitude, timing and
duration of the drought. My treatments experienced a severe drought in spring-summer 2012
when I found a strong ecosystem response that included the lower soil respiration, lower annual
yield, lower summer NDVI and EVI, lower plant heights and summer wilt in some research sites,
followed by the change of plant composition in 2012 and the following years in the switchgrass
and prairie fields (Abraha et al., 2016). Two main events with biological feedback caused the
severe spring-summer water deficit. First, the thaw was earlier due to unusually hot March (Fig.
2.3 & Table S.2.1). The warm ground stimulated the early germination of weeds and thus
consumed soil water via high March evapotranspiration (ET) (Abraha et al., 2015). Second, the

100

low May-August precipitation did not fully replenish the soil water pool and therefore did not
satisfy the water demands of crops and soil microbes. The depletion of soil moisture in June
suspended the growth and development of crops. Both NDVI and EVI values significantly
decreased in June and July in all fields, and biomass did not rebound until August (Figs. 2.2 (C)
& (D)).
Savage and Davidson (2001) reported similar findings in their soil respiration research in
Harvard Forest âuplandâ (well and moderately well drained) mixed hardwood forest in
Massachusetts. They argued that the late summer low soil respiration rates were due to low soil
moisture and were dependent on the duration and severity of the preceding dry period and the
timing of the onset of soil respiration increase in the previous spring. The preceding inter-annual
variation of soil water availability may be a crucial factor affecting the soil respiration rate,
although the correlation between soil respiration and soil moisture is usually not clear and
confounded with other factors. The effect of soil moisture on soil respiration is more like a
threshold, rather than a linear or quadratic correlation, for a well-drained habitat. The soil
respiration rates were markedly suppressed when soil moisture was lower than the threshold. I
will discuss more about how the community structure, ecological processes and ecosystem
functions have adapted to growing-season drought (months) in the years in later chapters.
Compared to the short-term inhibition of soil respiration by water deficit in 2011, 2013
and 2014, the severe and long-term (months) drought of 2012 had great and multifaceted impacts
on agricultural ecosystems, affecting plant community structure, biodiversity, and ecological
processes and functions. My 2012 results revealed that there was a decreased growing-seasonal
PRCP, VWC, NDVI, EVI, annual yield, Rs and Rh, although the annual ET was relatively
consistent. The adaptation of ecosystems to growing-season drought may include changes in

101

ecosystem water use efficiency (eWUE) via shifts in plant species composition and specifically
the biomass ratio of C3/C4 species. The changes of plant composition and ecosystem processes in
response to the 2012 drought lasted into the years following (Abraha et al., 2016). I will discuss
how ecosystem functions respond to the common short-term drought in the late growing seasonal
and the occasional long-term growing-seasonal drought in Chapter 4.

3.4.3 The seasonality of Ra and Rh in humid springs and cool summers
Post-drought years 2013 and 2014 had similar GS climate patterns, with wet springs (or
wet spring-summer periods) and cool summers; although 2014 had a wetter and cooler summer
than 2013 (Fig. 3.1 and Table S.2.1). The wet spring fulfilled the water demands in the early GS
and supplied enough water to plants and soil microbes. The peaks of soil respiration in those two
years were in the hottest month, July, during the middle of the growing season. The humid and
cool summer brought high soil respiration in July, but the August and September rates were
lower. Surprisingly, soil respiration did not recover to the 2011 level. In contrast to the steady
decrease of R in 2012, the summer R peaks were flattened or did not occur (Fig. 3.5).

3.4.4 Seasonal variation in RC
Seasonal patterns of RC can be complex because they reflect the relative dynamics of Ra
and Rh and their different responses to environmental change. The three peaks of RC can be
explained separately by the phenology of plants and seasonal activity of soil microbial
communities. The early-spring peak represented rapid increases of Ra due to the emergence of
crops, even though raising temperature stimulates both Ra and Rh. The rise of Rh was slower than
Ra and that resulted in lower RC in the mid-spring to early summer. The second peak usually
happened in late July during the hottest and driest period. The lack of water depressed the soil

102

microbial activity more than root metabolism since roots can obtain water from deeper horizons
(Carbone et al., 2011). The RC decreased slightly after a couple of heavy rains in late July or
early August. The responses of soil microbes to rewetting were quicker and larger than root
response. The third peak occurred in November and December when the decrease of temperature
limited most activities of soil microbes and roots but the relative contribution of Ra to Rs was
higher than most of the growing season, except in the spring and summer peaks.
In addition to the general three-peak patterns, different treatments had their own patterns
in response to drought. The Ref field had relatively consistent RC through the growing season
regardless of the soil moisture, which was extremely low in summer 2012 (Fig. 3.1 (B)). The
responses of Ra and Rh to low soil moisture were thus similar (Fig. 3.6 (B)). The RC in CRP-Pr
and AGR-Sw had similar relationships to VWC (Fig. 3.3 (B)) due to the differential decreases of
Ra and Rh when VWC decreased. The magnitude of Ra decreases was higher than those of Rh
(Fig. 3.6 (A) and (B)). The Ra in these two sites dropped to near zero when soil moisture was
low. I found that most of the plants in these two sites wilted during the summer 2012 drought.
Combining field observations, the low Ra and the VI data, I believe that the 2012 water deficit
had decreased available soil water even deep belowground. Plants were barely getting enough
water to support their basic metabolism when soil water is very low. The CRP-Sw responded
differently in that both Ra and Rh decreased in the 2012 drought but Rh decreased more than Ra
(Fig. 3.6 (A) and (B)), resulting in increased RCs when soil moisture was low (Fig. 3.3 (B)). The
two corn sites contained higher soil water compared to perennial treatments since we seeded corn
in early May. The consumption of water from the corn crop in spring was not as large as that in
the perennial crops. Their RC in general does not change much; but in the summer of 2012, RC
in both sites were lower than in other years (Fig. 3.3 (A)).

103

The CRP fields may have experienced more severe drought stress than the AGR fields in
2012 because of soil differences. Three CRP-perennial cropping farms, Ref, CRP-Pr and CRPSw had almost zero VWC while AGR-Pr and AGR-Sw had around 6% and two corn sites had
almost 10% VWC at the driest time in 2012 (Fig. 3.1 (B)). Three CRP-perennial crop sites
showed a sudden drop in Ra when VWC fell below 5% (Fig. 3.6 (A)), suggesting that the root
system lost its ability to respire. A dramatic drop in Rh also happened in the Ref and CRP-Sw
sites. The combination of the decreases in Ra and Rh in different sites changed RCs in the
summer of 2012. At the driest time, CRP-Pr and AGR-Sw had very low RC because of the
extremely low Ra. Ref had a relatively stable RC since both Ra and Rh were low. The CRP-Sw
had a high RC since the decrease of Rh was more than that of Ra. The Ra can be higher than Rh
when VWC is low, since plants can uptake soil water from deeper soil layers. However, the 2012
drought seems to of deeply drained the soil water and both switchgrass and prairie plants had
difficulty to obtaining water.
Compared to Ra at low VWC (<10%) levels across different years, the Ra in Brome grass
(Ref) and prairie sites was lower in 2012-2014 than that in 2011. There are some possible
reasons for the lower Ra. First, 2013 and 2014 had lower summer temperatures while 2012 had
lower VI, which reflects lower biomass and photosynthetic rates. The lower temperature and
lower photosynthesis-derived carbon inputs to roots reduced Ra. The second is that the 2012
drought changed the species composition of vegetation and that the changes beyond that year.
Abraha et al. (2016) reported this dramatic decrease in the C3:C4 ratio of grasses in the CRP-Sw,
AGR-Pr and AGR-Sw sites (Fig. 3.7 (A)). Switchgrass and prairie in AGR had higher ecosystem
water use efficiency (eWUE) after 2012 (Fig. 3.7 (B)). The first reason seems robust while the
latter one needs more evidence to improve or modify it.

104

(A)

(B)

Figure 3.6. The responses of Ra and Rh to VWC. (A) Ra and (B) Rh were correlated to VWC
differently in different treatments. The responses of the components of Rs changed the temporal
patterns of RC during growing season. Different years are exhibited by different colors.

105

(A)

(B)

Figure 3.7. The changes of C3:C4 biomass ratio and ecosystem water-use efficiency
(eWUE). (A) The changes of biomass of C3 and C4 grasses in prairie and switchgrass sites after
land use change; and (B) the variation of eWUE across years in all seven sites. The eWUE was
calculated by GPP/ET, where GPP is gross primary production and ET is evapotranspiration.
The letters present the difference across years in the same site while the asterisks show the
difference between land use histories in the same crop. Notice that the experimental period was
2010-2013 (Source: Abraha et al., 2016).

3.5 CONCLUSIONS
Amplitude, duration and timing of extreme drought events leave consequences on
ecosystem processes, such as Ra and Rh. At the intra-annual scale, I found that soil temperature
and EVI had an exponential correlation with both Ra and Rh. However, soil water deficits limited
soil respirationâeven soil temperature and EVI were high. In addition, the severe drought,
which had a high amplitude, long duration and occurred during spring-summer, seriously
impacted Ra and Rh. The impact of the occasional severe drought altered plant composition (RC),
which had responded to biophysical variables at inter- and intra-annual scales due to the different
magnitudes and speeds of Ra and Rh responses to biophysical variables. Three peaks of RC
occurred in early spring plant emergence, mid-summer dry and the winter low temperature
106

period. Different land use histories (LUH) and crop types (CROP) had different responses to the
2012 severe drought. Most perennial sites had lower RC in 2012, demonstrated that severe
drought had stronger depression on Ra than on Rh. The especially low summer RCs in CRP-Pr
and AGR-Sw fields and very high RCs in CRP-Sw field reflected the different responses of Ra
and Rh to different levels of water deficit underneath different composition of LUH and CROP.
Furthermore, the opposite responses of RC to the regular summer low soil moisture and to the
occasional severe drought revealed that short-term (1-2 weeks) summer drought depressed Rh
more than Ra while occasional severe drought impacted Ra more than Rh.
The depression of both Ra and Rh can be predicted if the occasional severe drought events
increases in the future, especially for Ra. The decrease of root respiration ability implies that the
decrease of metabolism and photosynthetic ability may increase NEE (decrease carbon
sequestration ability). However, perennial crops may adapt to low water conditions and mitigate
the impact of frequent severe drought by altering plant composition and increasing eWUE, while
corn cannot. The benefit of perennial cellulosic biofuel crops will enlarge the difference between
itself and annual crops.

107

LITERATURE CITED

108

LITERATURE CITED

Abraha, M., Chen, J., Chu, H., Zenone, T., John, R., Su, Y.J., Hamilton, S.K., Robertson,
G.P., 2015. Evapotranspiration of annual and perennial biofuel crops in a variable
climate. Global Change Biology Bioenergy, 7, 1344-1356.
Abraha, M., Gelfand, I., Hamilton, S.K., Shao, C., Su, Y.J., Robertson, G.P., Chen, J.,
2016. Ecosystem water-use efficiency of annual corn and perennial grasslands:
contributions from land-use history and species composition. Ecosystems, 19,
1001-1012.
Andresen, J., S. Hilberg, K. Kunkel, 2012. Historical climate and climate trends in the
Midwestern USA. In: U.S. National Climate Assessment Midwest Technical Input
Report. J. Winkler, J. Andresen, J. Hatfield, D. Bidwell, and D. Brown,
coordinators. Available from the Great Lakes Integrated Sciences and Assessments
(GLISA) Center, http://glisa.msu.edu/docs/NCA/MTIT_Historical.pdf.
Billings, S.A., Richter, D.D., and Yarie, J., 1998. Soil carbon dioxide fluxes and profile
concentrations in two boreal forests. Canadian Journal of Forest Research, 28,
1773â1783.
Borken, W., Muhs, A., and Beese, F., 2002. Application of compost in spruce forests:
Effects on soil respiration, basal respiration and microbial biomass. Forest Ecology
and Management, 159, 49â58.
Carbone, M.S., Still, C.J., Ambrose, A.R., Dawson, T.E., Williams, A.P., Boot, C.M.,
Schaeffer, S.M., Schimel, J.P., 2011. Seasonal and episodic moisture controls on
plant and microbial contributions to soil respiration. Oecologia, 167, 265-278.
Curiel Yuste, Y., Janssens, I.A., Carrara, A., Meiresonne, L., Ceulemans, R., 2003.
Interactive effects of temperature and precipitation on soil respiration in a
temperate maritime pine forest. Tree Physiology, 23, 1263-1270.
Curiel Yuste, J., Janssens, I. A., Carrara, A., and Ceulemans, R., 2004. Annual Q10 of
soil res- piration reflects plant phonological patterns as well as temperature
sensitivity. Global Change Biology, 10, 161â169.
Davidson, E.A., Verchot, L.V., Cattanio, J.H., Ackerman, I.L., Carvalho, J.E.M., 2000.
Effects of soil water content on soil respiration in forests and cattle pastures of
eastern Amazonia. Biogeochemistry, 48, 53â69.
Deal, M.W., Xu, J., John, R., Zenone, T., Chen, J., Chu, H., Jasrotia, P., Kahmark, K.,
Bossenbroek, J., Mayer, C., 2013. Net primary production in three bioenergy crop
systems following land conversion. Journal of Plant Ecology, 6, 1-10.

109

Epron, D., Le Dantec, V., Dufrene, E., and Granier, A., 2001. Seasonal dynamics of soil
carbon dioxide efflux and simulated rhizosphere respiration in a beech forest. Tree
Physiology, 21, 145â152.
Flanagan, L.B., Johnson, B.G., 2005. Interacting effects of temperature, soil moisture and
plant biomass production on ecosystem respiration in a northern temperate
grassland. Agrucultural and Forest Meteorology, 130, 237-253.
Gelfand, I., Zenone, T., Jasrotia, P., Chen, J., S.K.Hamilton, Robertson, G.P., 2011.
Carbon debt of conservation reserve program (CRP) grasslands converted to
bioenergy production. Proceedings of the National Academy of Sciences, 108,
13864-13869.
Gomez-Casanovas, N., Matamala, R., Cook, D.R., Gonzalez-Meler, M.A., 2012. Net
ecosystem exchange modifies the relationship between the autotrophic and
heterotrophic components of soil respiration with abiotic factors in prairie
grasslands. Global Change Biology, 18, 2532-2545.
Goulden, M.L., Munger, J.W., Fan, S.M., Daube, B.C., Wofsy, S.C., 1996. Exchange of
carbon dioxide by a deciduous forest: Response to interannual climate variability.
Science, 271, 1576-1578.
Hanson, P.J., Edwards, N.T., Garten, C.T., Andrews, J.A., 2000. Separating root and soil
microbial contributions to soil respiration: A review of methods and observations.
Biogeochemistry, 48, 115-146.
Huete, A., Justice, C., van Leeuwen, W., 1999. MODIS Vegetation Index (MOD 13)
Algorithm Theoretical Basis Document, Center, N.G.S.F. (Eds.), p. 129.
Jassal, R.S., Black, T.A., Novak, M.D., Gaumont-Guay, D., Nesic, Z., 2008. Effect of
soil water stress on soil respiration and its temperature sensitivity in an 18-year-old
temperate Douglas-fir stand. Global Chang Biology, 14, 1305-1318.
Lark, T.J., Salmon, J.M., Gibbs, H.K., 2015. Cropland expansion outpaces agricultural
and biofuel policies in the United States. Environmental Research Letters, 10,
044003.
Luo, Y., Wan, S., Hui, D., Wallace, L.L., 2001. Acclimatization of soil respiration to
warming in a tall grass prairie. Nature, 413, 622-625.
Luo, Y., Zhou, X., 2006. Soil Respiration and the Environment. Elsevier.
Peel, M.C., Finlayson, B.L., McMahon, T.A., 2007. Updated world map of the KoppenGeiger climate classification. Hydrology and Earth System Sciences Discussions 4,
439-473.Post, W.M., Emanuel, W.R., Zinke, P.J., Stangenberger, A.G., 1982. Soil
carbon pools and world life zones. Science, 298, 156-159.

110

Post, W.M., Emanuel, W.R., Zinke, P.J., Stangenberger, A.G., 1982. Soil carbon pools
and world life zones. Science, 298, 156-159.
Raich, J.W., Schlesinger, W. H., 1992. The global carbon dioxide flux in soil respiration
and its relationship to vegetation and climate. Tellus, 448, 81-99.
Raich, J.W., Potter, C.S., Bhagawati, D.B., 2002. Interannual variability in global soil
respiration, 1980-94. Global Change Biology, 8, 800-812.
Ryan, M.G., Law, B.E., 2005. Interpreting, measuring, and modeling soil respiration.
Biogeochemistry, 73, 3-27.
Savage, K., Davidson, E., 2001. Interannual variation of soil respirationin two New
England forests. Global Biogeochemical Cycles, 15, 337-350.
Schlesinger, W. H. & Andrews, J. A., 2000. Soil respiration and the global carbon cycle.
Biogeochemistry, 48, 7-20.
Tang, J., and Baldocchi, D. D., 2005. Spatial-temporal variation in soil respiration in an
oak-grass savanna ecosystem in California and its partitioning into autotrophic and
heterotrophic components. Biogeochemistry, 73(1), 183â207.
Tucker, C., Pinzon, J., Brown, M., Slayback, D., Pak, E., Mahoney, R., Vermote, E., El
Saleous, N., 2005. An extended AVHRR 8-km NDVI dataset compatible with
MODIS and SPOT vegetation NDVI data. International Journal of Remote Sensing,
26, 4485-4498.
Zenone, T., Chen, J., Deal, M.W., Wilske, B., Jasrotia, P., Xu, J., Bhardwaj, A.K.,
Hamilton, S.K., Robertson, P.G., 2011. CO2 fluxes of transitional bioenergy crops:
effect of land conversion during the first year of cultivation. GCB Bioenergy, 3,
401-412.
Zenone, T., Gelfand, I., Chen, J., Hamilton, S.K., Robertson, G.P., 2013. From set-aside
grassland to annual and perennial cellulosic biofuel crops: Effects of land use
change on carbon balance. Agricultural and Forest Meteorology, 182-183, 1-12.
Zhang, Q., Lei, H.M., Yang, D.W., 2013. Seasonal variations in soil respiration,
heterotrophic respiration and autotrophic respiration of a wheat and maize rotation
cropland in the North China Plain. Agricultural and Forest Meteorology, 180, 3443.

111

CHAPTER 4
ECOSYSTEM RESISTANCE AND RESILIENCE TO SEVERE DROUGHTS: A
BAYESIAN MODELING OF SOIL RESPIRATION
ABSTRACT
The intensification of drought magnitude and frequency is predicted to occur in the future
due to global climate change. The altered precipitation patterns and increasing precipitation
extremes increase soil water stress and impact ecosystem structure and function with immediate
and prolonged effects. How the major carbon processes of grasslands and croplands respond to
the severe droughts play important roles in scientific understanding and evidence-driven policy
making for biofuel development. However, at present, most drought effects on ecosystem carbon
cycle studies have only focused on immediate effects based on manipulated lab or field
experiments. I monitored the responses of soil respiration components of multiple crops under
different land use histories for four years, which included one wet-hot year, one severe dry-hot
year, and two wet-cool years. This dataset allows us to examine the resistance and recovery of
one of the major carbon cycling processes, soil respiration, to a naturally occurring drought.
Bayesian models help us test the difference of the temperature sensitivity (Î˛) and
growing-season soil respiration (LR20) of autotrophic (Ra) and heterotrophic soil respiration (Rh)
across different crop types (CROP) and land use histories (LUH) and their responses after severe
drought. My results revealed that CROP affected Ra more whereas LUH influenced Rh more.
Both Rh and Rs had different Î˛ responses. Strong water stress depressed both Î˛ and LR20 of Ra in
the drought year but stimulated Î˛ of Rh. The increased Î˛ of Rh may be due to the change in the
soil microbial community structure, but more research is still required to quantify these
responses. The LR20 of Rs rose back to the 2011 levels after a one- or two-year lag in recovery,

112

while those of Rh did not return back to the values of 2011 yet. The Î˛ of Rh seems to require
longer recovery time or has reached a new equilibrium.

4.1 INTRODUCTION
It is predicted that future climate changes will include a greater intensity and a greater
variability in extreme events in many parts of the world. Heat waves and precipitation extremes
(droughts, heavy rains and floods) are projected to occur with higher frequencies and magnitudes
(IPCC, 2013; Smith, 2011; Knorr et al., 2007) due to the amplification of the hydrological cycle
(Knapp et al., 2008). Rising global temperatures and the alteration of precipitation patterns have
already been reported in many regions, significantly impacting various ecosystems. The
intensified precipitation regimes, including larger individual precipitation events and longer
intervening dry periods, on both intra- and the inter-annual scales, induce stronger soil water
stress in mesic ecosystems and modify ecosystem functions by changing community structure
and individual traits, seriously disturbing ecosystems (Knapp et al., 2008).
Smith (2011) and Reichstein et al. (2013) redefined extreme climatic events (ECE) to not
only consider the statistically unusual climate periods (driver) but also the changes of ecosystem
functions (response) during or after the period of extreme climatic events. The new definition is
not only based on statistics, but also considers the responses of an ecosystem related to important
ecological processes and their drivers, and thus is better suited for ecological and environmental
research.
The resilience of an ecosystem to an external disturbance has multiple definitions.
Holling (1996) identified âecological resilienceâ as the magnitude of the disturbance that the
system can absorb before the system changes its structure and the processes that control the
behavior of the system; he also defines âengineering resilienceâ as the speed of return to the pre-

113

disturbance equilibrium. These two definitions of resilience were developed from different
perspectives: the former focuses on the persistence of the behavior and whether the change of
ecosystem functions and the relative processes and structure exist or not, while the latter
concerns the recovery speed and the time to return to the stable status (Holling, 1973). They
represent two important aspects of the systemâs behavior after the system experiences a
disturbance, reflecting the immediate and prolonged effects of the disturbance.
This chapter addresses the resilience of soil respiration in biofuel cropping systems. To
mitigate the impacts of CO2 emissions from fossil fuel combustion, the production of bioenergy
has dramatically increased in the past decades. The growing bioenergy industry has stimulated
worldwide land use change, converting many land use types into biofuel cropping farms or into
agricultural land to address food shortage, which is caused by biofuel production. The expansion
of biofuel crop production plays an important role in land use change now and in the future (Lark
et al., 2015). Major biofuel crops include corn and several perennial cellulosic herbaceous
grasses, such as switchgrass, Miscanthus and elephant grass (Lemus & Lal, 2005; Johnson et al.,
2007).
Drought, as one of the most import climate-induced extremes, critically impacts the
carbon cycling of grasslands and croplands (Reichstein et al., 2013; Ciais et al., 2005). It may
bring strong soil water stress, affect the plantâs growth and soil microbial activities, and
ultimately impact ecosystem carbon dynamics. On the other hand, grasslands and cropping
systems have high recovery plant growth potential in both managed and unmanaged ecosystems
and thus, may perform better in dry years than forests (Zavalloni et al., 2008; Gilgen &
Buchmann 2009; Reichstein et al., 2013). Most studies of drought consequences on ecosystems
emphasize immediate responses. However, the protracted recovery of ecosystem functions lacks

114

documentation, even though it may be more important (Hoover et al., 2016). In addition, most
drought or drought-rewetting studies are manipulation experiments in the lab or in the field. The
manipulation of water input may change some physical conditions and the radiation balance, and
ultimately affect ecosystem responses.
The naturally occurring inter-annual climate variability (wet-hot, severely dry-hot, wetcool, wet-cool, in sequence) in 2011-2014 at my experimental sites provided me with a good
opportunity to examine how ecosystems respond to the severe drought in both the disturbance
year and the following recovery years. In Chapter 2, I looked for differences in Q10 between
different sites in each year but found no significant difference among sites or years because of
high variability in each treatment. The Bayesian modeling approach outperforms the
conventional frequentist statistical approach for testing the changes of temperature sensitivities
of soil respiration among LUH and CROP, the year-to-year trend after LUC or the immediate
and prolonged effects after severe drought. In this chapter, I develop a new method, based on a
probabilistic Bayesian approach, to estimate how the relationships of soil respiration, including
Rh and Rs, and soil temperature respond to intra-annual climate patterns, severe drought and
long-term trends after LUC.
Specifically, I developed Bayesian models to test the R-Ts relationship among CROP and
LUH and clarified their immediate and prolonged responses to severe drought. I hypothesized
that perennial crops, which develop their dense root system after lands use change, have higher
probabilities to weaken the impact of drought, have higher resistance and lower sensitivity to
drought and have a high rate to recover after drought. The late successional grasslands may have
high resistance and resilience of soil respiration temperature sensitivity to severe drought due to
its highly developed root system while those in corn fields are weak.

115

4.2 METHODS

4.2.1 Study area
My experimental sites are located at the Great Lakes Bioenergy Research Center scale-up
fields of the Kellogg Biological Station (KBS, 42Â°40âN, 85Â°40âW), established in association
with the KBS Long-Term Ecological Research (LTER) site. The sites are located in the
southwest of Michigan, USA (Fig. 1.2.). The climate is humid continental (warm summer)
climate (Dfa; Peel et al., 2007). The mean annual air temperature and mean annual precipitation
at KBS are 10.1 Â°C and 1005 mm yr-1 (1981-2010), respectively (Robertson & Hamilton, 2015).

4.2.2 Experimental design and schedule
Seven experimental plots were located at two locations with their own land use histories
(LUHs): (1) Conservation Reserve Program (CRP) grasslands at Marshall Farm, and (2) cornsoybean rotation agricultural fields (AGR) at Lux Arbor Reserve. The CRP sites have been in a
monoculture of smooth brome grass (Bromus inermis Leyss) since 1987, while the AGR fields
have been under conventional corn-soybean rotation cultivation for several decades (Fig. 1.6).
The soil texture at all sites is sandy loam, except for a sandy clay loam at one field. However,
soil carbon and nitrogen contents at the CRP sites were significantly higher than those at the
AGR sites before land conversion (Table 1.1).
The experiment was conducted at seven scale-up fields ranging from 9 to 17 hectares.
Four fields are located in Marshall Farm (i.e., CRP sites) and three in Lux Arbor Reserve (i.e.,
AGR sites). All sites, except the reference site (Ref), were sprayed with herbicide at the end of
2008 to prepare the lands for soybean planting in 2009. The CRP and AGR sites were then
cultivated with either continuous corn (Zea mays, Dekalb DK-52), switchgrass (Panicum

116

virgatum), or a mixture of native prairie grasses, which have been dominated by Canada wild rye
(Elymus Canadensis), little bluestem (Schizachyrium scoparium), Indian grass (Sorghastrum
nutans), big bluestem (Andropogon gerardii), and switchgrass (Panicum virgatum) since 2010.
A particular CRP grassland dominated by brome grass was not disturbed and was retained as the
reference site (Gelfand et al., 2011; Zenone et al., 2011; Deal et al., 2013; Zenone et al., 2013). I
used an experimental code for these sites by abbreviating them as âLUH-CROPâ. The CRP-C,
CRP-Sw and CRP-Pr represent the CRP sites that were converted to corn, switchgrass and
prairie mixture respectively, while AGR-C, AGR-Sw and AGR-Pr are AGR farms were
converted to corn, switchgrass and prairie mixture. In 2010, when the crops were established, the
perennial crops switchgrass and prairie were accompanied by oats as a nurse crop with and
without fertilization, respectively. No other management practices were applied beyond
harvesting at the end of each growing season (GS). No-till continuous corn was seeded in midMay with a one-time herbicide mix (Lumax, Atrazine 4L, Honcho Plus and (NH4)2SO4).
Phosphorus, potassium and nitrogen fertilizers were applied during April and June in each year
(Zenone et al., 2013).

4.2.3 The climatic and microclimatic measurements and growth season identification
Climate, such as air temperature (Ta) and daily precipitation (PRCP), and microclimate,
such as soil temperature and soil water content, were recorded as the independent variables. They
were recorded at different spatial and temporal scales based on the requirement of the research
and the limitation of the equipment and labor.
The daily mean Ta and daily accumulative PRCP were collected and calculated from the
KBS LTER weather station (42Â°24â47.1â N, 85Â°22â15.3â W; www.lter.kbs.msu.edu). The air
temperature was recorded hourly by a thermometer at 3 m height. The mean daily air

117

temperature was calculated from the hourly data. Precipitation was measured by a NOAH IV
total precipitation gauge (ETi Instrument Systems Inc., Fort Collins, CO).
Soil temperature (Ts) and volumetric soil water content (VWC) were measured
simultaneously with soil respiration rate measurements biweekly during late April to September
and monthly during other times. Soil temperature was measured at 10 cm depth using a Taylor
8940N digital thermometer (Taylor Precision Products, Las Cruces, NM, USA), while VWC was
monitored below the ground surface at 10 cm depth using a HydroSense II with a CS659 sensor
(Campbell Scientific, Inc. (CSI), Logan, UT).
The length of the GS is an important index for climate change and agriculture. Several
identifications are based on climate and crop growth. I used temperature-based identification,
which is the most common in North America (a.k.a. frost-free season) instead of crop-based GS
since soil respiration includes the soil microbe-derived carbon efflux, not only plant-derived soil
carbon efflux. The onset and end of the climatic GS each year were calculated based on a nineday moving average of daily mean air temperature (9d-Ta) to filter short-term warm or cold
events, which may not induce the start of plant growth and development in spring. The onset of
the GS was identified as the first day exceeding 0ď°C 9d-Ta in spring, while the end of the GS was
recognized as the latest day exceeding 0ď°C 9d-Ta in fall. The lengths of the GS were the number
of dates between the onset and the end of the GS each year.
To compare the soil respiration rates, climate, microclimate and the vegetation indices, I
separated the data into different years based on the USGSâ âwater yearâ concept October 1st to
September 30thâas it is more meaningful than a calendar year in agricultural research. I
identified the water year in my dissertation from the end of the previous GS until the end of this
GS because post-GS precipitation may subsidize the soil water pool and support plant and soil

118

microbe consumption in the next GS. The length of the water year may be different due to the
different seasonal patterns of air temperature in fall and winter.

4.2.4 The total and heterotrophic soil respiration measurements
Four plots at each site were randomly selected to install 1x1 m root exclusion plots for
measuring heterotrophic soil respiration (Rh). We dug a 1 m deep trench around the edges of
each square (Fig. 2.1 (B)). The trenches were lined with root-barrier sheets before the soil was
refilled according to its soil profile (Fig. 2.1 (C); Tang et al., 2005). The plants and roots inside
the square were manually removed and/or killed by herbicide (Monsanto RoundupÂŽ). Two PVC
collars (10 cm diameter) were installed inside the square to accommodate flux chambers to
measure Rh. Another two collars were installed surrounding the square to measure total soil
respiration (Rs) (Fig. 2.1(D)). A total of 112 measurements (including Rh and Rs) were taken
between 10AM and 7PM when soil respiration is at its diel maximum.
Soil respiration was measured biweekly during the GS and monthly during the non-GS
when snow cover is minimal to allow the measurements (OctoberâDecember). The chamberbased infrared gas analyzer (IRGA) approach of soil respiration measurement was done with an
LI-6400 portable photosynthesis system with a 10 cm diameter 6400-09 soil chamber or LI-8100
with a 10 cm diameter 8100-102 soil chamber (LI-COR Biosciences, Lincoln, NE).

4.2.5 Vegetation Index data
Vegetation indices, including normalized difference vegetation index (NDVI) and
enhanced vegetation index (EVI), were calculated based on satellite remote sensing using band 1
(nir), band 2 (red), and band 3 (blue) of the Moderate Resolution Imaging Spectroradiometer
(MODIS). The NDVI was determined by the ratio of the difference between near-infrared

119

reflectance and red reflectance to their sum (Tucker et al., 2005). The EVI was further modified
with a soil adjustment factor, L, and two atmospheric aerosol scattering coefficients, C1 and C2
(Huete et al., 1999), and has higher sensitivity at the âgreenâ portion of the spectrum when
vegetation is very dense. These data were developed from MODIS daily products by selecting
the optimal day (i.e., low cloud cover) out of a 16-day interval. Missing data may appear when
clouds obscured the site over an entire 16-day period. NDVI and EVI data were downloaded
from the Oak Ridge National Laboratory Distributed Active Archive Center for Biogeochemical
Dynamics (ORNL DAAC, http://daac.ornl.gov/MODIS/modis.shtml). I used a single pixel from
the MODIS image that fell within each experimental site at the finest spatial resolution of 250 m.
The dates of NDVI and EVI values were paired with the nearest dates of soil respiration
measurement.

4.2.6 The concept of Bayesian modeling
In this chapter, I use a Bayesian approach to estimate the parameters of the posterior
models and analyze the changes of log10-transformed soil respiration temperature sensitivity (Î˛)
and soil respiration value at 20Â°C (LR20), both of which are used as indices to observe the
changes in the Rs-Ts and Rh-Ts relationships.
Temperature sensitivity of soil respiration (Î˛) and log10 soil respiration at 20Â°C (LR20)
The Pr(C|D) was calculated with the following equation:
Pr(đś|đˇ) = PrâĄ(đś) Ă

PrâĄ(đˇ|đś)

Eq. 4-1

PrâĄ(đˇ)

where Pr(C|D) is the probability of event C occurring, given that event D had occurred and that
Pr(D|C) is the inverse. Here, I can consider Pr(C), the first term on the right side of the equation,
as prior knowledge from previous studies, empirical data or broad-scale data. The second term
on the right side, Pr(D|C)/Pr(D), is the likelihood that can be based on new data, specific data or
120

even new estimates from different mechanisms, relationships with other variables or a new
hypothesis. The posterior model on the left side of the equation is then estimated from the
combination of the prior model and the likelihood function.
For continuous data, the posterior models of Bayesian statistics were estimated from the
combination of prior information and new knowledge (likelihood). The Bayesâ rule for
continuous data can be rewritten as below:
Pr(đś|đˇ) =

Pr(đś)ĂâĄPr(đˇ|đś)

Eq. 4-2

â

âŤ0 Pr(đĽ)ĂPr(đˇ|đĽ)đđĽ

where C is the specific value for the parameter. The replacement of the denominator Pr(D) to the
integration of over all possible events, which represents the sum of the conditional probabilities
of the parameter (x).
The Bayesian model, thus, can separate into three major parts: prior model, likelihood
and posterior model. The prior model can be the estimated distribution of probabilities from
empirical data, previous knowledge or a broader scale dataset. The likelihood can be a set of new
data, a relationship from other independent variables, or the specific group of data contained in
the broad scale data. The new probability distribution was created from the prior model and
likelihood.

4.2.7 The estimations of prior models
To better develop the Bayesian model and compare R-T relationships in various CROP or
LUH and their responses to climate patterns and severe climate events, as well as the shift of R-T
relationship after LUC, I separated Rh and Rs data by year and then created Bayesian models
independently. The probabilities of the values of log10 soil respiration were assumed to have a
natural distribution and were fitted by the Maximum Likelihood Estimation (MLE) approach:
đđ ~đ(đ, đ 2 )

Eq. 4-3
121

where Îź and Ď2 were the mean and variance, respectively, to the probabilities of a particular soil
respiration rate Ri.
The starting points of the suitable MLE models were determined from the means and
standard variations in the raw data of soil respiration rates. The results of the well-fitted MLE
models were my prior models of the Bayesian models. The MLE models were run by the
âbbmleâ package (https://cran.r-project.org/web/packages/bbmle/index.html) for R language (R
Development Core Team, 2010).

4.2.8 The equations of likelihood
To elucidate the R-T relationship based on traditional exponential hypothesis and my
previous analysis (Chapter 2), I assumed a linear correlation between the log10 soil respiration
rates (log10R) to soil temperature (Ts). Therefore, I assumed the likelihood is as below:
đđđđ = đź + đ˝ â đđ  + đ

Eq. 4-4

where Îą, intercept of the regression line, is the value of log10R at 0Â°C; and Î˛, the slope of the
regression line, is the raising value of log10R when Ts increase by 1Â°C, representing temperature
sensitivity of soil respiration.
In order to incorporate the different effects of CROP and LUH on R-T relationships when
they respond to climate, I added âsiteâ as a latent discrete term in the likelihood equation:
đđđđ = đź + đ˝ â đđ  + đ đđĄđ + đ

Eq. 4-5

Both Îą and Î˛ were assumed to have normal distributions with small variances.

4.2.9 Calculation of estimations of posterior models
After setting up the whole model, the âR2jagsâ package (https://cran.rproject.org/web/packages/R2jags/index.html) for R language ran the Bayesian model using the

122

Markov Chain Monte Carlo (MCMC) method to estimate the posterior models for 10,000 runs
with 1,000 discarded initial samples to obtain the values of Îą and Î˛ in each iteration. The
distributions of the parameters were checked by the âmcmcplotsâ package (https://cran.rproject.org/web/packages/mcmcplots/index.html) to diagnose whether posteriors converged to
their stationary distribution or not. Then, LR20, the log10R value at 20Â°C, was calculated based on
the values of Îą and Î˛:
đżđ20 = đź + 20 â đ˝

Eq. 4-6

In this equation, Î˛ and LR20 are plotted on a bi-plot. The locations of the predictions
revealed the temperature sensitivity of soil respiration and the GS soil respiration. The location
of the âcloudsâ, which are composed of the predictions of Bayesian models, on the Î˛-LR20 biplot
presents the R-T relationship of each field. The behaviors of clouds across years display its
responses to climate. The means and standard deviations of each cloud were calculated and used
to recognize the shift among years.
4.2.10 Estimation of Î˛ and LR20 of Ra
After simulating Î˛ and LR20 of Rs and Rh, I calculated the Î˛ and LR20 of Ra after
weighing the rootâs contribution to soil respiration (RC). This is because both Î˛ of Ra (Î˛Ra) and Î˛
of Rh (Î˛Rh) contribute to Î˛ of Rs (Î˛Rs), according to their relative contribution, which can be
presented as:
đ˝đđ  = đđś â đ˝đđ + (1 â đđś) â âĄ đ˝đâ

Eq. 4-7

where RC is the root contribution of total soil respiration (i.e., Ra:Rs ratio) for each field in each
year (Supplement 4-1).
The values of Î˛Ra can be calculated by the following equation, derived from Eq. 4-7
above:

123

đ˝đđ =

đ˝đđ  â(1âđđś)âđ˝đâ

Eq. 4-8

đđś

The LR20 of Ra was calculated by the same methods in the equation below:
đżđ20 đ =
đ

đżđ20 đ â(1âđđś)âđżđ20 đ
đ 

â

Eq. 4-9

đđś

where LR20Ra is the LR20 of Ra, LR20Rs is the LR20 of Rs and LR20Rh is LR20 of Rh.

4.3 RESULTS

4.3.1 Bayesian models for the R-Ts relationship
The prior models of Rs and Rh for each year were estimated by MLE based on the log10transformed soil respiration rates that were assumed to have normal distributions. The starting
points of each model simulation were the means and the standard deviations of raw soil
respiration data (Table 4.1 & Fig. 4.1). The MLE models fit raw data well, although the raw data
of log10 soil respiration were slightly left skewed. The year 2012 had the greatest variance while
2014 had the smallest variance for both Rs and Rh.

Table 4.1. The mean and standard deviation of log10 soil respiration by year. The values of
mean and standard deviation (sd) were used as the starting point of Maximum Likelihood
estimation (MLE) natural distribution. Rs is total soil respiration and Rh is heterotrophic soil
respiration.
year
mean
sd
Rs
2011
0.495
0.358
2012
0.298
0.468
2013
0.469
0.391
2014
0.592
0.280
Rh

2011
2012
2013
2014

0.374
0.108
0.164
0.362

124

0.405
0.502
0.448
0.301

(A)

(B)

Figure 4.1. The probabilities of log10 soil respiration in each year. For Rs (A) and Rh (B), the distribution of MLE models (dark
colors) were based on normal distributions, with the means and standard deviations derived from the raw data. The distribution of raw
data (light colors) were slightly skewed left, but most of them were well fitted. Green, red, blue and purple curves represent the
distributions of log soil respiration data in 2011, 2012, 2013 and 2014, respectively.

125

4.3.2 The differences of Î˛ and LR20 between CROP and LUH
CROP and LUH play important roles on the R-T relationship. I recognize the different
locations of Î˛Rs-LR20Rs and Î˛Rh-LR20Rh between CROP (i.e., C, Sw, Pr and Ref) and between
LUH (i.e., AGR, CRP and Ref). For total soil respiration, the Ref site had the highest LR20 and a
relatively high Î˛ in both Rs and Rh across all sites. In contrast, corn sites had the lowest LR20 in
both Rs and Rh compared to other perennial sites (switchgrass and prairie). Prairie had higher
LR20 and Î˛ than switchgrass. For the effect of LUH for Rs, CRP had lower Î˛ and higher LR20
than AGR in perennial crops while AGR always had higher Î˛ in CRP fields than those in AGR
fields in the annual crop (Table 4.2 & Fig. 4.2(A)). The effect of CROP and LUH on the R-T
relationship for Rh are similar to that for Rs. CRP sites had higher LR20 than AGR sites. Ref and
CRP-Pr had high LR20, followed by CRP-Sw and then CRP-C. The LR20 in AGR-Sw was near
that of AGR-Pr and higher than that of AGR-C (Table 4.2 & Fig. 4.2(B)).

126

(A)

(B)

Figure 4.2. The distributions of log10 soil respiration temperature sensitivity (Î˛) and log10
soil respiration rates at 20 Â°C (LR20). The Bayesian models were developed based on the
Bayesian approach, with data from the four growing seasons (2011-2014), which include Rs (A)
and Rh (B). The predictions were simulated from Bayesian models using the Markov Chain
Monte Carlo (MCMC), which is a method based on the probabilities of Î˛ and LR20. Each cloud
represents one site for different crop types and land use histories.
Table 4-2. The values of LR20 and Î˛ in the Rs-Ts and Rh-Ts relationships. The means and
standard deviations were calculated from the Bayesian estimation of each site, which include
data from all four growing seasons (2011-2014). CROP: crop types; LUH: former land use
history; C: corn; Sw: switchgrass; Pr: prairie; Ref: reference; CRP: Conservation Reserve
Program; AGR: corn=soybean rotation agriculture.
CROP
C
Sw
Pr
Ref

Rs

LUH

Rh

CRP

LR20
0.442 Âą 0.051

Î˛
0.027 Âą 0.002

LR20
0.289 Âą 0.053

Î˛
0.026 Âą 0.002

AGR

0.340 Âą 0.056

0.019 Âą 0.003

0.081 Âą 0.064

0.018 Âą 0.003

CRP

0.622 Âą 0.059

0.009 Âą 0.003

0.374 Âą 0.061

0.014 Âą 0.003

AGR

0.541 Âą 0.067

0.022 Âą 0.003

0.296 Âą 0.073

0.029 Âą 0.003

CRP

0.695 Âą 0.057

0.018 Âą 0.003

0.441 Âą 0.054

0.019 Âą 0.002

AGR

0.591 Âą 0.068

0.031 Âą 0.003

0.267 Âą 0.062

0.024 Âą 0.003

0.804 Âą 0.054

0.025 Âą 0.003

0.456 Âą 0.054

0.027 Âą 0.003

127

4.3.3 The distribution of Î˛ and LR20 of Ra and Rh and their trajectories across years
When comparing the locations of the Î˛-LR20 plots, I found that Rh had relatively smaller
amounts in LR20 than Rs (Fig. 4.3). I calculated the Î˛ and the LR20 of Ra based on the weighed Î˛
and the LR20 of Rs and Rh (Eq. 4-7 & Eq. 4-8, Supplement 4-1). Generally speaking, Ra and Rh in
the Î˛-LR20 plots had very different responses to the 2012 drought (Fig. 4.4). In the drought year,
all Î˛Ra decreased at different levels. They then rebounded in the following years but at variable
rates. The Ref site had high resistance but slow recovery. The Î˛Ra decreased in 2012 and 2013
and then rose back to the 2011 level in 2014. Other sites had stronger decreases of Î˛Ra in 2012,
but also stronger rebounds in 2013 and 2014 from the 2011 levels. The Î˛Ra in AGR-Sw, AGR-Pr
and CRP-C fell back in 2014. The shifts of the LR20Ra varied in 2012. Some sites increased while
some sites decreased. However, all of them increased in 2013, the first year after drought. The
magnitudes of increases of the LR20 of Ra varied. Some sites fell back in 2014 while others still
increased (Fig. 4.5 (A)).
Unlike Ra, the Î˛ and the LR20 of Rh told very different stories. The Î˛ of Rh in all sites
increased and the LR20 of Rh decreased in 2012, although the increasing magnitudes of Î˛ in CRPSw and CRP-Prâthe most dry sites (Fig. 3.1 (B))âwere very small. The levels of Î˛ and LR20 of
Rh did not change much after 2012. Some sites had slightly increasing Î˛ and LR20 while some Î˛
and LR20 rates fluctuated near the 2012 levels without returning to the 2011 levels (Fig. 4.5 (B)).
It had either required more time to recover or it had reached a new dynamic equilibrium after the
2012 drought.

128

(A)

(B)

Rs

Rh

Figure 4.3. The distributions of the log10 soil respiration temperature sensitivities (Î˛) and the log10 soil respiration rates at
20 Â°C (LR20) across years. The clouds show Î˛ and LR20 of Rs (A) and Rh (B), which were predicted by Bayesian models. Each
cloud contained 10,000 predictions and presented the values of Î˛ and LR20 for each site in each year. The locations of green, red,
blue, and purple dots represent the shifts from 2011 to 2014, respectively.

129

0.08

AGR-C

AGR-Sw

AGR-Pr

2011 Ra
2012 Ra

0.06

2013 Ra
2014 Ra

Î˛
0.04

2011 Rh
2012 Rh

0.02

2013 Rh
2014 Rh

0

-0.02

0.08

CRP-C

CRP-Sw

CRP-Pr

Ref

0.06

Î˛
0.04

0.02

0

0

0.5

1.0

LR20

1.5

0

0.5

1.0

1.5

0

LR20

0.5

1.0

LR20

1.5

0

0.5

1.0

1.5

LR20

-0.02

Figure 4.4. The inter-annual shifts of the log10 soil respiration temperature sensitivities (Î˛) and the log10 soil respiration rates
at 20 Â°C (LR20) of Ra and Rh for each site across all years. The solid dots and lines represent Î˛ and LR20 of Ra while the empty
circles and dash lines show those of Rh. The green, red, blue, and purple colors represent 2011â2014, respectively.

130

(A)
0.08
0.06

CRP-C Ra
CRP-C Rh
AGR-C Ra
AGR-C Rh

C

CRP-Sw Ra
CRP-Sw Rh
AGR-Sw Ra
AGR-Sw Rh

CRP-Pr Ra
CRP-Pr Rh
AGR-Pr Ra
AGR-Pr Rh

Pr

Ref Ra
Ref Rh

Sw

0.04
0.02
0

Î˛
0.08
0.06

Ref

0.04
0.02
0
2011

2012

2013

2011

2014

2012

2013

2014

year
(B)
1.2

CRP-C Ra
CRP-C Rh
AGR-C Ra
AGR-C Rh

C

CRP-Sw Ra
CRP-Sw Rh
AGR-Sw Ra
AGR-Sw Rh

Pr

Ref Ra
Ref Rh

Sw

0.8
0.4
0
-0.4

LR20
1.2

CRP-Pr Ra
CRP-Pr Rh
AGR-Pr Ra
AGR-Pr Rh

Ref

0.8
0.4
0

2011

2012

2013

2014

2011

2012

2013

2014

-0.4

year
Figure 4.5. The inter-annual shifts of the log10 soil respiration temperature sensitivities (Î˛)
and the log10 soil respiration rates at 20 Â°C (LR20) of Ra and Rh for different crops and land
use histories across all years. The Î˛ (A) and LR20 (B) present different crops. The solid dots
and lines display Ra, while the empty circles and dashed lines show Rh.

131

4.4 DISCUSSION
4.4.1 The distribution of Î˛ and LR20 across CROP and LUH combinations
The annual soil respiration rates varied widely across treatments, as shown in Chapter 2
(Fig. 2.5). The Ref field had the highest LR20, followed in sequence by Pr, Sw, and C (Fig. 4.2).
The effect of LUH is similar. Ref had the highest LR20, followed in sequence by CRP and AGR
(Fig. 4.2). When observing the distribution of the clouds of estimates on the LR20-Î˛ plots, it is
evident that the same CROP (with different LUH) has similar R-T relationship in Rs while the
same LUH (with different CROP) has similar R-T relationship in Rh. The results support my
hypothesis that CROP affects Ra more because its substrate supply is mainly regulated by
photosynthesis and the carbon allocation of plants. On the other hand, LUH affects Rh by
controlling the characteristics and processes in soil.
4.4.2 The stability of Î˛ and LR20 in the severe spring-summer drought in the Ref field
Ref had the smallest shift in both Î˛ and LR20 of Rs across all sites inter-annually. This
implies that the soil-root system in Ref was the most resistant to perturbation out of all the sites,
even with the severe drought in 2012. However, for Rh, the LR20 decreased and Î˛ increased
enormously due to the drought (Fig. 4.4). Brome grass may have been able to take up water from
deeper soil horizons and maintain its basic metabolism during the drought, whereas the soil
microbial activities were impacted in the upper 10 cm soil layer. The Î˛Rh in 2013 and 2014
dropped a little, but values were still higher than that in 2011. The LR20Rh values were very low
in 2013 and 2014. The Ref field exhibited a very different Rh-Ts relationship between 2011 and
the later three years.

132

4.4.3 Different responses of Î˛-LR20 on Rs and Rh after severe spring-summer drought
On the inter-annual scale, the effects of a disturbance on an ecosystem may be observed
as changes in the behavior of the ecosystem functions immediately (the same year) and in the
recovery years (the following years) (Holling, 1973; Holling, 1996; Smith, 2011). The climate
patterns I observedâwet-hot, dry-hot, wet-cool and wet-cool in the continuing four growing
seasons (2011-2014)âgave me an opportunity to test the short- and long-term responses of the
R-Ts relationship to a severe spring-summer drought. I found that the two components, Ra and
Rh, had different responses to the drought (Figs. 4.5 (A) & (B)).

4.4.3.1 The resistance of Rs-Ts and Rh-Ts relationships to the drought
In the drought year, all Î˛Ra decreased while Î˛Rh increased regardless of the fact that LR20
responses varied (Fig. 4.4 & Fig. 4.5). The opposite responses of Ra and Rh temperature
sensitivities were due to their own distinctive mechanisms. Most studies have indicated that the
temperature sensitivity of soil CO2 efflux is depressed under low water moisture (Savage &
Davidson, 2001; Jassal et al., 2008; Harper et al., 2005). My results of Ra supported the
conclusion of these studies.
Surprisingly, Î˛Rh increased in all sites in 2012 compared to 2011 (the red dashed lines in
Fig. 4.4; Fig. 4.5 (A)) when the water stress was strong. The climate patternâearly onset and a
warm spring with spring-summer drought in 2012âseemed to depress all Rh and increase
temperature sensitivity. I noted a similar phenomenon in the Rh-Ts relationship in Chapter 2.
Most Q10 of Rh in 2012 was higher than those in 2011 (Supplement 2-3). My results of Rh were
opposite to most studies, which claimed that the temperature sensitivity of Rh would decrease
when soil moisture is low in mesic ecosystems (Suseela et al., 2012; Wang et al., 2014; Sampson
et al., 2007; Jassal et al., 2008; Manzoni et al., 2012; Rey et al., 2002). In contrast, the driest

133

experimental sites (i.e., CRP-Sw and CRP-Pr) had lower magnitudes of increasing Î˛Rh in 2012
than other sites. The effects of water stress on log10 Rh temperature sensitivity within 2012
supported the previous studies that also claimed water deficit reduces Rh temperature sensitivity.
The contrast between the intra-annual (among sites) and inter-annual (2011 vs. 2012)
comparisons may be due to other causes.
Heterotrophic soil respiration results from the decomposition of soil organic matter by
soil microbes. Decomposition is a biochemical reaction and is assumed to have an exponential
relationship with temperature (Lloyd & Taylor, 1994; vanât Hoff, 1885). The decomposition rate
and its temperature sensitivity were regulated by its intrinsic kinetic properties and
environmental constraints (Davidson et al., 2006). The former mainly depends on the complexity
of carbon compounds (i.e., labile compounds and recalcitrant compounds), the concentration of
substrates, the physical protection (i.e., soil aggregates) and the chemical protection (i.e., the
absorption of carbon compounds into mineral surface by chemical bonds) (Davidson et al.,
2006). The latter restrains intrinsic temperature sensitivity of Rh and represents the apparent
temperature sensitivity (Suseela et al., 2012) and may be related to surrounding temperature,
water availability (Schimel et al., 1994; Ise & Moorcroft, 2006), soil microbial community
structure (Sheik et al., 2011; Zhou et al., 2014) and the affinities of the enzymes for the
substrates (Davidson et al., 2006). The decrease of Rh in all sites in 2012 was more possibly due
to the change in the soil microbial community structure or their activities since soil physical and
chemical properties are less likely to change much in the short-term in no-till fields.
In 2012, not only was there higher Î˛Rh but also lower âÎąâ, log10 Rh at 0 Â°C and LR20Rh
(Eq. 4-5) compared to 2011. The high Î˛Rh and low Îą and LR20Rh means that the overall Rh in
2012 was lower than those in 2011, regardless of temperature. The impact of water stress on the

134

Rh-Ts relationship may result from the changes in the soil microbial community structure that
shifted their intrinsic temperature sensitivity. Lipson et al. (2002) found that soil microbial
community structure responded to the seasonal variation of climate and substrate supply and
exhibited different sensitivities to temperature. It is likely that severe water stress depressed most
species in the soil microbial community, but that some drought-tolerant and high temperaturesensitive species became dominant. The drought-tolerant species became dominated and thus
displayed high Rh temperature sensitivity when the water stress had been mitigated. The effect of
severe drought may change soil microbial community structure (Chen et al., 2007) or only alter
the strength of their activities without significant shifts in their community structure (Castro et
al., 2010). For example, Zhou et al. (2014) reported an increase in the fungi to bacteria ratio (F:B
ratio) during drought, since fungi have hyphae, that help them to move, colonize and decompose
surface litter, and a cellular wall, containing melanin and chitin that leave them unaffected by
drought (Orchard & Cook, 1983; Denef et al., 2001). Zhou et al. (2014) also found that the F:B
ratio was tightly correlated with Q10 of Rh (r=0.617, p=0.014). Bradford et al. (2008) reported
that the changes of soil microbial community composition may affect the ecosystem processes
but that the mechanisms were yet to be confirmed. However, some opposite studies reported no
significant changes in the F:B ratio during droughts (Castro et al., 2010). More research is
needed to clarify the mechanisms of the responses of soil microbial decomposers to drought.

4.4.3.2 The changes of Rs-Ts and Rh-Ts relationships in the recovery years
In 2013 and 2014, the Î˛s were higher than those in 2011, although the Ref site and two
corn sites had smaller differences compared to 2011 (Fig. 4.4). The higher differences of Î˛Rh
between 2014 and 2011 in Sw and Pr sites represent the shift in the soil microbial community
structure. Differences of Î˛Ra, which is the combinations of Î˛Rs and Î˛Rh, exhibit the well-

135

established root systems in perennials. The trajectories of LR20 in Rh and Ra were different. Most
LR20 of Ra in 2014 had rebounded to the levels of LR20 in 2011, as some LR20 (such as Prairie
LR20) were even higher (Fig. 4.4). The well-established root system stimulated not only Î˛ but
also LR20 in perennial cropping systems. LR20 of Rh in all sites in 2014 were still far less than
those in 2011. The soil microbial community may be dominated by drought-tolerant, high
temperature-sensitive species that result in lower decomposition rates. However, we need more
studies of the soil microbial community structure and its response to changes in precipitation
patterns, and how those responses affect biogeochemical cycles.

4.4.3.3 The systematic responses of the R-T relationship to the severe drought
After a two-year recovery, I observed that the spring-summer severe drought depressed
both Ra and Rh in the drought year and recovered in the following years. However, different
CROP and LUH had different levels of resistance and speed to recovery, implying perennial
crops acclimated severe drought while annual crops did not. The high resistance of the Ref field
to severe drought revealed that the climax community of perennial crops may have a wellestablished root-soil microbial cooperation system, which resists the severe drought well. In
addition, the opposite response of Ra- and Rh-Ts relationships to drought implies there were shifts
in soil microbial activities in response to the changes in soil temperature. The different location
of the clouds on the Î˛RhâLR20 biplots between 2011 and 2012-2014 revealed that the structure of
soil microbial community might change and may shift into a new state.

136

4.5 CONCLUSIONS
The increasing frequencies, intensity and amplitude of climate extremes, such as heat
wave, periods of heavy rains and severe drought, is already apparent and is expected in the future
(IPCC, 2013; Andresen et al., 2013; Schar et al., 2004). Many studies explored the impacts of
increasing ECEs on individual/population mortality and fitness, community structure, ecological
processes and ecosystem functions (Thibault and Brown, 2008; MacGillivray et al., 1995;
Haddad et al., 2002; Ciais et al., 2005). The responses of ecosystems to these ECEs and their
sensitivities varied widely due to different underlying mechanisms and thresholds in different
ecosystems and climate regimes (Smith, 2011). However, the societal and ecological impacts of
these increasing climatic extremes still need more research to clarify the underlying mechanisms
and the consequences (Smith, 2011; Ciais et al., 2005).
Considering both the immediate resistance and prolonged recovery after a severe drought
underneath different crop types and land use histories helps us better understand the different
responses of roots and soil microbes to severe drought, and ultimately how drought affects the
behaviors of Ra and Rh. Perennial crops have the potential to acclimate to severe drought via its
dense root system (and the associated microbial community) and shift in plant composition. The
different plant composition at different successional stages may have different resistance and
recovery abilities in response to severe drought. The annual crops have a weaker ability to buffer
the impact of severe drought on soil respiration and are more vulnerable to severe drought, since
they do not have a root system like the perennials. The shifts of LR20 and Î˛ of Ra after the severe
drought, thus, were larger, especially under high soil nutrient content.
Perennial crops have a stronger ability than annual crops to buffer the perturbation of
occasional severe drought. The ability of buffering increased through the succesional stages

137

when its dense root system developed. The perennial crop system, can not only reduce the input
of fertilizer and reduce soil erosion and nutrient leaching, but can also reduce the droughtinduced fluctuation of soil respiration. I will discuss the impacts of drought on NEE, GPP and
Reco and the difference responses of these major carbon fluxes facing severe drought in Chapter
5.

138

APPENDIX

139

Table S.4.1. The root contribution of soil respiration (RC) for each site in each year.
The RCs (Ra/Rs ratio) were calculated from the TWV model estimates in chapter 2.
The data of Ra and Rs were grouped in pairs into each site in each year for the annual
RC.
CROP

LUH

C
Sw
Pr
Ref

year
2011

2012

2013

2014

CRP

0.3195Âą0.0513

0.3171Âą0.0449

0.3065Âą0.0526

0.3310Âą0.0633

AGR

0.4650Âą0.0983

0.4342Âą0.0611

0.4902Âą0.1159

0.5125Âą0.1282

CRP

0.4678Âą0.0826

0.5657Âą0.1712

0.4563Âą0.0772

0.4700Âą0.0974

AGR

0.4512Âą0.1379

0.3452Âą0.2438

0.4331Âą0.1427

0.4399Âą0.1724

CRP

0.4477Âą0.0596

0.2893Âą0.1770

0.3864Âą0.1200

0.4057Âą0.0879

AGR

0.5135Âą0.0339

0.5121Âą0.0430

0.5247Âą0.0475

0.5321Âą0.0376

0.5639Âą0.0363

0.5610Âą0.0410

0.5375Âą0.0559

0.5411Âą0.0559

140

LITERATURE CITED

141

LITERATURE CITED

Bradford, M.A., Davies, C.A., Frey, S.D., Maddox, T.R., Melillo, J.M., Mohan, J.E., Reynolds,
J.F., Treseder, K.K., Wallenstein, M.D., 2008. Thermal adaptation of soil microbial
respiration to elevated temperature. Ecology Letters, 11, 1316â1327.
Castro, H.F., Classen, A.T., Austin, E.E., Norby, R.J., Schadt, C.W., 2010. Soil microbial
community responses to multiple experimental climate change drivers. Applied and
Environmental Microbiology, 76(4), 999-1007.
Chen, M.-M., Zhu, Y.-G., Su, Y.-H., Chen, B.-D., Fu, B.-J., Marschner, P., 2007. Effects of soil
moisture and plant interactions on the soil micorbial community structure. European
Journal of Soil Biology, 43, 31-38.
Ciais, Ph., Reichstein, M., Viovy, N., Granier, A., Ggee, J., Allard, V., Aubinet, M., Buchmann,
N., Bernhofer, Chr., Carrara, A., Chevallier, F., De Noblet, N., Friend, A.D.,
Friedlingstein, P., Grunwald, T., Heinesch, B., Keronen, P., Knohl, A., Krinner, G.,
Loustau, D., Manca, G., Matteucci, G., Miglietta, F., Ourcival, J.M., Papale, D., Pilegaard,
K., Rambal, S., Seufert, G., Soussana, J.F., Sanz, M.J., Schelze, E.D., Vesala, T., Valentini,
R., 2005. Europe-wide reduction in primary productivity caused by the heat and drought in
2003. Nature, 437, 529-533.
Davidson, E.A., Janssens, I.A., 2006. Temperature sensitivity of soil carbon decomposition and
feedbacks to climate change. Nature, 440(9), 165-173.
Denef, K., Six, J., Paustian, K., Merchkx, R., 2001. Importance of macroaggregate dynamics in
controlling soil carbon stabilization: short-term effects of physical disturbance induced by
dryâwet cycles. Soil Biology & Biochemistry, 33, 2145â2153.
Gilgen, A.K., Buchmann, N., 2009. Response of temperate grasslands at different altitudes to
simulated summer drought differed but scaled with annual precipitation. Biogeosciences, 6,
2525-2539.
Haddad, N.M., Tilman, D. & Knops, J.M.H., 2002. Long-term oscillations in grassland
productivity induced by drought. Ecology Letters, 5, 110-120.
Harper, C.W., Blair, J.M., Fay, P.A., Knapp, A.K., Carlisle, J.D., 2005. Increased rainfall
variability and reduced rainfall amount decreases soil CO2 flux in a grassland ecosystem.
Global Change Biology, 11, 322-334.
Holling, C.S., 1973. Resilience and stability of ecological systems. Annual Reviews, 4, 1-23.
Holling, C.S., 1996. Engineering resilience cersus ecological resilience. In: Schulze, P., (Eds.),
Engineering within Ecological Constraints. National Academy Press, Wshington, D.C., pp.
31-44.
142

Hoover, D., Knapp, A.K., Smith, M.D., 2016. The immediate and prolonged effects of climate
extremes on soil respiration in a mesic grassland. Journal of Geophysical Research:
Biogeosciences, 121, 1034-1044.
IPCC, 2013: Climate Change 2013: The Physical Science Basis. Contribution of Working Group
I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
[Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y.
Xia, V. Bex and P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, United
Kingdom and New York, NY, USA, 1535 pp, doi: 10.1017/CBO9781107415324.
Ise, T., Moorcroft, P.R., 2006. The global-scale temperature and moisture dependencies of osil
organic carbon decomposition: an analysis using a mechanistic decomposition model.
Biogeochemistry, 80, 217-231.
Jassal, R.S., Black, T.A., Novak, M.D., Gaumont-Guay, D., Nesic, C., 2008. Effect of soil water
stress on soil respiration and its temperature sensitivitu in an 18-year-old temperate
Douglas-fir stand. Global Change Biology, 14:1305-1318.
Johnson, J.M-F., Coleman, M.D., Gesch, R., Jaradat, A., Mitchell, R., Reicosky, D., Wilhelm,
W.W., 2007. Biomass-bioenergy crops in the United States: a changing paradigm. The
Americas Journal of Plant Science and Biotechnology, 1(1), 1-28.
Knapp, A.K., Beier, C., Briske, D.D., Classen, A.T., Luo, Y., Reichstein, M., Smith, M.D.,
Smith, S.D., Bell, J.E., Fay, P.A., Heisler, J.L., Leavitt, S.W., Sherry, R., Smith, B., Weng,
E., 2008. Consequences of more extreme precipitation regimes for terrestrial ecosystems.
BioScience, 58(9), 811-821.
Knorr, W., Gobron, N., Scholze, M., Kaminski, T., Schnur, R., Pinty, B., 2007. Impact of
terrestrial biosphere carbon exchanges on the anomalous CO2 increase in 2002â2003.
Geophysical Research Letters, 34, L09703.
Lark, T.J., Salmon, J.M., Gibbs, H.K., 2015. Cropland expansion outpaces agricultural and
biofuel policies in the United States. Environmental Research Letters, 10, 044003.
Lemus, Q., Lal, R., 2005. Bioenergy crops and carbon sequestration. Critical Reviews in Plants
Sciences, 24, 1-21.
Lipson, D.A., Schadt, C.W., Schmidt, S.K., 2012. Changes in soil microbial community and
function in an Alpine dry meadow following spring snow melt. Microbial Ecology, 43:
307-314.
Lloyd, J., Taylor, J.A., 1994. On the temperature dependence of soil respiration. Functional
Ecology, 8, 315-323.
MacGillivray, C.W., Grime, J.P. & The Integrated Screening Programme (ISP) Team, 1995.
Testing predictions of the resistance and resilience of vegetation subjected to extreme
events. Functional Ecology, 9, 640-649.

143

Manzoni, S., Schimel, J.P., Porporato, A., 2012. Responses of soil microbial communities to
water stress: results from a meta-analysis. Ecology, 93, 930-938.
McCarthy, M.A., 2013. Bayesian methods for ecology. Cambridge University Press.
Orchard, V.A., Cook, F.J., 1983. Relationship between soil respiration and soil moisture. Soil
Biology and Biochemistry, 15, 447-453.
R Development Core Team, 2010. R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria.
Reichstein, M., Bahn, M., Ciais, P., Grank, D., Mahecha, M.D., Seneviratne, S.I., Zscheischler,
J., Beer, C., Buchmann, N., Frank, D.C., Papale, D., Rammig, A., Smith, P., Thonicke, K.,
van der Velde, M., Vicca, S., Walz, A., Wattenbach, M., 2013. Climate extremes and the
carbon cycle. Nature, 500, 287-296.
Rey, A., Pagoraro, E., Tedeschi, V., De parri I., 2002. Annual variation in soil respiration and its
components in a coppice oak forest in central Italy. Global Change Biology, 8: 851-866.
Sampson, D.A., Janassens, I.A., Curil Yuste, J., Ceulemans, R., 2007, Basal rates of soil
respiration are correlated with photosymthesis in a mixed temperate forest. Global Change
Biology, 13, 2008-2017.
Savage, K.E., Davidson, E.A., 2001. Interannual variation of soil respiration in two New English
forests. Global Biogeochemical Cycles, 15(2), 337-350.
Schar, C., Vidale, P.L., Luthi, D., Frei, C., Haberli, C., Liniger, M.A. & Appenzeller, C., 2004.
The role of increasing temperature variability in European summer heatwaves. Nature,
427, 332â336.
Sheik, C.S., Beasley, W.H., Elshahed, M.S., Zhou, X., Luo, Y., Krumholz, L.R., 2011. Effect of
warming and drought on grasslad microbial communities. The International Society for
Microbial Ecology Journal, 5,1692-1700.
Shimel, D.S., Braswell, B.H., Holland, E.A., McKeown, R., Ojima, D.S., Parton, W.J.,
Townsend, A.R., 1994. Climatic, edaphic, and biotic controls over storage and turnover of
carbon in soils. Global Biogeochemical Cycles, 8(3), 279-293.
Smith, M.D., 2011. An ecological perspective on extreme climatic events: a synthetic definition
and framework to guide future research. Journal of Ecology, 99, 656-663.
Suseela, V., Conant, R.T., Wallenstein, M.D., Dukes, J.S., 2012. Effects of soil moisture on the
temperature sensitivity of heterotrophic respiration vary seasonally in an old-field climate
change experiment. Global Change Biology, 18, 336-348.
Thibault, K.M., Brown, J.H., 2008. Impact of an extreme climatic event on community
assembly. Proceedings of the National Academy of Science USA, 105, 3410-3415.

144

vanât Hoff, J. H., 1884. Ătudes de dynamique chimique (Studies of chemical dynamics). Frederik
Muller and Co., Amsterdam, the Netherlands.
Wang, Y., Hao, Y., Cui, X.Y., Zhao, H., Xu, C., Zhou, X., Xu, Z., 2014. Responses of soil
respiration and its components to drought stress. Journal of Soils Sediments, 14, 99-109.
Zavalloni, C., Gielen, B., Lemmens, C.M.H.M., De Boeck, H.J., Blasi, S., Van den Bergh, S.,
Nijs, I., Ceulemans, R., 2008. Does a warmer climate with frequent mild water shortages
protect grassland communities against a prolonged drought? Plant Soil, 308, 119-130.
Zhou, W., Hui, D., Shen, W., 2014. Effects of soil moisture on the temperautre sensitivity of soil
heterotrophic respiration: A laboratory incubation study. PLoS One, 9(3), e92531.

145

CHAPTER 5
THE RESPONSE OF NET ECOSYSTEM EXCHANGE (NEE) OF CARBON DIOXIDE
AND ITS PARTITIONING TO THE BIOPHYSICAL VARIABLES
ABSTRACT
The dynamics of carbon dioxide net ecosystem exchange (NEE), which is determined by
gross primary production (GPP) and ecosystem respiration (Reco), decides the direction and
amount of carbon flux between ecosystems and the atmosphere. The sum of regional carbon
fluxes contributes to global atmospheric CO2 balances and decides the rate of atmospheric CO2
increase. Managed ecosystems, including grasslands and croplands, can be managed to sequester
carbon dioxide in soil carbon pools. The understanding of the seasonality of GPP and Reco and
how they respond to climate events and human activities is critical to mitigate global climate
change.
Major carbon processes, GPP and Reco, have different dominant drivers and these drivers
have different seasonal patterns. The asynchronous seasonality of GPP and Reco determine the
carbon fluxes between the ecosystem and the atmosphere; however, crop type and soil nutrient
content also play important roles. My results revealed that the different phenology of annual and
perennial crops can affect carbon sequestration capacity according to the differences in the
length of the growing season. Moreover, the different phenology between annual and perennial
crops may affect crop tolerance to different seasonal patterns of climate events. The early onset
of the growing season in perennial crop ecosystems heightens the risk of severe the spring and
summer drought. In contrast, the delay of a regular summer water deficit may dramatically
increase carbon sequestration ability. Changing land use from undisturbed conservation lands to
corn may release a huge amount of carbon via ecosystem respiration, while the perennial crops
would be better able to maintain the carbon.

146

5.1 INTRODUCTION
Regional net ecosystem exchange (NEE) of carbon dioxide, defined by the difference
between gross primary production (GPP) and ecosystem respiration (Reco), determines global
carbon cycling. However, the rules of biotic and abiotic reactions on NEE have complicated and
uncertain interactions and feedbacks across different climate regions and different biomes. The
understanding of how biophysical variables and their interactions influence GPP and Reco is
crucial to estimating the contribution that biofuel cropping ecosystems have on future climate
GHG mitigation in different climate scenarios.
Eddy-covariance and chamber-based approaches have been widely used in the
monitoring ecosystem CO2 exchange across different regions with different climate regimes and
ecosystems. However, compared to the large number of NEE studies on forests, croplands have
received far less attention even though they may have important contributions to regional carbon
budgets, especially in mid-latitude areas (Soegaard et al., 2003). This kind of managed
ecosystem has large variation on its functions, such as carbon sequestration, since agricultural
practices (i.e., fertilization, plantation, harvest, irrigation and tillage) determine important
ecosystem structure and biogeochemical processes, such as soil nitrogen content, the type of
crops, the land cover and the water management, which in turn determine the carbon, nitrogen,
water and energy cycles of croplands (Schimel et al., 2000). Research on managed grasslands
and croplands can contribute to how we understand the responses of carbon fluxes and how to
determine when an ecosystem converts between a carbon source and sink.
NEE of CO2 in an ecosystem depends on the interplay of gross primary production (GPP)
and ecosystem respiration (Reco). The GPP, determined by photosynthetic rate, is mainly
controlled by photosynthetically active radiation (PAR) when temperature and water are

147

adequate. Light intensity and patterns determine GPP and alter daytime NEE in most
ecosystems. Ecosystem respiration rates are exponentially correlated to temperature and regulate
nighttime NEE. Temporal variations in two physical variables, PAR and temperature, alter the
variation of GPP and Reco and the NEE (Falge et al., 2002; Gilmanov et al., 2007; Zhang et al.,
2015; Chen et al., 2015). Asynchrony of light and temperature seasonal variations, combined
with the plantâs phenology and the activity fluctuation of soil microbial activity, determine the
annual NEE of grasslands and croplands.
In this chapter, I will focus on how the seasonal variations of GPP and Reco affect the
annual NEE and the seasonality of NEE across different crop types and land use histories,
especially after the impact of severe drought. I hypothesized that annual and perennial crops
have different seasonal patterns and different annual NEE due to their different phenology and
the length of growing season. I also expected that annual and perennial crops have different
magnitudes of NEE increases in the growing seasonâs drought period and different recovery
speeds after severe drought, due to different water use efficiencies and drought-adaptation
ability. I will also discuss the Rs:Reco ratio in different crop types after land use change. I
assumed that the Rs:Reco ratio in perennial croplands would increase from 2011 to 2014 due to
root systems development. The Rs:Reco ratio in Ref and corn fields will not change significantly
since Ref fields have had a well-established root system and corn does not have a perennial root
system.

148

5.2 METHODS

5.2.1 Study site
My experimental sites are located in the Great Lakes Bioenergy Research Center scale-up
fields of the Kellogg Biological Station (KBS, 42Â°40âN, 85Â°40âW), and are established in
association with KBS Long-Term Ecological Research (LTER). The sites are located in south
Michigan, USA (Fig. 1.5). The climate is humid continental (warm summer) climate (Dfa) (Peel
et al., 2007). The mean annual air temperature and mean annual precipitation at KBS are 10.1
Â°C and 1005 mm yr-1 (1981-2010), respectively (Robertson & Hamilton, 2015).

5.2.2 Experimental design and schedule
Seven experimental plots were located at two locations with their own land use histories
(LUHs): (1) Conservation Reserve Program (CRP) grasslands at Marshall Farm, and (2) cornsoybean rotation agricultural fields (AGR) at Lux Arbor Reserve. The CRP sites had been in a
monoculture of smooth brome grass (Bromus inermis Leyss) since 1987, while the AGR fields
had been under conventional corn-soybean rotation cultivation for several decades (Fig. 1.6).
The soil texture at all sites is sandy loam, except for a sandy clay loam at one field. However,
soil carbon and nitrogen contents at the CRP sites were significantly higher than those at the
AGR sites before the land conversions (Table 1.1).
The experiment was conducted at seven scale-up fields ranging in size from 9 to 17
hectares. Four fields are located in Marshall Farm (i.e., CRP sites) and three in Lux Arbor
Reserve (i.e., AGR sites). All sites, except the reference site (Ref), were sprayed with herbicide
at the end of 2008 to prepare the lands for soybean planting in 2009. The CRP and AGR sites
were then cultivated with either continuous corn (Zea mays, Dekalb DK-52), switchgrass

149

(Panicum virgatum), or a mixture of native prairie grasses that was dominated by Canada wild
rye (Elymus Canadensis), little bluestem (Schizachyrium scoparium), Indian grass (Sorghastrum
nutans), big bluestem (Andropogon gerardii), and switchgrass (Panicum virgatum) since 2010.
One CRP grassland, dominated by brome grass, was not disturbed and retained as the historical
reference site (Gelfand et al., 2011; Zenone et al., 2011; Deal et al., 2013; Zenone et al., 2013). I
used an experimental code for these sites by abbreviating them as âLUH-CROPâ. CRP-C, CRPSw and CRP-Pr represent the CRP sites that were converted to corn, switchgrass and prairie
mixture respectively, while AGR-C, AGR-Sw and AGR-Pr are AGR farms converted to corn,
switchgrass and prairie mixture. In 2010, when the crops were established, the perennial crops
(switchgrass and prairie) were accompanied by oats as nurse crop with and without fertilization,
respectively (Fig. 1.6). Switchgrass fields were applied 55 kg N ha-1 (28% liquid urea
ammonium nitrate) on DOY 172. No other management practices were applied beyond
harvesting at the end of each growing season. No-till continuous corn was seeded in mid-May
with a one-time herbicide mix (Lumax, Atrazine 4L, Honcho Plus and (NH4)2SO4). Phosphorus,
potassium (P2O5 + K2O, 168.5 kg ha-1 on DOY 95, 2010) and nitrogen fertilizers (112 kg N ha-1
on DOY 165 & 160 in AGR-C and CRP-C, respectively) were applied in 2010. Phosphorus,
potassium (P2O5 + K2O, 294 kg ha-1 on DOY 104, 2011) and nitrogen fertilizers (liquid
nitrogen168 kg N ha-1 on DOY 172, 2011) were applied (Zenone et al., 2013).

5.2.3 The calculations of NEE, GPP and Reco by eddy-covariance approach

5.2.3.1 The measurements of eddy-covariance and microclimatic data
One open-path eddy-covariance (EC) system was established in each site (Figs. 1.5 &
5.1) at the end of 2008 before land use change (2009) to measure the carbon, water and energy

150

fluxes. Carbon dioxide, water vapor concentration and three-dimensional wind speed were
recorded for the calculation of net ecosystem CO2 exchange and water vapor between the
ecosystems and the atmosphere. The CO2 and water vapor concentrations were measured by a
LI-7500 open-path infrared gas absorption analyzer (IRGA, LI-COR Biosciences, Lincoln, NE).
The three-dimensional wind speed was measured by a CSAT3 anemometer (Campbell Scientific
Inc. (CSI), Logan, UT). Inbound (top-down) and the outbound (bottom-up) short- and long-wave
radiation were measured by a CNR1 radiometer (Kipp & Zonen, Delft, The Netherlands), while
the photosynthetically active radiation (PAR) was measured by a LI-190SB (LI-COR
Biosciences). Air temperature (Ta) and relative humidity (RH) were recorded by a HMP45C
(CSI). Most of the above instruments were mounted at about 1.5 m above the canopy surface.
Soil temperature (Ts at 2, 5, and 10 cm depths belowground) and soil water content (0-30 cm
depth VWC) were measured by CS107 temperature probes (CSI) and a CS616 Time Domain
Reflectometer (TDR) probe (CSI), respectively. Three HFT3 flux plates (CSI) were placed
randomly 2 cm below the soil surface to measure soil heat flux. A Campbell CR5000 datalogger
(CSI) was used to collect the eddy-covariance (10 Hz) and microclimatic (30 min) data (Fig.
5.1).

151

Figure 5.1. Eddy-covariance tower with instruments for carbon, water and energy cycle
monitoring. The eddy-covariance (EC) tower was established in late 2008 to measure high
frequency CO2 and water vapor concentration and 3D wind velocity (u). It also records incoming
and reflecting long- and short-wave radiation, photosynthetically active radiation (PAR), air
temperature (Ta), relative humidity (RH), soil temperature (Ts), soil water moisture (VWC) and
soil heat storage. Most aboveground instruments were mounted 1.5 m above the canopy surface.
5.2.3.2 Data processing and gap-filling
The eddy-covariance data were processed by the EdiRe software (University of
Edinburgh, UK). Out-of-range data (i.e., spikes greater than four standard deviations), which
may have been generated by bad weather or instrument failure, as well as time lags between CO2
and water vapor concentrations and the vertical wind speed were removed (McMillen, 1988).
The CO2 and H2O fluxes were adjusted for frequency response and air density fluctuations
(Webb et al., 1980), including the warming of the IRGA (Burba et al., 2008). Coordinate
rotation was applied to the wind components using planar fit rotation (Wilezak et al., 2001). The
high frequency data were processed in 30-minute blocks averages without detrending (Moncrieff
152

et al., 2004). The missing or poor quality data were replaced based on the standardized
FLUXNET gap-filling approach (Reichstein et al., 2005, http://www.bgcjena.mpg.de/~MDIwork/eddyproc/). More details of the data processing can be found in Abraha
et al. (2015).

5.2.3.3 Estimation of NEE, GPP and Reco
The net ecosystem exchange (NEE) of CO2, defined by the net CO2 exchange between an
ecosystem and the atmosphere, is the difference between gross primary production (GPP) and
ecosystem respiration (Reco) (Eq. 5-1; Chapin et al., 2006):
đđ¸đ¸ = đđđđ â đşđđ

Eq. 5-1

where GPP and Reco are the absolute amounts of carbon captured and lost from the ecosystem to
the atmosphere. The values of nighttime NEE were recognized as Reco since there is no
photosynthesis at night. An exponential model based on the correlation between Reco and air
temperature was developed, and daytime Reco was adjusted by the Reco âTa model. The GPP then
calculated the difference between daytime NEE and the adjusted daytime Reco (Reco â).
đşđđ = đđđđ â˛ â đđ¸đ¸

Eq. 5-2

Thirty-minute GPP, Reco and NEE were recorded and the daily GPP, Reco and NEE were
calculated for later analysis.
5.2.4 Estimation of Rsâ, Raâ and Rhâ

5.2.4.1 The TWV model and its variables
The adjusted 30-minute Rs (Rsâ) were estimated by the TWV model developed in Chapter
2 (Eq. 2-2 & Supplement 2-4). The 30-minute soil temperature (Ts) data were measured at each
plot (four plots in each site) by a HOBO datalogger (Onset Computer Co., Bourne, MA) with

153

two thermal couples. Thirty-minute soil water content (VWC) data were recorded at 0-30 cm
depth by a TDR at each siteâs eddy-covariance tower. Enhanced Vegetation Index (EVI) data
from the MODIS satellite were downloaded from the Oak Ridge National Laboratory Distributed
Active Archive Center for Biogeochemical Dynamics (ORNL DAAC,
http://daac.ornl.gov/MODIS/modis.shtml; details can be found in the methods section of Chapter
2).

5.2.4.2 Data processing and gap filling
Out-of-range data of Ts, which was mainly generated by failured thermocouples, HOBO
datalogers or power outages, includes extremely high or low values, abnormal jumpy data (>5ď°C
changes between two measurements) and high magnitudes of fluctuation. The out-of-range data
were recognized and removed manually. The whole data set in a measured period (weeks to
months) may be removed if most of data of that sensor was abnormal. The gap-filling of data
followed different methods depending on its length. The short-term (hours) gaps were replaced
based on the normal data before and after the gap with the following equation:
đđ  đ = đđ  đâ1 +

đđ  đ+1 âđđ  0

Eq. 5-3

đ+1

where đđ  đ was the ith cell of the gap and n was the number of the gap cells. The đđ  0 was the last
normal datum before the gap, while đđ  đ+1 was the first normal datum after the gap. The i were
calculated from 1 to n.
The long-term (days) gaps were filled based on the fluctuation trend at the nearest sensor,
either within the same site or between sites. The calculation followed the equation 5-3:
đđ  đ = đđ  đ
đ

đâ1

+ (đđ  đ â đđ  đ
đ

đâ1

)

Eq. 5-4

154

where đđ  đ was the ith cell of the gap and đđ  đ
đ

was the (i-1)th cell in the thermal couple âaâ,

đâ1

which contained the data gap. The đđ  đ and đđ  đ
đ

synchronized to đđ  đ and đđ  đ
đ

đâ1

đâ1

were the ith cell and the (i-1)th cell that

, respectively, in the nearest thermal couple âbâ. Equation 5-4

can mimic the diurnal temperature fluctuations at local scale.
The post-processing of VWC was similar to Ts. The missing data due to the failure of the
CR5000 datalogger were filled depending on the length of gaps. The VWC data were filled
based on Eq. 5-3 and Eq. 5-4.
The quality of EVI data was good, although there were some one-cell gaps due to
continuous 16-day cloudy periods. The gaps were filled by the average of the EVI values
immediately before and after the gap.
The 30-minute Rsâ values were estimated based on the TWV model and the 30-minute Ts,
VWC and EVI data. The daily average Rsâ were calculated for later statistical analysis and chart
drawing.

5.2.4.3 The components of ecosystem respiration and their calculation
Ecosystem respiration can be partitioned into aboveground (Rabove) and belowground (soil
respiration; Rs):
đđđđ = đđđđđŁđ + đđ 

Eq. 5-5

The aboveground and belowground contributions to the ecosystem respiration depend on the
carbon allocation of plants, the phenology and its different responses and time lags to changes in
environmental conditions.
The 30-minute Rsâ data were thereafter combined with NEE, GPP and Reco data. The
aboveground respiration (Rabove) was calculated by the difference of Reco and Rs.

155

đđđđđŁđ = đđđđ âđđ  â˛

Eq. 5-6

The chronological variations of daily mean NEE, GPP, Reco, Rabove and Rs were plotted to
observed inter- and intra-annual fluctuations, response to climate patterns and their resistance
and recovery after a severe drought. There are still some long-term missing data of Rsâ since the
late start of the first year (21 May 2011) and early stop of the second year summer (20 August
2012) measurements of Ts. The mid-growing season (21 May to 20 August) data were selected to
calculate the Reco:GPP ratio, Rsâ:GPP and Rsâ:Reco ratio since only this period had data with all
components of the carbon cycle in all years. The ratios across years were compared by one-way
ANOVA and the Tukeyâs HSD test.

5.3 RESULTS
5.3.1 The seasonal variations of NEE, GPP, Reco and Rsâ
The variations of the components of carbon cycle, including NEE, GPP, Reco, Rabove and
Rsâ, were presented in Fig. 5.2. The seasonality of NEE, which is the difference between GPP
and Reco, was affected by GPP, Reco and the interplay between the two. They always showed
hump-shaped curves that increased in spring with the growth of plants and decreased with the
senescence or harvest of the plants. However, annual and perennial crops had different lengths of
photosynthetic active periods and influenced the seasonal patterns of NEE.
The NEE in corn fields started to decrease in mid-June, around two weeks after the corn
was planted. Theoretically, the decreases of NEE were determined by the variations of Reco and
GPP. The onset of the Reco active period was in April depending on the temperature. The date of
GPP increase started after the emergence of corn in late May to early June. The values of NEE
were positive (carbon source) before mid-June when Reco > GPP. The NEE then became negative

156

and decreased sharply due to the dramatic increase of GPP in late June, whereas the increase of
Reco was less over the same period. The NEE attained its minimum (negative value, carbon sink)
in mid-July before the soil water deficit became a limiting factor (Figs. 5.2 (A) & (B); Fig. 3.1
(B)) in 2011, 2013 and 2014. The decrease of the NEE values was slower in 2012 due to the poor
growth of corn in July (low NDVI & EVI, Fig. 3.1 (C)). Regardless of year, a remarkable
decrease in the absolute NEE value and low VWC occurred in phase, although it occurred at
different times in different years. The regular concaves of absolute NEE values were due to the
decreases of both GPP and Reco due to water deficit. For example, the concave happened in late
June in 2011, 2012 and 2013 and in mid-August in 2014 (Figs. 5.2 & 3.1 (B)). The water deficitderived low values of NEE, GPP and Reco in the middle of the growing season (GS) separated the
hump-shaped seasonal curves into two periods: early GS and late GS. The seasonality of Rsâ
highly coincided with GPP and Reco but generally had smaller amplitude.
Corn usually had smaller minimum NEE with large maximum GPP comparing to other
crops (Table 5.1), especially the CRP-C field. The minimum GPP in CRP-C was 91.86 g CO2 m2

d-1 occurring in 2014. It had a shorter growing season (Fig. 5.2) with higher peaks of GPP in

the late GS.
Like the seasonality of Rs described in Chapter 3, perennial and annual crops had
different onset dates of active NEE, GPP and Reco. In perennial croplands, the onsets of GPP and
Reco active periods, depending on the date of spring warming, were in phase but different amount
of increase. The date when the ecosystem converted from a carbon source (positive NEE) to a
sink (negative NEE) were several weeks after the onsets of GPP and Reco active periods, ranging
from late March (2012) to late May (2014) when the amount of GPP was higher than Reco.
Perennial crops and annual crops have different seasonal patterns of NEE due to their different

157

phenology and length of growing season. In spring (April â May), perennial cropâs NEE is small
while GPP is higher than Reco (Figs. 5.2 (C)-(G)). The NEE in corn fields was high since we
planted corn in late May, thus April-May NEE was high (Figs. 5.2 (A)-(B)). The June-July NEE
in corn fields was lower than that in perennial croplands, since there was a rapid corn growth
after it emerged. The high GPP in corn ecosystems maintained itself until the late summer water
deficit, which occurred in late July in 2011-2013 and mid-August in 2014. In sum, perennial
crops have lower NEE early on (April-July) with a longer GS overall, while annual crops have
lower NEE mid-GS (June-July).
Table 5.1. The minimum NEE, maximum GPP and maximum Reco among sites across
years. The unit is g CO2 m-2d-1. Negative NEE means carbon flow from the atmosphere to the
ecosystem (the ecosystem was carbon sink).
CROP
C
Sw
Pr
Ref
CROP
C
Sw
Pr
Ref

2011

LUH

2012

Min NEE

Max GPP

Max Reco

Min NEE

Max GPP

Max Reco

CRP

-44.89

83.28

58.2

-29.55

59.92

39.28

AGR

-38.01

67.23

32.81

-24.16

54.14

33.25

CRP

-22.24

47.57

33.8

-22.87

48.48

34.44

AGR

-25.16

48.06

37.18

-25.33

54.73

35.13

CRP

-19.92

48.27

31.27

-19.56

45.28

33.8

AGR

-12.59

38.53

27.77

-41.09

70.03

40.87

CRP

-32.86

61.2

56.08

-25.5

57.45

40.93

2013

LUH

2014

Min NEE

Max GPP

Max Reco

Min NEE

Max GPP

Max Reco

CRP

-46.6

83.42

56.24

-57.21

91.86

46.48

AGR

-45.82

83.06

51.61

-62.49

75.87

32.82

CRP

-37.45

73.75

37.06

-46.81

77.62

40.08

AGR

-42.72

69.05

42.15

-35.25

62.36

33.93

CRP

-24.38

53.78

38.97

-34.97

65.96

37.07

AGR

-41.09

70.03

40.87

-43.65

69.55

33

CRP

-30.21

59.22

41.55

-23.79

45.24

37.42

158

25
25

NEE
GPP
Reco
Rabove
Rs

(A) CRP-C
NEE, GPP, Reco, Rabove & Rs (Âľmol s-1m-2)

20
20

15
15

10
10

55

00
J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

M

A

M

J

J

A

S

O

N

D

Month
-5
-5

2011

2012

2013

2014

-10
-10

-15
-15
25
25

(B) AGR-C

NEE
GPP
Reco
Rabove
Rs

NEE, GPP, Reco, Rabove & Rs (Âľmol s-1m-2)

20
20

15
15

10
10

55

00
J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

Month
-5
-5

2011

2012

2013

2014

-10
-10

-15
-15

Figure 5.2. The seasonal and interannual variations of daily NEE, GPP, Reco, Rabove and Rs among sites during 2011-2014. Grey
lines indicate net ecosystem exchange (NEE) of CO2. The negative NEE means a carbon sinks. Red dotted lines show gross primary
production (GPP) while black solid lines display ecosystem respiration (Reco). Green and brown lines present the aboveground (Rabove)
and belowground (Rs) respiration. (A) CRP-C; (B) AGR-C; (C) CRP-Sw; (D) AGR-Sw; (E) CRP-Pr; (F) AGR-Pr; and (G) Ref.
159

Figure 5.2. (contâd)
25

25

NEE
GPP
Reco
Rabove
Rs

(C) CRP-Sw
20

NEE, GPP, Reco, Rabove & Rs (Âľmol s-1m-2)

20

15

15

10

10

5

0

-5

5

0

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

Month

-5

-10

F

2011

2012

2013

2014

-10

-15

-15
25

25

(D) AGR-Sw

NEE
GPP
Reco
Rabove
Rs

NEE, GPP, Reco, Rabove & Rs (Âľmol s-1m-2)

20
20

15
15

10
10

5
5

00

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

Month
-5
-5

-10
-10

2011

2012

2013

-15
-15

160

2014

J

J

A

S

O

N

D

Figure 5.2. (contâd)
25
25

(E) CRP-Pr

NEE
GPP
Reco
Rabove
Rs

NEE, GPP, Reco, Rabove & Rs (Âľmol s-1m-2)

20
20

15
15

10
10

55

00
J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

M

A

M

J

J

A

S

O

N

D

Month
-5
-5

2011

-10
-10

2012

2013

2014

-15
-15
25

(F) AGR-Pr

NEE
GPP
Reco
Rabove
Rs

NEE, GPP, Reco, Rabove & Rs (Âľmol s-1m-2)

20
20

15
15

10
10

55

00
J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

Month
-5
-5

-10
-10

2011

2012

2013

-15
-15

161

2014

Figure 5.2. (contâd)
25
25

(G) Ref

NEE
GPP
Reco
Rabove
Rs

NEE, GPP, Reco, Rabove & Rs (Âľmol s-1m-2)

20
20

15
15

10
10

55

00
J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

M

J

J

A

S

O

N

D

J

F

Month
-5
-5

-10
-10

2011

2012

2013

-15
-15

162

2014

M

A

M

J

J

A

S

O

N

D

5.3.2 The mid-growing season NEE, GPP, Reco, Rabove and Rs among CROP and LUH
Generally, perennial croplands had lower NEE (higher carbon sequestration rates) due to
higher GPP with lower Reco (Figs. 5.3 (A)-(C)). The NEE increased in the order Sw ~ Pr < C <
Ref except during the drought year (2012) when GPP and Reco both showed no differences
between annual and perennial crops. Compared NEE across years, the Ref site had low NEE
since it had a high GPP in 2011. The value of NEE, on the contrary, became the highest among
all sites 2012 â 2014 due to its low GPP. In the experimental perennial crops (Sw and Pr), the
CRP-Pr had highest NEE and the lowest GPP out of any other perennial crops in 2012 and 2013.
Its Reco did not show any significant difference from other perennial crops. The CRP-Sw had a
higher GPP and Reco among perennial crops in non-drought years. Corn always had higher
temporal variations of NEE, GPP and Reco than any other crops in my analytical period (21 May
â 20 August).

5.3.3 The mid-growing season Reco:GPP, Rs:GPP and Rs:Reco ratios among CROP and LUH
There is no significant difference in the Reco:GPP ratio among sites due to high temporal
variation in the mid-GS. The within site variation was larger than the between-site variation (Fig.
5.3 (A)). The within-site temporal variation of Reco:GPP ratios can be very large or very small
when GPP becomes small or large, but Reco has different responses or time lags.
Unlike Reco:GPP ratio, Rs:Reco and Rs:GPP ratios have very different values between the
undisturbed Ref site and other sites. Ref always had high values of Rs:Reco and Rs:GPP ratios,
following CRP-Sw , AGR-Pr and others in order (Figs. 5.4 (B) & (C); Table 5.2). Annual crops
always had lowest Rs:Reco and Rs:GPP ratios among all crops regardless of year.

163

(A)

(B)

(C)

Figure 5.3. The mid-growing season (21 May-20 August) average NEE, GPP, Reco, Rabove
and Rs for each of the four years. The boxplots show the quartiles, median values and outliers.
The letters indicate the results of Tukeyâs HSD test for different sites. The letters display the
means from small to large alphabetically. Different colors show different CROP (yellow: C; light
green: Sw; bright green: Pr; forest green: Ref) while the dashed and transparent column shows
different LUH (non-textured: CRP; dotted: AGR). The mid-growing season is identified as 21
May â 20 August due to the limitation of available data.

164

Figure 5.3. (contâd)
(D)

5.3.4 The shift of mid-growing season Reco:GPP and Rs:Reco ratios across years
Most sites did not have substantial changes in the Reco:GPP ratio across years because the
within-GS temporal variations of the Reco:GPP ratio were larger than between-GS variations.
Only two sites had statistically significant shifts during the experimental period. The CRP-Sw
showed a decrease 2012â2013 while CRP-Pr decreased 2011â2014. The decreases in the
Reco:GPP ratio to less than one means the system converted from a carbon source to a carbon
sink mid-GS. The Reco:GPP ratio in the Ref site in 2012 exceeded one, which means it was a
carbon source in the drought year (Fig. 5.5 (A)).

165

(A)

(B)

(C)

Figure 5.4. The Reco:GPP, Rs:GPP and Rs:Reco ratios among CROP and LUH over four
years. The boxplots show the quartiles, middle values and outliers. The letters indicate the
results of Tukeyâs HSD test for different sites. The letters display the means from small to large
alphabetically. Different colors show different CROP while different: yellow: corn; light green:
Sw; bright green: Pr & dark green: Ref. Textured column show AGR LUH. The panels without
letters indicate there is no significant difference among treatments ((A) 2012-2014).

166

Table 5.2. The Reco:GPP, Rs:GPP and Rs:Reco ratios among CROP and LUH over four
years. CROP: current crop type; LUH: land use history; C: corn; Sw: switchgrass; Pr: prairie;
Ref: reference; CRP: Conservation Reserve Program; AGR: corn: soybean conventional rotation
agriculture; R: ecosystem respiration; GPP: gross primary production; R: soil respiration. The
data presents mean Âą 1 S.D.
Reco:GPP
CROP

C
Sw
Pr

LUH

C
Sw
Pr

C
Sw
Pr
Ref

2013

2014

1.180

Âą

1.181

0.851

Âą

1.973

-3.090

Âą

40.181

0.797

Âą

2.006

AGR

0.972

Âą

0.875

0.963

Âą

2.175

1.377

Âą

4.935

-0.081

Âą

9.657

CRP

0.801

Âą

0.247

0.897

Âą

1.089

0.668

Âą

0.357

0.610

Âą

0.467

AGR

0.837

Âą

0.348

0.900

Âą

1.435

0.581

Âą

0.188

0.719

Âą

1.737

CRP

1.870

Âą

5.993

1.305

Âą

2.395

0.767

Âą

0.275

0.613

Âą

0.432

AGR

1.030

Âą

0.498

0.605

Âą

4.237

0.620

Âą

0.259

0.587

Âą

0.569

0.847

Âą

0.282

1.142

Âą

3.230

0.962

Âą

0.468

0.851

Âą

0.342

Rs:GPP

LUH

2011

2012

2013

2014

CRP

0.243

Âą

0.378

0.098

Âą

0.380

-0.037

Âą

1.933

0.140

Âą

0.398

AGR

0.085

Âą

0.103

0.059

Âą

0.171

0.093

Âą

0.220

0.071

Âą

0.227

CRP

0.382

Âą

0.209

0.406

Âą

0.589

0.621

Âą

0.378

0.510

Âą

0.309

AGR

0.213

Âą

0.129

0.233

Âą

0.423

0.170

Âą

0.067

0.324

Âą

0.721

CRP

0.321

Âą

1.137

0.199

Âą

0.240

0.265

Âą

0.174

0.179

Âą

0.174

AGR

0.354

Âą

0.184

0.096

Âą

1.262

0.419

Âą

0.232

0.465

Âą

0.502

0.762

Âą

0.397

1.026

Âą

1.289

0.746

Âą

0.325

0.881

Âą

0.321

Ref
CROP

2012

CRP

Ref
CROP

2011

Rs:Reco

LUH

2011

2012

2013

2014

CRP

0.191

Âą

0.085

0.130

Âą

0.052

0.159

Âą

0.053

0.270

Âą

0.160

AGR

0.089

Âą

0.080

0.049

Âą

0.033

0.088

Âą

0.033

0.350

Âą

0.542

CRP

0.491

Âą

0.249

0.524

Âą

0.429

0.947

Âą

0.429

0.894

Âą

0.307

AGR

0.250

Âą

0.095

0.270

Âą

0.134

0.294

Âą

0.080

0.470

Âą

0.152

CRP

0.198

Âą

0.110

0.209

Âą

0.205

0.377

Âą

0.283

0.368

Âą

0.377

AGR

0.354

Âą

Âą

0.102

0.717

Âą

0.291

0.800

Âą

0.231

0.956

Âą

0.129
0.473

0.237
0.911

Âą

0.456

0.835

Âą

0.316

1.127

Âą

0.555

The Rs:Reco ratios had stronger inter-GS trends across years. Over the longer-term period
(four years), the trends of Rs:Reco ratios were increasing, except in Ref. The Rs:Reco ratios in four
perennial sites (CRP-Sw, CRP-Pr, AGR-Sw and AGR-Pr) significantly increased. The CRP-Sw
had the largest amplitude of increase, rising from 0.49 to 0.89, whereas AGR-Pr increased from
0.35 to 0.80. The ratios in annual crops increased slightly while Ref did not have a significant
increase. The drought year had a small or no change in Rs:Reco ratios (Fig. 5.5 & Table 5.1).
167

(A)
2

2

1

1

Reco/GPP

0
2012

2013

2014

2011

-1

AB

2012

2013

2014

2013

2014

-1

C

-2

-2

CRP-C
AGR-C

-3
-4

B

Sw
CRP-Sw
AGR-Sw

-3
-4
2

AB
AB

1

A

1
0

0

Reco/GPP

A

0
2011

2

B

AB

2011

2012

2013

2014

2011

year

-1

Pr

-2

-2

CRP-Pr
AGR-Pr

-3
-4

2012

year

-1

Ref
Ref

-3
-4

(B)

C

1.6

Rs/Reco

1.2

Sw

1.2

CRP-C
AGR-C

CRP-Sw
AGR-Sw
A

0

A

1.2

AB

B

a
2011

1.6

2012

a

b
C

0

B

2013

2011

2014

1.6

2012

2013

Ref

2014

Ref

c

d

B
0.8

2011

a

a

1.2

CRP-Pr
AGR-Pr

A

b

0.4

0

Pr

a

A

a

a

0.8

0.4

B

0.8

0.8

0.4

Rs/Reco

1.6

b

A

A

0.4

B

A
2012

AB

B
0

year

2013

2014

2011

2012

year 2013

2014

Figure 5.5. The mid-growing season Reco:NPP and Rs:Reco ratios across years in different
sites. The yearly variations of Reco:GPP (A) and Rs:Reco ratios (B) are indicated by their CROP
(between panels) and LUH (within panel). Solid dots show CRP while empty circles show AGR
sites. The letters present the results of Tukeyâs HSD test for different sites, where upper-case
letters present the statistical results of CRP and the lowercase letters represent the results of AGR
sites. The letters display the means from small to large alphabetically. The curves without letters
mean there is no significant difference between all years. The Reco:NPP ratio (A) does not show
the high or low standard deviations completely since the range is too large.

168

5.3.5 The shift in annual NEE, GPP, Reco and Rs among CROP and LUH across years
In all treatments, NEE increased in the drought year (2012) and then decreased in the
following years (2013 and 2014). In the drought year, the increases in NEE are more likely due
to the decrease of GPP than the increase of Reco. The NEE decreased in most fields in 2013 and
2014. However, GPP and Reco contributed differently to the NEE decrease in two yearsâ2013
and 2014. In 2013, the decrease of NEE was due to the different magnitude of increase on GPP
and Reco, although both of them had remarkable increases. The decrease of NEE in 2014 was
likely due to the coincident decrease of Reco, while the change of GPP varied (Figs. 5.6 (B) &
(C)).
Comparing the change of NEE across years, each field had its own distinct trajectories
due to different GPP and Reco responses to climate. The Ref always had the highest annual NEE.
It was positive in 2012, 2013 and 2014, meaning it was a carbon source. At other sites, most
NEE was negative, implying most biofuel cropping fields were carbon sinks (Fig. 5.6 (A)). The
CRP-C, the annual crop with high soil carbon and nitrogen content, had the highest Reco out of
any other perennial crop or AGR-C fields (close to Ref site, Fig. 5.6 (C)), resulting in its high
NEE. The CRP sites generally had a higher GPP and Reco than AGR sites. The mid-GS Rs had a
similar yearly trend to GPP. Annual GPP explained 74% of mid-GS Rs and 38% of annual Reco
(Fig. 5.7).
The AGR sites always had a smaller NEE than CRP sites under the same crop type. The
higher soil carbon and nitrogen content may increase Reco in CRP fields although GPP in CRP
fields was generally higher than those in AGR fields. The CRP-Pr field had a remarkably low
NEE with high GPP and Rs (Figs. 5.6 (A), (B) & (D)), implying that its well-established root

169

system may have helped the growth of plants and stimulated a higher photosynthetic rate and
higher root metabolism.

170

(A)

(B)
500
500

0

-500
â500

AnnNEE

NEE (g CO2m-2y-1)

0

-1000
â1000

-1500
â1500

-2000
â2000

2011
2011

2012

2012

2013
2013

year

2014
2014

year

(C)

(D)

Figure 5.6. The cumulative annual NEE, GPP, Reco and mid-growing season Rs across years
in different sites. The net annual NEE (A), GPP (B), Reco (C) and mid-GS Rs (D) in all sites
were presented in different panels. The stellate figures presented Ref sites while circles, squares
and triangles shows corn, switchgrass and prairie, respectively. The solid figures with solid lines
display CRP sites and the empty figures with dash lines show AGR sites. The mid-growing
season is identified as May 21 - August 20 due to the limitation of available data.

171

(A)

(B)

Figure 5.7. The linear regression models of Reco and mid-growing season Rs to annual GPP.
The correlation between mid-GS Rs and annual GPP (A) and between Reco and annual NEE (B).
The GPP (B), (C) and (D) in all sites were presented in different panels. The stellar figures
present Ref sites while circles, squares and triangles shows corn, switchgrass and prairie,
respectively. The solid symbols display CRP sites and the empty symbols show AGR sites. The
mid-growing season is identified as May 21 - August 20.

172

5.4 DISCUSSION

5.4.1 The seasonality of NEE, GPP & Reco between annual and perennial crops
The seasonality of NEE can be mostly explained by the different phenology between corn
and other perennial crops. We always planted corn in May, resulting in a short GS due to late
onset. The low GPP before June determined the positive spring NEE in corn fields (carbon
source) during this period. Perennial crops, in contrast, germinated when the soil temperature
rose in spring, generally in April. The two-month difference of carbon assimilation explained the
different seasonality of NEE and GPP between annual and perennial crops, which also reflected
annual NEE. Corn sites had a weaker carbon sequestration capacity than switchgrass and
restored prairie, with a lower annual GPP (Figs. 5.6 (A) & (B)) at an annual scale. However it
had a very strong peak of NEE in July and August (Fig. 5.2) and its minimum daily NEE was
always lower than that in perennial crops (Table. 5.1).
Falge et al. (2002) analyzed FLUXNET data from 35 different sites across diverse
biomes, climate regimes and latitudes. They inferred that the future climate warming in
temperate and high latitude areas would alter the terrestrial ecosystem carbon balance by
increasing the net primary production (NPP) by changing the difference between GPP and
autotrophic respiration and the length of growing season. The latter will have a stronger impact
on the ecosystem carbon budget than the former. In My experimental farms, the length of GS is
important for the annual carbon budget, however, the climate benefits from annual and perennial
cropping ecosystems can be very different resulting from different mechanisms. On one hand,
the increasing temperature would lengthen the growing season of perennial crops, but would
have little effect on corn in the croplands. The climate warming would enhance the difference in
GS length between corn and perennial crops and would enlarge the difference in their annual

173

carbon sequestration capacities. On the other hand, the advanced onset of a GS may increase the
risk of a summer water crisis in perennial crop fields, since the high spring evapotranspiration
may consume most of the soil water. The impacts that future climate will have on NEE on annual
or perennial crop ecosystems are âcontext-dependentâ.
The timing of unusual climate events or the seasonality of temperature and precipitation
may also play different roles on the annual NEE in crop ecosystems. For example, low summer
temperature may increase the value of NEE in an annual crop more than in a perennial crop,
since corn has higher GPP in a short summer period while perennial grasses have a relatively
gentle variation and a longer active period of GPP. In contrast, the spring drought may impact
perennial more than annual crops since early emergence, which is characteristic of perennial,
may drain soil water and induce a high risk of water deficit in the summer.
In sum, the integration of temperature and water availability interplay, the timing of
climate events and their gradients should be considered when we estimate effects of the future
climate scenarios on different ecosystems.

5.4.2 The mid-growing season Reco:GPP and Rs:Reco ratios
The seasonality of NEE is comprised of the seasonality of respiration assimilatory fluxes
and processes (Falge et al., 2002). Carbon assimilation and respiration have distinct drivers.
Daytime NEE was more controlled by photosynthetically active radiation (PAR) while nighttime
NEE and ecosystem respiration were dominated by temperature (Gilmanov et al., 2007; Zhang et
al., 2015; Chen et al., 2015). The time lags between the response of GPP to PAR and the
responses of Reco to temperature may increase the Reco:GPP ratio dramatically. The high
temporal variation of the Reco:GPP ratio reveals time lags in my systems. Only the Reco:GPP ratio
in CRP-Pr between 2011 and 2014 and in CRP-Sw between 2012 and 2013 had significant

174

differences. CRP-Pr seems to have a long-term decrease in the Reco:GPP ratio from 2 in 2011 to
less than 1 in 2014, while CRP-Sw obviously decreased in the Reco:GPP ratio between 2012 and
2013 (Fig. 5.5 (A)). Both of them were due to the significant increase of GPP after 2012. There
are several possible reasons for the dramatic increase of GPP after 2012. First is the shorter and
delayed summer water deficit. The summer water deficit occurred mid- to late July in 2011 and
2012 (Fig. 3.1 (B)), which depressed all components of NEE, including GPP, Reco and Rs (Fig.
5.2). This is because the low availability of water limited biological activities and reduced the
photosynthetic carbon source indirectly. The weak and delayed water deficit mitigated the
limitation of photosynthesis and brought high GPP and Reco. The amplitude of increase in GPP
was greater than Reco, resulting in lower NEE. Second, the decrease of the Reco:GPP ratio in all
Sw and Pr sites coincided with the increase of C4:C3 biomass after the severe drought in 2012
(Fig. 3.7, Abraha et al., 2016). The increased dominance of C4 grasses may explain the higher
photosynthetic rates in 2013 and 2014.

5.4.3 The interannual trend of mid-GS Rs:Reco ratios
One of the most important differences between annual and perennial crops is the
development of their root systems. Perennial crops develop dense root systems to increase their
water and nutrient uptake ability and store carbon belowground. The establishment of the root
system requires more than one year and may strongly affect soil respiration. The contribution of
belowground respiration may increase year by year due to the development of root systems in
perennial crop ecosystems. My results across different crops proved this hypothesis and revealed
that the development of root systems increase the Rs contribution to Reco after land use change.
The Ref field had a high Rs:Reco ratio across treatments, while corn fields had a low Rs:Reco ratio.
Perennial crops Pr and Sw had remarkable increasing trends from 2011 to 2014 (Fig. 5.5 (B)).

175

The Rs:Reco ratios in CRP-Sw and AGR-Pr sites increased and arrived at similar levels to the Ref
site, implying that they had established their root systems five years after land use change.

5.4.4 The limitation of the TWV model
For the estimation of soil respiration, I assumed that total soil respiration (Rs) is a
function of soil temperature (Ts), soil moisture (VWC) and the Enhanced Vegetation Index (EVI)
(i.e., TWV model in Chapter 2; Eq. 2-2 & Supplement 2-4). There are some uncertainties that
may affect the estimates of Rs.
First, I combined the four-year data set to develop the parameters of the equation.
However, we experienced a severe drought (2012) and two cool summers (2013 and 2014). The
effects of climate regimes in a certain year and the immediate and lagged responses to severe
drought may be complex and hard to recognize because the model ignored the differences in
parameters between years. For example, my TWV model may have underestimated Rs in the
disturbed year (2012) and overestimate the Rs in the recovery years (2013 & 2014). It would be
better to add a âyearâ term in the equation to correct the problem.
Second, the extrapolation of soil respiration from the growing season to the whole year
may introduce uncertainties. The soil respiration data were collected biweekly during the
growing season. However, the soil respiration may show different patterns in different seasons.
The cross-seasonal estimation brought uncertainty. Based on my model, the winter Rh was
always higher than Rs since Rh had a smaller sensitivity to temperature change (Supplement 2-4).
It would have been better to monitor soil respiration in winter to address this problem.
Third, the EVI data were collected from MODIS and had low spatial (250 by 250m) and
temporal resolution (16 day). My experimental site fits into only one grid cell of the image and
had no replication of the EVI value. The 16-day temporal resolution is adequate to follow the

176

dynamics of vegetation, but that is too long of a time period for soil respiration estimation. The
seasonal curves of Rs were stepped. Higher spatial and temporal resolution of vegetation index or
leaf area index may improve the TWV model. The in situ measurement of LAI may reduce the
uncertainty of EVI.

5.5 CONCLUSIONS
Most studies on ecosystem carbon fluxes response to climate focused on interannual scale
and ignored the seasonality of climate variables. My research reveals that the impacts of climate
change on NEE of CO2 are context-dependent and require emphasize at the intra-annual level.
The different phenology between annual corn and perennial grasses affect the seasonality of GPP
and Reco. The timing of specific climate events, such as common summer drought, the occasional
spring severe drought or cool summer may impact annual and perennial crops differently,
whereas the annual mean temperature and precipitation are similar.
Severe drought increases NEE by decreasing both GPP and Reco. The responses of GPP
and Reco in the following years were different, although both GPP and Reco rebounded
remarkably. Huge amounts of dead organic carbon entered the soil due to the 2012 drought and
stimulated the growth of plants. This resulted in high GPP and Reco in 2013. Reco dropped
dramatically in 2014 while GPP in CRP fields maintained a high level. The different responses
of GPP and Reco two years after the severe drought supported the high ability of carbon
sequestration in CRP fields, especially in the CRP-Pr field. The combination of high soil
nutrients and perennial crops stimulated a quick development of a root system with a strong
carbon sequestration ability.
The net increase of NEE in the drought year for annual crop fields was higher than
perennial crops, implying that there is a low resistance in cornâeven though the NEE decreased

177

in the following year. In contrast, perennial crops increased their tolerance to drought by
developing root systems that can absorb deeper water sources and store carbon and nutrients for
further use. Soils in perennial and annual crops have different strategies to tolerant severe
drought.

178

LITERATURE CITED

179

LITERATURE CITED

Abraha, M., Chen, J., Chu, H., Zenone, T., John, R., Su, Y.J., Hamilton, S.K., Robertson,
G.P., 2015. Evapotranspiration of annual and perennial biofuel crops in a variable
climate. Global Change Biology Bioenergy, 7, 1344-1356.
Abraha, M., Gelfand, I., Hamilton, S.K., Shao, C., Su, Y.J., Robertson, G.P., Chen, J.,
2016. Ecosystem water-use efficiency of annual corn and perennial grasslands:
contributions from land-use history and species composition. Ecosystems, 19,
1001-1012.
Burba, G.G., McDermitt, D.K., Grelle, A., Anderson, D.J., Xu, L.K., 2008. Addressing
the influence of instrument surface heat exchange on the measurements of CO2 flux
from open-path gas analyzers. Global Change Biology, 14, 1854â1876.
Chapin, F.S., Woodwell, G.M., Randerson, J.T., Rastetter, E.B., Lovett, G.M., Baldocchi, D.D., Clark, D.A., Harmon, M.E., Schimel, D.S., Valentini, R., Wirth, C., Aber,
J.D., Cole, J.J., Goulden, M.L., Harden, J.W., Heimann, M., Howarth, R.W.,
Matson, P.A., McGuire, A.D., Melillo, J.M., Mooney, H.A., Neff, J.C., Houghton,
R.A., Pace, M.L., Ryan, M.G., Running, S.W., Sala, O.E., Schlesinger, W.H.,
Schulze, E.-D., 2006. Reconciling carbon-cycle concepts, terminology, and
methods. Ecosystems, 9, 1041-1050.
Chen, C., Li, D., Gao, Z., Tang, J., Guo, X., Wang, L., Wan, B., 2015. Seasonal and
interannual variations of carbon exchange over a rice-wheat rotation system on the
north China Plain. Advances in Atmospheric Sciences, 32, 1365-1380.
Deal, M.W., Xu, J., John, R., Zenone, T., Chen, J., Chu, H., Jasrotia, P., Kahmark, K.,
Bossenbroek, J., Mayer, C., 2013. Net primary production in three bioenergy crop
systems following land conversion. Journal of Plant Ecology, 6, 1-10.
Falge, E., Baldocchi, D., Tenhunen, J., Aubinet, M., Bakwin, P., Berbigier, P., Bernhofer,
C., Burba, G., Clement, R., Davis, K.J., Elbers, J.A., Goldstein, A.H., Grelle, A.,
Granier, A., GuĂ°mundsson, J., Hollinger, D., Kowalski, A.S., Katul, G., Law, B.E.,
Malhi, Y., Meyers, T., Monson, R.K., Munger, J.W., Oechel, W., U, K.T.P.,
Pilegaard, K., Rannik, Ă., Rebmann, C., Suyker, A., Valentini, R., Wilson, K.,
Wofsy, S., 2002. Seasonality of ecosystem respiration and gross primary
production as derived from FLUXNET measurements. Agricultural and Forest
Meteorology, 113, 53-74.
Gelfand, I., Zenone, T., Jasrotia, P., Chen, J., S.K.Hamilton, Robertson, G.P., 2011.
Carbon debt of conservation reserve program (CRP) grasslands converted to
bioenergy production. Proceedings of the National Academy of Sciences, 108,
13864-13869.

180

Gilmanov, T.G., Soussana, J.F., Aires, L., Allard, V., Ammann, C., Balzarolo, M.,
Barcza, Z., Bernhofer, C., Campbell, C.L., Cernusca, A., Cescatti, A., CliftonBrown, J., Dirks, B.O.M., Dore, S., Eugster, W., Fuhrer, J., Gimeno, C.,
Gruenwald, T., Haszpra, L., Hensen, A., Ibrom, A., Jacobs, A.F.G., Jones, M.B.,
Lanigan, G., Laurila, T., Lohila, A., Manca, G., Marcolla, B., Nagy, Z., Pilegaard,
K., Pinter, K., Pio, C., Raschi, A., Rogiers, N., Sanz, M.J., Stefani, P., Sutton, M.,
Tuba, Z., Valentini, R., Williams, M.L., Wohlfahrt, G., 2007. Partitioning European
grassland net ecosystem CO2 exchange into gross primary productivity and
ecosystem respiration using light response function analysis. Agricultural,
Ecosystems and Environment, 121, 93-120.
McMillen, R.T., 1988. An eddy correlation technique with extended applicability to nonsimple terrain. Boundary-Layer Meteorology, 43, 231â245.
Moncrieff, J., Clement, R., Finnigan, J.J., Meyers, T., 2004. Averaging, detrending, and
filtering of eddy covariance time series. In: Handbook of Micrometeorology: A
Guide for Surface Flux Measurement and Analysis, Lee X, Massman W, Law B
(Eds). Kluwer Acadamic: Dordrecht, Netherlands, 7â32.
Peel, M.C., Finlayson, B.L., McMahon, T.A., 2007. Updated world map of the KoppenGeiger climate classification. Hydrology and Earth System Sciences Discussions 4,
439-473.Post, W.M., Emanuel, W.R., Zinke, P.J., Stangenberger, A.G., 1982. Soil
carbon pools and world life zones. Science, 298, 156-159.
Reichstein, M., Falge, E., Baldocchi, D., Papale, D., Aubinet, M., Berbigier, P.,
Bernhofer, C., Buchmann, N., Gilmanov, T., Granier, A., Grunwald, T.,
Havrankova, K., Ilvesniemi, H., Janous, D., Knohl, A., Laurila, T., Lohila, A.,
Loustau, D., Matteucci, G., Meyers, T., Miglietta, F., Ourcival, J.M., Pumpanen, J.,
Rambal, S., Rotenberg, E., Sanz, M., Tenhunen, J., Seufert, G., Vaccari, F., Vesala,
T., Yakir, D., Valentini, R., 2005. On the separation of net ecosystem exchange into
assimilation and ecosystem respiration: review and improved algorithm. Global
Change Biology, 11, 1424â1439.
Robertson, G.P., Hamilton, S.K., 2015. Long-term ecological research at the Kellogg
Biological Station LTER site: conceptual and experimental framework. In:
Hamilton, S.K., Doll, J.E., Robertson, G.P. (Eds.), The Ecology of Agricultural
Landscapes. Oxford University Press, New York, pp. 1-32.
Schimel, D., Melillo, J., Tian, H., McGuire, A.D., Kicklighter, D., Kittel, T., Rosenbloom,
N., Running, S., Thornton, P., Ojima, D., Parton, W., Kelly, R., Sykes, M., Neilson,
R., Rizzo, B., 2000. Contribution of increasing CO2 and climate to carbon storage
by ecosystems in the United States. Science, 287(5460), 2004-2006.
Soegaard, H., Jensen, N.O., Boegh, E., Hasager, C.B., Schelde, K., Thomsen, A., 2003.
Carbon dioxide exchange over agricultural landscape using eddy correlation and
footprint modeling. Agricultural and Forest Meteorology, 114, 153-173.

181

Webb, E.K., Pearman, G.I., Leuning, R., 1980. Correction of flux measurements for
density effects due to heat and water vapor transfer. Quarterly Journal of the Royal
Meteorological Society, 106, 85â106.
Wilczak, J.M., Oncley, S.P., Stage, S.A., 2001. Sonic anemometer tilt correction
algorithms. Boundary-Layer Meteorology, 99, 127â150.
Zenone, T., Chen, J., Deal, M.W., Wilske, B., Jasrotia, P., Xu, J., Bhardwaj, A.K.,
Hamilton, S.K., Robertson, P.G., 2011. CO2 fluxes of transitional bioenergy crops:
effect of land conversion during the first year of cultivation. GCB Bioenergy, 3,
401-412.
Zenone, T., Gelfand, I., Chen, J., Hamilton, S.K., Robertson, G.P., 2013. From set-aside
grassland to annual and perennial cellulosic biofuel crops: Effects of land use
change on carbon balance. Agricultural and Forest Meteorology, 182-183, 1-12.
Zhang, L., Sun, R., Xu, Z., Qiao, C., Jiang, Q., 2015. Diurnal and seasonal variations in
carbon dioxide exchange in ecosystems in the Zhanfye oasis area, northwest China.
PLoS ONE, 10(3), e0120660.

182

CHAPTER 6
CONCLUSIONS AND IMPLEMENTATIONS
Global climate change, which mainly results from the past centuriesâ anthropogenic
greenhouse gas emissions, alters Earthâs energy, hydrological and carbon cycles. This impacts
both natural and human systems. To address the critical challenges on human well-being and the
ecosystem health and to achieve the purposes of Millennium Development Goals (MDGs; United
Nations, 2015) and the Convention on Biological Diversity (CBD; United Nations, 1992), the
jointly research were incorporated (Sachs et al., 2009) to mitigate the impacts of climate change
on human well-being, biodiversity loss and ecosystem service degradation. The integration and
the convergence of ecological and socio-economic sustainability has quickly become the largest
challenge in contemporary sustainability science. The human-natural nexus system has
dramatically developed to address climate change-derived problems from different perspectives,
such as the coupled human and nature system (CHANS; Liu et al., 2007a, b; McConnell et al.,
2011; Chen & Liu, 2014; Chen et al., 2015), water-energy-food nexus (NEF Nexus; FAO, 2014)
and ecosystem stewardship (Chapin et al., 2010). However, these frameworks are still in their
initial stages, and they require fundamental knowledge to fill the gaps between how subsystems
mechanisms and their respond to climate scenarios.

183

Figure 6.1. The framework of coupled human and nature systems (CHANS) on the impacts
that bioenergy development has on managed ecosystems. Bioenergy-induced land use change
and agricultural practices may change the community structure of vegetation, affect their traits,
modify biogeochemical processes, and ultimately alter ecosystem functions, such as carbon
sequestration ability. The changing climate, such as temperature and precipitation patterns, may
affect biogeochemical processes via community traits with or without altering community
structure. The framework focuses on the impact that bioenergy has on agricultural ecosystem
developments at local scaleâdisincluding economic and telecoupling effects. LUH: land use
history; CROP: crop type; GHG: greenhouse gas; R-T: respiration-temperature correlation; WUE:
water use efficiency; NUE: nutrient use efficiency; LUE: light use efficiency; AOA: ammoniaoxidizing archaea; AOB: ammonia-oxidizing bacteria; GPP: gross primary production; Reco:
ecosystem respiration; Rs: total soil respiration; Ra: autotrophic soil respiration; and Rh:
heterotrophic soil respiration.

184

The purpose of biofuel development is to reduce total carbon dioxide emission per unit of
fuel used. However, direct and indirect land use conversion and agricultural practices in the
biofuel croplands may affect the benefits that bioenergy has on climate. Development of
bioenergy and bioenergy-derived land use change have been the most hotly disputed topics in the
past decade (Liu et al., 2015), leaving many gaps in current knowledge. I developed a nexus
framework of bioenergy, including coupled human and nature systems (Fig. 6.1). My study
mainly focused on the responses of the nature system under some settings of land use conversion
and agricultural practices. I explored the response of community traits, community structure,
major carbon processes and ecosystem carbon sequestration underneath different crop types and
land use histories. This framework does not include socio-economic and telecoupling effects, but
can be expended in the future if required.
My dissertation emphasized the relative contributions that the components of carbon
dioxide processes have on net ecosystem CO2 flux and how they respond to climate patterns in
different lands use histories (before land use conversion) and crop types (after land use
conversion) at inter- (Chapter 2) and intra-annual (Chapter 3) scalesâespecially in regards to
immediate and prolonged effects on a drought during the growing-season. I also developed
Bayesian models to explore the soil respirationâs temperature correlations and its shift after
severe drought (Chapter 4). Chapter 5 discussed the different responses that gross primary
production (GPP) and ecosystem respiration (Reco) have in regards to NEE and its amount as well
as the seasonality of net ecosystem CO2 exchange (NEE) caused by the asynchrony of GPP and
Reco.
In Chapter 2 and Chapter 3, I found that both soil temperature and vegetation index were
positively correlated with total and heterotrophic soil respiration. The temperature and

185

photosynthetic-derived carbon via carbon allocation determines the intrinsic potential of soil
respiration. Soil water content has weak or no correlation with soil respiration. However, it plays
crucial roles during summer drought periods and limits soil respiration.
In my studied ecosystems, different effects and magnitudes of three different natural and
anthropogenic forces on soil respiration were recognized at different temporal scales. First, the
seasonal patterns of temperature and water availability in a specific year affected the annual soil
respiration. Not only did temperature and precipitation affect vegetation growth âand therein
total and heterotrophic soil respirationâbut also the timing of the microclimate. The spring and
summer drought had stronger impacts on annual soil respiration than those occurring in fall since
the spring-summer drought degraded the vegetation growth and photosynthesis, which ultimately
impacted total and heterotrophic soil respiration. Second, the severe drought in 2012 not only
depressed soil respiration within the same year, but also affected one or more following years by
changing vegetation composition in perennial crop fields and ecosystem water use efficiencies
(eWUE; Abraha et al., 2016). The effect of severe drought was strong and lasted more than one
year by changing the vegetation community structure. Third, the long-term trend of autotrophic
soil respiration (Ts) was remarkable in CRP-Pr and AGR-Pr field in the 5th year (2014) following
land use conversion due to well-established root systems.
Generally speaking, annual and perennial crops had different strategies to adapt to severe
drought. Corn has late onset and a shorter growing season to alleviate summer water stress on
vegetation and soil microbes. It also has a higher eWUE (4.1 g C m-2/mm H2O in CRP-C & 3.8
in AGR-C) than perennial crops (2.5-3.3 in prairie and switch grass croplands) (Abraha et al.,
2016). In contrast, perennial crops had highly developed dense root systems helping them absorb
deep water storages. The mature perennial ecosystem at the Ref site had the highest resistance

186

and soil respiration recovery after a severe drought, even though it had the lowest eWUE (2.3;
Abraha et al., 2016). Perennial crops also increased eWUE to acclimate to severe drought by
increasing its C4 plant composition. The different strategies between annual and perennial crops
to cope with drought determined their resistance to increasing risks caused by extreme
temperatures and drought events. Under different climate scenarios and different combinations of
temperature and precipitation gradients, annual and perennial crops may exhibit different
benefits. The increase of temperature may improve photosynthesis rates, elongate the growing
season and stimulate the carbon sequestration abilities of perennial crops, but may also increase
the risk of summer a water crisis. Also, the presence of heavy rain in late July may convert a
cropland from a carbon source to a carbon sink in a hot year. Vegetation composition, which
reflects the cross-year acclimation to the temperature-precipitation regimes, and the succession
of perennial grasslands/croplands also determine the resistance and resilience of soil respiration
to future climate regimes.
After understanding the soil respiration between different crop types and how they
respond to three different forces, I studied the correlation between soil respiration and its major
driver, soil temperature (R-T correlation), in different LUH and CROP and how it responds to
severe drought using Bayesian models. I found that the LR20, which denotes the growing season
mean soil respiration, ranks from high to low in the following ecosystem types: Ref > prairie >
switchgrass > corn. This order coincides with root density, where: late succession perennial crop
(Ref) > early succession perennial crop (Pr and Sw) > annual crop (C). Soil respiration in Ref
had the highest soil respiration temperature sensitivity (Î˛) resistance and recovery and LR20 on
Ra. The well established LR20 and Î˛ of Ra remained even in severe drought conditions. All Î˛ of
Ra decreased while those of Rh increased in Pr, Sw and C fields. The decrease of Î˛Ra coincided

187

with most previous studies that found severe drought depresses soil respiration and its
temperature sensitivity by inhibiting cellular biochemical reaction when water is not available.
The increase of Î˛Rh was surprising. The possible reason is the change of soil microbial
community structure, which caused the high Î˛ with low R when severe drought was experienced.
After a two-year recovery from the 2012 drought, the R-T relationship of Ra and Rh seemed have
different trajectories. The Ra was likely returning to equilibrium while Rh moved toward a new
equilibrium with higher Î˛ and lower LR20. Both the immediate and prolonged reactions of the RT relationship implied that the soil microbial community exhibited higher âplasticityâ in adapting
severe drought while vegetation roots were likely to return to pre-drought condition. When
comparing the locations of different LUH-CROP treatments on the Î˛-LR20 biplots between 2014
and 2011, the Euclidean distances of annual crop and the Ref fields were very small. This
implies no long-term trends in the R-T relationship in annual crops and in late successional grass
ecosystems. Other perennial crop fields had larger Euclidean distances that revealed long-term
trends toward higher growing season Ra with higher temperature sensitivities due to the
development of dense root systems.
The combination of Chapters 2-4 tells an interesting story on how late successional
grassland ecosystems have well-established root systems and high resistance and recovery to
drought disturbancesâeven with high respiration and low eWUE. Early successional grasslands
have slightly a lower stability (higher magnitude of fluctuation after disturbance and longer
duration to recover) than Ref fields. However, the development of root systems, the shift of
vegetation composition and the possible change in soil microbial community structures may help
an ecosystem better adapt to severe drought in the future.

188

GPP and Reco have different driving forces. The asynchrony of the drivers and the time
lags for carbon reallocation result in the seasonality of NEE. Differences in phenology between
annual and perennial crops determine NEE, GPP and Reco seasonality. Corn has a late growing
seasonal onset with low NEE in the mid-summer while perennial grasses have early onset and a
longer growing season. However, early onset of the GS may increase the risk of the summer
water deficit, since high spring evapotranspiration may drain soil water in the early growing
season. The climate regimes and the timing of extreme events may stimulate or dampen the
carbon sequestration ability based on the different ecosystem characteristics. For example, spring
drought may depress perennial grasslands while cool summers may increase the NEE in annual
crop ecosystems by dramatically reducing GPP. Perennial grasses have lower eWUE and higher
soil respiration. Both increase their risk in the face of extreme drought. However, wellestablished root systems may tap into deep-water storages and strengthen drought resistance and
recovery. The carbon and nutrients stored in roots may also support future use.

189

LITERATURE CITED

190

LITERATURE CITED

Abraha, M., Gelfand, I., Hamilton, S.K., Shao, C., Su, Y.J., Robertson, G.P., Chen, J.,
2016. Ecosystem water-use efficiency of annual corn and perennial grasslands:
contributions from land-use history and species composition. Ecosystems, 19,
1001-1012.
Chapin III, F.S., Carpenter, S.R., Kofinas, G.P., Folke, C., Abel, N., Clark, W.C., Olsson,
P., Smith, D.M.S., Walker, B., Young, O.R., Berkes, F., Biggs, R., Grove, J.M.,
Naylor, R.L., Pinkerton, E., Stefen, W., Swanson, F.J., 2010. Ecosystem
stewardship: sustainability strategies for a rapidly changing planet. Trends in
Ecology and Evolution, 25(4), 241-249.
Chen, J., Liu, Y., 2014. Coupled natural and human systems: a landscape ecology
perspective. Landscape Ecology, 29, 1641-1644.
Chen, J., John, R., Zhang, Y., Shao, C., Brown, D.G., Batkhishig, O., Amarjargal, A.,
Ouyang, Z., Dong, G., Wang, D., Qi, J., 2015. Divergences of two coupled human
and natural systems on the Mongolian Plateau. BioScience, 65(6), 559-570.
Liu, J., Mooney, H., Hull, V., Davis, S.J., Gaskell, J., Hertel, T., Lubchenco, J., Seto,
K.C., Gleick, P., Kremen, C., Li, S., 2015. Systems integration for global
systainability. Science, 347, 1258832.
McConnell, W.J., Millington, J.D.A., Reo, N.J., Alberti, M., Asbjornsen, H., Baker, L.A.,
Brozovic, N., Drinkwater, L.E., Drzyzga, S.A., Fragoso, J., Holland, D.S., Jantz,
C.A., Kohler, T.A., Maschner, H.D.G., Monticino, M, Podesta, Jr., G., Pontius,
R.G., Redman, C.L., Sailor, D., Urquhart, G., Liu, J., 2011. Research on coupled
human and natural systems (CHANS): approach, challenges, and strategies. The
Bulletin of the Ecological Society of America, 92, 218â228.
FAO, 2014. The Water-Energy-Food Nexus: A New Approach in Support of Food
Security and Sustainable Agriculture. Food and Agriculture Organization of the
United Nations, Rome.
Liu, J., Dietz, T., Carpenter, S.R., Folke, C., Alberti, M., Redman, C.L., Schneider, S.H.,
Ostrom, E., Pell, A.N., Lubchenco, J., Taylor, W.W., Ouyang, Z., Deadman, P.,
Kratz, T., Provencher, W., 2007a. Coupled human and natural systems. Ambio,
36(8), 639-649.
Liu, J., Dietz, T., Carpenter, S.R., Alberti, M., Folke, C., Moran, E., Pell, A.N., Deadman,
P., Kratz, T., Lubchenco, J., Ostrom, E., Ouyang, Z., Provencher, W., Redman,
C.L., Schneider, S.H., Taylor, W.W., 2007b. Complexity of coupled human and
natural systems. Science, 317, 1513-1516.

191

Sachs, J.D., Baillie, J.E.M., Sutherland, W.L., Armsworth, P.R., Ash, N., Beddington, J.,
Blackburn, T.M., Collen, B., Gardiner, B., Gaston, K.J., Godgray, H.C.J., Green,
R.E., Harvey, P.H., House, B., Knapp, S., Kumpel, N.F., Machonald, D.W., Mace,
G.M., Mallet, J., Matthews, A., May, R.M., Petchey, O., Purvis, A., Roe, D., Safi,
K., Turner, K., Walpole, M., Watson, R., Jones, K.E., 2009. Biodiversity
conservation and the Millennium Development Goals. Science, 325, 1502-1503.
United Nations, 2014. The Millennium Development Goals Report 2015. United Nations,
New York.

192