5:5? 32..
h. 3.

r.

.. w . ..
as... gowns...
.9
V 5‘

Q;
a . _

 

‘13,!
.....VQ¢H .ra
, 41.0.1“.

1... ..
H515 a“?
I.J...fl.s (.1,
2.....1. J...

. seams.

, 5:... .‘ a. .2

3!! n In
In... .

.95
El‘ninzvl
1. §

2 '— lIBRARY j
4 F Michigan State I
140")? University __|

This is to certify that the
dissertation entitled

EFFECT OF TOPOGRAPHY AND SOIL PROPERTIES
ON SPATIAL VARIABILITY OF SOIL CARBON(C)
LOSS IN DIFFERENT CROP MANAGEMENT
SYSTEMS OF A LONG-TERM EXPERIMENT

presented by

SENTHILKUMAR SUBRAMANIAN

has been accepted towards fulfillment
of the requirements for the

PhD degree in Crop and Soil Sciences

 

 

Major Professor’s Signature

[2/0 2/0?

Date

MSU is an Affinnative Action/Equal Opportunity Employer

 

---u--.---.

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Aoc&Pres/ClRCIDateDue.indd

 

 

EFFECT OF TOPOGRAPHY AND SOIL PROPERTIES ON SPATIAL VARIABILITY
OF SOIL CARBON(C) LOSS IN DIFFERENT CROP MANAGEMENT SYSTEMS OF
A LONG-TERM EXPERIMENT

By

Senthilkumar Subramanian

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Crop and Soil Sciences

2008

ABSTRACT

EFFECT OF TOPOGRAPHY AND SOIL PROPERTIES ON SPATIAL
VARIABILITY OF SOIL CARBON(C) LOSS IN DIFFERENT CROP
MANAGEMENT SYSTEMS OF A LONG-TERM EXPERIMENT

By
Senthilkumar Subramanian
Accurate quantification of total soil C is needed to assess the sequestration potential of
different management systems; however the accuracy in estimating soil C to determine
best management practices is affected by the inherent variability of soil properties. I
hypothesized that spatial variability present in soil C distribution across landscape to a
large extent is related to topographical gradients; and that tillage and management practices
might further buffer or intensify the influence of topography on C variability. The study
was carried out at the Long Term Ecological Research site at Kellogg Biological Station,
Michigan. In this work, I analyzed the effects of topographical features, soil properties and
different management practices on soil C dynamics and characterized spatial variability
pattern in soil C distribution in each management system. Treatments studied were short
rotation woody perennial system (poplar) and agronomic systems including chemical input
based chisel-plow (CT), no-till (NT) system and organic based chisel-plow system with
cover crops (CT-cover). As hypothesized, spatial variability characteristics of the soil C in
the agronomic treatments, but not in the poplar systems, were strongly affected by
topographical gradients. Within agronomic system, topography effect was stronger in

organic based chisel-plow system with cover crops (CT-cover) compared to chemical input

A“.

based chisel-plow (CT) and no-till (NT) system. After controlling the topographical effect,
the greatest difference in terms of spatial variability characteristics between poplar and
agronomic systems was observed in variogram values near the origin. Results indicated
that in poplar, C is much more variable at very short distances as compared to agronomic
systems. Total C and overall variability in C observed at poplar was found to be similar to
that of NT, and slightly greater than CT-cover and differs substantially only with CT.
Among the agronomic systems, CT-cover was as efficient as the NT system in restoring C
and N and the largest C and N benefits of CT-cover in the landscape were observed in
valleys as compared to upper slope positions. The net changes in soil C content that
occurred in the past 15-20 years under different agricultural management practices and in
never tilled soils were estimated after accounting for the variations in baseline C, soil
texture and topographical features among the studied treatments. Results indicated that soil
C has decreased by about 10 percent at conventional chisel-plow management soil since
1986. No-till and organic based chisel-plow system neither gained nor lost C as compared
to the baseline values (1986-88). Baseline C was found to be negatively related with
changes in C content and C loss was found to increase with increase in baseline values.
Higher silt and clay contents were associated with greater gains in C in no-till system. Over
all results demonstrated that the conservational management practices have prevented total
C loss as compared to conventional management but there was no net gain in C over the

past two decades in the studied Michigan soils.

'A‘nifi " “ a.
l _

ACKNOWLEDGMENTS

I extend my deep sense of gratitude and sincere thanks to Dr.Sasha Kravchenko,
my major advisor for her teaching, guidance and support all through the course of this
study. She spent countless hours in guiding me with data analysis and documenting the
results which have facilitated the production of this dissertation in a timely manner. It
was a great opportunity and I feel very proud for working under your supervision. I am
greatly indebted to you Sasha.

I extend my sincere thanks to Dr. Philip Robertson, for providing the logistic
support required during the field work and data collection at KBS and for his constructive
suggestions during this study. I always admire him for his great love towards science.

I express my appreciation to my committee members, Dr. Alvin Smucker, Dr. Kurt
Thelen and Dr.Jiaguo Qi for providing valuable comments during the course of this
research. I also thank Brain Graff, Andrew Corbin, Suzanne Sippel and other LTER staff,
for their support during geo-referencing sample locations and soil sampling. Special
thanks also extended to the faculty members and secretarial staff of Crop and Soil
Science Department for their support during my time at Michigan State University.

I am grateful for the financial support received from Kellogg Biological Station-
Long Term Ecological Research, Eugene Pedology award and Dissertation Completion
Fellowship offered by the Department of Crop and Soil Sciences and the Graduate
School during the course of my study.

I would like to thank my fellow colleagues Dr.Xinmei, Dr.Xuewen, Wei, Juan
David, Katya and Mike for their strong support and timely help during my stay at

Michigan State. I am grateful to my friends Madhu, Vijay, Kannan, Kumaran and Srini

Mi

for their great moral support, and encouragement in professional and personal aspects of
my life.

I express my deep sense of indebtedness to my beloved parents Sri Subramanian,
Smt. Kamala, my chittappa Natarajan and my sisters and brother, whose unparallel
affection and constant encouragement in shaping my career will go a long way

throughout my life. Thank you all!.

find a”.‘
‘.'
I

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ........................................................................................................... xi
CHAPTER I

INTRODUCTION ............................................................................................................... 1
CHAPTER II

SPATIAL VARIABILITY OF SOIL C, N CONTENT IN POPLAR AND IN
DIFFERENT ANNUAL CROPPING SYSTEMS IN A LONG TERM EXPERIMENT

ABSTRACT ............................................................................................................. 5
INTRODUCTION ................................................................................................... 7
MATERIALS AND METHOD
Test Site ........................................................................... 11
Experimental Layout and Agronomic Protocols .............................. 12
Soil Sampling ..................................................................... I3
Terrain Attributes ................................................................. 19
Data Analysis ..................................................................... 20
Spatial variability pattern in poplar plots
RESULTS AND DISCUSSION ....................................................... 21
REFERENCES ............................................................................. 42
CHAPTER III

RELATIONSHIPS BETWEEN TOTAL SOIL CARBON AND TOPOGRAPHY OF
DIFFERENT MANAGEMENT SYSTEMS

ABSTRACT ................................................................................. 45
INTRODUCTION ......................................................................... 46
MATERIALS AND METHODS
Test Site ........................................................................... 52
Experimental Layout and Agronomic Protocols .............................. 53
Soil Sampling ........................................................................ 54
Terrain Attributes ....................................................................................... 58
Data Analysis ............................................................................................. 59
Terrain attributes as continuous variables .............................. 60
Analysis of landform classes ........................................... 62
RESULTS
Descriptive Statistics and Landform Classification ........................... 64
Relationship between Terrain Attributes and Soil C and N at 0-5 cm
depth ................................................................................ 66

vi

 

Relationship between Terrain Attributes and Soil C and N at 20-30 cm

depth ................................................................................. 73

Relationship between Terrain Attributes and Soil C and N at 0-30 cm

depth ................................................................................. 78
DISCUSSION .............................................................................. 83
SUMMARY AND CONCLUSIONS ................................................... 86
ACKNOWLEDGEMENTS ............................................................... 87
REFERENCES ............................................................................. 87

CHAPTER IV

CHANGES IN TOTAL SOIL C UNDER DIFFERENT CROP MANAGEMENT
SYSTEMS IN A LONG TERM EXPERIMENT

ABSTRACT ................................................................................. 93
INTRODUCTION ......................................................................... 95
Land Use Control on Soil C Changes .......................................... 97
Management Control on Soil C Changes ....................................... 97
Relationship between Changes in C and Baseline C ....................... 100
Relationship between Changes in C and Soil Texture ...................... 101
MATERIALS AND METHOD ......................................................... 103
Main Site ........................................................................... 104
Interaction Site ................................................................... 106
Data analyses ..................................................................... 110

Relationships between changes in C content and baseline C soil
texture and topographical features in different treatments. . . . . ...1 10

Reanalyzing the baseline values from the main site ............... 1 1]

Regression to the Mean (RTM) correction .......................... 111

RESULTS AND DISCUSSION ........................................................ 115
Long Term Trends in Changes in Soil C under Different

Management Systems ........................................................... 115

Relationship between Changes in C and Soil Texture ...................... 128

SUMMARY ............................................................................... 131

REFERENCES ........................................................................... 132

CHAPTER 1V
SUMMARY AND CONCLUSIONS ................................................. I37
vii

"‘"filhl
a

LIST OF TABLES

Table 2.1a. Description of soil sampling data collected during 2003 and 2004 ............ 14

Table 2.1b. Description of soil sampling data collected at 20-30 cm depth during
May 2006 ............................................................................................. 14

Table 2.2. Treatment mean estimate of total C, N (g kg") in CT, NT, CT-cover and
Poplar treatments of LTER site (standard error in paranthesis) .............................. 22

Table 2.3a. Variances and spatial variability parameters for the total C data (0-5 cm
depth) and the residuals of the regression model with topography under poplar and
three different agronomic management systems ................................................ 24

Table 2.3b. Variances and spatial variability parameters for the total N data (0-5 cm
depth) and the residuals of the regression model with topography under poplar and
three different agronomic management systems ................................................ 24

Table. 2.4a. Summary of soil and topographical properties of the studied plot. Results of
multiple regression for total C, N at the 0- to 5 cm depth with topographical variables
also shown. For total C and N coefficients of variation (%) are shown in

parentheses ......................................................................................................................... 3 1

Table. 2.4b Summary of soil and topographical properties of the studied plot. Results of
multiple regression for total C at the O- to 5 cm and 20-30 cm depth with topographical
variables also shown. For total C coefficients of variation (%) are shown in

parentheses ........................................................................................... 35

Table. 2.5a. Total C content at surface (0-5 cm) and below plow depth (20-30 cm)
at CT and CT-cover treatments .................................................................... 37

Table 2.5b. Variances and spatial variability parameters for the total C data at

surface (0-5 cm) and below plow depth (20-30 cm) and the residuals of the
regression model with topography at CT and CT-cover treatments .......................... 37

Table 3.1. Mean C and N for all the studied depth at CT, NT, CT-cover crop
treatment along with percentages of coefficients of variation in parentheses ............... 65

viii

Table 3.2. Mean bulk density values for all the studied depths and a summary of terrain
properties for the three studied treatment sites .................................................. 65

Table 3.3. Mean values of the terrain variables and soil texture values for the
upperslope and valley positions selected by stepwise discriminant procedure
(PS 0.10) in the study site .......................................................................... 67

Table 3.4. Pearson correlation coefficients between the total C and N at 0- to 5-cm depth
with terrain attributes significant at 0.05 level ................................................... 67

Table 3.5. Results of multiple regressions for total C, N at the 0 to 5cm depths with
terrain attributes significant at 0.05 level ......................................................... 69

Table 3.6. Linear regression slopes, intercepts, r2 and MSE resulting from ANCOVA of
total C, N at 0- to 5-cm with wetness index as covariate ....................................... 70

Table 3.7. Total C and N content at 0- to 5-cm for the upperslope and valley positions in
CT, NT and CT-cover .............................................................................. 70

Table 3.8. Total C and N difference between upperslope and valley landform positions
for CT, NT and CT-cover .......................................................................... 72

Table 3.9. Pearson correlation coefficients between the total C and N at 20- to 30-cm
depth with terrain attributes significant at 0.05 level .......................................... 72

Table 3.10. Results of multiple regressions for total C, N at the 20- to 30-cm depth with
terrain attributes significant at 0.05 level ......................................................... 76

Table 3.11. Linear regression slopes, intercepts, r2 and MSE resulting fiom ANCOVA of
total C, N at 20- to 30-cm with wetness index as covariate .................................. 76

Table 3.12. Total C and N content at 20- to 30-cm at upperslope and valley positions in
CT, NT and CT—cover .............................................................................. 77

Table 3.13 . Pearson correlation coefficients between the total C and N at O- to 30-cm
depth with terrain attributes significant at 0.05 level .......................................... 77

Table 3.14. Results of multiple regressions for total C, N at the 0- to 30-cm depth with
terrain attributes significant at 0.05 level ......................................................... 79

Table 3.15. Linear regression slopes, intercepts, r2 and MSE resulting from ANCOVA of
total C, N at O- to 30-cm with wetness index as covariate .................................... 81

Table 3.16. Total C and N content at O- to 30-cm at upperslope and valley positions in
CT, NT and CT-cover ............................................................................... 81

ix

Table 4.1. Summary of data collection in 1988 and 2006 for LTER mainsite and in 1986
and 2007 for the Interaction site ................................................................. 107

Table 4.2a. Mean baseline(1988) C and contemporary(2006) C in CT, NT and CT-cover
at LTER main site. Standard errors are shown in parentheses .............................. 1 17

Table 4.2b. Mean changes in C content adjusted for standardized baseline C content
along with raw Means (unadjusted for baseline) in CT, NT and CT-cover at LTER main
site. Standard errors are shown in parentheses ................................................ 117

Table 4.3a. Mean baseline C values (g kg ") and contemporary(2007) C content at
fertilized and unfertilized plots of CT and NT at LTER-interaction site treatments.
Standard errors are shown in parentheses ...................................................... 119

Table 4.3b. Mean changes in C values (g kg 'l)adjusted for the standardized baseline C
content in CT and NT of LTER-interaction site treatments. Standard errors are shown in
parentheses .......................................................................................... 119

Table 4.4. Results of observed and RTM adjusted slope estimates for baseline C effect on
changes in C content in the LTER-mainsite treatments ....................................... 127

Table 4.5. Results of observed and RTM adjusted slope estimates for baseline C effect on
change in C content of LTER- interaction site treatments ................................... 127

Table 4.6. Relationships between changes in C, N and silt+clay content at O- to 15- cm
depth for the LTER main site ..................................................................... 129

Table 4.7. Relationships between changes in C and silt+clay content at O- to 20- cm depth
for the LTER interaction site ..................................................................... 129

LIST OF FIGURES

Fig.2. 1. Sampling scheme for each l-ha plot followed at CT, NT and CT-cover
treatments ............................................................................................. 15

Fig.2.2. Sampling scheme for each l-ha plot followed at Poplar treatment ................. 17

Fig.2.3. Sample variograms of the total C for all the plots at poplar along with the model
and overall sample variogram(green solid circle) obtained including data from all the
plots. (a) With all the plots ; (b) With out plot 6 ................................................ 27

F ig.2.4. Sample variograms and variogram models for: (a) original C data (b) residuals
from topographical regression models at Poplar and three agronomic management
systems. Model parameters are shown in Table 3a ............................................. 29

Fig.2.5. Sample variograms and variogram models for: (a) original N data (b) residuals
from topographical regression models at Poplar and three agronomic management
systems. Model parameters are shown in Table 3b ............................................. 33

Fig.2.6. Sample variograms and variogram models for: (a) original C data

(b) residuals fi'om topographical regression models at conventional chisel-plow(CT) and
organic chisel-plow with cover cropping (CT-cover). Model parameters are shown in
Table 4 ................................................................................................ 39

Fig.3.]. Layout of the experimental site along with the locations of the 18 experimental
plots used in the study, the 15x15 m interpolated elevation map ........................... 55

Fig.3.2. Total C and N content at 20- to 30-cm at upperslope and valley positions in CT,
NT and CT-cover. The same letters within each landform positions shows that treatments
are not significantly different from each other (p<0.05) ....................................... 72

F ig.3.3. Total C and N content at O- to 30-cm at upperslope and valley positions in CT,
NT and CT—cover. The same letters within each landform positions shows that treatments
are not significantly different from each other (p<0.05) ........................................ 80

xi

’ 4

F ig.4.1. Comparison of baseline C values obtained from KBS database value with
baseline samples reanalyzed using current procedure for the LTER Main site along with
1:] line .............................................................................................. 112

Fig.4.2. Baseline C (1988) and contemporary(2006) soil C content (0-15 cm) in CT, CT-
cover(organic) and NT treatments at the LTER main site .................................... 118

F ig.4.3. Baseline (1986) and contemporary (2007) soil C content (0-20 cm) in CT, NT
and Never tilled grassland plots (undisturbed) at the long term interaction site
experiment ........................................................................................... 118

Fig.4.4. Decadal pattern of number of days above freezing point at the KBS-LTER
site ................................................................................................... 123

xii

Chapter I
INTRODUCTION

Soil carbon is considered as an index for soil quality and crop productivity and
also plays an important role in the global carbon(C) cycle. C content varies across
landscape as a result of complex interactions among different factors including
topographical attributes, inherent soil properties and land use and management practices.
A well replicated long term experiment with known site and management histories and
organized along different management intensity gradient based on soil disturbance,
chemical inputs and cropping intensity will provide greater advantage for understanding
the complex interactions between various factors that control the variability in soil C
distribution in a landscape.

Among the various soil forming factors, topography or local relief is one of the
major factors contributing to the spatial variability of soil carbon (C) and nitrogen (N)
content at the field level. Local relief regulates soil C distribution by means of altering
soil moisture, nutrient transport, leaching and mineralization rate. General topographical
influences on soil C related processes are likely to differ in magnitude under different
tillage and agricultural systems. With the advancement in GIS and GPS technology,
topographical information can now be more precisely obtained and even a small
depression or a micro level undulation in a field plot can be well represented in digital
elevation model (DEM) based on the topographical information. This information
provides greater advantage for exploring the relationship between topography effects on
soil C variability at a field scale. The overall goal of the present study was to study the

effects of topographical features, soil properties and different management practices on

soil C dynamics and to characterize spatial variability pattern in soil carbon distribution
in each management system.

The first objective was to assess the potential of a short rotation woody perennial
(poplar) system in managing soil C as compared to agronomic systems and to
characterize the effects of management practices and topographical features on spatial
variability patterns at each of the studied systems. The magnitude of spatial variability
varies mainly with level of tillage disturbance and cropping intensity. The inclusion of
spatial variability characteristics will improve the precision of the treatment estimate,
while comparing different management systems.

In the present study, to characterize spatial variability patterns about 600 soil
samples from each agronomic treatment and 500 samples from poplar were collected
from the surface soil (0-5 cm) during 2003-2004. Treatments studied were chisel plow
(CT) and no-till (NT) with conventional chemical inputs, chisel-plow organic
management practice with cover crops (CT-cover), and poplar. In addition to surface
depth, C measurements were also obtained from the soil samples collected at below plow
depth (20-30 cm) during 2006 from CT and CT-cover and compared with the spatial
distribution in soil C observed at surface depth (0-5 cm).

Spatial variability for each treatment was characterized using geostatistical
techniques. Sample variograms were computed for the raw C, N data and for the residuals
obtained from the regression models with topographical features using PROC
VARIOGRAM procedure in SAS. Variogram models were fitted using weighted least
square in PROC NLIN with spherical, Gaussian, and exponential models and the model

that provided the smallest mean square error was selected. Overall treatment variance, sill

values, variogram values at 1.5 m distance to sill ratio and spatial correlation range were
used to compare the spatial variability pattern among the treatments.

The second objective was to determine interactions between topographical
features and tillage and management systems on soil C and N and to identify the
landform elements that lead to most rapid and effective C sequestration in each
management system. At a field scale, the magnitude of soil C variability depends on the
tillage practice and is strongly affected by topography and amounts of plant residue
returned to soil. Topography, tillage gradient and C residues inputs may interact in a
number of ways on soil C and the influence of topography will be either subsided or
intensified by different tillage and crop management practices. Though a number of
studies addressed the relationships between topography and soil C and tillage effects on
soil C, still there is only limited information. on how topography and soil C are related
under different tillage practices and different management systems. In this study,
relationships between topography and total C and N levels under a no-till system,
conventional tillage system, and organic based tillage systems were compared in a
replicated long-term field experiment. The study was conducted at the Long Term
Ecological Research (LTER) site, at Kellogg Biological Station, in southwest Michigan.
Treatments used in the study were conventional tillage (chisel plowed) with conventional
chemical inputs (CT), no-till with conventional chemical inputs (NT), and organic chisel
plowed system with a winter leguminous cover crop (CT-cover). Interaction effects
between topographical features and tillage and management systems on soil C and N was
tested first, with wetness index (WI) a compound topographical index as a continuous

variable using analysis of covariance (ANCOVA) and second, using topographical

information as categorical variable by grouping similar class of terrain attributes into
specific landform elements namely, upperslope and valley using discriminant analysis.
Then the relationship between soil C and N with topography in different treatments was
assessed via a two factor analysis of variance with landform elements as a categorical
variable. Landform elements that lead to most rapid and effective C sequestration in each
management system were identified.

The third objective was to examine the net changes (gain or losses) in C contents
under different management systems that have taken place over a period of 15-20 years
from the historic (baseline) C values in Southern Michigan. Various soil properties such
as baseline C levels, soil texture, topographical features and climatic factors that drive the
change in C content were also explored. Dense geo-referenced historic C values (1986-
88) for C measurements were available for both the experimental sites collected prior to
the establishment of the experiment, which served as baseline C in this study.
Contemporary samples collected during 2006-2007 were compared with historic C values
(1986-88) to assess the absolute changes, such as losses or gains in different management
systems.

Relationships between the changes in soil C and baseline values were conducted
with accounting for the regression to the mean phenomenon (RTM). Changes in soil C
and potential differences in relationships between the changes in soil C and the baseline
C values under different management systems were studied using analysis of covariance
(ANCOVA) with baseline C as a covariate to control the variability in baseline C status

and to improve the precision of treatment estimates.

Chapter II

SPATIAL VARIABILITY OF SOIL C, N CONTENT IN POPLAR AND IN
DIFFERENT ANNUAL CROPPING SYSTEMS IN A LONG TERM
EXPERIMENT

ABSTRACT

Lack of accounting for the spatial variability in soil carbon distribution
may cause uncertainty in treatment estimates of soil C gains or losses. In the present
work, the relative efficacy of a poplar system in managing soil C was compared to
agronomic systems under row crop agriculture and characterized the spatial variability
pattern at all the studied management systems. The objectives of the study were to: (i)
compare the surface total C content (0-5 cm) of a 15 year old poplar plantation with
three commonly practiced annual cropping systems, chisel plow (CT), no-till (NT) with
conventional chemical inputs, and a chisel plow organic system with a winter leguminous
cover crop (CT-cover); (ii) characterize the effects of management practices and
topographical features on spatial variability patterns in soil C with the poplar system and
compare its variogram characteristics with those of the agronomic treatments; and (iii)
compare the spatial variability patterns in C observed at the soil surface (0-5 cm) and
below plow depth (20-30 cm) in CT and CT-cover. The study was conducted at the Long
Term Ecological Research site at the Kellogg Biological Station in southwest Michigan.
Total C measurements were obtained from the soil samples collected at 0-5 cm depth

during 2003-2004 and at 20-30 cm from CT and CT-cover during 2006. Total soil C

under poplar management (11.8 g kg '1 soil) was significantly greater than that in CT (7.4

g kg '1 soil) and CT-cover (10.2 g kg '1 soil) however the difference with NT (11.4 g kg '1

soil) was not statistically significant (p<0.05). The mean N values for poplar (1.1 g kg '1
soil) was significantly greater than that of CT (0.67 g kg '1 soil) and CT—cover (0.93 g

kg '1 soil), but not of NT treatment (1.01 g kg '1 soil) (p<0.05). At the soil surface depth

(0-5 cm), overall total C, N variability was found to be closely similar to that of NT
followed by CT-cover system and substantially different from that of CT.

Spatial variability for each treatment was characterized using geostatistical
techniques. In poplar, variogram values of C at 1.5 m distance to sill ratios were greater
compared to agronomic treatments which show that C is highly variable at very shorter
distance, compared to other agronomic systems. Variability characteristics of the total
soil C in the studied plots were strongly affected by topographical gradients except in the
poplar system.

Comparing the variability between surface (0-5 cm) and below plowing
(20-30 cm) depth, with CT, spatial correlation range, nugget and sill values at surface
depth is substantially greater compared to below plowing depth indicating greater
variability in the surface depth compared to below plowing depth. However with CT-
cover, variability in C distribution pattern was found to be greater even at below plow
depth. Topographical features have a greater effect on controlling the spatial correlation
with CT-cover system compared to CT at both studied depths.

In summary, poplar system did not differ substantially with the annual crop no-till
system in terms of soil C, N content and in overall C and N variability pattern in a

landscape.

INTRODUCTION

Currently the Green House Gas (GHG) reduction initiative has taken a new
paradigm of research by testing the feasibility of different crops in energy production as
sources for biofuel feedstock. With the growing energy crisis and increased awareness of
global warming issues, the spectrum of bioenergy crops may be expanding in the near
future. Bioenergy crops have the greatest potential to supply a major portion of US and
global energy needs while reducing the rate of enrichment of atmospheric C02 (Lemus
and La] 2005). They projected about 60 million hectares (Mha) of land are available in
US to grow bioenergy crops, which can sequester approximately 318 Million Metric tons
or Tg C per year. In this context, growing poplar (Populus Sps.) is considered to be one
of the conservative land management alternatives that have good potential in enhancing
soil quality in marginal lands and mitigating C02 emissions. At the same time, poplar is
considered to be one of the potential bioenergy crops suitable for North Central region.
The US. Great Lakes Bioenergy Research Center (GLBRC) included poplar systems as
one of the novel production system for biofuel production along with various agronomic
systems.

Accurate quantification of total soil C is needed to assess the sequestration
potential of poplar along with that of other agronomic systems. Currently only a few
studies reported the relative efficacy of carbon sequestration potential of a poplar
plantation compared with conventional agricultural system (Coleman et al., 2004;

Degryze et al., 2004; Sartori et al., 2006, Grandy and Robertson, 2007). Poplar system of

12- to 18- year-old sequestered 1.6 Mg C ha '1 yr '1 more C than the adjacent agronomic

treatments based on 11 sites across North Dakota, Minnesota, Iowa and Wisconsin
(Sartori et al., 2006). In contrast, at North central region on compiling data from 27 study
sites, Coleman et al. (2004) reported SOC from poplar fields did not differ significantly
from that of agricultural crops except when poplars are placed in marginal soils.
Accuracy in estimating C content in poplar systems is greatly hindered by the larger
spatial variability in soil C content as compared to agronomic systems. Robertson et al.
(1997) mentioned that spatial variability in soil properties in cultivated soils varied
substantially from the uncultivated sites both in magnitude and spatial scale. Stoyan et al.

(2000) reported that plant root and plant residue patterns are the main factors that control
soil heterogeneity at micro scales of about 2 m 2. Based on the study conducted in the

same site, Kravchenko et al. (2006) characterized the spatial variability of soil C in
agronomic treatments and demonstrated that spatial correlation was strongest in no-till
(NT) followed by cover crop based organic tillage (CT-cover) and conventional till (CT).
The authors found that lack of soil mixing in NT and greater variability in biomass inputs
with cover crop based organic tillage systems contribute to the increase in spatial
variability in both treatments. Even though the spatial variability of soil C in individual
fields and in different tillage and management system has been assessed in a number of
studies (Robertson et al., 1993,1997, Bergstorm et al., 2001; Kravchenko et al., 2006) the
information about how the spatial variability of soil C in a perennial system differs from
the soils under agricultural treatments is largely unknown. Gathering such information
will be useful in improving the precision of the C estimate while comparing poplar
system with other agronomic management system and for adopting an efficient sampling

strategy.

Poplar systems differ from agronomic treatments in a number of ways. Poplar
receives no soil disturbance compared even with no till system, where the soil is
subjected to minimum soil disturbance during planting, harvest and herbicide application.
Elimination of soil mixing with NT was reported to increase C variability in the topsoil C
distribution (Paustian et al., 2000; Kravchenko et al., 2006). Quantity and quality of
above ground and belowground biomass produced by poplar system was found to be

much different from agronomic crops. Cook and Beyea (2000) reported that poplar

produced 8.0 Mg C ha '1 yr '1 of above ground biomass as compared to 5.4 Mg C ha '1

yr 4 produced by com (Zea mays). In poplar, the trees are planted up to 4 meters apart as

compared to closely and unifome planted corn or soybean plants in agronomic
treatments, which might also substantially influence the spatial distribution of C inputs
within the plot. In poplars, the soil surface close to the main tree stem will have a dense
network of roots which might receive more C inputs from the root exudates and dead
roots compared to the soil surface farther from the trunk. Aboveground C inputs from the
leaf litter in poplar will be generally clustered in small pockets within the experimental
plot distributed based on the preferential wind directions, as compared to relatively
evenly spread of residues on the surface in NT or in conventional tillage systems. This
heterogeneity in C inputs will influence the mineralization and C turnover ratio within the
landscape/plot which in turn may result in greater variability of soil C in poplar as
compared to agronomic systems.

Considering lack of soil disturbance and larger spatial heterogeneity in
distribution of both aboveground and belowground C inputs, I hypothesize that overall

variability and spatial dependence in soil C distribution in poplar system is greater than

that in agronomic systems including NT and organic based conventionally tilled
treatment with cover crops. To test this hypothesis, the spatial variability patterns in C, N
content (0-5 cm) in the poplar system was compared with 3 agronomic treatments in a
replicated long-term (15 years) experiment at Kellogg Biological Station Long Term
Ecological Research (LTER) site.

The variability in soil C distribution also expected to be different between the soil
and deeper soil horizons. Many long term experiments reported substantial decrement in

C content from surface to subsurface depth (Yang and Wander, 1999; Dou et al., 2006).

ma?“- -: I

The stratification of C content at different depths is mainly influenced by differences in
tillage and cropping intensity. In this study, in addition to the soil surface, the spatial
variability pattern at 20—30 cm depth was also examined in CT and CT-cover.
Considering the lack of soil disturbance and lesser fluctuations in soil moisture and
temperature at below plow depth (20-30 cm) compared to the soil surface, I hypothesize
that overall variability and spatial dependence in soil C distribution will be much less at
the subsurface depth compared to the soil surface, and the variability pattern at organic
based conventionally tilled treatment with cover crops must be greater than the
conventional tillage system. To test this hypothesis, the spatial variability in C observed
at the soil surface (0-5 cm) was compared with the spatial variability in C observed at

below plowing depth (20-30 cm) with CT and CT-cover.

10

The objectives of this study were 2
1. To compare changes in soil C at 0-5 cm depth in a 15-year second ratoon poplar
system with three agronomic treatments, namely:

a. chisel-plowed with conventional chemical inputs (CT),

b. no-till with conventional chemical inputs (NT), and

c. organic chisel plowed system with a winter leguminous cover crop and zero
chemical inputs (CT-cover).
2. To characterize spatial variability patterns in soil carbon at 0-5 cm depth under a 15-
year poplar system and compare its variogram characteristics with those of the agronomic
treatments.
3. To characterize spatial variability patterns in soil carbon at 20—30 cm depth under
organic conventional system with cover crops (CT-cover) and to compare it with chisel-

plowed conventional chemical inputs (CT).

MATERIALS AND METHODS
Test Site

The study was conducted at the Long Term Ecological Research (LTER) site, at
Kellogg Biological Station (KBS), in southwest Michigan (85°24' W, 42°24' N)
established in 1988. Soils at this site belong to Typic Hapludalfs of the Kalamazoo (fine-
loamy, mixed, mesic) and Oshtemo (coarse-loamy, mixed, mesic) series (Mokma and
Doolittle, 1993). The site receives approximately 90 cm annual precipitation, with about

half as snow and the mean annual temperature is 9°C (Grandy and Robertson, 2007).

ll

Experimental Layout and Agronomic Protocols

The LTER experimental layout is a one-factor randomized complete block design
with six replications and seven treatments organized along a management intensity
gradient based on tillage intensity, external inputs and net primary productivity (Grandy
and Robertson, 2007). The experiment was started at 1989, prior to that the studied site
had been conventionally managed in row-crop agriculture for at least a century
(Robertson et al., 1997). Treatments relevant to this study include three agronomic annual
systems on corn (Zea mays L.)—soybean [Glycine max (L.) Merr.]—wheat (Triticum
aestivum L.) rotation and one perennial system under poplar trees (Populus x
euramericana c. v Eugenei.) The agronomic treatments studied were chisel-plowed with
conventional chemical inputs (CT); no-till with conventional chemical inputs (NT); and
certified organic chisel plowed system with winter leguminous cover crops and zero
chemical inputs (CT-cover). The poplar system consists of short-rotation poplar clones
which are grown since 1988 on a 10 year rotation cycle. Among the three studied
systems, poplar is subjected to the least soil disturbance and CT-cover management had
the greatest amount of mechanical disturbance. CT-cover in addition to chisel-plowing, is
subjected to soil disturbance due to weed control achieved by weekly-biweekly rotary
cultivations early in the growing season. The cover crops of the CT-cover treatment were
hairy vetch (Vicia villosa Roth) from 1989 to 1993 and red clover (Trifolium pratense L.)
from 1994 onwards. The cover crops are generally incorporated in April or May before
corn or soybean planting. The complete site description, experimental design, and
management protocols are available at the KBS website (Kellogg Biological Station,

2007)

12

Soil Sampling

Soil sampling was carried out during 2003-2006 and consisted of 2 sample sets.
The first set of soil samples was collected at 0-5 cm depth in 2003-04 and the second set
of samples was collected at 20-30 cm depth in 2006. The first set included soil samples at
0-5 cm depth collected from replications 1 through 4 during May of 2003 for CT, NT and
CT-cover; and soil samples collected from replication 5 and 6 from CT, NT and CT-
cover and from all the replications from poplar treatment (except replication 3) in May of
2004. Description of soil sampling during 2003 and 2004 is provided in Table 2.1a.

In poplar treatment soil samples were not collected from replication 3 since the
tree population was not well established due to severe erosion during establishment year.
In both years CT and CT-cover treatments were plowed approximately 2 weeks prior to
sampling. The sampling scheme followed at the agronomic treatments is shown on Fig.
2.1 (Kravchenko et al., 2006). Each experimental plot is about 80 by 100 m in size. To
avoid plot boundary effects and the long-term protected portions of the plots restricted to
sampling, the sample collection has been limited to the center 60- by 60-m portion in
each experimental plot. Approximately 100 georeferenced soil samples were collected
from each plot. Among the 100 sampling locations within each plot, 20 samples were
collected on a 15- by 18-m triangular grid and the remaining 80 samples were collected at
distances 1.0, 2.5, 5.0, and 7.5 m from the grid points. At each sampling point, the sample
was taken between the plant rows and was composited from five 2.5-cm-diameter cores

collected within a 0.2-m radius.

13

Table 2.1a. Description of soil sampling data collected during 2003 and 2004

 

 

 

 

Time of sampling
Treatment May-2003 Maj-2004 Depth N
CT Rep 1, 2, 3, 4 5, 6 0-5 cm 600
NT Rep 1, 2, 3, 4 5, 6 0-5 cm 600
CT cover Rep 1, 2, 3, 4 5, 6 0-5 cm 600
Poplar Rep 1, 2, 3, 4, 5, 6 0-5 cm 500
Total 2300

 

CT - Conventional chisel-plow; NT-no-till; CT-cover —Organic chisel plow with cover cropping.
Poplar — Poplar trees of 15 years old; N - Number of soil samples collected in each treatment

Table 2.1b. Description of soil sampling data collected at 20-30 cm depth during
May 2006

 

 

Treatment Replication sampled Depth N

CT Rep 1, 4 20 - 30 cm 200
CT cover Rep 3, 4 20 - 30 cm 200
Total 400

 

N - Number of soil samples collected in each treatment

14

 

 

 

 

 

80 -
’° 1 .- ..
' o O. o o - fie o O.
60 a "
E O : . O
:E .0 o o? o .0 ’ o
O e
Z ' o
40 - . ' .
.0 o ‘ o a +1.: 0 .0
20 _ o. 5 . .- .. .
20 4O 60 80
0 Grid Points EaSt’m
0 1m Points
- 2.5m Points
0 5m Points
0 7.5m Points

 

 

 

Fig. 2.1. Sampling scheme for each l-ha plot followed at CT, NT and CT-cover

treatments. Images in this dissertation are presented in color.

15

The above sampling scheme provided a sufficient number of samples for more
accurate spatial variability characterization with minimal sampling effort (Kravchenko et
al., 2006). In CT replication-3 plot topography was represented by a deep bowl-shaped
depression and distinctly different from the any of the plots in the other treatments in the
rest of the LTER site. This distinctly different shape led the data from this plot to differ
notably from the rest of the site both in terms of the total C and N values, spatial patterns,
and relationships with topography. Since it could not be meaningfully compared with the
other treatments the data from this plot were excluded from further analyses.

The sampling layout for poplar was slightly modified from that used in agronomic

 

treatments by taking into account the spatial arrangement of trees and spacing between
them. Similar to agronomic treatments, within each plot 20 samples were collected on a
15- by l8-m triangular grid and an additional 80 off-grid points were collected at four
distances l-m, 2.5-m, 5.0-m and 7.5-m from several randomly selected grid points. The
sampling scheme followed with the poplar treatment is shown on Fig. 2.2

At the agronomic treatments the off-grid points were randomly selected from
eight directions such as north, north-east and east etc. In poplar instead of eight
directions, sampling was restricted to four directions namely North, South, East and West
such as to be aligned with rows of poplar trees. To reduce the small scale variations due
to tree presence and to maintain the same distance from a main trunk, the samples were
taken from the middle of the row. The second set of soil samples was collected at 20-30
cm depth in May of 2006 from replications 1 and 4 for CT and replication 3 and 4 for

CT-cover (Table 2.1b).

l6

 

 

 

 

 

 

 

60 L
I l
I u M u u l
50 -f . l
l - 1
l l
40 _I I I I I I ‘ .
, _ _ . . _ l . Bigredpopl
E t _ ' I . 1m point popl
g 30 I A“ I I M I I M I m I 2m pair“
2 I ' l - 4m points
[ I I am DOInIS
20 "I, ' I I I I I :
I . l
', on m I u 105- n I m I
10 «1 l
l _ ' l
j 0‘ A. CA . OI - l
o ~—~~— -— __, —— ———.-~————— ________ ‘
0 20 4O 6O 80
East,m

 

 

 

Fig. 2.2. Sampling scheme for each l-ha plot followed at poplar treatment. Images in this

dissertation are presented in color.

17

These plots were selected for 20-30 cm sampling because of their relative
proximity to each other and relative topographical diversity of that portion of the LTER
site. At each plot approximately 100 samples were collected from the same locations as
the 0-5 cm samples of 2003. Among the 100 samples within each plot, 20 undisturbed
cores of 5.2 cm-diameter were obtained using hydraulic soil core unit from Geoprobe
systems (Geoprobe Inc. Salina, Kansas) to a depth of 0-40 cm on a 15- by 18-m
triangular grid. The samples were air dried at room temperature and cleaned from all
plant residue and stones. The samples were air dried at room temperature and cleaned
from all plant residues and stones. Proper care was taken to exclude coarse roots, twigs,
seeds and other visible plant residues from each sample. Smaller plant material was
removed by gentle air blowing and the samples were ground on a shatterbox machine for
one minute to pass a 250-um sieve. Total C was measured using an automatic Carlo-Erba
CN analyzer (Carlo Erba Instruments, Milan, Italy).

In each plot, about 10% of the samples were replicated thrice for total C
measurements to account for the variability associated with laboratory measurement
error. Previous studies conducted at LTER site reported that inorganic C at the soil
surface was found to be negligible (DeGryze et al., 2002) thus the values obtained from

chemical analysis of total C were considered to be representative of the soil organic C.
Total C and N contents at 0-5 and 20-30 cm depths were reported in g kg '1 of soil which

was obtained by multiplying percent of C by 10.

C (g C kg '1 of soil) = C percent (g C 100g '1 of soil) * 10

18

Terrain Attributes

Topographical survey for measuring elevation data was conducted at the LTER
site twice. The first survey was conducted using a land-based laser in 1988 generating
597 elevation measurements (Robertson et al., 1997) and second survey was carried out
in 2004 using 12-channel Leica SR530 real-time kinematic DGPS (dual-frequency global
positioning system) receiver within measurements collected every 2 m in rows 5 m apart
(Kravchenko et al, 2006) at all the treatments. The elevation data from both surveys were
merged and converted into a cell-based terrain map on a 4 by 4 m grid using inverse
distance weighting with power of two and six nearest neighbors in ArcGIS 9.0 Spatial
Analyst (ESRI, 2004). Then topographical features such as terrain slope, aspect,
curvature, and flow accumulation were derived from the elevation map using surface and
hydrology functions of ArcGIS 9.0 Spatial Analyst. Relative elevation (m) for each cell
was obtained by calculating the difference in height of specific cell of interest from the
minimum elevation in each plot. A more detailed description of terrain attributes and

scale of measurements were provided by Kravchenko and Bullock (2000).

The wetness index (WI) was computed based on flow accumulation and slope

data for each map grid cell (Gallant and Wilson, 2000) as:

WI = ln(AS /tan ,8) [1]

where A S is specific catchment area derived from flow accumulation, and tan 6 is a

tangent function of slope at percent-rise.

19

Data Analysis
Data analysis was conducted using SAS (SAS Institute, 2001). First, comparison

of total C, N variances was tested using Levene’s test (Milleken and Johnson, 1992).
Estimation of different variance components due to measurement error, plot and block
effects was carried out in PROC MIXED using the restricted maximum likelihood
method (SAS Institute, 2001).

Second, to evaluate the variability imposed by topography, multiple regression
analysis was carried out between soil C and N with the terrain attributes in each treatment
separately using PROC REG procedures (SAS Institute, 2001). The multiple regression
models initially included either total C or total N as the response variable and terrain
attributes such as elevation, slope, curvature, flow accumulation, aspect, and W1 as
independent variables. The need for quadratic terms for elevation, slope, curvature, flow
accumulation, aspect, and W1 were checked separately in each plot and the quadratic
terms that were found to be significant (p < 0.05) were added to the regression equation.
The residuals obtained from the topographical regression models are the variability in
total C after controlling the topography effect on soil C changes.

Third, geostatistical analysis was carried out to evaluate the spatial variability in
the raw C, N data and the spatial components of the variability that remained after
controlling the topographical effects. Here the spatial structure obtained using raw C, N
depicts the topography and management practices induced changes in variability pattern
in C and N; whereas spatial variability observed using residuals obtained from the

regression models with topography is the variability in soil C, N afier removing the

20

topographical effect. Sample variograms were computed for the C, N data and for the
residuals obtained from the regression models with topography using

PROC VARIOGRAM procedure in SAS. Each sample variogram was computed with
lag distance of 4 meters and 15 distance classes. Variogram models were fitted using
weighted least square in PROC NLIN (Schabenberger and Pierce, 2002) with spherical,
Gaussian, and exponential models and the model that provided smallest mean square
error was selected. Sill values, variogram values at 1.5 m distance to sill ratio and spatial
correlation range were used to compare the spatial variability pattern among the

treatments.
RESULTS AND DISCUSSION

Total C (0-5 cm) at 15-year old poplar compared to three agronomic treatments

Total soil C accumulated by poplar (1 1.8 g/kg) was significantly greater than that
in CT (7.4 g/kg) and CT-cover (10.2 g/kg) under annual crop rotation. However the
difference with NT treatment (11.4 g/kg) was not statistically significant (p<0.05) (Table
2.2). The mean N value for poplar (1.1 g/kg) was significantly greater than that of CT
(0.67 g/kg) and CT-cover (0.93 g/kg), but not of the NT treatment (1.01 g/kg) (p<0.05).

These results were consistent with other studies who reported greater soil C and N
levels with poplar compared with conventional tillage treatments (Sartori et al., 2006).
Currently literature comparing poplar with NT is limited. However two other studies
conducted in the same site also reported no significant difference between poplar and NT
when the poplar trees were 10 to 12- year-old (Robertson et al., 2000; Degrye et al.,
2004; Grandy and Robertson, 2007). The lack of significant differences between poplar

and NT crop systems might be attributed to the fact that both systems were subjected to

21

Table 2.2. Treatment mean estimate of total C, N (g C kg -1 soil) in CT, NT, CT cover
and Poplar treatments of LTER site (standard error in paranthesis).

 

 

TRT Total C (8.133) Total N (SE)
CT 7.4 (0.4) a" 0.7 (0.06) a*
NT 11.4 (0.5) c 1.0 (0.06) bc
CT cover 10.2 (0.5) b 0.9 (0.06) b
Poplar # 11.8 (0.6) c 1.1 (0.08) c
Poplar # # (with Rep 6) 12.3 (0.6) 1.2 (0.07)

 

CT - Conventional chisel-plow; NT-no-till; CT-cover —Organic chisel plow with cover cropping.
Poplar - Poplar trees of 15 years old

"‘ Means within the same column followed by the same letter are not significantly different
(a=0.05)

# - Means estimates of poplar calculated based on the data from plot], 2, 4 and 5

# # - Means estimates of poplar calculated based on the data from plot], 2, 4, 5 and 6

22

none or very small soil disturbance. In addition, 15 years may be still be not sufficient
time for substantial differences in soil C to develop between afforested system and
agronomic treatments under temperate conditions. Degryze et al. (2004) hypothesized
that 10-40 years may be required to find any significant difference in poplar system
compared to agronomic treatments.

These results were consistent with other studies who reported greater soil C and N
levels with poplar compared with conventional tillage treatments (Sartori et al., 2006).
Currently literature comparing poplar with NT is limited. However two other studies
conducted in the same site also reported no significant difference between poplar and NT
when the poplar trees were 10 to 12- year-old (Robertson et al., 2000; Degrye et al.,
2004; Grandy and Robertson, 2007). The lack of significant differences between poplar
and NT crop systems might be attributed to the fact that both systems were subjected to
none or very small soil disturbance. In addition, 15 years may be still be not sufficient
time for substantial differences in soil C to develop between afforested system and
agronomic treatments under temperate conditions.

Degryze et al. (2004) hypothesized that 10-40 years may be required to find any
significant difference in poplar system compared to agronomic treatments. Overall total
C, N variability observed with poplar was found to be very similar to that of NT and
cover crop system and significantly greater to that of CT treatments (Table 2.3a, 2.3b).
Total soil C variances at poplar (0.5 g/kg) was significantly greater than that in CT (0.1
g/kg) however the difference with CT-cover (0.4 g/kg) and NT (0.5 g/kg) was not

statistically significant (Table 2.3a; p<0.05).

23

Table 2.3. Variances and spatial variability parameters for the (a) total C (b) total N and
the residuals of the regression model with topography under poplar and three different
agronomic management systems (0-5 cm depth)

 

 

 

 

 

a)
Residuals from topographical
Original Total C regression models
Variog Varl.5 Variog Var1.5
TRT Var“ model m Sill Range model m Sill Rang:
CT 0.1a Sph 0.05 0.1 l 20 Sph 0.05 0.1 17
NT 0.5b Gau 0.18 0.69 30 Gau 0.2 0.55 26
CT cover 0.4ab Gau 0.09 0.52 38 Gau 0.1 0.26 31
Poplar # 0.5b Sph 0.31 0.6 35 Sph 0.31 0.6 30
Poplar # # 0.8 Gau 0.28 1.04 32 Gau 0.3 0.9 23
b)
Residuals from topographical
Original Total N regression models
Variog Var1.5 Variog Var1.5
TRT Var“ model m Sill Rarrge model m Sill Range
CT 0.001a Sph 0.0008 0 16 Sph 0.0008 0 12
NT 0.005ab Gau 0.0014 0.01 25 Gau 0.002 0.01 26
CT cover 0.003a Gau 0.001 I 0 42 Gau 0.001 0 32
Poplar # 0.006b Sph 0.0039 0.01 32 Sph 0.004 0.01 28
Poplar # # 0.008 Sph 0.0041 0 28 Gau 0.004 0.01 20

 

CT - Conventional chisel-plow; NT-no-till; CT cover —Organic chisel plow with cover cropping.
Poplar — Poplar trees of 15 years old

Variog- Variogram Models: sph - spherical; Gau -— Gaussian

* Var - Variances followed by the same letter are not significantly different (p< 0.1, Levene’s

test)

# - Means estimates of poplar calculated based on the data from plotl , 2, 4 and 5
# # - Means estimates of poplar calculated based on the data from plotl, 2, 4, 5 and 6

24

The variances of total N decreased in the order of poplar~NT~ CT-Cover >CT
(Table 2.3b). I hypothesized that overall variance at poplar will be greater than
agronomic treatments considering the larger heterogeneity in C inputs and no tillage
disturbance, but surprisingly the variance with poplar was found to be very similar to that
of the no-till crop system. In the present study, the soil disturbance gradient follows the
order of CT-Cover> CT>NT> Poplar. The results show that total C variability increased
with lack of soil mixing due to reduction in disturbance level during the past 15-yr period
after conversion to poplar and NT. However, CT-cover treatment which is subjected to
heavy soil disturbance in this study site (more soil mixing) than CT still showed higher
variability than the CT. With the CT-cover system the amount of residue returned from
the cover crop and main crop was found to vary over the landscape, which in turn results
in greater variability in soil C content. This indicates that apart from the soil disturbance
gradient, greater variability in the biomass inputs also increases the variability pattern in
soil C.

The measurement error variance component for both C and N at poplar was found
to be significantly higher over all agronomic crop system (p<0.05). Total C variance due
to laboratory measurements constituted 0.02, 0.1, 0.05 and 0.2 at CT, NT, CT-cover and
poplar respectively. Similarly for N, variance component due to laboratory measurements
constituted 0.0003, 0.002, 0.002 and 0.02 at CT, NT, CT-cover and poplar respectively.
Here the variance observed with poplar was found to be at least twice that of NT and CT-
cover treatment. Possibly because even after removing the smaller plant material by

gentle air blowing, there were still very finer pieces of organic material abundant in

25

poplar soil samples compared to agronomic treatments, which might have resulted in

greater within sample variance.

Spatial variability pattern in poplar plots

A sample variogram of total C for all the individual plots sampled at poplar is
shown in the Fig. 2.3a . The model and sample variogram was obtained by pooling data
from all the plots and is depicted by solid circles. The sample variogram from the plot 6
is found to be distinctly different from the other plots in poplar as well as with overall
sample variogram (Fig. 2.3a). The sample variogram from plot 6 also differed notably
from rest of the plots in other treatments. Further the overall C variance from plot 6(0.18)
was also found to be much higher compared to plot 1(0.06), plot 2(0.06), plot 4(0.02) and
plot 5(0.06). The model and sample variograms obtained combining data from all the
plots except plot 6 is found to be well representative of sample variogram of the
individual plots (Fig. 2.3b).

Since the spatial pattern observed at plot 6 is found to be totally dissimilar with
other poplar plots, for more meaningful comparison the data from this plot 6 were
excluded while comparing with other treatments. However the results obtained including

plot 6 is also presented but not discussed.

26

(a)

 

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Rep1
Po Iar-With all lots ‘
0.30 p p x Rep2
O O
0.25~ ° A Rep4
o o x Rep5
E 0.20« o o Rep6
g, 0.15 4 o 0 Poplar
E 0 ———Model
> 0.104 0 9 f x
i
005 taflAIIfifit
. '1‘
A A A A A A A a A 6
0.00 a . - , , ,
0 10 20 30 40 50 60 7o
Lag distance,m
0))
Poplar -With out plot6 ° Rem
0.20 x Rep2
A Rep4
016 4 x Rep5
E I Poplar
g 0.12 —Model
2 X if
' 0.08 « X
g o x 3 § §< X .
o O o
0.04 ~
X A A
§§ X A A A A A A A A A A A
0.00 r T r T . .
0 10 20 30 40 5O 60 70
Lag distance, m

 

 

 

Fig. 2.3. Sample variograms of the total C for all the plots at poplar along with the model
and overall sample variogram (green solid circle) obtained including data from all the

plots. (a) With all the plots ; (b) With out plot 6

27

Spatial variability patterns in soil C, N under a lS-year poplar and three agronomic
treatments at 0-5 cm

Stronger spatial structure was observed in poplar, NT and CT-cover which is
clearly evident from the variograms of original C and N data with all treatments except
CT (Fig. 2.4a, Fig. 2.4b). Similar to soil C and N levels and overall variability, spatial
variability parameters such as nugget and sill value of C and N with poplar system were
slightly greater or equal to that of NT and CT-cover (Table 2.3a, 2.3b). Greater
variogram values near the origin with poplar shows that C is highly variable even at very
shorter distances compared to agronomic treatments. In this study, to explain the spatial
dependence, the ratio obtained by variogram values at 1.5 meters and sill values were
used as outlined by Kravchenko et al., (2006). This ratio represents the amount of an
unexplainable component in the dataset. The smaller the ratio, the greater the spatial
dependence over the range of separation distances modeled.

In this study, the ratio between variogram values at the smallest lag distance (1.5
m) to sill values with poplar was found to be 52 percent whereas at CT, NT and CT-
cover were 42, 26 and 18 percent respectively. This trend shows that a larger portion of
the variability with poplar is occurring in shorter distance compared to NT and CT-
cover. With NT and CT-cover, a larger portion of the overall variability is more spatially
structured. The laboratory measurement error constituted 63 percent of variance
observed at smallest lag distance of 1.5 m at poplar compared to 45 and 50 percent at NT

and CT-cover respectively (Table 2.3a).

28

a)

b)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 CT
0.08 I Q NT
0.07 . x CT cover
E 0.06« A Poplar
g 0.05. I ——Model
8' l
t 0.04.
N
> o.03~ I
0.02 4 l
0.01 - i
l
0.00 . . . . . . l
0 10 20 30 40 50 60 70
Distance, m
. o CT
TC Resrduals
0.08- 0 NT
0.07 . x CT cover
Po ar
0.06- A pl
——Mode|
E 0.05-
g, 0.04 ~
5
> 0.03-
0.02«
0.01 - 0
0.00 . . . . . .
o 10 20 30 4o 50 60 70
Distance, m

 

Fig. 2.4. Sample variograms and variogram models for: (a) original soil C data (b)

residuals from topographical regression models at poplar and three agronomic

management systems. Model parameters are shown in Table 2.3a.

29

 

 

In the present study, topographical features explained a substantial portion of the
variability in total C and N distribution in all the studied treatments except poplar. The R2
value for the topographical regression model for total C ranged from 0.07 (NT-Rep6 and
Poplar-Rep1) to 0.73 (CT-cover-Repl) (Table 2.4a). Similarly for total N the R2 ranged
from 0.08 (NT-Rep6 and poplar-Rep1) to 0.79 (CT-cover-Repl) (Table 2.4a).

Combining data from all the plots in each treatment, topography was found to
explain 26, 18, 38 and 7 percent of the variability in soil C at CT, NT, CT-cover and
poplar respectively. The results shows that the relationship between topography and soil
C was stronger at organic based system (CT-cover) and weaker at poplar and no till
system. Similarly for total N, topography was found to explain 24, 16, 33 and 10 percent
of the variability with CT, NT, CT—cover and poplar respectively. Stronger relationships
between tepography and soil C observed in this study was consistent with the results
reported in other studies conducted at the comparable scale (Kravchenko and Bullock
2000; Muller and Pierce 2003; Terra et al., 2004). Compared to agronomic treatments,
the topographical effect was less pronounced in poplar due to following reasons.
Topography affects soil carbon mainly in two different ways. It influences erosion and
redistribution of soil particles and it drives vertical and horizontal water redistribution
patterns within the landscape (Creed et al., 2002). With poplar, the general topography
soil C relationship is greatly subsided because the distribution of soil particles within the

plot is greatly restricted due to dense plant cover and less soil disturbance.

30

.96. no... «a 8.855% on 8 6.58 :2...» £0.68 05 8 .62... 203 3:8 05.3.2.0 .8on cam .82.. $0503

. no.3...asooa30c wo— .o..3a>..=o 6qu 59:8 @022 038.2 .«o 38...... 2.2.: vows—o... 3o... .3 .«o 22.2.. 5.3022 18.....an H H

2.... w A a 2.82%.... 82 : 2. .95. 2.382.. 8... 2.. 3 2823.5 : .25. 2.338.. 2.... 2.. a 502...»: ..

 

 

:2... :3... .3 SN .3 829. 82...: A. 2.3.
:2... :8... 2... 8.. 2... .2... .22.... 2. 2.3.
:2... :2... S... 2.. .5 6...... .23.: a. 2.3.
:2... :2... 2.... 2... 2.. 82... 92.2 2. 2.3.
2...... ”22.... .m... 2... SN 82.... $2.... a. 33.
.295

:2... 2...... 8... a... 3... .25. .23... .3 2.3.
:3... :2... A: 2.... 3.... 2...... 8:2: 3 an...
:2... :2... a... 2... on... .25.. .22.... .3 we...
:2... :8... G... 2.. a... .23.. .52.. 8 2.3.
:2... :2... 2.... an... 2... .82. .2... 8 2.3.
:2... :2... 3.. n. 8.. 2.2... $2.... .3 as.
€969 HOV ”5.398 .960 5.3 to... .02.... 2.390

:2... :2... 2.. t... E... .25. 2...... 3 2.3.
.8... wit... 2... 3 2 .5»... .22.... o... 2.3.
:3... :2... 2... 2.... 2.... 8...... 3292 c... 2.3.
:3... :3... 2... 2... .5... 82.... 6...... 8. as.
:8... :2... a... z... o... 82... .23: 8 2.3.
:2... :2... e: 2.... 2... .3... .43.: 2. as.
92......2

.8... :2... 2... 2... 2... .23.. 82.... .3 83.
:2... :2... a... A... 3.. .25. 8...... 8. 2.3.
.8... :2... Rs 2.... 2.... ..2...... 2...... 3 1.3.
:8... :2... 8.. a... :.. 32... .592 a 33.
:2... :2... 2.. we... 2.... .22.... .2... a 2.3.
33... :32... .3... ..... 2... .25... .2. 3 3 :5.
Z... 0... E. to... .839 .anoceoEoU
”a”. “a”. .3 38.. 2.2m 3.3. .29 .3 z E...

 

232.258.. 5 :32... 0.8 3... 5:22... .«o 35.0508 .7. .09... O .8... .852... on? Sinus.» 303.3325 5.3 5%.. Bo m o.
-o 0.: 2. Z .0 8.. 5.3032 2.53.: mo 8.28% do... 3.me 0... .0 8.9.2.8.. 30293028 US. so... mo magnum a: 63¢...

31

From the regression results, terrain attributes were found to explain a substantial
portion of over all variability in the C and N content in all the treatments. To assess the
spatial component of variability remaining after controlling the topography effect
variogram of the residuals for total C and N were calculated from the respective multiple
regression topographical models. The sill values and spatial correlation ranges obtained
from the residual variograms were decreased and the 1.5 m variogram value/sill ratio
increased marginally compared to the variogram models of original C, N data at all the
studied tillage and cropping treatments except poplar. The 1.5 m variogram value/sill
ratio constituted 50, 36, 39 and 56 of the overall variability of the residuals for CT, NT,
CT-cover and poplar treatments respectively. With CT-cover, more substantial increase
in 1.5 m value/ sill ratio with residual variograms shows stronger influence of
topography on total C on these treatments. For example, Kravchenko et al. (2005)
reported stronger relationships between topography and crop yield in the CT-cover
system compared to CT and NT. Similarly Munos et al., (2008) observed strong
relationship between cover crop biomass yields with elevation data collected in the same
site. After removing the topography effect, the spatial variability structure greatly
dropped with CT-cover, which depicts that the spatial structure in C and N distribution at
CT-cover is mainly controlled by topographical gradient (Fig. 2.4b; 2.5b).

With poplar, there is no substantial change in variogram parameters because
topography has less impact on soil C distribution in poplar compared to other agronomic
treatments. Comparison of variograms based on the raw C data, the variogram values of

NT were found to be similar or greater than poplar at most of the lag distances.

32

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TN 0 CT
0.0009
0 NT
0.0008 1
0.0007 4 A CT cover
E 0.0006 - 0 Poplar
g 0.0005 - ___Mode|
E 00004 -
0.0003 l
0.0002 -
0.0001 ~
0.0000 . r t . t .
0 10 20 30 40 50 60 70
Distance, m
b)
. 0 CT
TC Resnduals
0.08 0 NT
0.07 « x CT cover
0.06 4 A Poplar
——Mode|
E 0.05 -
E
3’ 0.044
E
> 0.03 ~
0.02 «
0.01 « 0
0.00 I
0 10 20 30 40 50 60 70
Distance, m

 

 

 

Fig. 2.5. Sample variograms and variogram models for: (a) original N data (b) residuals
from topographical regression models at poplar and three agronomic management

systems. Model parameters are shown in Table 2.3b.

33

However after removing the topographical effect, the variogram values with
poplar were found to be higher than NT and CT-cover at all the lag distances (Fig. 2.4b).
Greater C variability with poplar might be attributed due to the following mechanism.
Especially afier land use conversion from conventional tillage to any conservative
management system, the rapid change in the soil C pool will be mainly attributed from
the changes in light fraction and particulate organic matter. Greater variability in soil C
distribution in poplar, NT and CT-cover might be attributed due to the increase in
particulate organic matter (POM) content afier conversion from conventional tillage
system. Poplar systems sequester C mainly at the fine intra aggregate particulate organic
matter (POM) (Degryze et al., 2004). Fine intra aggregate-POM estimated at poplar in
the same study site during 2001 was reported to be greater than NT and CT-cover system
(Grandy and Robertson, 2007).

Similarly many previous studies reported NT (Needelman et al., 1999) and CT-
cover (Marriott and Wander, 2006) hold 30—40 percent more POM than conventional
tillage treatment. POM was found to be more variable than the total C (Burke et al.,
1999). Greater overall variability with poplar and NT is attributed due to a greater

amount of POM and lesser soil disturbance compared to CT-cover.

Spatial variability patterns in soil C, N at surface (0-5 cm) and below plow
(20-30 cm) depth among two management systems

Total C with CT-cover was significantly greater than CT at both the studied
depths (Table 2.5a; p<0.05). Within each treatment the C content at the surface depth

was significantly greater than the subsurface depth.

34

Table. 2.4b Summary of soil and topographical properties of the studied plot. Results of
multiple regression for total C at the 0- to 5 cm and 20-30 cm depth with topographical
variables also shown. For total C coefficients of variation (%) are shown in parentheses.

 

 

 

TC (g C kg " soil) R 2:

TRT N R relief Slope WI 0-5 cm 20-30 cm 0-5 cm 20-30 cm
CT

Rep] 95 1.48 1.41 6.48 6.5(17) 3.0(29) 0.43“ 0.23"

Rep4 97 0.81 0.86 7.27 8.0(11) 3.1(20) 0.12" 0.06NS
CT cover

Rep] 97 0.94 1.17 6.61 10.7(17) 3.7(34) 0.24“ 0.39"

Rep4 98 0.56 0.76 6.81 10.1(15) 4.3(50) 0.70" 0.59"

 

* Significant at the 0.05 probability level. ** Significant at the 0.01 probability level.

NS —— Non significant at P S 0.05

CT - Conventional chisel-plow; CT-cover —Organic chisel plow with cover cropping.
I Topographical regression models of all plots included linear effects of relative relief, terrain
slope, curvature, log flowaccumulation, wetness index and aspect.

35

At both 0-5 and 20-30 cm depth, the overall total C variability at CT cover was
found to be substantially greater than CT (Table 2.5b). At CT, total C variance (0.12) at
the soil surface was twice greater than the subsurface depth (0.05), whereas with CT-
cover, C variance at the surface (0.29) and sub-surface (0.30) depth were found to be
similar. Total C variability was also strongly spatially structured with CT-cover
compared to CT, which is depicted by the variograms of original C data (Fig. 2.6a).

Among the studied depth with CT, variogram values at 0—5 cm depth were
found to be greater than the variogram values from 20-30 cm depth at all the lag
distances (Fig. 2.5a). This depicts that, C is more spatially variable at the soil surface
compared to subsurface depth with CT. With CT-cover, variogram values at surface and
subsurface depth were found to be similar in most of the lag distance which indicates
greater variability in C even below plow depth (Fig. 2.6a). The spatial correlation range
of C at the 20-30 soil depth at CT-cover (27m) was much greater than CT (10m).
Compared to the soil surface (0-5 cm) the spatial correlation range of C at 20-30 cm
depths was found to be relatively shorter with both CT and CT-cover, which depicts that
C values are correlated at larger distances at the surface soil depth compared to below
plowing depth (Table 2.5b). Similar to surface depth, at 20-30 cm, relationships between
topography and total C were much stronger with CT-cover than CT similar to the surface
depth. At CT, topography was found to explain a lesser amount of variability in C at 20-

30 cm depth compared to 0-5 cm depth.

36

However after removing the topographical effect, the variogram values with
poplar were found to be higher than NT and CT-cover at all the lag distances (Fig. 2.4b).
Greater C variability with poplar might be attributed due to the following mechanism.
Especially after land use conversion from conventional tillage to any conservative
management system, the rapid change in the soil C pool will be mainly attributed from
the changes in light fraction and particulate organic matter. Greater variability in soil C
distribution in poplar, NT and CT-cover might be attributed due to the increase in
particulate organic matter (POM) content after conversion from conventional tillage
system. Poplar systems sequester C mainly at the fine intra aggregate particulate organic
matter (POM) (Degryze et al., 2004). Fine intra aggregate-POM estimated at poplar in
the same study site during 2001 was reported to be greater than NT and CT-cover system
(Grandy and Robertson, 2007).

Similarly many previous studies reported NT (Needelman et al., 1999) and CT-
cover (Marriott and Wander, 2006) hold 30-40 percent more POM than conventional
tillage treatment. POM was found to be more variable than the total C (Burke et al.,
1999). Greater overall variability with poplar and NT is attributed due to a greater

amount of POM and lesser soil disturbance compared to CT-cover.

Spatial variability patterns in soil C, N at surface (0-5 cm) and below plow
(20-30 cm) depth among two management systems

Total C with CT-cover was significantly greater than CT at both the studied
depths (Table 2.5a; p<0.05). Within each treatment the C content at the surface depth

was significantly greater than the subsurface depth.

37

Table. 2.5a. Total C (g C kg '1 soil) at surface (0-5 cm) and below plow depth (20-30
cm) at CT and CT-cover treatments.

 

 

TRT 0-5 cm 20-30 cm
CT 7.2 (0.6)a*A"‘* 3.1 (0.1) aB
CT cover 10.5 (0.8)b A 4.2 (0.4) bB

 

* Comparisons of mean effects (Treatment effect within each depth) within the same column
followed by the same lowercase letter are not significantly different (a=0.05)
“Comparisons of mean effects (Depth effect within each treatment) within the same row
followed by the same upper case letter are not significantly different (a=0.05)

Table 2.5b. Variances and spatial variability parameters for the total C data at surface (0-
5 cm) and below plow depth (20-30 cm) and the residuals of the regression model with
topography at CT and CT-cover treatments.

 

Residuals from topographical

 

 

Original Total C regression models
Var Var
Variog at 1.5 Variog at 1.5
TRT Var # model m* Sill Range model m* Sill Range
0-5 cm
CT 013 Sph 0.05 0.11 20 Sph 0.05 0.09 17
CT cover 0.3b Gau 0.12 0.39 45 Gau 0.12 0.21 31
20-30 cm
CT 0.1a Gau 0.02 0.06 10 Gau 0.02 0.05 8
CT cover 0.3b Gau 0.11 0.37 27 Sph 0.11 0.2 17

 

CT - Conventional chisel-plow; CT-cover —Organic chisel plow with cover cropping.

Variog- Variogram Models: Sph - spherical; Gau — Gaussian

* Variogram values at 1.5 m distance

# Variances followed by the same letter are not significantly different (p< 0.05, Levene’s test)

38

However with CT—cover, the amount of variability explained at 20-30 depth was
found to be similar to that of 0-5 depth (Table 2.5b). Topographical features explained
about 41 percent of variability in total C with CT-cover compared to 10 percent with CT
at 20-30 cm depth. The spatial component of variability unexplained by the terrain
attributes was calculated from the variogram of the residuals for total C from the
respective multiple regression topographical models. The variogram values at 1.5m
distance to sill ratio constituted 40 and 55 of the overall variability of the residuals for
CT and CT-cover treatments respectively.

At CT-cover, more substantial increase in nugget to sill ratio values and decrease
in spatial correlation range with residual variograms show the stronger influence of
topography on total C in these treatments. Because with CT-cover, the amount of crop
residue inputs is strongly controlled by topography driven redistribution of soil moisture,
resulting in a stronger relationship between topography and soil C. Topography in this
study generally varied smoothly and had larger correlation ranges. The larger correlation
range in CT-cover appears to form in response to topographical gradients. After
removing the topography effect, the spatial variability structure greatly dropped at CT-
cover compared to CT at both the depths. Still, CT-cover showed greater variability and
stronger spatial structure compared to CT at all the lag distance at both the sampling

depths.

39

a)

 

 

 

 

 

 

 

 

 

 

 

o CTO5
TC
005 MW... .______ 1 o CTcover 05
l 0 CT20 30
0'04 l g A CT cover 20 30
a
E 0 03 1 l
a 1
F. 0.02 « i
> 1
0.01 1 l
1
0.00 . . . . . -
0 10 20 30 40 50 60 70
Lag didance, m
b)
TC - Residuals ° C705
0.05 A CT cover 05
O CT20 30
0‘04 q o CTcover 20 30
E 0'03 ‘ —|\DDE.
%
E 002«
>
0.01 4
0
O 70

Lag distance, m

 

Fig. 2.6. Sample variograms and variogram models at 0-5 cm and 20-30 cm for: (a)

original C data (b) residuals from topographical regression models at conventional chisel-

plow (CT) and organic chisel-plow with cover cropping (CT-cover). Model parameters

are shown in Table 2.5b.

40

Greater variability in subsurface depth with CT-cover may be attributed due to the
following mechanisms. Belowground biomass input, which is the major source for soil C
at subsurface depth (Ussiri et al., 2006) varies with fertilization, cropping intensity and
crop species. In this study, CT is a chemical input treatment and CT-cover is legume
based organic tillage treatment. Yield variability with organic based treatments was
reported to be higher compared to the conventional chemical input system (Kravchenko
et al., 2005; Smith et al., 2007). These greater differences in above ground biomass
yields might also produce greater differences at below ground biomass residue return at
the lower sampling depth resulting in greater variability at below plow depth in CT-
cover. Apart from the amount of C inputs, CT-cover receives greater diversity of
biomass returned to the soil through cover crops and weeds compared to CT. Each crop
species has varying rooting depths which may cause stratification of C at different depths
(Holonda et al., 1988; Lorenz and La], 2005).

In CT-cover, among the soil C pools, POM was enriched at least two times
compared to any other fraction and POM contributes major proportion to the total C with
the organic system compared to CT (Marriott and Wander, 2006). POM contents within
the soil profiles will be highly variable (Burke et al., 1999; Hook and Burke, 2000),
which might partly contribute to the greater variability in the soil C distribution at deeper
layers with the CT-cover system.

In summary, surface soils from 0-5 cm depths in poplar management system was
found to be more efficient in maintaining and enhancing C compared to CT and CT-
cover cropping systems. However, no substantial differences were observed between

poplar and NT cropping systems. Overall total C and N variability and spatial variability

41

patterns observed with poplar were found to be similar to that of NT followed by CT-
cover system and substantially different from that of CT. Spatial correlation of total C
was strongest with CT-cover followed by poplar and NT and then CT. Variability
characteristics of the total soil C in the studied plots were strongly affected by
topographical gradients with CT-cover and NT and least with the poplar system.

At the 20-30 cm depth, overall variability in C distribution was much greater in
CT-cover compared to CT. Comparing the spatial variability pattern between the studied
depth, with CT, spatial correlation range and nugget and sill values at the soil surface
were substantially greater compared to below plowing depth indicating greater variability
at the soil surface compared to the 20-30 cm depth. However with CT-cover, variability
in C distribution pattern was found to be similar at both studied depths.

To conclude, 15 years after land use conversion, total C, N content was greatest at
poplar and NT followed by CT-cover and least at CT. In addition to increasing C, N
content the land use conversion also affected spatial variability pattern, where the C, N
content at poplar and NT was more spatially variable followed by CT-cover and least in

CT.

42

REFERENCES

Bergstorm, D.W., C.M. Monreal, and E. Jacques. 2001. Spatial dependence of soil

organic carbon mass and its relationship to soil series and topography. Can. J. Soil
Sci. 81 :53—62

Burke, I.C., W.K. Lauenroth, R. Riggle, P. Brannen, B. Madigan, and S. Beard. 1999.
Spatial variability of soil properties in the shortgrass steppe: The relative
importance of topography, grazing, microsite, and plant species in controlling
spatial patterns. Ecosystems 2:422—438

Coleman, M.D., J .G. Isebrands, D.N. Tolsted, V.R. Tolbert. 2004. Comparing Soil
Carbon of Short Rotation Poplar Plantations with Agricultural Crops and Woodlots

in North Central United States. Environmental Management Vol. 33, Supplement 1
: 299—308.

Cook, J ., and Beyea, J. 2000. Bioenergy in the United States: Progress and Possibilities.
Biomass and Bioenergy 18: 441-445.

Creed, I.F., C.G. Trick, L.E. Band, and l.K. Morrison. 2002. Characterizing the spatial
heterogeneity of soil carbon and nitrogen pools in the Turkey Lakes Watershed: A
comparison of regression techniques. Water Air Soil Poll. FOCUS 2: 81-102.

DeGryze, S., J. Six, K. Paustian, S.J. Morris, E.A. Paul, and R. Merckx. 2004. Soil
organic carbon pool changes following land-use conversions. Glob. Change Biol.
10:1120—1132.

ESRI, 2004 Environmental Systems Research Institute. ArcGIS Spatial Analyst. ESRI,
Redlands, CA.

Gallant, J .C., and J .P. Wilson. 2000. Primary t0pographic attributes. In:Wilson, J .P.,

Gallant, J .C. (Eds.), Terrain Analysis: Principles and Applications. Wiley, New
York, pp. 51-86.

Grandy, A. S., and GP. Robertson. 2007. Land-use intensity effects on soil organic
carbon accumulation rates and mechanisms. Ecosystems 10: 58-73

Holanda, F.S.R., D.B. Mengel, M.B. Paula, J .G. Carvaho and J .C Bertoni. 1998.
Influence of crop rotations and tillage systems on phosphorus and potassium

stratification and root distribution in the soil profile. Commun. Soil Sci. Plant Anal.
29: 2383-2394.

Hook, RB, and LC. Burke. 2000. Biogeochemistry in a shortgrass landscape: Control by
topography, soil texture, and microclimate. Ecology 81:2686—2703.

43

Kravchenko, A. N., G. P. Robertson, X. Hao, and D. G. Bullock. 2006. Management
Practice Effects on Surface Total Carbon: Differences in Spatial Variability Patterns.
Agron. J. 98: 1559-1568.

Kravchenko, A.N., and D.G. Bullock. 2000. Correlation of corn and soybean grain yield
with topography and soil properties. Agron. J. 92:75—83.

Kravchenko, A.N., G.P. Robertson, K.D. Thelen, and RR. Harwood. 2005. Management,
topographical, and weather effects on spatial variability of crop grain yields. Agron.
J. 97:514—523.

Lemus, R., and La], R. Bioenergy crops and carbon sequestration. 2005. Critical Reviews
in Plant Sciences, 24 (1): 1-21.

Lorenz, K., and R. Lal. 2005. The depth distribution of soil organic carbon in relation to
land use and management and the potential of carbon sequestration in subsoil
horizons, Adv. Agron. 88: 35—66.

Marriott, BE, and M.M. Wander. 2006. Total and labile soil organic matter in organic
and conventional farming systems. Soil Sci. Soc. Am. J. 70:950—959.

Milliken, G.A., and DE. Johnson. 1992. Analysis of messy data: Designed experiments.
CRC Press, Boca Raton, FL.

Mueller, T.G., and F.J. Pierce. 2003. Enhancing spatial estimates of soil carbon with
terrain attributes. Soil Sci. Soc. Am. J. 67:258—267.

Needelman, B.A., M.M. Wander, G.A. Bollero, C.W. Boast, G.K. Sims, and D.G.
Bullock. 1999. Interaction of tillage and soil texture: Biologically active soil
organic matter in Illinois. Soil Sci. Soc. Am. J. 63:1326—1334.

Paustian, K., J .Six, E.T.Elliott and H.W.Hunk. 2000. Management options for reducing
C02 emissions from agricultural soils. Biogeochemistry 48: 147-163

Robertson, G.P., J .R. Crum, and BC. Ellis. 1993. The spatial variability of soil resources
following long-term disturbance. Oecologia 96:451—456.

Robertson, G.P., K.M. Klingensmith, M.J. Klug, E.A. Paul, J .C. Crum, and BC. Ellis.
1997. Soil resources, microbial activity, and primary production across an agricultural
ecosystem. Ecol. App]. 7: 1 58—170.

Robertson, G.P., E.A. Paul, and RR. Harwood. 2000. Greenhouse gases in intensive
agriculture: Contributions of individual gases to the radiative forcing of the
atmosphere. Science 289: 1922—1925.

44

Sartori, F., R. Lal, M.H. Ebinger and DJ. Parrish. 2006. Potential soil carbon
sequestration and CO2 offset by dedicated energy crops in the USA. Critical
Reviews in Plant Sciences 25(5): 441-472.

SAS Institute. 2001. SAS user's guide. Version 9.1. SAS Inst., Cary, NC.

Schabenberger, O., and F.J. Pierce. 2002. Contemporary statistical models for the plant
and soil sciences. CRC Press, Boca Raton, PL.

Smith, R.G.. F. D. Menalled, and G. P. Robertson 2007.Temporal Yield Variability under
Conventional and Alternative Management Systems Agron. J ., November 6, 2007;
99(6): 1629 — 1634

Stoyan, H., H. De-Polli, S. Bohm, G. P. Robertson, and E. A. Paul. 2000. Spatial
variability of soil respiration and related soil properties at the plant scale. Plant and
Soil 222: 203-214

Terra, J .A., J .N. Shaw, D.W. Reeves, R.L. Raper, E. van Santen, and PL. Mask. 2004.
Soil carbon relationships with terrain attributes, electrical conductivity, and a soil
survey in a Coastal Plain landscape. Soil Sci. 169:819—83 1.

Ussiri, D.A, R. Lal and RA. Jacinthe 2006 Post-reclamation land use effects on
properties and carbon sequestration in minesoils of southeastern Ohio. Soil Sci. 171:
261—271.

Yang, X.M, and M.M. Wander.1999 Tillage effects on soil organic carbon distribution
and storage in a silt loam soil in Illinois. Soil Tillage Res. 52: 1—9.

45

Chapter III
RELATIONSHIPS BETWEEN TOTAL SOIL CARBON AND TOPOGRAPHY OF
DIFFERENT MANAGEMENT SYSTEMS
ABSTRACT

Topography or local relief is one of the major factors affecting spatial variability of
soil carbon (C) and nitrogen (N) contents at the field/landscape level. General
topographical influences on soil C and N related processes, controlled by soil water
contents, are likely to differ in magnitude under different agricultural systems. In the
present Study, our objective was to examine the interactions between terrain attributes
and management systems on soil C and N and to identify the landform elements that were
most effective in C sequestration under each management system. The study was
conducted at the Long Term Ecological Research (LTER) site, at Kellogg Biological
Station, in southwest Michigan. Treatments used in the study were chisel-plow(CT) and
and no-till(NT) with conventional chemical inputs and a chisel-plow organic
management system with winter leguminous cover crops (CT-cover). Interaction effect
between topographical features and tillage and management systems on soil C and N was
tested first, with wetness index (WI) a compound topographical index as a continuous
variable using analysis of covariance (ANCOVA) and second, topographical information
was used as categorical variable by grouping similar class of terrain attributes into
specific landform elements namely, upperslope and valley using discriminant analysis.
Upperslopes are high relative relief regions in the landscape, characterized by less soil
wetness index (water shedding region in a landscape). Valley positions are low lying

areas characterized by greater soil wetness index (water receiving region in a landscape).

46

Interaction between treatment effects and W1 was found to be significant at both
(0-5, 20-30 cm) the studied depths. The site’s topography effect was found to be stronger
with the CT-cover treatment compared to CT and NT. Among the landform elements, C
and N content was found to be greater at valley areas compared to upperslope position in
all the studied treatments. Greater differences in C and N contents between the landform
positions were more pronounced at CT-cover followed by conventional CT and NT
treatments and the differences were more pronounced at 20-30 cm depth compared to
surface depth (0-5 cm).

In overall at 0-30 cm depth, CT-cover and NT were found to accumulate 5.5 and
4.61 Mg more C per ha than CT at upperslope position and 9.12 and 15.39 Mg per ha
more C than CT in the valleys. Similarly with N, CT-cover and NT were found to retain
0.73 and 0.84 Mg more N per ha than CT at upperslope position and 0.84 and 1.59 Mg
per ha more N than CT in the valleys. In summary, tillage based organic cover crop
system (CT-cover) was found to be equally efficient as the NT system in restoring C and
N at 0—30 cm depth and the largest C and N benefits of CT-cover were observed in valley

position as compared to upperslope position in the landscape.

INTRODUCTION

Soil organic matter (SOM) content is considered an index for soil quality
(Seybold et al., 1997) and crop productivity and also plays an important role in the global
carbon cycle by acting as a source as well sink for the atmospheric COz emissions and C
uptake (Smith et al., 1999). Thus today soil carbon management is a key component in

most sustainable agriculture programs and global climate change studies. The net C

47

content in a soil system is a tradeoff between the C additions into the system through crop
biomass, cover crops, weeds and microbial biomass and C losses through soil respiration
and erosion. The level of soil disturbance in the system has a substantial influence on the
magnitude of C losses. Among the agricultural systems even with similar levels of carbon
inputs, such as conventional tillage and no till systems, the greater soil disturbance
generated by conventional tillage will lead to greater C losses. Even within the same
management system, soil C sequestration potentials vary in response to variations in
topography, parent materials, and microclimatic conditions. Increasing the carbon content
in the soil from its current levels requires a thorough understanding of various factors
controlling soil carbon and interactions among these factors. At a field scale, the
magnitude of soil C variability depends on the tillage practice and is strongly affected by
topography and amounts of plant residue returned to soil.

Topography affects soil carbon mainly in two different ways. It influences erosion
and redeposition of fine soil particles, organic matter, and nutrients resulting in
accumulation of nutrient rich, finer soil particles at valley regions as compared to soils at
upperslope positions in a landscape (Ovalles and Collins.,l986; Pennock and de
Jong.,l990; Kravchenko and Bullock, 2000). Topography drives vertical and horizontal
water redistribution, resulting in differences in leaching, infiltration rates, and runoff
affecting plant growth through varying moisture availabilities (Creed et al., 2002). Effects
of topography on SOM have been well documented over the past few decades. Especially
in the recent past, with the advancement in GIS and GPS technology, many researchers
reported a strong relationship between terrain attributes and soil properties at a field scale

(Ziadat, 2005). Topographical features such as landscape position, wetness index (WI)

48

(Moore et al., 1993; Moonnan et al., 2004), terrain curvature (Yoo et al., 2006, Papiemik
et al., 2007) and slope (Hao et al., 2002) were considered to be important factors
influencing amount and turnover of SOC. In several studies slope and W1 were found to
account for more than half of the variability in SOM (Moore et al., 1993; Terra et al.,
2004). W1 is a compound topographical index which depicts the relative likelihood of
saturation or wetness. Areas with higher WI were found to have higher soil organic
carbon (SOC) due to higher sediment deposition and higher biomass production
compared to lower WI zones. Curvature influences SOC by controlling the rate of soil
erosion and soil thickness (Yoo et al., 2006). Aspect direction influences SOC by
creating local variation in soil temperature and moisture which results in altering the
decomposition rates of soil organic matter as well as affecting plant growth and biomass
inputs (Yimer et al., 2006).

General topographical influences on soil C related processes are likely to differ in
magnitude under agricultural systems with different tillage. Tillage controls soil organic
matter dynamics by three major actions such as, periodic disruption of soil structure,
incorporating plant residues within soil horizon, and altering soil microclimate (Balesdent
et al., 2000). These three major mechanisms in turn influence various soil processes, such
as soil aggregation, erosion, mineralization rate, wetting and drying cycles, soil moisture,
temperature, and aeration (F ranzluebbers et al., 1994, Hernanz et al., 2002).

Periodic disruption of soil structure due to tillage tends to reduce C and N content
in the soil. No till management was introduced to reduce C loss associated with tillage
disturbance, and Conservation Technology information Center (CTIC) reported that land

area under no till increased from 2.2 Million hectares in 1973-74 to 21.1 million hectares

49

in 2001-02. In no—till, higher organic carbon at the soil surface leads to high aggregate
stability, controls erosion, and buffers temperature and moisture fluctuations.
Franzlubbers (2004) in a review of studies that compared no-till with other tillage
systems concluded that water-stable macro aggregates tend to be more abundant in soils
under no-till as compared to other tillage systems. Increased numbers of water stable
aggregates in no-till leads to accumulation of soil organic matter within macro aggregates
and protects SOC from faunal action and microbial consumption (Beare et al., 1994a,
1994b; Six et al., 2000a) producing aggregates enriched in fine particulate organic matter
which are more resistant to decomposition (Six et al., 2000b). In heavily disturbed soil
such as chisel plowed systems, tillage interferes with the soil aggregation process and
exposes more area for microbial decomposition. However the effectiveness on reducing
C losses via no-till strongly depends on no-till being maintained for a long time with soils
remaining unplowed to protect aggregation and physically stabilized C pools (Grandy and
Robertson, 2006).

Tillage affects mixing of aboveground residue and decomposition rate within the
soil profile, which in turn influences varying concentration of SOM content with in the
soil profile. Tillage by a chisel plow to a depth of 15 cm was found to incorporate nearly
60 percent of the residues at a depth of 0-6 cm, whereas in a no till system all the residues
remained on the surface (Allmaras et al., 1996). Crop residue on the soil surface in no-till
generally decomposes slower than residue incorporated in soil via tillage because of less
contact with micro organisms (Reicoskt et al., 1997) and differences in soil microclimate.
This leads to an increase in the particulate organic carbon in no-till over tilled systems

(Franzluebbers et al., 2004).

50

Tillage alters the soil microclimate and creates local variation in soil temperature
and moisture conditions (Balesdent et al., 2000). SOC losses through soil respiration
greatly depend on soil temperature and soil moisture. In no-till, the presence of previous
crop residues have been reported to decrease topsoil temperature, by 0-3° C, increase soil
moisture content, and reduce soil aeration (Coote and Macolm-McGovem, 1989). Based
on Rothemsted C model (J enkinson et al., 1992) for every 2°C decrease in soil
temperature the decay rates decrease by a factor of 0.8, while the decay rate increases 1.2
to 1.8 times for every 10% increase in soil moisture of the pore space corresponding to a
specific range of temperature and moisture conditions (Balesdent et al., 2000).

Organic based systems with winter cover crops (WCC) is one of the potential
management systems for increasing soil carbon and protecting soil quality in Midwest
agriculture. Organic based systems with WCC were reported to be as effective in
sequestering C as no-till despite heavy soil disturbance by tillage in such systems
(Teasdale et al., 2007). Systems with WCC enhance SOC by increasing the amount and
diversity of biomass returned to the soil and by increasing the time during which the land
is covered by growing plants (Follett, 2001). Benefits of growing winter annual cover
crops include increased soil aggregate stability (McVay et al., 1989) and increased soil N
content (Blevins et al., 1990).

It is apparent that the effects of topography, tillage and cover crops on soil C and
N may interact in a number of ways and that the soil C relationship with topography may
vary in magnitude in different tillage and management systems. The influence of
topography can be expected to be either buffered or intensified by different tillage and

crop management practices (Bergstrom et al., 2001). Though a substantial number of

51

studies addressed the relationships between topography and soil C and N and the tillage
effects on soil C and N separately, there is only limited quantitative information on
interactions between management systems and topography and their effects on soil C and
N. Tillage induced soil loss or displacement are reported to be highest in crest, shoulder
and upperslope positions which may result in truncated soil profiles in convex slope
positions and inverted soil profiles with subsoil material deposited over original surface
horizons at convex slope position (Kosmos et al., 2001, Heckrath et al., 2005 Papiemil et
al., 2007). In no-till systems, VandenBygaart et al. (2002) found that landscape position
plays a significant role in the ability of no-till soils to sequester SOC. After adoption of
NT, the redistribution of soil particles from upperslope to valley positions was found to
be greatly reduced and C content was shown to increase in upper and middle slope
positions similar to that of lower slope positions due to increased moisture retention
capabilities resulting in better crop yields and greater C inputs to the soil (VandenBygaart
et al., 2002). Bergstorm et al. (2001) mentioned that no-till had more SOC than
conventional tillage only at well drained upperslope positions. The above studies
illustrate that upperslope positions are susceptible to erosion under tillage systems and
under no-till systems particle movement is greatly reduced due to lower soil disturbance
and more surface crop residues. These findings demonstrate that the topographical effect
on soil particle redistribution varies under different management systems. The
disadvantage of most of the studies that examined topographical effects on soil C in
different management systems is that they were based on data being collected from a
limited number of sites without replication for management systems. Thus the effects of

management on the relationships between topography and soil C could not be properly

52

separated from the individual characteristics of the specific study sites. In addition, there
is only little such information reported studying the relationship between topography and
soil C, N in an organic management system (Kasper et al., 2006).

In this study, topography modifications of total C and N levels under a no-till
system, conventional tillage system, and organic based tillage systems were compared in
a replicated long-term field experiment. The large experimental plots of 1 ha size and
dense elevation data provided reliable estimates of various topographical features within

each plot.

The objectives of this study were:

(i) To assess the strength of the relationships between topographical features and
soil total C and N and study the interactions between topography and
management systems on soil C and N;

(ii) To identify the landform elements that lead to most rapid and effective C

sequestration in each management system.

MATERIALS AND METHODS
Test Site

The study was conducted at the Kellogg Biological Station’s Long Term
Ecological Research (KBS-LTER) site, in southwest Michigan (85°24' W, 42°24' N)
established in 1988. Soils at the site have been developed on glacial outwash and belong
to Typic Hapludalfs of the Kalamazoo (fine-loamy, mixed, mesic) and Oshtemo (coarse-
loamy, mixed, mesic) series (Mokma and Doolittle, 1993). Climate is temperate with

cool, moist winters and warm, humid summers and receives approximately 90 cm annual

53

precipitation, with about half as snow. Mean annual temperature is 9°C (Grandy and

Robertson, 2007).

Experimental Layout and Agronomic Protocols

The LTER experimental layout is a one-factor randomized complete block design
with six replications and seven treatments organized along a management intensity
gradient based on tillage intensity, external inputs and net primary productivity (Grandy
and Robertson, 2007). Each experimental plot is about 90 by 110 m in size. The
experiment was started at 1989, before that, the studied site had been conventionally
managed in row—crop agriculture for at least a century (Robertson et al., 1997). In this
study, three LTER agronomic com-soybean-wheat rotation treatments were studied,
namely, chisel-plowed with conventional chemical inputs (CT); no-till with conventional
chemical inputs (NT); and organic chisel plowed system with winter leguminous cover
crops and zero chemical inputs (CT-cover). The cover crops of the CT-cover treatment
was hairy vetch (Vicia villosa Roth) from 1989 to 1993 and red clover (T rifolium
pratense L.) from 1994 onwards. The cover crops are generally incorporated in April or
May before corn or soybean planting. Weed control in CT-cover for corn and soybean
was achieved by weekly-biweekly rotary cultivation from May to early July. For wheat,
cultivation prior to wheat planting was conducted in the fall. Thus CT-cover management
had the greatest amount of mechanical disturbance among the three studied systems. The
complete site description, experimental design, and management protocols are available at

the KBS website (Kellogg Biological Station, 2005).

54

Soil Sampling

Soil sampling was carried out over the period from 2003-2006 and consisted of 3
sample sets. The first set of soil samples was collected at 0-5 cm depth from replications
1 through 4 in May of 2003 and from replication 5 and 6 in May of 2004. In both years
CT and CT-cover treatments were plowed approximately 2 weeks prior to sampling. The
sampling scheme for 2003-04 is shown on Fig. 3.1. To avoid plot boundary effects and
the long-term protected portions of the plots restricted to sampling, the sample collection
area was limited to the middle 60- by 60-m portion in each experimental plot.

Approximately ~100 georeferenced soil samples were collected at each plot for a
total of 1714 samples. Among the 100 sampling locations within each plot, 20 samples
were collected on a 15- by 18-m triangular grid and the remaining 80 samples were
collected at distances 1.0, 2.5, 5.0, and 7.5 m from the grid points (Kravchenko et al.,
2006). At each sampling point, the sample was taken between the plant rows and was
composited from five 2.5-cm-diameter cores collected within a 0.2-m radius.

In the CT plot, replication—3 (bottom right comer plot in the Fig. 3.1), topography
was represented by a deep bowl-shaped valley. As such, it did not have analogues among
any of the plots of the other treatments in the rest of the LTER site. This distinctly
different shape led the data from this plot to differ notably from the rest of the site both in
terms of the total C and N values, their spatial patterns, and relationships with
topography. Since it could not be meaningfully compared with the other treatments the

data from this plot were excluded from further analyses.

55

 

  
 
 
 
 
 
 

Elevation (m)

D 282.8 - 284.0
(:1 284.0 - 285.4
- 285.4 -286.5
- 286.5 — 287.6
- 287.6 - 289.2
- 289.2 - 291.5

 

 

 

 

Fig 3.1. Layout of the experimental site along with the locations of the 18 experimental

plots used in the study, the 15x15 m interpolated elevation map.

56

The second set of soil samples was collected at a 20-30 cm depth in May of 2006
from replications l and 4 for CT and replication 3 and 4 for CT-cover. These plots were
selected for 20-30 cm sampling because of their relative proximity to each other and
relative topographical diversity of that portion of the LTER site. At each plot
approximately 100 samples were collected from the same locations as the 0-5 cm samples
of 2003. Among the 100 samples within each plot, 20 undisturbed cores of 5.2 cm-
diameter were obtained using a hydraulic soil core unit from Geoprobe systems
(Geoprobe Inc, Salina, Kansas) to a depth of 0-40 cm on a 15- by 18-m triangular grid
and the remaining 80 off-grid samples were collected at four distances, 1.0, 2.5, 5.0 and
7.5m, from the grid points at 20- to 30-cm depth from between the plant rows and was
composited from five 2.5-cm-diameter cores collected at 0.2-m- radius circle
(Kravchenko et al., 2006).

The third set of soil samples was collected along with the second set in May, 2006
from the locations that were sampled in 1988 prior to LTER establishment. The samples
were undisturbed cores obtained using the hydraulic soil core unit of 5.2 cm in diameter
to a depth of 0-40 cm. There were 11 sample cores collected for CT, 17 for CT-cover and
22 for NT. The cores were segmented at the following depth of 0-1 5, 15-20, 20-30, 30-40
and bulk density were determined for all the segments. For the bulk density
measurements, soil from each horizon of depth was carefully transferred in to polythene
bag and total wet mass of the sample weighed. Soil gravimetric water content was
determined by weighing, oven drying, and reweighing subsamples.

Based on the moisture content of the subsample, the dry weight of the total

sample was calculated for each depth. The bulk density of each soil depth was computed

57

by dividing the oven dry weight into the volume of each segment which is obtained by
measuring the length and diameter of tube from which the sample was taken for each
depth. The bulk density values were then used to assess soil C and N stocks on an aerial
basis. The measurements from the cores that were relevant to this study are the bulk
density and total C and N for the depths of 0-15 cm and 20-30 cm.

For 20-30 cm depth, total C and N data from sampling set two and three were
combined into a single data set, thus, a total number of samples for 20-30 cm depth were
equal to 159 and 182 in CT and CT—cover treatments, respectively, and 22 in NT. The
samples were air dried at room temperature and all large plant residues removed
manually. Smaller plant material was removed by gentle air blowing and the samples
were ground on a shatterbox machine for one minute to pass a 250-pm sieve. Total C was
measured using Carlo-Erba, model NA1500, series 2 Nitrogen-Carbon-Sulfur Analyzer
(Carlo-Erba, Milan, Italy). In each plot, about 10% of the samples were replicated thrice
for total C measurements to account for the variability associated with laboratory
measurement error. Previous studies conducted at LTER site reported that inorganic C at
the soil surface was found to be negligible thus the values obtained from chemical
analysis of total C are considered to be representative of the organic C (Degryze et al.,
2004). Total C and N contents at 0-5 and 20-30 cm depths were reported as g/100g of
soil. Total C and N stocks were calculated on an aerial basis for 0-30 cm depth based on

the sampling locations where 20-30 cm data were available.

58

For 0-20 cm values, in CT and CT-cover, data obtained from 2003-04 samples
and 2006 samples were extrapolated to 0-20. In NT, data obtained from 2006 samples at
0-15 cm were extrapolated to 0-20 cm. Bulk density values of 0-30 were derived from
bulk density measurements of 2006 samples obtained using the hydraulic soil core unit.

Total C and N stocks at 0-30 cm depth were reported in Mg per ha.
Terrain Attributes

A topographical survey for measuring elevation data was conducted at the LTER
site twice. The survey was conducted using a land-based laser in 1988 generating 597
elevation measurements (Robertson et al., 1997) and a second survey was carried out in
2004 using a 12-channel Leica SR530 real-time kinematic DGPS (dual-frequency global
positioning system) receiver with measurements collected every 2 m in rows 5 m apart
(Kravchenko et al, 2006). The elevation data from both surveys were merged and
converted into a cell-based terrain map on a 4 by 4 m grid using inverse distance
weighting with power of two and six nearest neighbors in ArcGIS 9.0 Spatial Analyst
(ESRI, 2004). Then topographical features such as terrain slope, aspect, curvature, and
flow accumulation were derived from the elevation map using surface and hydrology
functions of ArcGIS 9.0 Spatial Analyst. Previous reports (Kravchenko et al., 2006) on
aspect effects on total C at this site showed that the major difference in total C existed
between southern aspects (south, southeast, and southwest) and the rest of the aspects.
Slope aspects measured in degrees was reclassified into two categories, namely 0 for
southern aspect which ranges from 135 to 225 degrees and 1 for the other orientations.
Relative elevation (m) for each cell was obtained by calculating the difference in height

of specific cell of interest from the minimum elevation in each plot. Slope represents

59

gradient between the land surface and a horizontal plane at a given point. Curvature
depicts the measure of convexity or concavity of the land surface in the direction of
slope. By convention, convex curvatures are assigned positive values and represents hills
in the landscape and concave curvatures are assigned negative values and represents
valley areas. Flow accumulation is calculated for each pixel based on the accumulated
weight for all pixels that flow into each downslope pixel. More detailed descriptions of
each terrain attribute and scale of measurements were provided by Kravchenko and
Bullock (2000) and Pennock and Coore (2001). The wetness index (WI) was computed
based on flow accumulation and slope data for each map grid cell (Gallant and Wilson,

2000; Huang et al., 2007) as:

WI = ln(AS /tan ,3) [1]

where As is specific catchment area derived from flow accumulation, and tan ,6 is a

tangent function of slope at percent-rise.
Data Analysis

Data analysis was conducted using SAS (SAS Institute, 2001). Relationships
between terrain attributes and soil C and N were studied using two different approaches.
First, the relationship between soil C and N with the terrain attributes in different
treatments were evaluated by treating the terrain attributes as continuous variables via
correlation and multiple regression analyses and analysis of covariance. The purpose of

using terrain attributes as continuous variables is to quantify relationships between

60

specific terrain attribute or their combinations with soil C and N and to study how much
of variability in soil C and N can be explained by terrain attributes in each treatment.

In the second approach, I grouped locations with similar terrain attributes into
specific landform elements using discriminant analysis. Then the relationship between
soil C and N with topography in different treatments was assessed via two factor analysis
of variance with landform elements as a categorical variable. Each landform element has
a distinct set of quantitatively defined range of morphological and positional slope
attributes, which in turn results distinct hydrological and soil properties (Pennock and
Coore, 2001). The goal of landform segmentation is to find out the landform elements

that lead to most rapid and effective C sequestration in each management system.

Terrain attributes as continuous variables

In the first approach, terrain attributes were considered as continuous variables
and were related to soil C using correlation, multiple regression, and analysis of
covariance (ANCOVA) methods. Correlation analysis and multiple regression were
performed separately for each treatment using PROC CORR and PROC REG procedures
(SAS Institute, 2001). The multiple regression models initially included either total C or
total N as the response variables and terrain attributes such as elevation, slope, curvature,
flow accumulation, aspect, and W1 as independent variables. Then the model selection
was conducted using stepwise regression (or=0.10). Prior to model selection, all variables
were checked for normality and presence of outliers and transformed if found necessary.
Specifically, slope and flow accumulation was found to be heavily skewed, thus both

variables were log-transformed prior to analyses. The need for quadratic terms for

61

elevation, slope, curvature, flow accumulation, aspect, and W1 were checked separately
in each plot and the quadratic terms that were found to be significant at the p = 0.05 level
were added to the regression equation. Presence of multicollinearity was examined using
variance inflation factors and condition index. Some multicollinearity was anticipated
because most of the terrain variables were the primary and secondary derivatives of
elevation. Topographical variables that appeared to be collinear and provided the least
contribution to the model were excluded from further analysis. The model fit was
evaluated by looking at residual plots and presence of outliers was assessed using Cook’s
D values.

The interactions between terrain attributes and studied agronomic treatments on
soil C, N were studied using analysis of covariance. Wetness index is a compound
topographical index that integrates information from specific catchment area and slope,
and several previous studies used WI as an important covariate while predicting soil C,
yield and other soil attributes (Skidmore et al., 1991; Kasper et al., 2006; Huang et al.,
2007). In this study, WI was found to be the terrain variable most consistently related to
soil C and N. Thus WI was used as the primary terrain covariate in the subsequent
ANCOVA and evaluations of the interactions between topography and treatments for
soil C and N. The analysis was carried out using PROC MIXED procedure in SAS
(Milliken and Johnson, 2002). The relationships between C and N with WI were found to
be linear in all treatments, thus higher order terms were not used.

The statistical model included treatments as the fixed factor and W1 as a
covariate. Replications and plots nested within treatments were treated as random factors

and plots were used as an error term for testing the treatment effects. Degrees of freedom

62

for error for the covariate were set to be equal to those used for treatments. Estimates of
linear regression slopes from ANCOVA were obtained for each treatment for both total C
and N. When an interaction between treatments and W1 was found to be statistically
significant (0r=0.05), pair-wise comparisons was conducted between the regression slopes
for the three treatments.

There was a strong spatial correlation present in C and N data, thus 1 accounted
for spatially correlated residuals (Schabenberger and Pierce, 2002). For that sample
variograms for the residuals were examined separately for each treatment and fitted with
variogram models using PROC NLIN procedure in SAS. Then the variogram model
parameter estimates were used in ANCOVA to describe spatial structure of the correlated
residuals. Such an approach ensured easier convergence and helped to reduce the
computation time. The decisions on whether models with spatially correlated residuals
were warranted were made based on the Akaike Information Criterion (AIC) values with

lower AIC values indicating preferable model.

Analysis of landform classes

In this study, landscape within each plot was classified into two landform classes,
i.e. upperslopes and valleys. Similar approach was carried out for landform classification
at field scale by Macmillan et al. (2000). The land form classification was conducted
using discriminant analysis. Discriminant analysis has been employed earlier in
predicting soil properties and found to be effective by Bell et al. (1994), Kravchenko et

al. (2002) and Bakhsk et al. (2007).

63

In this study site, landform class information and terrain attributes were available
for 258 locations. These landform classes were identified through a direct field survey by
traversing along the contour lines in the landscape and the point where we can make a
clear judgement about the landform, the specific landform class was recorded using GPS.
Among the 258 known landform locations observed, 156 observations were located in
upperslope positions and 102 in low lying valleys. This constituted a model dataset.
Using the model dataset, classification rule was developed to predict each landform class
for all the other locations that had terrain attribute data but where the landform class was
unknown. Stepwise discriminant procedure, STEPDISC (SAS Institute, Cary, NC) was
used to select the topographical variables that had significant contribution to landform
class prediction. Relative elevation, WI and Flow accumulation were the variables found
to be significant at 0.15 significance level and used in building the classification rule.

A predictive model developed using the model dataset was then applied to the
remaining dataset to place each point into the appropriate landform class. Cross-
validation was applied to evaluate the accuracy of landform class prediction. For cross-
validation each value from the model data set was eliminated in turn and, then, estimated
using information from the rest of the data (Khattree and Naik, 2000; Kravchenko et al.,
2002). For each data location, discriminant analysis calculated the posterior probability
of an observation for each of the landform classes and the specific observation was
assigned a landform class that had the highest posterior probability (Johnson,1998 and
Khattree and Naik, 2000).

Effects of treatments and landform elements on total C and the interaction

between them was quantified via PROC MIXED procedure. The statistical model

64

included the treatments, landform elements and their interaction as fixed factors and
replications, plots, and interaction between landform elements, treatments and
replications as random factors. Plots nested within treatments were used as error terms for
testing treatment effects and interaction between landform elements, treatments and
replications was used as an error term for testing the landform and landform by treatment
interaction effects.

At 20-30 cm, there were only 22 observations in NT as compared to much larger
numbers of data points in CT and CT-cover (158 and 182, respectively) which
complicated obtaining reliable comparisons between them. Thus for the 20-30 and 0-30
cm depth, I mainly focused on CT and CT-cover treatment results however the results for

NT treatments were also shown but not discussed in detail.

RESULTS

Descriptive Statistics and Landform Classification

Total C and N values were significantly different among the studied treatments
(p< 0.01; Table 3.1) at all the studied depth. In overall, CT-cover and NT had the highest
C and N in soil depths from 0 to 30 cm. Relative elevation of these treatment plots ranged
from 2.38 to 5.59 meters (Table 3.2). Elevation gradient of about six meters within 50
hectares of the experimental site provided a reasonably large topographical diversity to
study the relationship between topography and soil C and N under different management
systems. In all three treatments the curvature and W1 values were relatively similar,
whereas the elevation range, slope and flow accumulation were found to be somewhat
lower in CT-cover than in CT and NT treatments. However, the differences among the

three treatments in terms of any of the terrain attributes were not

65

however after removing the topographical effect, the variogram values with
poplar were found to be higher than NT and CT-cover at all the lag distances (Fig. 2.4b).
Greater C variability with poplar might be attributed due to the following mechanism.
Especially after land use conversion from conventional tillage to any conservative
management system, the rapid change in the soil C pool will be mainly attributed from
the changes in light fraction and particulate organic matter. Greater variability in soil C
distribution in poplar, NT and CT-cover might be attributed due to the increase in
particulate organic matter (POM) content after conversion from conventional tillage
system. Poplar systems sequester C mainly at the fine intra aggregate particulate organic
matter (POM) (Degryze et al., 2004). Fine intra aggregate-POM estimated at poplar in
the same study site during 2001 was reported to be greater than NT and CT—cover system
(Grandy and Robertson, 2007).

Similarly many previous studies reported NT (Needelman et al., 1999) and CT-
cover (Marriott and Wander, 2006) hold 30-40 percent more POM than conventional
tillage treatment. POM was found to be more variable than the total C (Burke et al.,
1999). Greater overall variability with poplar and NT is attributed due to a greater

amount of POM and lesser soil disturbance compared to CT-cover.

Spatial variability patterns in soil C, N at surface (0-5 cm) and below plow
(20-30 cm) depth among two management systems

Total C with CT-cover was significantly greater than CT at both the studied
depths (Table 2.5a; p<0.05). Within each treatment the C content at the surface depth

was significantly greater than the subsurface depth.

66

Table 3.1. Mean C and N for all the studied depth at CT, NT, CT-cover crop treatment

along with percentages of coefficients of variation in parentheses.

 

 

 

 

 

 

TRT N Total C Total N
0-5 cm #

CT 485 7.6(19) a* 0.70(22) a

NT 548 11.5 (24) b l.01(30) b

CT cover 583 10.3 (20) c 0.91 (22) b
20—30 cm #

CT 158 3.1(25) a 0.33 (25) a

NT 22 5.0(59) b 0.50 (50) b

CT cover 182 4.3 (44) b 0.41 (34) b
0-30 cm # #

CT 151 26.84(15) a 2.65(20) a

NT 21 39.64(31) b 4.1 1(26) b

CT cover 167 35.01(19)b 3.45 (17)b

 

*Means within each column followed by the same letter are not significantly different (0t=0.05)

#-ngg'l soil

# # Soil C (Mg C ha ") = (soc (g c kg " soil) /10) x 10,000 (m2) x depth x bulk density

(Mg m '3)

Table 3.2. Mean bulk density values for all the studied depths and a summary of terrain
properties for the three studied treatment sites.

 

 

 

Bulk Densich) Terrian attributes
Elevation
TRT 20-30 0—30 range Slope Curvature Flow
CT 1.70a 1.52a 5.59a 1.26a 0.047a 0.5053
NT 1.69a 1.58a 5.323 1.15a -0.015a 0.483a
CT cover 1.56a 1.43a 2.38a 0.86a 0.018a 0.335a

 

*Means within each column followed by the same letter are not significantly different (0:0.05)

Elevation in meters; Slope in degrees; Curvature: Curvature of the terrain, 10 ' m.

WI : Wetness index, m 2 m

-1

Flow : Log transformed flow accumulation calculated as accumulated weight for all cells that
flow into each downslope cell.

67

statistically significant (p<0.05; Table 3.2). Although bulk densities at all soil depths of
the three treatments were not significantly different (Table 3.2), there was a trend toward
lower bulk densities for CT-cover treatments.

Mean values of the terrain variables and soil texture for each landform class
classified via discriminant analysis are shown in Table 3.3. The upperslope element was
characterized by higher relative elevation, lower WI and lower log flow accumulation
values and higher sand content. Valley positions are low lying areas characterized by
greater soil wetness index, lower relative relief and log flow accumulation. Upperslope
element in this study includes geomorphic units such as summit, sideslope, backslope and
shoulder. Valley element includes footslope and valleys. Upperslopes are water shedding

regions and valleys are water receiving region in the landscape.

Relationship Between Terrain Attributes and Soil C and N at 0—5 cm Depth

Pearson correlation coefficients of total C and N with WI tended to be higher
among those studied terrain attributes (Table 3.4). Correlation coefficients varied fiom
treatment to treatment, however in general, total C and N at 0-5 cm depth were negatively
correlated with elevation, terrain slope and curvature; and positively correlated with WI
and flow accmnulation in all three treatments. Positive correlations between soil C and
W1 and negative correlations between C and elevation and slope, observed in this study,
were consistent with results reported by Moore et al., (1993); Florinsky et al., (2002);
Muller and Pierce (2003); Terra et al., (2004). Differences in strength of relationship
between topography and soil C under different land use system was also reported by Tan

et al. (2004) under cropland, grassland, and forestland in Ohio.

68

Table 3.3. Mean values of the terrain variables and soil texture values for the upperslope
and valley positions selected by stepwise discriminant procedure (PS 0.10) in the study
site.

 

 

Landform

Class N Rel. Elevation Flow WI Sand # Silt # Clay #
Upperslope 1 1 11 0.704 0.217 6.40 413 453 134
Valley 505 0.18 0.924 8.09 322 517 160

 

. . . 2 -l . .
N — number of samples; Elevation in meters ; WI : Wetness index, m m . # - g kg ' sorl

Flow : Log transformed flow accumulation calculated as accumulated weight for all cells that
flow into each downslope cell.

Table 3.4. Pearson correlation coefficients between the total C and N at 0- to 5-cm depth
with terrain attributes significant at 0.05 level.

 

 

 

TC (0-5 cm) TN (0-5 cm)

TRT Elev Slope Curv Flow WI Elev Slflre Curv Flow WI
CT NS -0.38 -0.13 0.17 0.38 -0.13 -0.37 -0.12 0.14 0.34
NT -0.24 -0.22 -0.12 0.16 0.26 -0.30 -0. 18 -0.1 l 0.15 0.23
CTcover -0.18 -0.18 -0.11 0.37 0.44 -0.16 -0.11 -0.11 0.33 0.36

 

NS - Not significant at P S 0.05

69

Terrain attributes explained about 26, 11 and 26 percent of variation in total C and
21, 12 and 21 percent of variation in total N in CT, NT and CT-cover treatments,
respectively (Table 3.5). Topographical models were significant in all the studied plots (p
< 0.05). At 0-5 cm depth, the relationship between total N and terrain attributes followed
the same trend as that for total C in all the three treatments. The amount of variability
explained by terrain attributes for total C and N was found to be relatively higher in
tillage treatments (CT and CT-cover) as compared to the NT treatment. Significant
positive contributions of WI to the regression equations for total C were observed in all
three treatments. Contributions of the other topographical variables were significant in
some but not other treatments.

Based on the correlation and regression results, I decided to run ANCOVA with
WI as a covariate. Several previous studies also used WI as a variable representing
topography when discussing topographical effects on total C (Moore et al., 1991, Kasper
et al., 2006). These ANCOVA results demonstrated significant interactions between
treatment effects and W1 (P<0.01) (Table 3.6). The regression slopes of both tillage
treatment i.e., CT-cover crop and CT for total C were significantly greater than that of the
NT treatment for total C (Table 3.6).

The regression slope for CT-cover WI was numerically higher than that of the CT,
though not significantly different (p<0.05). For total N, the regression slope estimate of
the organic based treatment (CT-cover) was significantly higher than that of both

chemical input treatments (CT and NT).

70

Table 3.5. Results of multiple regressions for total C, N at the 0 to 5cm depths with
terrain attributes (P _<_ 0.05)

 

Total C (0-5 cm) Total N (0-5 cm)

 

 

Variables CT NT CT cover CT NT CT cover
Intercept 0.465 0.88 0.56 0.062 0.094 0.05
Elevation-

Linear -0.047 NS NS NS -0.01 1 0.006
Quadratic -0.023 0.14 NS NS NS 0.02]
Slope -0.077 NS NS -0.01 1 NS NS
Flow NS NS NS 0.006 NS NS
Aspect 0.061 NS NS 0.005 0.01 NS
W1 0.033 0.036 0.06 NS NS 0.005
MSE 0.117 0.259 0.176 0.013 0.027 0.018
R 2 0.26 0.11 0.26 0.21 0.12 0.21

 

Elevation in meters; Slope in degrees; Curvature: Curvature of the terrain, 10 '2 m.
Aspect: Reclassified aspect South = 0 (135—225 degrees); North =1

. 2 -l .
WI : Wetness index, m m .; Flow : Log transformed flow accumulation calculated as
accumulated weight for all cells that flow into each downslope cell. ;

R 2 : Coefficient of determination for multiple regression, including only the variables significant
at 0.05 significance level.

71

Table 3.6. Linear regression slopes, intercepts, 12 and MSE resulting from ANCOVA of

total C, N at 0- to 5-cm with wetness index as covariate.

 

 

TRT N Slope (SE) Intercept (SE) R MSE
Total C

CT 485 0.015*a**(0.005) 0.658 (0.056) 0.14 0.13

NT 548 0.003 b NS(0.005) 1.164 (0.056) 0.07 0.26

CT cover 583 0.017*a (0.005) 0.902 (0.059) 0.19 0.19
Total N

CT 485 0.001NS(0.001) 0.068(0.007) 0.12 0.01

NT 535 0.0005NS (0.001) 0.101 (0.007) 0.06 0.03

CT cover 548 0.002" (0.001) 0.080 (0.007) 0.13 0.02

 

* Regression slopes are significantly greater than zero (a=0.05).

** Regression slopes in the equations followed by the same letter are not significantly different

from each other (01=0.05).

NS - Regression slope of NT is not significantly greater than zero

Table 3.7. Total C and N content at 0- to 5-cm for the upperslope and valley positions

 

 

 

in CT, NT and CT-cover.
TC (g €4ng soil) TN (g N kg " soil)
TRT Upperslope Valley Upperslope Valley
CT 7.44 a* 7.71 a* 0.69 a" 0.72 a*
NT 11.4c 11.5b 0.99b 1.03b
CT cover 9.94 b 10.6 b 0.89 c 1.01 b

 

*Means followed by same letter within each column are not significantly different at P<0.05.

72

Even though relationships similar to those observed here between terrain
attributes and soil C and N have been reported before (Moore et al., 1993; Moorman et
al., 2004; Terra et al., 2004), this present study highlights that the magnitudes of the
observed relationships were found to be different among the treatments. At 0-5 cm depth,
the terrain attributes appeared to affect C stronger in CT-cover and CT than in NT

treatments.

Among the landform classes, as expected, C and N contents in valleys were
substantially higher as compared to those in upperslope elements, Table 3.7. The
interactions between the landform elements and treatments for both C and N were not
statistically significant. Contrast estimation was carried out to better assess the
differences in C and N contents between upperslope and valley positions at different

treatments. Differences in C contents at 0-5 cm depths (Table 3.8) between upperslope

and valley was numerically greater with CT-cover (0.7 g kg '1 soil) followed by CT (0.3

g kg '1 soil) NT (0.1 g kg '1 soil). However, these differences were not significantly

different among the treatments. The above trend shows that at 0-5 cm depth, topography
tended to have a stronger effect in the CT-cover and CT treatments as compared to NT.
The greater numeric differences in total C in CT—cover and lower differences in NT
between upperslope and valley positions were in agreement with greater regression
slopes observed in ANCOVA with WI for CT-cover and CT and a smaller slope from

ANCOVA for NT.

73

Table 3.8. Total C and N difference between upperslope and valley landform positions
for CT, NT and CT-cover.

 

 

 

Total C* Total N*
Depth CT NT CT cover CT NT CT cover
g C kg.l soil g C kg.I soil
0-5 # 0.3 a" 0.1 a 0.7 a 0.03 a 0.04 a 0.12 a
20-30 0.2 a 4.2 b 2.0 b 0.00 a 0.34 b 0.14 b
Mg C ha.l soil
0-30 # # 1.6 a 12.3 b 5.2 b 0.25 a 1.0 b 0.36 b

 

* C, N values for 0-5 and 20-30 were measured in g kg '1 soil and for 0-30 cm in Mg ha '1 soil

IM'Means within each row followed by the same letter are not significantly different (0:0.05)
# C differences at CT (0-5 cm) = C at valley (0-5 cm) — C at upperslope (0-5 cm)

(Similarly differences in values are obtained for all treatments at all the three studied depth)

# # Soil C (Mg C ha '1 soil) = (SOC (g kg '1) / 10) x 10,000 (m 2) x depth x bulk density(Mg m '
3
)

Table 3.9. Pearson correlation coefficients between the total C and N at 20- to 30-cm
depth with terrain attributes significant at 0.05 level.

 

 

 

 

 

Two-30 cm) TN (20-30 cm)

TRT Elev Slope Curv Flow WI Elev Slope Curv Flow WI
CT NS -0.23 -0. 19 0.26 0.30 -0. 17 -0.24 -0.20 0.25 0.30
NT -0.52 -0.53 NS NS 0.38 -0.55 -0.54 NS NS 0.39
CTcover -0.32 -0.21 NS 0.43 0.47 -0. l 9 -0. 16 NS 0.41 0.43

 

NS - Not significant at P S 0.05

74

Similarly, for total N, the differences between the upperslope and valley positions

were found to be numerically greater with CT-cover (0.12 g kg '1 soil) followed by NT

(0.04 g kg '1 soil) and CT (0.03 g kg '1 soil) although the differences were not

significantly different among the treatments. Greater numeric differences in total N in
CT-cover and lower differences in CT and NT between upperslope and valley positions
were in agreement with greater slopes observed in ANCOVA with W1 for CT-cover and

an insignificant slope from ANCOVA for NT and CT.

Relationship Between Terrain Attributes and Soil C and N at 20-30 cm Depth

Similar to the results at 0-5 cm depth, Pearson correlation coefficients between
total C and N and the terrain attributes varied from treatment to treatment and correlation
coefficients with WI tended to be higher than those with the other terrain variables(Table
3.9). Unlike the results for 0-5 cm depth, terrain attributes explained larger portions of
variability in total C and N at CT-cover (34% and 21%) as compared to CT (9 % and
13%). The R2 values and regression coefficients from step-wise multiple regression for
soil C and N with topographical variables are shown in Table 3.10. The r2 values of NT
treatment was based on the 22 samples as compared to that of CT(158) and CT-
cover(182) which makes comparisons with the other two rather difficult. So for the 20-
30, I mainly focused on CT and CT—cover results.

Based on the ANCOVA results with WI as the covariate (Table 3.11), the
interaction between treatments and W1 for total C was found to be significant (P<0.01),
implying that relationships between WI and soil C at 20-30 cm depth were different

under different management systems. The regression slope for the CT-cover was

75

significantly greater than that of the CT treatment. In NT the regression slopes was not
significantly different from zero, most likely reflecting the small sample size. For total N
the regression slopes of CT-cover and CT were not significantly different.

Soil C and N contents at 20-30 cm depth for different landscape positions for CT,
NT and CT-cover are shown in Table 3.12 (P<0.05) and Fig. 3.2. There was a significant
interaction between the landform elements and treatments for both C and N. As expected,
C and N contents in valleys were higher than in upperslope areas. At upperslope
positions, there was no significant difference between total C and N of the CT, CT-cover
and NT treatments. However at valleys, CT-cover and NT had higher C and N than CT.

Contrast results carried out with the differences in C and N contents between
upperslope positions and valleys of each treatment are shown in Table 3.8. Unlike 0-5
cm, at 20-30 cm the difference in C content between upperslope and valleys was greater
with CT-cover as compared to the CT treatment. Differences in C and N contents
obtained between the two landforms were about ten-fold greater in CT-cover as
compared to CT (Table 3.8). This depicts that topography effects were more pronounced
at 20-30 cm depth as compared to 0—5 cm and much stronger in CT-cover compared to
CT. These relative gains in C and N content at 20-30 cm, while adopting cover crop
treatment over CT were significantly higher at upperslope positions compared to valleys.
The greater differences in total C and N with CT-cover over CT were in agreement with
greater regression coefficient values and greater slope from ANCOVA with WI for CT-

cover compared to CT.

76

- cr Cl cr Cover NT
1

15' '2‘ 20.30 cm
20-30 cm

b

   
 
 

  
 

///////////

/
Total N (Mg per ha)

    
  

/////

      
 

 

 

V//////////

Upperslope Valle

    

\

S
Upperslo

pe Valley

'<

Fig. 3.2. Total C and N content at 20- to 30-cm at upperslope and valley positions in CT,
NT and CT-cover. The same letters within each landform positions shows that treatments
are not significantly different from each other (p<0.05).

77

Table 3.10 Results of multiple regressions for total C, N at the 20- to 30-cm depth with
terrain attributes (P S 0.05)

 

 

 

Total C (20-30 cm) Total N (20-30 cm)
Variables CT NT CTcover CT NT CTcover
Intercept 0.30 0.77 0.30 0.03 0.06 0.04
Elevation—Linear NS NS 0.20 -0.001 NS NS
Quadratic NS NS 0.34 NS NS NS
Slope NS -0. 15 0.06 NS -0.02 NS
Flow NS NS NS NS 0.002 NS
Aspect NS NS NS NS 0.003 NS
WI 0.02 NS NS NS 0.003 NS
R 2 0.09 0.29 0.34 0.13 0.3 0.21

 

R2 : Coefficient of determination for multiple regression, including only the variables significant
at 0.05 significance level.

Elevation in meters; Slope in degrees; Curvature: Curvature of the terrain, 10 '2 m.
Aspect: Reclassified aspect South = 0 (135-225 degrees); North =1

WI : Wetness index, m 2 m '1; Flow : Log transformed flow accumulation calculated as
accumulated weight for all cells that flow into each downslope cell.

Table 3.11. Linear regression slopes, intercepts, r 2 and MSE resulting from ANCOVA of
total C, N at 20- to 30-cm with wetness index as covariate.

 

 

TRT N Slope(SE) lntercein (SQ R MSE
Total C (20-30 cm)

CT 158 0.014*(0.005) a" 0.206 (0.035) 0.09 0.07

NT 21 0.048NS(0.033) ab 0.111 (0.240) 0.15 0.30

CT cover 182 0.046 * (0.006) b 0.045 (0.054) 0.22 0.1.6
Total N (20-30 cm)

CT 158 0.0017*(0.001) a 0.021 (0.005) 0.18 0.01

NT 21 0.0043NS (0.003) a 0.022 (0.20) 0.15 NS 0.03

CT cover 182 0.0024“ (0.001) a 0.025 (0.006) 0.18 0.01

 

* Regression slopes were significantly greater than zero (01:0.05).

** Regression slopes in the equations followed by the same letter are not significantly different
from each other (a=0.05).

NS - Regression slope of NT is not significantly greater than zero

78

Table 3.12. Total C and N content at 20- to 30-cm at upperslope and valley positions in
CT, NT and CT-cover.

 

 

 

TC (g Gig '1 soil) TC (g C kg '1 soil)
TRT Upperslope Valley Upperslope Valley
CT 3.04 a* 3.22 a* 0.33 a* 0.33 3*
NT 2.87 a 7.05 b 0.33 a* 0.67 b
CT cover 3.34 a 5.29 b 0.34 a 0.48 b

 

*Means followed by same letter within each column are not significantly different at P<0.05.

Table 3.13. Pearson correlation coefficients between the total C and N at 0- to 30-cm
depth with terrain attributes significant at 0.05 level.

 

 

TC (0-30 cm) TN (0—30 cm)

TRT N Elev Slope Curv Flow WI Elev Slope Curv Flow WI
CT 151 -0.49 -0.68 -0.31 0.34 0.52 -0.45 -0.57 -0.24 0.26 0.42
NT 21 -0.59 -0.58 -0.41 NS 0.39 -0.56 -0.55 -0.45 NS 0.43
CT cover 167 NS -0.16 NS 0.43 0.44 NS -0. 14 NS 0.43 0.43

 

NS - Not significant at P S 0.05

79

Relationship between Terrain Attributes and Soil C and N at 0—30 cm depth

Total C and N stocks at 0-30 cm depth were strongly negatively correlated with
elevation and terrain slope; and positively correlated with WI, log flow accumulation and
curvature in all the three treatments (Table 3.13). For both C and N, correlations with
slope and W1 were found to be significant in all the three treatments. At 0-30 cm, the
proportion of variability explained by terrain attributes in multiple regression was higher
with CT compared to CT-cover (Table 3.14). No single variable was consistently
significant in multiple regression in all the three treatments. WI was used as a primary
terrain covariate in the subsequent ANCOVA analysis for consistency purpose.

There was no significant interaction between treatment effect and W1 for total C
and N (Table 3.15). Total C and N contents at different landscape positions for CT, NT
and CT-cover are shown in Table 3.16 (P<0.05) and Figure 3.3. The ANOVA results,
demonstrated a significant interactions among the landform elements and treatments for
both C and N. CT-cover had higher C and N than CT at both landscape positions. In
valleys, NT had higher C and N than CT but was not significantly different from CT-
cover. In upperslope positions for NT were not significantly different from either CT or
CT-cover. The differences in C content between upperslope positions and valleys at 0-30
cm were 1.55, 12.33, and 5.17 Mg/ha for total C and 0.25, 1.0, and 0.36 Mg/ha for total N

with CT, NT and CT-cover, respectively.

80

Table 3.14. Results of multiple regressions for total C, N at the 0- to 30-cm depth with
terrain attributes (P S 0.05)

 

 

 

Total C (0-30 cm) Total N (030 cm)
Variables CT NT CT cover CT NT CT cover
Intercept 0.68 0.85 0.78 0.068 0.088 0.075
Elevation -0.01 -0. 16 NS -0.002 -0.013 NS
Slope -0. 10 -0.045 -0.01 NS NS NS
Flow 0.032 NS NS NS NS 0.004
Aspect 0.03 NS NS NS NS 0.004
WI NS NS 0.094 0.002 NS 0.003
R 2 0.54 0.35 0.24 0.39 0.31 0.22

 

R 2 : Coefficient of determination for multiple regression, including only the variables significant
at 0.05 significance level.

Elevation in meters; Slope in degrees; Curvature: Curvature of the terrain, 10 '2 m.

Aspect: Reclassified aspect South = 0 (135-225 degrees); North =1

WI : Wetness index

Flow : Log transformed flow accumulation calculated as accumulated weight for all cells that
flow into each downslope cell.

81

    

8N\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\N

y
e
bl"
a
m V
C
0 e
a b p
O a m
m
e
p
p
U
_ A
5 4 3 2 1

I: CT-Cover- NT

-cr

 

 

   

8N\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\N

 

V.
m
w
m
C e
0
e m
o n
0
D.
D.
U
a a _ _ _
o 2 4 6 8 0
5 4 3 2 1 1
$52. 250 .83

Fig. 3.3 Total C and N content at 0- to 30-cm at upperslope and valley positions in CT,

NT and CT-cover. The same letters within each landform positions shows that treatments

are not significantly different from each other (p<0.05).

82

Table 3.15. Linear regression slopes, intercepts, r 2 and MSE resulting from AN COVA of
total C, N at 0- to 30-cm with wetness index as covariate.

 

 

TRT N Slope(SD Intercem(SE) R MSE
Total C

CT 151 0.021* (0.005) a" 0.430 (0.053) 0.27 0.08

NT 21 0.038NS (0.019) a 0.566 (0.137) 0.15NS 0.25

CT cover 167 0.023 * (0.009) a 0.649 (0.072) 0.19 0.14
Total N

CT 151 0.002* (0.001) a" 0.044 (0.007) 0.18 0.01

NT 21 0.004NS (0.003) a 0.058 (0.014) 0.19NS 0.02

CT cover 167 0.003“ (0.001) a 0.064 (0.007) 0.18 0.01

 

* Regression slopes are significantly greater than zero (01=0.05).

** Regression slopes in the equations followed by the same letter are not significantly different
from each other (0r=0.05).

NS - Regression slope of NT is not significantly greater than zero

Table 3.16. Total C and N content at 0-30 cm at upperslope and valley positions in CT,
NT and CT-cover.

 

 

 

TC Mglha (0-30 cm) TN Mg/ha (0-30 cm)
TRT Upperslope Valley Upperslope Valley
CT 26.74 a“ 28.29 a 2.59 a 2.84 8
NT 31.35 ab 43.68 b 3.43 ab 4.43 b
CT cover 32.24 b 37.41 b 3.32 b 3.68 b

 

*Means followed by same letter within each column are not significantly different (P<0.05)
TC (Mg C ha ") = (soc (g C kg " soil) /10) x 10,000 (m 2) x depth x bulk density(Mg m '3)

83

For both C and N, the differences with CT-cover and NT were significantly
greater than those with the CT treatment. Thus the results at 0-30 cm indicate that
topography produced a greater effect on total C and N in CT-cover than in the CT
treatment and the potential benefits of adopting cover crops were found to be greater in

the valleys compared to upperslope positions.

DISCUSSION

In this study, at the surface depth (0-5 cm), the relationship between topography
and total C were regarded to be affected primarily by tillage disturbance. Total C and N
contents were strongly related to topography in the tillage treatments, CT-cover and CT,
than in NT as indicated by higher multiple regression coefficients and greater regression
slopes from ANCOVA with WI. This effect can be attributed to the greater influence of
moisture and particle redistribution that occurs under constant soil disturbance by tillage
at 0-5 cm depths. It has been shown in a number of studies that over a time, tillage leads
to soil losses or displacements from upper landscape positions and its accumulation at
lower landscape positions (Li and Lindstorrn, 2001; Ritchie et al., 2004, Kaspar et al.,
2006). After conversion to no-till, topography influence on soil erosion and particle
redistribution can be expected to be greatly reduced due to lower soil disturbance,
increased water permeability and a larger amount of crop residues remaining on the
surface.

Moreover it has been noted that after implementation of NT, landforms
susceptible to erosion, such as upperslope regions, may also improve in soil quality due

to increased moisture retention characteristics and improved soil structure, leading to

84

greater C accumulations along the downsloped positions within the landscape

(V andenBygaart et al. 2002). Both reductions in topography-driven material transport and
improved soil C sequestration at upperslope positions were consistent with the observed
weaker relationship between soil C and topography in NT at 0-5 cm depth, including
lower correlation and regression coefficients; lower slope from ANCOVA with WI; and
lower difference between upperslope and valley soil C values.

The greater regression slope between total N and W1 from ANCOVA in CT-cover
probably reflects the effect the cover crop. CT-cover is a zero chemical input treatment,
where soil nitrogen inputs come mainly from decomposition of crop and cover crop
residues. The presence of cover crops affects N processes as compared to CT and NT due
to differences in quantity and quality of residues and differences in plant rooting pattern
and activities (Power et al., 1998; Villarnil et al., 2006). Red clover and rye residues have
narrower C/N ratio (11 to 37) than main crop residues (57 to 80) (Brady and Weil.,
2002). Cover crop residues with narrow CN ratio provide a positive priming effect on N
mineralization rate as well as an increase in microbial biomass turnover in the rhizophere
by plant rhizodeposition (Kuzyakov et al., 2000) and symbiotic N-fixation from red
clover. In the studied area, water redistribution along the landscape and water availability
to plants affects the cover crop performance along with that of the main crops. Thus, in
dry years lower located sites have a substantially greater cover crop stand than
upperslope areas. The cover crop stand within the studied landscape is strongly controlled
by topography driven redistribution of soil moisture, reflected in WI (Munos et al., 2008),
thus explaining why the regression slope for N relationship with WI at CT-cover was

greater than those of CT and NT. The smaller regression slopes in CT and NT may be

85

due to the fact that, N was supplied via chemical fertilizers, thus uniform external N
fertilization intensity prevailed over topography effect. Moreover, among the landform
elements valleys had numerically higher C and N compared to upperslopes in CT and
CT-cover. These higher levels may be due to the fact that, in a landscape position, valleys
are generally fine textured (Table 3.2) and subject to less erosion compared to
upperslopes (Franzluebbers, 2004). Furthermore C loss due to soil respiration which is
speculated to be higher during spring from the drier, coarse textured soil of the upper
slope positions compared to wetter, colder fine textured soils at footslope positions
(Parkin and Kaspar, 2003, Kasper et al., 2006) contributes to the higher C and N in
valleys relative to upperslopes.

At 20-30 cm, relationships between topography and total C and N were much
stronger with CT-cover than CT. Since in a chisel plow system the tillage zone lies within
the 0-20 cm depth, at 20-30 cm the C and N contents can be hypothesized to be driven by
quantity and quality of below ground biomass inputs rather than by the tillage disturbance
gradient. The stronger relationship between C, N and topography with CT-cover
compared to CT may be attributed due to following mechanisms. First, with CT-cover, as
mentioned earlier the amount of cover crop residue inputs is strongly controlled by the
topography driven redistribution of soil moisture, resulting in stronger relationship
between topography and soil C. Moreover, upperslopes were characterized by coarser
textured soil (Table 3.3) with unfavorable moisture conditions with low inherent organic
matter content, which might have restricted rooting pattern and activities (V illamil et al.,
2006) to a much shallower depth. This would produce lower belowground biomass and C

inputs at 20-30 cm in upperslopes compared to dense and deeper rooting pattern with

86

more C inputs in valley positions which explain the lack of differences in carbon content
between CT-cover and CT at upperslope positions. Fine textured soils in the valleys have
a higher inherent level of macroaggregates inspite of heavy soil disturbance due to the
cohesive nature of the reactive clays (Franzluebbers, 2004). The additional biomass with
a narrow CN ratio that comes from winter cover crop roots and weeds in CT—cover might
have enhanced the water stable aggregates and aggregation process resulting in more
carbon and nitrogen sequestration with less soil C and N losses. CT-cover tends to have
lower bulk density values at both 0-5 and 20-30 cm depths even though it was not
statistically significant over CT and NT, and the lower bulk density in the cover crop
treatment might have resulted in enhanced total porosity, which leads to increased
macropores and micropores. Increases in macropore and micropores reduces moisture
tension and increases plant available water capacity resulting in better crop stands and
subsequently more C inputs in CT-cover. Similar results in a chisel plowed treatment
with winter cover crops were reported by Villarnil et al. (2006).

At 0-30 cm depth (Table 3.15), CT-cover and NT were reported to accumulate 5.5
and 4.61 Mg per ha more C than CT on the upperslopes and 9.12 and 15.39 Mg per ha
more C than CT in the valleys. CT-cover and NT were found to retain 0.73 and 0.84 Mg
per ha more N than CT in the upperslopes and 0.84 and 1.59 Mg per ha more N than CT
in the valleys. Other studies reported a winter cover crop effect resulting in more soil
mineral nitrogen content at 0-30 cm depth compared to chisel till treatment without cover
crop Varco et al ., (1987) and Sainju and Singh (2001). Similar results in an organic
based tillage system that include winter cover crops were reported to be as effective in

gaining organic C as NT despite the heavy soil disturbance by

87

tillage (Teasdale et al., 2007). In this study the site topography effect was found to be
stronger with CT-cover treatment compared to CT and NT and; the CT-cover treatment
and NT system were more efficient in restoring C and N at 0-30 cm than the CT system.
The potential benefits of adopting cover crop treatment found to be greater at valleys

compared to upperslope positions.

SUMMARY AND CONCLUSIONS

Tillage and management practices significantly interacted with topographical
features as represented by WI with both C and N. Topography was found to explain
larger proportion of variation in C at organic based tillage treatment compared to the
chemical input based no-till (NT) and chisel plow (CT) systems. Among the landform
elements, C and N content was found to be greater at valley areas compared to
upperslope position in all the studied treatments. At 0-30 cm depth, CT-cover and NT
was found to accumulate 2.2 and 1.81 Mg more C per acre than CT at uphill position and
3.7 and 6.2 Mg per acre more C than CT in the depressions. In overall, CT-cover was
found to be equally efficient in maintaining C and N content as compared to NT
treatments. Largest C and N benefits of CT-cover in the landscape were observed in
valleys as compared to uphill position in the landscape. This may be attributed to greater
gradients in spatial distribution of plant nutrients in organic management as well as
greater influence of moisture redistribution through the landscape, which in CT-cover
affects not only the main crop but as well performance of the cover crops. The results
indicate that CT-cover has greater potential for C sequestration at lower landscape

positions.

88

Acknowledgements:

I thank Andrew Corbin, Suzanne Sippel and other LTER staff, for their support
during geo-referencing sample locations and soil sampling. I also thank Matias Ruffo and
Ignacio Colonna from Dr.Don Bullock lab at University of Illinois, for their help on

collecting elevation data.

89

REFERENCES

Allmaras, R.R., S.M. Copeland, P.J. Copeland, and M. Oussible. 1996. Spatial relations
between oat residue and ceramic spheres when incorporated sequentially by tillage.
Soil Sci. Soc. Am. J. 60:1209-1216

Bakhsh, A., R.S. Kanwar and R.W. Malone. 2007. Role of landscape and hydrologic
attributes in developing and interpreting yield clusters. Geoderrna 140: 235-246.

Balesdent, J ., C. Chenu and M. Balabane. 2000. Relationship of soil organic matter
dynamics to physical protection and tillage. Soil Till. Res. 53: 215-230

Beare, M.H., P.F. Hendrix, and DC. Coleman. 1994a. Water-stable aggregates and

organic matter fractions in conventional and no-tillage soils. Soil Sci. Soc. Am. J.
58:777—786.

Beare, M.H., M.L. Cabrera, P.F. Hendrix, and DC. Coleman. 1994b. Aggregate-
protected and unprotected pools of organic matter in conventional and no-tillage
soils. Soil Sci. Soc. Am. J. 58:787—795.

Bell, J .C., R.L. Cunningham, and M.W. Havens. 1994. Soil drainage probability mapping
using a soil-landscape model. Soil Sci. Soc. Am. J. 58:464-470.

Bergstorm, D.W., C.M. Monreal, and E. St. Jacques. 2001. Spatial dependence of soil

organic carbon mass and its relationship to soil series and topography. Can. J. Soil
Sci. 81 :53—62

Blevins, R.L., J .H. Herbeck, and W.W. Frye. 1990. Legume cover crops as a N source for
no-till corn and grain sorghum. Agron. J. 82:769—772

Brady, NC, and RR. Weil. 2002. The nature and properties of soils. 13th ed. Pearson
Education, Inc.

Coote, DR. and Malcolm-McGovem, C.A., 1989. Effects of conventional and no-till
corn grown in rotation on three soils in Eastern Ontario. Soil Till. Res. 14: 67—84.

Creed, I.F., C.G. Trick, L.E. Band, I.K. Morrison. 2002. Characterizing the spatial
heterogeneity of soil carbon and nitrogen pools in the Turkey Lakes Watershed: A
comparison of regression techniques. Water Air Soil Poll. FOCUS 2: 81-102.

DeGryze, S., J. Six, K. Paustian, S.J. Morris, E.A. Paul, and R. Merckx. 2004. Soil
organic carbon pool changes following land-use conversions. Glob. Change Biol.
10:1120—1132.

Environmental Systems Research Institute. 2004. ArcGIS Spatial Analyst. ESRI,
Redlands, CA.

F lorinsky, I.V., R.G. Eilers, G.R. Manning, and LG. Fuller. 2002. Prediction of soil
properties by digital terrain modelling. Environ. Modell. Softw. 17:295-311

90

Franzluebbers , K., R.W. Weaver, A.S.R. Juo, and A.J. Franzluebbers. 1994. Carbon and
nitrogen mineralization from cowpea plants part decomposing in moist and in
repeatedly dried and wetted soil. Soil Biol.Biochem. 26: 1379-1387

Franzluebbers, A.J. 2004. Tillage and residue management effects on soil organic matter.
p. 227—268. In. F. Magdoff and RR. Weil (ed.) Soil organic matter in sustainable
agriculture. CRC Press. Boca Raton, FL.

Follett, RP. 2001. Soil management concepts and carbon sequestration in cropland soil,
Soil Tillage Res. 61: 77—92.

Gallant, J .C., Wilson, J .P., 2000. Primary topographic attributes. In:Wilson, J .P., Gallant,

J .C. (Eds.), Terrain Analysis: Principles and Applications. Wiley, New York, pp.
51—86.

Grandy, A. S., and G. P. Robertson. 2006. Aggregation and organic matter protection
following rillage of a previously uncultivated soil. Soil Sci. Soc. Am. J. 70: 1398-
1406

Grandy, A. S., and G. P. Robertson. 2007. Land use intensity effects on soil C
accumulation rates and mechanisms. Ecosystems 10: 59-74

Hao, Y., R. Lal, L.B. Owens, R.C. Izaurralde, W.M. Post, and BL. Hothem. 2002. Effect
of cropland management and slope position on soil organic carbon pool at the North
Appalachian Experimental Watersheds. Soil Tillage Res. 68:133—142

Heckrath, G., J. Djurhuus, T.A. Quine, K. Van Oost, G. Govers and Y. Zhang, 2005 :
Tillage erosion and its effect on soil properties and crop yield in Denmark, J.
Environ. Qual. 34: 312—324.

Hemanz, J .L., R.Lopez, L. Navarrete, and V. Sénchez-Girén. 2002. Long-term effects of
tillage systems and rotations on soil structural stability and organic carbon
stratification in semiarid central Spain. Soil Till. Res. 66:129—141.

Huang, X., S. Senthilkumar, A. Kravchenko, K. Thelen and J. Qi 2004 Total carbon
mapping in glacial till soils using near-infrared spectroscopy, Landsat imagery and
topographical information. Geoderrna. 141: 34-42.

Jenkinson, D.S., D.D. Harkness, E.D. Vance, D.E. Adams, and AF. Harrison. 1992.
Calculating net primary production and annual input of organic matter to soil from

the amount and radiocarbon content of soil organic matter. Soil Biol. Biochem. 24:
295—308.

Johnson, D. E, 1998. Applied Multivariate Methods for Data Analysts. Duxbury Press.

91

Kaspar, T.C., T. B. Parkin, D. B. Jaynes, C. A. Cambardella, D. W. Meek, and Y. S.
Jung. 2006. Examining Changes in Soil Organic Carbon with Oat and Rye Cover
Crops Using Terrain Covariates. Soil Sci Soc Am J 70: 1168-1177.

Kellogg Biological Station. 2005. Long-terrn ecological research in row-crop agriculture.
Available at www.kbs.msu.edu/lter (accessed 18 July 2006; verified 11 Oct. 2006).
KBS, Hickory Comers, Ml.

Khattree, R., and D.N. Naik. 2000. Multivariate data reduction and discrimination with
SAS software. SAS Institute, Cary, NC.

Kosmas, C., S. Gerontidis, M. Marathianou, B. Detsis, T. Zafiriou, W. Van Muysen, G.
Govers, T. Quine and K. Van Oost 2001. The effects of tillage displaced soil on soil
properties and wheat biomass, Soil Tillage Res. 58: 31-44.

Kravchenko, A.N., and D.G. Bullock. 2000. Correlation of grain yield with topography
and soil properties. Agron. J. 92:75-83.

Kravchenko, A.N., G.A. Bollero, R.A. Omonode, and D.G. Bullock. 2002. Quantitative
mapping of soil drainage classes using topographical data and soil electrical
conductivity. Soil Sci. Soc. Am. J. 66:235—243

Kravchenko, A.N., G.P. Robertson, X. Hao, and D.G. Bullock. 2006. Management
practice effects on surface total carbon: Differences in spatial variability patterns.
Agron. J. 98:1559—1568.

Li, Y., and M.J. Lindstrom. 2001. Evaluating soil quality—soil redistribution relationship
on terraces and steep hillslope. Soil Sci. Soc. Am. J. 65:1500—1508.

Macmillan, R. A., W.W. Pettapiece, S. C. Nolan and T. W. Goddard. 2000. A generic
procedure for automatically segmenting landforms into landform elements using
DEMs, heuristic rules and fuzzy logic. Fuzzy Sets and Systems 113: 81—1 10

McVay, K.A., D.E. Radcliffe, and W.L. Hargrove. 1989. Winter legume effects on soil
properties and N fertilizer requirements. Soil Sci. Soc. Am. J. 53:1856—1862.

Milliken, G.A., and DE. Johnson. 1992. Analysis of messy data: Designed experiments.
CRC Press, Boca Raton, FL.

Moore, I.D., R.B. Grayson, and A.R.Ladson. 1991. Digital terrain modeling: A review of
hydrological, geomorphological and biological applications. Hydrol. Processes 5: 3-
30.

Moore, 1. D., P. E. Gessler, G. A. Nielsen, and G. A. Peterson. 1993. Soil attribute
prediction using terrain analysis. Soil Sci. Soc. Am. J. 57:443-452.
Moorman, T.B., C.A. Cambardella, D.E. James, D.L. Karlen, and LA. Kramer. 2004.

Quantification of tillage and landscape effects on soil carbon in small Iowa
watersheds. Soil Tillage Res. 78:225—236.

92

Mokma, D.L., and J .A. Doolittle. 1993. Mapping some loamy Alfisols in southwestern
Michigan using ground-penetrating radar. Soil Surv. Horiz. 31 :71—78.

Mueller, T.G., and F.J. Pierce. 2003. Enhancing spatial estimates of soil carbon with
terrain attributes. Soil Sci. Soc. Am. J. 67:258—267.

Munos, J .D., R. Gehl. S. Snapp, and AN. Kravchenko. 2008. Identifying the factors
affecting cover crop performance in row crops. To be presented at 9th International
conference on precision agriculture, Denver, Colarado, USA July 20-23.

Ovalles, F .A., and ME. Collins. 1986. Soil—landscape relationships and soil variability in
North central Florida. Soil Sci. Soc. Am. J. 50:401-408

Papiemik, S.K., M.J. Lindstrom, T.E. Schumacher, J .A. Schumacher, D.D. Malo and
DA. Lobb 2007 Characterization of soil profiles in a landscape affected by long-
term tillage Soil and Tillage Research. 93(2): 335-345

Parkin, TB, and TC. Kaspar. 2003. Temporal variability of soil C02 flux: Effect of
sampling frequency on cumulative carbon loss estimates. Soil Sci. Soc. Am. J.
68:1234—1241.

Pennock, D.J., and MD. Corre, 2001. Development and application of landform
segmentation procedures. Soil & Tillage Research 58: 151—162.

Pennock D.J., and E. de Jong. 1990. Spatial pattern of soil redistribution in boroll
landscapes, southern Saskatchewan, Canada. Soil Sci.1502867-873.

Power J.F., P.T. Koemer, J.W.Doran, W.W. Wilhelm 1998. Residual effects of crop
residues on grain production and selected soil properties. Soil Sci. Soc. Am. J. 62:
1393-1397.

Reicosky, D.C., W.A. Dugas, and HA. Torbert. 1997. Tillage-induced soil carbon
dioxide loss for different cropping systems. Soil Till. Res. 41 :105—1 18

Ritchie, J.C., G.W. McCarty, E.R. Venteris, T.C. Kaspar, L.B. Owens, and M. Nearing.
2004. Assessing soil organic carbon redistribution with fallout I37Cesium. In
Conserving soil and water for society: Sharing solutions. Proceedings of the ISCO

2004—13th International Soil Conservation Organization Conference, Brisbane,
Australia, 4—8 July 2004. Paper No. 613, 4 pages. (CD-ROM)

Robertson, G.P., K.M. Klingensmith, M.J. Klug, E.A. Paul, J .C. Crum, and BC. Ellis.
1997. Soil resources, microbial activity, and primary production across an

agricultural ecosystem. Ecol. Appl. 7: 1 58—170.

Sainju, U.M., and B.P. Singh. 2001. Tillage, cover crop, and kill-planting date effects on
corn yield and soil nitrogen. Agron. J. 93:878—886.

SAS Institute. 2001. SAS user's guide. Version 9.1. SAS Inst., Cary, NC.

93

Schabenberger, O., and F.J. Pierce. 2002. Contemporary statistical models for the plant
and soil sciences. CRC Press, Boca Raton, FL.

Seybold, C. A., M. J. Mausbach, D. L. Karlen, and H. H. Rogers. 1997. Quantification of
soil quality. In:Soil processes and the carbon cycle (ed. R. Lal), pp. 387-404. CRC
Press LLC, Boca Raton, FL.

Six, J ., K. Paustian, E.T. Elliott, and C. Combrink. 2000a. Soil structure and organic
matter: 1. Distribution of aggregate-size classes and aggregate-associated carbon.
Soil Sci. Soc. Am. J. 64:681—689.

Six, J ., E.T. Elliott, and K. Paustian. 2000b. Soil macroaggregate turnover and
microaggregate formation: A mechanism for C sequestration under no-tillage
agriculture. Soil Biol. Biochem. 32:2099—2103.

Skidmore, A.K., P.J. Ryan, W. Dawes, D. Short, and E. O'loughlin. 1991. Use of an
expert system to map forest soils from a geographical information system. Int. J.
Geographical Information Syst. 5:431—445.

Smith, K. A. 1999. After the Kyoto protocol: Can soil scientists make a useful
contribution? Soil Use Manage. 15: 71-75

Tan, Z.X., R. Lal, N.E. Smeck and PG. Calhoun. 2004 Relationship between surface soil
organic carbon pool and site variables. Geoderrna 121: 187—195.

Teasdale, J .R., C B. Coffman, and R.W. Mangum. 2007 Potential Long-Term Benefits of
No-Tillage and Organic Cropping Systems for Grain ProduCtion and Soil
Improvement Agron. J. 99: 1297-1305.

Terra, J .A., J .N. Shaw, D.W. Reeves, R.L. Raper, E. van Santen, and PL. Mask. 2004.
Soil carbon relationships with terrain attributes, electrical conductivity, and a soil
survey in a Coastal Plain landscape. Soil Sci. 169:819—83 1.

VandenBygaart, A. J ., X. M. Yang, B. D. Kay, and J. D. Aspinall. 2002. Variability in
carbon sequestration potential in no-till soil landscapes of southern Ontario. Soil
Tillage Res. 65:231-241.

Varco, J.J., W.W. Frye, M.S. Smith, and J .H. Grove. 1987. Legume nitrogen
transformation and recovery by com as influenced by tillage. p. 1—40. In J .F. Power
(ed.) The role of legumes in conservation tillage systems. Soil Conserv. Soc. of

Am., Ankeny, IA.
Villamil, M.B., G.A. Bollero, R.G. Darmody, F.W. Simmons, and Bullock D.G., 2006

No-Till Corn/Soybean Systems Including Winter Cover Crops: Effects on Soil
Properties. Soil Sci. Soc. Am. J. 2006 70: 1936-1944.

94

Yoo K., R. Amundson, A.M. Heimsath, E. Dietrich. 2006. Spatial patterns of soil organic
carbon on hillslopes: Integrating geomorphic processes and the biological C cycle
Geoderma.130: 47-65

Yimer, F., S. Ledin and A. Abdelkadir. 2006. Soil organic carbon and total nitrogen
stocks as affected by topographic aspect and vegetation in the Bale Mountains,
Ethiopia Geoderrna. 135: 335-344

Ziadat, F M. 2005 Analyzing Digital Terrain Attributes to Predict Soil Attributes for a
Relatively Large Area Soil Sci. Soc. Am. J69: 1590-1599.

95

Chapter IV
CONTEMPORARY EVIDENCE FOR SOIL C LOSS UNDER DIFFERENT
CROP MANAGEMENT SYSTEMS IN A LONG TERM EXPERIMENT IN THE
US CORN BELT
ABSTRACT

Monitoring soil carbon (C) plays an important role in understanding the influence
of different land uses and management practices on C sequestration. Changes in C
content vary across landscape as a result of complex interactions among different factors
including historical soil C (baseline) levels, soil texture, topographical attributes, and land
use and management practices. Recent changes in global temperature patterns may
enhance or mediate the changes in soil C by interacting with different land use and
management practices. In this study, first I reported the changes in soil total C contents
that occurred in the past 15-20 years in soils under agricultural management and in non
agricultural soils of Southern Michigan, USA, and, second, the relationships between
these changes with soil properties such as baseline C levels, soil texture and
topographical features was explored.

The data were collected from two long-term experiments namely; Long Term
Ecological Research (LTER) Interaction site and Main site established in 1986 and 1988,
respectively. Geo-referenced samples were collected from both experiments prior to
establishment and then the locations were resampled in 2006-2007. Studied agricultural
treatments included conventional chisel plowed management with and without N
fertilizing, conventional no-till management with and without N fertilizing, and organic

chisel plowed management. The non-agricultural treatment was a never tilled soil kept at

96

present under grass vegetation. Total C in the conventional chisel plowed fertilized
management system has decreased by 1.15 g/kg and 0.22 g/kg at main site and interaction
site respectively. At the interaction site within the conventional tillage system, C losses
were found to be greater with unfertilized (1.54 g/kg) compared to fertilized (0.22 g/kg)

treatments.

In comparing management practices at Main site, soil C under no-till (9.9 g kg '1
of soil) management and under organic management (9.8 g kg 4 of soil) with cover crops

tends to be higher than that of the conventional management (8.2 g kg '1 of soil) however;

it has not changed as compared to 1988. That is, the conservation management practices
appeared to have prevented total C loss as compared to conventional management but

they did not lead to gains in absolute amount. Likewise, total C in the never tilled
grassland soil has decreased by 7.4 (g kg '1 of soil) since 1986. Increased decomposition

rate as a result of winter warming in Michigan might be one of the major factors that
drive soil C losses in this study site. Baseline C was found to have a negative relationship
with changes in C content and C loss was found to increase with increase in baseline
values. Increase in silt and clay content leads to positive changes (gain in C) in C content
in no-till system. Topographical effect was not found to have a significant relationship
between changes in C content, probably due to small sample size and low variations in
topography. In summary, none of the studied treatments showed an increase in C content

from the baseline values during the past two decades.

97

INTRODUCTION

Soil carbon is considered one of the major reservoirs for the carbon stocks on
earth. It holds about 1550 Gt as organic carbon (th = 10 '5 g) and 950 Gt as inorganic C

(Lal, 2004). Currently there is growing concern that rising global temperature might
possibly affect C content in the soil and influences the global C cycle (Bellamy et al.,
2005). Reduction in global soil C stock due to global warming phenomenon has been
already addressed by number of authors (J enkinson et al., 1991; Schimel et al., 1994;
Jones et al., 2005). If the rising global temperature were to cause even a little increase in
the release of stored organic carbon as C02 to the atmosphere it may further intensify the
present global warming potential.

Recently, Bellamy et al. (2005) and Schipper et al. (2007) reported widespread
soil C losses across all land uses, soil types and management systems when comparing

the contemporary C values with the baseline C values. Bellamy et al. (2005) reported C
losses of about 0.12 kg C m '2 yr '1 with a net C loss of about 15 g kg '1 (~2.93 kg C m '2)

across England and Wales during the past 25 years. The study was based on the data

obtained from more than 2000 sites sampled at 0-15 cm depth. Similarly Schipper et al.
(2007) reported soil C losses of about 2.1 kg C m '2 during last 20 years based on 31

whole soil profiles sampled at 1 meter depth in the pasturelands of New Zealand. The
findings of Bellamy et al. (2005) and Schipper et al. (2007) of losses in soil C occurring
at a wide range of soils, land uses and management practices and appearing to accelerate
over past couple of decades point to possible effect of the recent increases in global

temperature.

98

 

Increase in global temperature affects soil C by influencing soil organic matter
decomposition rate (Knorr et al., 2005), soil and root respiration (Kirschbaum, 2004;
Jones et al., 2005) and mineralization rate (Davidson et al., 2000). Data analysis based on
12 soil warming experiments showed that an increase in soil temperature by up to 3°C
increased soil respiration rate by 20 %, mineralization rate by 46 % and the plant
productivity by 19 % (Rustad et al., 2001; Richter et al., 2007). Rising global temperature
effect on soil C losses has been reported to be higher in temperate regions as compared to
C stored in the warmer regions, since the temperature sensitivity of organic matter
decomposition decreases with increase in temperature (Kirschbaum, 1995, 2000). For
instance, increase in global temperature by 1° C could lead to a loss of over 10 percent of
soil C in regions with an annual mean temperature of 5° C whereas the same temperature
increase in tropical climate could lead to C loss of only 3 percent (Kirschbaum, 1995).

On a regional scale, Michigan’s mean winter temperature was found to be
increasing during the past few decades. Global warming effect on soil C changes at
Michigan was earlier predicted with equilibrium analysis using Century model by
Paustian et al. (1995). In this context, changes in temperature patterns may enhance or
mediate the changes in soil C by interacting with different land use and management
practices. For monitoring the dynamics of soil C in agricultural soils many long term
studies have been established under different management practices. In the following

sections, I discussed land use and management driven changes in soil C content.

99

Land Use Control on Soil C Changes

Among the land use induced C changes, conversion from natural vegetation to
arable agriculture leads to a rapid depletion of soil carbon. Major decline in soil C content
was reported at Morrow Plots (established at 1876) in Illinois and Sanbom field
(established at 1888) at Columbia on conversion from native prairie to arable agriculture
(Huggins et al., 1998). Global meta data analysis of the studies carried out under different
climatic regions and soil types revealed that land use conversion from forest to

agricultural crops results in average loss of C up to 24% (Murty et al., 2002).

Management Control on Soil C Changes
Management induced soil C changes depend on factors such as tillage

disturbance, amount of crop residue return, cropping pattern and cropping intensity.

Tillage

Conventional tillage system affects soil C by disrupting the soil structure and
aggregation, influencing soil microclimate and finally mixing of crop residues within the
soil matrix (Balesdent et al., 2000). No till (NT), where the soil disturbance is minimal
provides an overall positive effect on soil C sequestration compared to conventional
tillage system. Global meta data analysis based on results compiled from long term
studies revealed that the adoption of NT leads to an increase in soil C as compared to
conventional tillage treatments (Franzluebbers, 2004; Puget and Lal, 2005). However to
maintain or enhance the soil C storage, NT has to be maintained continuously without

tillage (Grandy et al., 2006).

100

 

Crop residues

Crop residues are the main source of C input into the soil. The quantity and
quality of crop residue returned within a management system varies according to the
cropping pattern, cropping intensity and crop rotation. Many long-term studies reported
gradual increase in soil C with increase in C inputs across treatments (Paustian et al.,
1997). Removal of crop residues results in a decline in soil organic matter content.
Huggins et al., (1998) reported a drastic decline in C content in Morrow Plots at Illinois
and Sanbom field plots as a result of removal of crop residues during the initial 50 years
of the experiment. However losses were slightly reverted after the initiation of annual
return of straw and stover during 1950s and 1960s. An increase in C content was
observed after continuously returning the crop residues for 17 years at Sanbom plots and

at Morrow plots the loss was slower or leveled off after beginning of residue returns.

Organic based systems

Organically managed tillage systems are reported to accumulate more SOM
content compared to the conventional tillage systems (Pulleman et al., 2000; Stockdale et
al., 2001). Marriott and Wander (2006) reported a 14 % increase in soil organic C in
organically managed CT when compared with the conventional tillage systems in a 10
years period. Similarly Teasdale et al., (2007) reported greater C contents at chisel plow
based organic cover crop system compared to no-till system in a nine year old study
under corn soybean wheat rotation conducted in a coarse and fine loamy Typic Hapludult
soils. Based on KBS-LTER results, Robertson et al. (2000) reported that C levels at

organic based cover crop system in the KBS-LTER were midway between the

101

conventional and no till systems 10 years after treatment initiation. Differences in C
content between organic based CT and conventional CT is mainly driven by quantity and
quality of organic inputs (Stockdale et al., 2002). Inclusion of winter cover crops (WCC)
in the crop rotation increases the amount and diversity of biomass returned to the soil.
Leguminous cover crop residues with narrow CN ratio and rich in protein promote
mycorhiza which in turn results in increased microbial communities, production of
polysaccharides and aggregate stabilization (Haynes and Beare 1996).

General land use and management practices effect on soil C related processes are
likely to differ in magnitude under different baseline C contents and soil textures. Local
variations in soil properties such as changes in baseline C content and soil texture interact

with different land use and management practices may enhance or mediate the changes in

soil C.

Relationship Between Changes in C and Baseline C

Baseline soil C levels might be one of the critical factors influencing the rate of
change in soil carbon under any land use or management practice (Jansen et al., 1998;
Bellamy et al., 2005). For instance, VandenBygaart et al. (2002) observed negative
relationship between the baseline C content and the change in C afier 15 years of no-till
treatment. No-till was found to generate a greater increase in soil C at eroded locations
with lower baseline soil C than in the depositional areas with higher baseline C. Puget
and Lal (2005) reported no change in C content in a NT experiment carried out in a soil

which has high baseline C and high clay content.

102

Relationship between baseline C levels and the change in C content might be
controlled by soil’s saturation with C. C saturation capacity is defined as the maximum
amount of C that can be sequestered by a soil under specific climatic and management
conditions (Hassink 1997; Six et al., 2002). Saturation capacity depends on the sizes of
protected and unprotected soil C pools. In soils with lower baseline carbon content, the
sites able to protect soil C might be under-saturated with C. In this case adoption of NT
or a high C input system will show greater increase in C content, which is supported by
VandenBygaart et al. (2002) data. At higher baseline carbon content, most of the sites,
where C could be protected are saturated and additionally a lot of C is stored in
unprotected forms as light fractions and POM. This limits the increase in protected C
pool which is supported by Puget and Lal (2005) experiment. The amount of unprotected
C being held at equilibrium is affected by environmental conditions for decomposition
which prevents substantial increases in C storage in unprotected forms. Limited or no
increase in soil C with increased organic residue inputs observed in soils with relatively
high baseline C in several experiments support the saturation capacity hypothesis
(Hassink and Whitmore,1997; Six et al., 2002).

Soils with both low and high baseline C values on subjected to unfavorable
conditions, such as increase in temperature or heavy soil disturbance will lose C. High
baseline C soils, that have more unprotected C will lose more C compared to lower
baseline C soils. Bellamy et al. (2005) reported relative loss in soil C in top soil increased

with increase in baseline values.

103

Relationship between Changes in C and Soil Texture

Soil texture plays an important role in influencing soil C changes directly and indirectly
in many different ways. Soil texture controls SOM decomposition rate where the
decomposition rate will be greater at coarse textured soil compared to frne textured soil.
Increases in clay content interact with SOC and forms complexes and microaggregates,
which provides greater physical protection to the SOC stored in the soil (Tisdall and
Oades, 1982; Amato and Ladd et al, 1992). Positive relationship between soil C and clay
content has been reported in many previous studies (Hassink, 1997; Hao and
Kravchenko, 2007).

Magnitude of soil C and texture relationship may depend on the degree of
saturation of C in silt and clay protected soil C pools (Hassink and Whitrnore, 1997).
Silt+clay content increases soil C indirectly by increasing the physically protected storage
capacity. As a result, fine textured soils will have more protective capacity and retain
more organic matter than coarse-textured soils even on applying the same level of C
inputs (Kortleven, 1963; Jenkinson, 1988).

However the general relationship between SOC and soil texture vary as a function
of tillage and management systems. For instance, Campbell et al. (1996) reported stronger
positive relationship between clay and C contents at NT as compared to that in CT
management at three sites in western Canada. In contrast, Paustian et al. (1997) observed
no relationship between texture and the effects of tillage on SOM contents while
analyzing data from 27 long-term tillage trials. The relationship between silt+clay content
with soil C varies with management system because apart from silt+clay protection, C is

also protected in other forms such as microaggregate protected C, biochemically

104

 

protected C and unprotected C. In tilled treatments the silt+clay protected C will be
reduced due to tillage disturbance resulting in weaker relationship between C and soil
texture at tillage treatments.

Most of the C studies examine the differences in soil C between different
management systems at specific time points after initiation of the experiment. The
changes in soil C are thus assessed on a relative scale via comparison with a control
treatment, which is typically the conventional management practice. Contemporary

samples are rarely compared with baseline values (VandenBygaart et al., 2002; Bellamy

 

et al., 2005; Scipper et al., 2007) since geo-referenced soil C and N data are not available L
in most of the long term sites. However, to monitor temporal trends in soil C dynamics,
not only relative but also absolute changes in soil C over time must be assessed.

Contemporary samples are desirable to compare with historic values to know the
impact, such as losses or gains in different management systems, including the system
used as a control. Bellamy et al. (2005) and Scipper et al. (2007) on comparing the
contemporary C values with baseline values raised a concern that there are widespread
losses in C occurring across all land uses, soil types and management systems common to
temperate regions. It is important to determine whether soils in temperate regions of the
US also experience similar C losses.

This study examines the changes in C and N contents under different management
systems that have taken place over a period of 15-20 years in Southern Michigan. The
two long-term experiments studied were Long Term Ecological Research (LTER) site
and the Interaction study site located at Kellogg Biological Station. Both experiments

have dense geo-referenced baseline data and hence resarnpling the baseline locations can

105

be done at any point of time with an accuracy of <05 meters in identifying the sampling
sites. Moreover, for some of the locations, the archived baseline samples are still
available and could be reanalyzed using newer techniques to reduce experimental errors.
The large size of LTER experimental plots provided a diverse range of topographical and
soil characteristics for studying interactions between managements and soil properties on

the SOC dynamics.

The objectives of this study are to:

1) Examine the changes that have taken place in soil C under different tillage and
management practices and in never tilled grassland over the past two decades in Southern
Michigan.

2) Analyze the relationships between the changes in soil C and baseline soil C and soil

texture under different tillage and management systems and in never tilled grassland.

Materials and Method

The study was conducted at two long-term experiments from Kellogg Biological
Station’s Long Term Ecological Research (KBS-LTER) site in Southwest Michigan
(85°24' W, 42°24' N). Climate is temperate with cool, moist winters and warm, humid
summers with approximately 90 cm annual precipitation about half of which is snow.
Mean annual temperature is 9°C (Grandy and Robertson, 2007). Detailed description of
soil properties for the study site is presented in Chapter 2. Soil sampling for the main site
and interaction site were carried out during May of 2006 and May of 2007, respectively.

The sampling protocol and treatments sampled at both the sites are discussed below.

106

 

FIJI I- — __-

Main Site

The LTER main site experiment is a one-factor RCBD with seven treatments
organized along a management intensity gradient based on tillage intensity, external
inputs, and net primary productivity (Grandy and Robertson, 2007). All the agronomic
treatments are under corn (Zea mays L.)—soybean [Glycine max (L.) Merr.]—wheat
(Triticum aestivum L.) annual rotation. The experiment was started at 1988. Prior to that,
the site had been conventionally managed under row-crop system for at least a century 1
(Robertson et al., 1997). A set of 417 total C measurements was obtained from the area

that became blocks 1-5 of the main site experiment in 1988. Among the 417 sampling

 

locations, 192 sample locations were selected using stratified unaligned grid design
(Webster and Oliver 1990) based on 50 * 50 m sampling cells and the remaining samples
were collected randomly as singlets, doublets and triplets (Robertson et al., 1997).
Doublets and triplets were separated by 1m from one another. Two 0-15-cm deep soil
cores were collected using a 7.5 cm diameter root sampler (Elkjamp,Wageningen, The
Netherlands) during late May of 1988 and composited into a single sample for each
location. The locations were geo-referenced using laser stratigraphy that also provided
elevation information for each location (Robertson et al., 1997). Year prior to the
initiation of experiment, entire site was planted with corn, followed by fall-planted rye
(Lolium sp.). The 1988 data will be referred from now on as baseline samples for the
main site.

In May of 2006, 50 of the locations sampled in 1988 were found using GPS
(Trimble Receiver Type Pro XRS model 33302-51) and re-sampled. The sample

locations chosen for re-sarnpling were those located within central portions of the plots of

107

the present LTER experiment. Three of the LTER management systems were studied,
namely, chisel plowed (CT) and no-till (NT) with conventional chemical inputs, and
certified organic chisel plowed system with a winter leguminous cover crop and zero
chemical inputs (CT-cover). Overall, 11, 22 and 17 locations suitable for re-sampling
were found in CT, NT and CT-cover treatments, respectively, with approximately 3-4
locations per each plot.

At each sampling location a hydraulic soil core unit from Geoprobe systems
(Salina, Kansas) with 4.2 cm diameter cylinder was used to collect an undisturbed soil

core of 0 - 40 cm depth. The cores were segmented into 0-15, 15-20, 20-30, 30-40 cm

 

depths and bulk density was determined for all the segments. The measurements from the
cores that will be reported in this study are the bulk density and total C and N data for the
depth of 0-15 cm. The samples were air dried at room temperature and all plant residues
and stones were removed. Smaller plant material was removed by gentle air blowing and
the samples were ground on a shatterbox machine (Shatterbox model 8530) to pass
through 250-um sieve. Total C were measured using an automatic Carlo-Erba CN
analyzer (Carlo Erba Instruments, Milan, Italy) and soil texture analyses were carried out
using the hydrometer method. Bulk density measurements were obtained on dry weight

basis and then used to assess soil C stocks on an areal basis.

108

The 2006 sampling was carried out during same period of time as that of 1988
sampling (May) with corn being previous year crop in both 1988 and 2006. This allowed
reducing errors associated with seasonal effects and crop residue effects. Soil sampling
information for data collection during 1988 and 2006 for LTER main site is provided in
Table 4.1.To minimize the errors associated with differences in laboratory procedures,
the 1988 archived samples from all 50 locations were reanalyzed for total C and N using
the same procedure that was used for the analysis of 2006 samples. The changes in soil C

and N are reported based on the reanalyzed 1988 data.

Interaction Site

The LTER Interaction site is a RCBD experiment with four replications. The two
studied factors are the fertilization with two levels, fertilized and unfertilized, and tillage
with two levels, tilled and no tilled. Each experimental plot is 27 by 40 m in size. All the
agronomic treatments are under corn (Zea mays L.)—soybean [Glycine max (L.) Merr.]—
wheat (Triticum aestivum L.) annual rotation. In the fertilized plots, during wheat years
Urea (N H2-C0-NH2) is applied as a source of N (40 pounds per acre) and during corn
years liquid N (28 percent active ingredient) is applied as per Michigan recommended
fertilizer rates for corn. No fertilizer is applied when soybeans are grown. During the last
five years no phosphorous (P) or potassium (K) was applied even at the fertilized plots.
The experiment was started at 1986, before that, the tilled portions of the studied site
have been conventionally managed at least from 1950 prior to that it was under native

vegetation (oak-hickory (Quercus-Carya spp.) forest) (Robertson et al., 1993).

109

Table 4.1. Summary of data collection in 1988 and 2006 for LTER mainsite and in
1986 and 2007 for the Interaction site.

 

Sampling Sampling

 

 

TRT N depth time Previous crop C N analysis method
Main site
1988 2006* 1988 2006 1988 2006 1988 2006
CT I l 0'] 5 Carlo-

NT 22 0-15 15-20 May May Com Com Erba CN Carlo-Erba
CT cover 17 20-30 analyser CN analyser
Interaction site

1986 2007 1986 2007 1986 2007 1986 2007
CT F 14
CT NF 10
NT F 12 0-20 0-20 May May Soybean Soybean
NT NF 1 1 Walkley-
Never Black Carlo-Erba
tilled 13 method CN analyser

 

N - Number of samples collocated at each treatment
* At main site during 2006, samples were collected from 0-15, 15-20 and 20-30 at CT, NT and

CT-cover

110

Several never tilled grassland sites are located near the interaction experiment. In
the never tilled grassland, mowing was done every year during late Fall since 1960 and
the residues were left at the surface.

A set of 256 samples from the tilled portions and 65 samples from the never tilled
portions of the site were collected in May 1986 prior to the Interaction experiment
establishment. Stratified unaligned sampling design based on 15* 18 sampling cells was l
adopted to select the sampling locations and provided a sufficient number of samples '

separated by relatively small distances (< 15 m). For geooreferencing the sampling

 

locations, the most South-West comer cell of the rectangular grid was assigned the
starting value of (0,0) and set as a reference point. Then coordinates of each sampling
location was calculated based on the distance from the set reference point. For instance, if
sampling location has coordinates as (5,3) then the specific sample will be located 5
meters from East and 3 meters from North of the reference point (0,0).

At each sampling point, five 2.5-cm-diameter cores were collected at 0-20 cm
depth within 15 cm radius circle and composited into single homogenous sample
(Robertson et al., 1993). In the year prior to soil sampling, the tilled portion of the site
has been planted with winter cover of annual ryegrass (Lolium sp.) without tillage.
Carbon measurements were carried out with triplicates subsample from each sampling
point following Walkley and-Black wet combustion technique during 1986 (Nelson and
Sommers 1982). Soil sampling information for data collection during 1986 and 2007 for

LTER interaction site is provided in Table 4.1.

111

Unlike mainsite, at the interaction site archived samples were not available to
reanalyze the baseline samples with our current procedure. The 1986 data will be referred
from now on as baseline samples for the interaction site.

During 2007, we collected a total of 57 samples from 1986 sampled locations. At
each sampling point, the sample was taken between the plant rows and was composited
from three 2.5-cm-diameter cores collected within a 0.2-m radius to a depth of 20 cm.

Soybean was grown during 2006, i.e., year prior to the contemporary soil sampling. I
Approximately, 2-4 samples were collected in each plot of the Interaction experiment.

Locations for resampling were selected in such a way that they were located in central

 

portions of the plots.

The procedures for sample cleaning, processing, and analyses were the same as
those used for the main site and described earlier. Smaller plant material was removed by
gentle air blowing and the samples were ground on a shatterbox machine (Shatterbox
model 8530) to pass through 250-pm sieve. Total C and N were measured using an
automatic Carlo-Erba CN analyzer (Carlo Erba Instruments, Milan, Italy). Subsarnple of
20 g was used for particle size analysis. Soil was dispersed in 5 % sodium
hexarnetaphosphate solution and kept in a mechanical shaker for about 16 hours. Sand
content was then isolated by passing the dispersed sample through a 53-um sieve. For the
interaction site, silt+clay content was obtained by subtracting sand content from the total
weight of soil used for analysis. Bulk density measurements were obtained on dry weight

basis and then used to assess soil C stocks on an areal basis.

112

Data analyses
Assessing treatment effects and relationships between changes in C content and
baseline C, soil texture and topographical features in different treatments

Data analysis was conducted using SAS (SAS Institute, 2001). Changes in soil C
and potential differences in relationships between the changes in soil C and the baseline
C values, silt+clay content and relative elevation under different management systems i
were studied using analysis of variance (ANOVA) and analysis of covariance
(ANCOVA) approaches in PROC MIXED (Milliken and Johnson, 2002). For the LTER

main site, the statistical model included treatments as a fixed factor, and blocks and plots

 

nested within treatments as random factors with plots used as an error term for testing the
treatment effects. For the Interaction site, the statistical model included fixed effects of
tillage, fertilization, and their interaction and random effects of blocks and plots. Baseline
C was used as a continuous covariate in both experiments. When interactions between
fixed factors and covariates were found to be statistically significant (01:0.05),
comparisons were conducted at several different covariate levels. In both experiments
normality of the residuals and homogeneity of variances assumptions were checked and
analysis with heterogeneous variances was conducted, if found necessary, using
REPEATED /GROUP statement in PROC MIXED. When the fixed effects were found to
be statistically significant (a=0.05), the comparisons between the treatments were

conducted using t-tests.

113

Reanalyzing the baseline values from the main site

Correspondence between soil C measurement on 1988 samples conducted in 1988
and reanalyzed in 2006 was assessed using simple linear regression. The R2 was equal to
of 0.51 (Fig. 4.1). Noticeable scattering of the data might be due to differences in sample
processing and analytical procedures. Reanalyzing the archived samples with the current
procedure will serve as a robust method for detecting changes and provide more credence
to our reported results. Hence for the analysis of changes in C content in different
treatments, I used the baseline values obtained using our current procedure from archived
samples to overcome any errors associated with sampling and instrumental variation.

For interaction site, archived samples were not available for reanalysis, hence I

reported the changes in C content calculated using the baseline values as measured in

1986.

Regression to the Mean (R T M correction

Regression analysis of the relationships between the changes in soil C and
baseline values was conducted with accounting for the regression to the mean
phenomenon (RTM). Lark et al. (2006) mentioned that while relating a change in value in
any soil properties with its baseline value, at least part of the relation will include
regression to the mean (RTM) effect. RTM occurs when unusually large or small
measurements obtained by chance during baseline sampling are followed by the

measurements that are less extreme and closer to the mean in the subsequent sampling.

114

 

2.0 -~—-- -

 

 

 

 

Total C in 1988 (%) (reanalyaed in 2006)

R2=0.51
g 1.8 - a
2:5? 1.5- . .
A

A
g: E 1.3 a ‘ “A
.5 é ‘ A. g A ‘
Q 0 1.0 ~ ‘
_ E A f A
5 ‘ ‘
,2 0.8 — ‘

0.5 T T l T I
0.5 0.8 1.0 1.3 1.5 1.8 2

Fig. 4.1. Comparison of baseline C values obtained from KBS database value with
baseline samples reanalyzed using current procedure for the LTER Main site along with
1:1 line

 

115

Thus even in the absence of the real relationship between the changes and the
baseline levels, larger changes are often associated with more extreme baseline values
leading to negative sample correlation (Barnett, 2005). The proportion of RTM effect
may be more, when we resarnple only portion of the original sites in the subsequent
sampling. In our study both the main site and interaction site, only a portion of the
baseline points were resampled, so I used a method proposed by Blomqvist (1977) for
overcoming the RTM effect to get unbiased regression slope estimates and confidence *

intervals for regressions between changes in soil C and baseline values (Lark et al.,

 

2006). The sample regression slope obtained from an ordinary simple linear regression L

between the change and the baseline, ,8 is adjusted as following (Blomquist, 1977)

~ _fi+a-V>
’6 adjusted _ V [I]

where V = l- (Suz/SZZ),

 

fladjusted is the adjusted regression slope obtained afier accounting for RTM effect.

Su 2 is the independently determined estimate of the variance for the baseline data which

is due to all the sources of uncertainty in baseline measurements, i.e., those associated

with analytical error and the spatial variability between the nominal and true location of

resampling array (Lark et al., 2006). To obtain the Su 2 values, I calculated sample

variograms for the baseline values using PROC VARIOGRAM and fitted the variograms

model with PROC NLIN. The value of Su 2 was set to be equal to the variogram value at

3 meters distance as obtained from the fitted variogram models. This value was

considered to be representative of the spatial variability of the baseline C data within 3

116

meters area around the sampling site. This area was assumed to provide a conservative

estimate of the errors in identifying the precise location of the baseline sampling sites.

Being a cumulative estimate, Su 2 evaluates the error associated with relocating the
sample sites plus the variability occurring at distances less than 3 m, that is, accounts for
analytical errors as well (Lark et al., 2006). SZ 2 is the overall sample variance obtained
from the 1988 baseline samples used in this study. Substituting SZ 2 and Su 2 in Eq. [1] I l

obtained adjusted regression slopes and confidence intervals for the adjusted regression

slope as

 

‘1

_ A :1:
CI — ,Badjusted it(%)(n—l)df S.E [2]

Here the t (a/Z)(n-1)df is the tabulated t-value obtained for the (n-l) degrees of freedom at

significance level(01 = 0.05) and SE is obtained as

A A 2 ~ A ' A
Vadflaab'usted ) /(1 + '8 adjusted) ~ Var(,B Initial )/(1+ '61

.. ')2
mtzal
+(v/(1— v))2[CV2(S:) + CV2(SZZ)]

[3]

CV(S§) = std(sj)/E(Sf) [4]
C1453): Sid(S:)/E(Sf) 151

Here CV(Su 2) and CV(Sz 2) represent the dispersion in the value of independently

determined estimate of the variance for the baseline data and the true sample variance

respectively. CV (Su 2) and CV(SZ 2) were calculated using Eq. [4] and Eq. [5]. For

CV(Su 2), standard deviation of Su 2 is assumed to be about 6 % of the mean of Su 2 ,

117

since about 6 percent variability in C was associated with analytical errors and spatial

variability within 3 meters. For CV(SZ 2), standard deviation of $2 2 is about 21% and

45% of the mean of SZ 2 at main site and interaction site respectively which is obtained

based on true sample variance. To obtain the intercept, the regression line with

A

,Badjumd is passed through the point of mean baseline C value, a? and the mean 2006 C

value, 37. Then the intercept is obtained from the following equation.

Y = fladjusted X + .60 [6]
Relationships between the changes in soil C and baseline values presented and discussed
were the results obtained after accounting for the regression to the mean phenomenon

(RTM) effect.

RESULTS AND DISCUSSION

Long Term Trends in Changes in Soil C under Different Management Systems
No single treatment showed increase in C content over the baseline C level (1986-

88) at both main site and interaction site. In LTER Main site, the mean baseline C (1988)
values were found to be somewhat lower in CT (9.3 g C kg '1 soil) than in CT-cover (10.1
g C kg '1 soil) and NT (10.6 g C kg '1) treatments even prior to management
implementation (Fig. 4.2). Mean C content of contemporary samples (2006) under

NT(9.9 g C kg 4 soil) and CT-cover(9.8 g C kg '1 soil) were significantly greater than
CT(8.2 g C kg -1 soil) (Table 4.2a). Mean changes in C (2006-1988) adjusted for the

common level of baseline C values was equal to -1.15, -0.71 and -0.16 g C kg '1 soil for

CT, NT and CT-cover, respectively.

118

 

 

 

1.6‘1

...~ ° 1'

1.2‘
u _
1.0“ l +
i + ‘
0.8“ l
1 1

0.6 ‘

 

 

Total Carbon

 

 

 

 

 

 

 

I l l I

CT1988 CT2006 NT‘I988 NTZOO6 Org1988 Org2006

Fig. 4.2 Baseline C (1988) and contemporary (2006) soil C content (0-15 cm) in CT, CT-
cover(organic) and NT treatments at the LTER main site

 

3.0 ‘

2.5“

 

2.0 ‘

 

Total__Carbon
J
on
1

 

 

 

 

. q 0 T D-
1.0 + , %_E
0.57 J. é

I I l l I F

CT1986 CT2007 N T1986 N T2007 Undl1986 U nd12007

Fig.4.3. Baseline (1986) and contemporary (2007) soil C content (0-20 cm) in CT, NT
and Never tilled grassland plots (undisturbed) at the long term interaction site

experiment.

 

 

 

 

 

 

119

Table 4.2a. Mean baseline(1988) C and contemporary(2006) C in CT, NT and
CT-cover at LTER main site. Standard errors are shown in parentheses.

 

 

 

 

Baseline C 2006 C
TRT g kg " of soil
CT 9.3 (0.7) a“ 8.2 (0.3) a“
NT 10.6 (0.5) a 9.9 (0.5) b
CT-cover 10.1 (0.5) a 9.8 (0.3) b
P value 0.43 0.02

 

*Means within the same column followed by the same letter are not significantly different
(or=0.05)

Table 4.2b. Mean changes in C content adjusted for standardized baseline C content
along with raw means (unadjusted for baseline) in CT, NT and CT-covcr at LTER
main site. Standard errors are shown in parentheses.

 

 

 

Treatment Raw data Adjusted for standardized TC 1988
ng " of soil

CT -l.15*(0.36) a" -1.15*(o.27) a"

NT -0.71NS(0.52) a -0.71NS (0.48) a

CT—cover -0.17NS (0.49) a -0.16NS (0.29) a

P value 0.35 0.14

 

NS - the value is not significantly different from zero (p<0.05)

TC1988 — baseline C value during initiation of the experiment at 1988

* Means that were significantly different from zero at p<0.1

** Means within the same column followed by the same letter are not significantly different
(0L=0.05)

At NT and CT-cover the mean changes in C value were not different from

120

 

zero indicating no change in C content over the baseline values, while in CT the mean
change is significantly lower than zero indicating C losses prevailing at CT. The studied
treatments, CT, NT and CT-cover were not significantly different from each other in
terms of changes in C content (p<0.05) (Table 4.2b). In the Interaction site, mean

baseline C (1986) were somewhat different among the treatments prior to treatment

implementation, e. g. C in NT(8.2 g C kg '1 soil) tended to be lower than that in CT(9.8 g

C kg '1 soil). Baseline C values of never tilled sites (20.4 g/kg) were much higher than

those of the agronomic part of the experiment (Fig.4.3). Mean C content of contemporary
samples(2007) under NT(8.6 g C kg '1 soil) were similar to that of CT(8.8 g C kg '1 soil)

(Table 4.3a).The variability in the Interaction site baseline (1986) C data was found to be
much higher as compared to that in 2007.

For example, C variances in samples collected from NT and CT plots in 1986
were equal to 0.07 and 0.07, respectively, while they were equal to 0.03 and 0.01 in 2007
measurements. The difference could be attributed to different measurement procedures
used in 1986 (Walkley-Black) and 2007 (Carlo-Erba) and differences in sample
preparation and processing. There was a significant interaction effect between tillage and

fertilization for the change in C content (p<0.l) and the main effect of tillage also was

found to be significant (Table 4.3b; p<0.05). Mean changes in C values(2007-1986) for
the CT-F, CT-NF, NT-F and NT-NF were equal to -0.22, -1.54, 0.31 and 0.97 g C kg '1

soil, respectively (Table 4.3b). The mean C change in CT-NF was less than zero and less
than those of both NT-F and NT-NF treatments. The mean C changes of CT-F, NT-F and

NT-NF were not different from zero and not different from each other either.

121

 

Table 4.3a. Mean baseline C (g kg 4) and contemporary (2007) C content at
fertilized and unfertilized plots of CT and NT at LTER-interaction site treatments
from 0-20 cm depth. Standard errors are shown in parentheses.

 

 

 

 

 

 

 

Treatment Baseline C 2007 C

Main efi'ects

CT 9.8 (1.2) 8* 8.6 (0.3) a*

NT 8.2 (1.2) a 8.6 (0.3) a*

P Value 0.11 0.61

Interaction efgct L

CT-F 9.8 (1.4) a“ 9.5 (0.5) a* 7
CT-NF 9.8 (1.4) a 8.2 (0.5) a

NT—F 7.4 (1 .4) a 8.3 (0.5) a

NT-NF 8.9 (1.4) a 8.9 (0.5) a a"
P Value 0.48 0.07 i”

* Comparisons of mean effects (either among marginal means or among cell means)
within the same column followed by the same letter are not significantly different (0.=0.05

Table 4.3b. Mean changes in C values (g kg ")adjusted for the standardized
baseline C content in CT and NT of LTER-interaction site treatments. Standard
errors are shown in parentheses.

 

 

 

Treatment Raw data Adjusted for standardized TC 1986
Main effects

CT -1.15NS (0.9) a“ -0.88* (0.34) a“
NT 0.94NS (0.9) a 0.64NS (0.35) a

P value 0.08 0.013
Interaction effect

CT-F -0.28NS (1.1)a ** -0.22NS (0.46) ab"
CT-NF -202” (1.2)a -1.54* (0.51) a
NT—F 0.79NS (1.1)a 0.31NS (0.48) b
NT-NF 1.08NS (1 .2)a 0.97NS (0.51) b
P value 0.36 0.07

 

CT-F , Conventional tillage, F ertilized; CT-NF , Conventional tillage, non fertilized;

NT-F , No tillage, F ertilized; NT-NF , No tillage, non fertilized;

TC1986 — baseline C value content during initiation of the experiment at 1986

"‘ Means that were significantly different from zero at p<0.1

** Comparisons of mean effects (either among marginal means or among cell means)
within the same column followed by the same letter are not significantly different (01=0.05)

122

Mean changes in C content at CT and NT across both fertilization treatments were
equal to -0.9 and 0.6, g C kg '1 soil, respectively. The change in C at CT was significantly

less than zero representing C losses at CT. However the change in NT was not different
from zero indicating no change in C content over the baseline values, similar to the
results from the main site. Decline in soil C content over the baseline status at CT at 0-20
cm depth was previously reported by Hendrix et al. (1998) in a 16 year old experiment
carried out in a fine loamy soil in Georgia and by Doran et al. (1998) at 0-30 cm in a 22-

27 year old experiment in a silt loam soil in Nebraska. Both studies reported that the

 

magnitude of C losses was greater in CT as compared to NT. No change in C content at E
NT was similar to the results reported in other previous studies reported by Puget and Lal

(2005) at 0-30 depth in a 8 year old experiment carried out in a Mollisol of central Ohio;

and by Eynard et al.(2005) at 0—20 cm in a 6-16 years old long-term experiments studied

in Ustolls in South Dakota.

C losses at never tilled sites were equal to 1.7 kg C m '2 (0-20 cm) and were the
highest as compared to all treatments from the Interaction site and the main site.
Occurrence of greater C losses in New Zealand grasslands of about 2.1 kg C m '2 from

one meter depth during the past 2 decades was also reported by Scipper et al. (2007).
Summarizing the Main site and Interaction site results, I conclude that changes in C
content varied with the management system. Total C at conventional chisel plowed
management has decreased by 1.15 g C kg '1 soil and 0.22 g C kg '1 soil at main site and
interaction site respectively (treatment CT-F is used here for comparing CT of main and
interaction sites). When compared at present (in 2006 and 2007), total C values under no-

till management with conventional chemical inputs and under organic management with

123

cover crops are higher than those of the conventional management ((Table 4.2a, Table
4.3a). However, examination of the changes since 1986 and 1988 revealed that there has
not been a net gain in C under these systems. The observed differences with

conventionally tilled systems are reflecting C losses sustained by its soil. C in never tilled
soil kept for past >25 years under grass vegetation has decreased on 7.4 g C kg '1 soil

since 1986. Within the conventional tillage system, the C losses were found to be greater
in unfertilized compared to fertilized treatments.

Rate of change in C content and the size of the C pool in any management system
depend on the balance between soil C inputs through plant residues and C losses which
occur mainly through decomposition. The balance exists for particular environmental
settings affecting decomposition rates, including air and soil temperatures and moisture
availability. It appears that the observed C losses in never tilled sites might be due to
increased decomposition rate. The increased decomposition rate could be driven by rising
winter temperatures in Michigan. Two recent studies also linked widespread soil C losses
at England and Wales (Bellamy et al., 2005) and at pasturelands at New Zealand (Scipper
et al., 2007) with the increase in global climate change.

Latest report from the Intergovernmental Panel on Climate Change (IPCC)
predicts the average increase in global temperature of about 1.4-5.8° C for the period of
1990 to 2100 (Luo et al., 2001). Organic matter decomposition is very sensitive to rise in
temperature (Knorr et al., 2005; Conant et al., 2008). It is expected that the temperature
effect on changes in soil C will be greater in temperate regions than in tropical regions. In
temperate regions, rising temperatures will increase organic matter decomposition rates

but will not lead to substantial changes in net primary productivity (N PP) (Kirschbaum,

124

 

1995). The greater sensitivity for organic matter decomposition to rising temperature as
compared to that of NPP will play an important role affecting the balance between C
inputs and C outputs.

In Michigan, over the last century, the mean temperature has increased from
46.6°F (1876-1905 average) to 47.7°F (1962-1991 average; US EPA,1997). Over the next
century, based on IPCC and United Kingdom Hadley Centre’s climate model (HadCM2),
it was projected that by 2100, Michigan’s temperature might increase by 4°F in all
seasons with range of 2-8°F (US EPA,1997). Data from KBS weather stations indeed
indicate that the number of days per year with temperature daily average minimum above
freezing has been increasing during the past 15 years (Fig. 4.4).

Potential effects of global warming on soil C storage under CT and NT
management practices in KBS-LTER were reported earlier by Paustian et al. (1995).
They evaluated soil C storage projections under the scenario of the mean annual
temperature increasing by 2° C by 2050 based on IPCC estimates using Century model.
The simulation assumed that no changes in management or in C input levels were made
in conventional tillage system. Model predicted a decline in C storage from the baseline
status at CT by taking baseline C values as 3 kg C m '2 for 0-20 cm observed at mid
1980’s. Results from our current measured C levels support these model predictions.

In soils under agricultural management, soil C losses due to increase in

decomposition rate is partly compensated by increase in agricultural productivity due to

improvement in crop varieties and usage of external chemical inputs.

125

 

 

 

.3 250 —

g . 0

8 . O 0

0° 0 9 . . o

g 225 — . .

3

CI . . a

Q a

5'

3 200 —

o

E

E

g 175 I If I j ‘1
1985 1990 1995 2000 2005 2010

Year

 

Fig. 4. 4. Decadal pattern of number of days above freezing point at the KBS-LTER site
(R2 = 0.27).

126

 

In Michigan, the annual agricultural productivity was reported to increase by
about 2.6 percent per year during 1960-1999
(http://www.ers.usda.gov/Data/AgProductivity). At never tilled sites, the productivity
might be more stable and might have not increased over this time period unlike that in the
agricultural systems. Hence at undisturbed sites, due to winter warming C loss through
decomposition will be exceeding the amount of C from the inputs that can be stored in
soil under new temperature balance.

Pautian et al., (1995), predicted that the baseline C can be sustained or marginally
increased if the agricultural productivity increases annually by 1.4 percent which might
result in 40 percent more residue return as compared to present residue return rate. Our
study results from NT and CT-cover shows that C content has not changed since 1986-
1988. With the projected increase in global temperature, at practical standpoint, at least
adoption of NT or inclusion of cover crops in the crop rotation may be one of
prerequisites to sustain the present soil C levels. With the growing bioenergy market,
increased removal of crop residues for the biofuel feedstock may have serious negative
impact on the soil quality causing increase in soil C losses, in turn increasing the global
warming potential.

Climate induced soil C changes are likely to differ in magnitude under different
tillage and management systems. Moreover, within each management system local
variations in soil properties also influence C change values. Greater C losses at CT and
no change in C content at NT are mainly driven by difference in disturbance gradient at

these treatments. In this study, NT receives the least soil disturbance compared to CT,

127

however C inputs through annual residue return reported to be same under both CT and
NT (Robertson et al., 2000). Though both the treatments receive the same level of C
inputs, C losses due to organic matter decomposition might be relatively greater at CT
compared to NT. At CT, due to tillage disturbance the aggregate turnover rate will be
faster which in turn increases organic matter decomposition by means of exposing more
of relatively young and labile organic matter to microbial attacks (Pautian et al., 2000).
Moreover, in CT faster aggregate turnover also inhibits formation and stabilization of
more recalcitrant organic matter fractions which has longer mean residence time (Six et
al. 1999). Lack of soil disturbance in NT increases the number and stability of soil
aggregates which in turn protects SOC from rapid oxidation by encapsulating C within
the stable aggregates (Puget and Lal, 2005). NT also promotes the formation of
recalcitrant SOC within micro and macro aggregates and increases the mean residence
time (MRT) of the organic matter fractions compared to CT (Pautian etal., 2000).

The difference in changes of C content between CT-cover and CT are driven by
difference in quantity and quality of residue inputs from the primary crops and cover
crops of these treatments. At CT-cover, despite of heavy soil disturbance, magnitude of
soil C changes was found to be similar to that of NT. Similar results showing organic
based systems with cover crops to be as effective as no-till (Teasdale et al., 2007) and
more efficient than CT in maintaining C was also reported earlier in other long term
studies (Pulleman et al., 2000; Marriott and Wander 2006).

Legume based organic tillage system enhances or maintains soil C storage due to
following mechanisms. C inputs from cover crops have narrow CN ratio and thus

enhance soil C storage by means of improving the soil physical properties such as

128

increase in the total porosity and wet aggregate stability (V illamil et al., 2006).
Mycorrhizae associated with the leguminous cover crop secretes proteineous substance
called glomalin which acts as cementing agent in binding the soil particles and is found to
be effective in providing short term stabilization of large aggregates (Brady and Weil,
2002). Inclusion of cover crops in the crop rotation reduces the frequency of bare fallow
period in crop rotations, which in turn reduces organic matter decomposition rates. Thus
additional inputs from the cover crop and greater aggregate stabilization offset soil C

losses due to heavy tillage disturbance in an organic CT system.

Relationship between Changes in C content and Baseline C

At main site, baseline C content was found to have significant negative
relationship with the change in C values at CT-cover however at CT and NT the
relationship found to be insignificant (Table 4.4). At the interaction site, significant
negative relationship was observed between baseline C values and the change in C values
from 1986 to 2007 C values in all the studied treatments (Table 4.5).

Precision of the treatment estimates were also improved after adjusting with the
baseline value, which is reflected in lower standard error and lower p-values compared to
the unadjusted mean values (Table 4.2b). Negative relationship between baseline C and
change in C content reported in the study were consistent with the results reported in
literature (Bellamy et al., 2005; VanderBygaart et al., 2002). One possible mechanism
behind the negative relationship between baseline C levels and the change might be

related to soil’s saturation with C (Hassink 1997, Six et al., 2002).

129

 

Table 4.4. Results of observed and RTM adjusted slope estimates for baseline C
effect on changes in C content in the LTER-mainsite treatments.

 

 

TRT Observed slope Adjusted slope

CT -0.64*i0.56 -0.41N51.-0.91
NS NS

NT -0.47 $20.50 -0.31 $0.81

CT-cover -0.73i0.28 -0.58*:t0.45

 

NS - Regression slopes that were not significantly lesser than zero (01=0.05).
* Regression slopes that were significantly less than zero at p<0.05

Table 4.5. Results of observed and RTM adjusted slope estimates for baseline C
effect on change in C content of LTER— interaction site treatments

 

 

 

TRT N Observed slope Adjusted slope
CT-F 14 -0.95*:t0.35 -0.93*:1:0.49
CT-NF 10 -0.82*i0.18 -0.76*i0.28
NT-F 12 -0.89*i0.18 -0.85*d:0.26
NT-NF 1 1 -l.04*i0. 15 -1.05i0.22
Never tilled 13 -0.88*:1:0.41 -0.84*i0.58

 

CT-F , Conventional tillage, Fertilized; CT-NF , Conventional tillage, non fertilized;
NT-F , No tillage, Fertilized; NT-NF , No tillage, non fertilized;

Native: Annually mowed, never-tilled grassland.

"‘ Regression slopes that were significantly different from zero (p<0.05)

130

In our study, the baseline C content at the Native sites was relatively higher than
the conventional tillage system. The major proportion of the higher baseline C content in
the never tilled sites will be occupied at the protected C sites and additionally a lot of C is
stored in unprotected forms as light fractions and POM. The amount of unprotected C
being held at equilibrium at never tilled sites, on subjected to change in environmental
conditions such as increase in temperature might have experienced greater losses as a
result of increased decomposition rate. At conventional tillage system, most of the labile
state C that comes from the native vegetation might have already been depleted and the
baseline C (1988) mentioned in this study might be the equilibrium state of most stable C.
These most stable, low baselines C at conventional tillage system would have been
occupied in the protected sites and only small amount of C is present at unprotected C
pools which resulted in relatively lower soil C losses. Similar to the results reported in
our study, Bellamy et al. (2005) observed lower soil C losses in cultivated lands and

greater C losses in grasslands.

Relationship Between Changes in C and Soil Texture (silt+clay)

Among the studied treatments, significant positive relationship between changes
in C content and silt+clay content was observed only at NT with r2 value of 0.32 at the
main site (Table 4.6) and 0.13 for the interaction site. Increase in SOC with increase in
silt+clay content in NT was consistent with the previous studies (Needelman et al., 1999;

Hao and Kravchenko, 2007). Silt +clay content increases SOC possibly because NT

131

favors formation of micro- and macro-aggregates which provides greater stability to the
silt and clay associated C stored as the protected pool.

Table 4.6. Relationships between changes in C, N and silt+clay content at 0- to 15-
cm depth for the LTER main site.

 

 

TRT lggression model R 2
CT AC = -0.11 — 0.00003 Silt+Clay ”5 0.01
NT AC = -0.68 + 0.01 Silt+Clay m 0.32
CT-cover AC = -0.12 +0002 Silt+Clay ”5 0.01

 

AC=2006C~1988C; ‘
NS - Regression slope is not significantly greater than zero (01=0.05).
Regression coefficient that were significantly less than zero p<0.01(***)

 

Table 4.7. Relationships between changes in C and silt+clay content at 0- to 20- cm
depth for the LTER interaction site.

 

 

TRT Iggression model R2

CT F AC = -1.8 +0.03 Silt+Clay “5 0.14
CT NF AC = 1.22 - 0.023 Silt+Clay “5 0.29
NT F AC = -024 +0003 Silt+Clay ”5 0.003
NT NF AC = -4.3 + 0.08 Silt+Clay m 0.44
Never tilled AC = -0.49 + 0.004 Silt+Clay “5 0

 

CT F , Conventional tillage, F ertilized; CT NF , Conventional tillage, non fertilized;
NT F , No tillage, Fertilized; NT NF , No tillage, non fertilized;

Never tilled: Annually mowed undirturbed grassland.

AC — 2007 C -l986 C

NS - Regression slopes that were not significantly greater than zero (a=0.05).

132

In NT, changes in C content were found to be positively correlated with silt+clay content
which shows that protective capacity of our soils has not been saturated.

In CT and CT-cover, relationships between changes in C content and silt+clay
were not significant. Conventionally tilled soils have limited potential to protect C in the
aggregates since the aggregate formation is greatly disrupted by plowing (Puget et al.,
1995; Paustian et al., 1997). For example, Beare et al., (1994) mentioned that the pool of
physically protected C in macroaggregates accounts for 10 % of soil C stocks in the tilled
soil and 19% in the no-tilled soils.

In previous studies, greater C content in organic based systems is mainly
attributed to increases in POM. Organic based system is reported to have 30-40 percent
more POM than the conventional system (Marriott and Wander, 2006). Earlier studies
reported that POM content is not directly related with soil texture (Plante et al., 2006;
Franzluebbers and Arshad ,1997). Soil C at CT and CT-cover is protected mainly through
physico-chemically stabilization (Krull et al., 2003) rather than physical stabilization
alone (Hao and Kravchenko, 2007). Above results shows that soil texture information
might explain greater amount of variation in soil C changes in NT soils compared to CT

and CT-cover.

SUMMARY
In summary, increase in winter temperature at Michigan might be one of the
major factors that cause soil C losses in this site. Baseline C was found to have negative

relationship with changes in C content, and C loss was found to increase with increase in

133

baseline values. Increase in silt and clay content leads to bring positive changes (gain in
C) in C content in no-till system. The effect of relative relief was not found to have any
relationship between changes in C content, probably due to small sample size and low
variations in topography among the sampled locations. With respect to long term trend in
soil C storage, at conventional chisel plowed management, total C was found to decline
by about 10 percent C from that baseline C value (1986-2007).Within conventional
tillage system, C losses were found to be greater at unfertilized compared to fertilized
treatments. As comparing management practices at present, soil C under no—till
management and under organic management with cover crops tends to be higher than that
of the conventional management, however, it has not changed as compared to 1988. That
is, the conservational management practices appeared to have prevented total C loss as
compared to conventional management but they did not lead to gains in absolute amount.
Likewise, the total C of the never tilled grassland soil has decreased by about one third
from the baseline C of 1986. In summary, none of the studied treatment showed increase
in C content from the baseline values. With the projected increase in global temperature,
at practical standpoint, at least adoption of NT or inclusion of cover crops in the crop
rotation may be one of prerequisites to sustain the present C levels at soils in temperate

region.

ACKNOWLEDGEMENTS

This research was sponsored by the NSF Long-Term Ecological Research Program
(Project No. DEB98-10220) at the Kellogg Biological Station, Michigan Agricultural
Experiment Station, USDA Sustainable Agriculture Grant and MSU Intramural Research
Grant. I thank Xuewan Huang and Xinmei Hao for his support during soil sampling and

analyses and Suzanne Sippel for geo-referencing sample locations.

134

REFERENCES

Amato, M.A., and J .N. Ladd. 1992. Decomposition of l4C-labelled glucose and legume
material in soils: Properties influencing the accumulation of organic residue C and
microbial biomass C. Soil Biol. Biochem. 24:455—464.

Balesdent, J., C. Chenu and M. Balabane. 2000. Relationship of soil organic
matterdynamics to physical protection and tillage. Soil Till. Res. 53: 215-230

Barnett, A.G., J.C. van der P013, and A.J. Dobson. 2005. Regression to the mean: what it
is and how to deal with it. Int J Epidemiol. 34:215—20.

Beare, M.H., Cabrera, M.L., Hendrix, PF. and DC. Coleman. 1994. Aggregate-protected
and unprotected organic matter pools in conventional-tillage and no-tillage soils. Soil
Sci. Soc. Am. J. 58, pp. 787—795.

 

Bellamy, P.H., Loveland, P.J., Bradley, R.I., Lark, R.M., and G.J.D. Kirk 2005. Carbon
losses from all soils across England and Wales. Nature 437: 245—248.

Bergstorm, D.W., C.M. Monreal, and E. St. Jacques. 2001. Spatial dependence of soil
organic carbon mass and its relationship to soil series and topography. Can. J. Soil
Sci. 81 :53—62

Blomqvist. N. 1977. On the relation between change and initial value. Journal of the
American Statistical Association. 72(360): 746-749

Brady, NC, and RR. Weil. 2002. The nature and properties of soils. 13th ed. Pearson
Education, Inc.

Campbell, C.A., B.G. McConkey, R.P. Zentner, F. Selles, and D. Curtin. 1996. Long-
term effects of tillage and crop rotation on soil organic C and total N in a clay soil in
southwestern Saskatchewan. Can. J. Soil Sci. 76:395—401.

Conant. R.T., R A. Drijber, M. L. Haddix, W. J. Parton, E. A. Paul, A. F. Plante, J. Six
and J. M. Steinweg 2008. Sensitivity of organic matter decomposition to warming

varies with its quality 14: 868—877

Davidson, B. A., S. E. Trurnbore, and R. Amundson. 2000. Soil Warming and organic
carbon content. Nature 408: 858-861

Doran, J.W., E.T. Elliott and K. Paustian. 1998. Soil microbial activity, nitrogen cycling,
and long-term changes in organic carbon pools as related to fallow tillage
management. Soil and Tillage Research 49: 3—18.

135

Franzluebbers, A.J. 2004. Tillage and residue management effects on soil organic
matter.p. 227—268. In. F. Magdoff and RR. Weil (ed.) Soil organic matter in
sustainable agriculture. CRC Press. Boca Raton, FL.

Franzluebbers, A.J., and MA. Arshad. 1997. Particulate organic carbon content and

potential mineralization as affected by tillage and texture. Soil Sci. Soc. Am. J.
61:1382—1386.

Grandy, A.S., G.P. Robertson, and K.D. Thelen. 2006. Do productivity and
environmental trade-offs justify periodically cultivating no-till cropping systems?
Agron. J. 98:1377—1383.

Grandy, A. S., and G. P. Robertson. 2007. Land use intensity effects on soil C
accumulation rates and mechanisms. Ecosystems 10: 59-74.

Hao, X., and A. N. Kravchenko. 2007. Management Practice Effects on Surface Soil
Total Carbon: Differences along a Textural Gradient. Agron. J 99: 18-26

Hassink. J. 1997. The capacity of soils to preserve organic C and N by their association
with clay and silt particles. Plant Soil. 191: 77-87.

Hassink, J ., and AP Whitrnore. 1997. A model of the physical protection of organic
matter in soils. Soil Sci. Soc. Am. J. 61 : 131-139.

Haynes, R.J., and Beare, M.H. 1996. Aggregation and organic matter storage in meso-
thennal, humid agricultural soils. In: Carter MR, Stuart BA, Structure and organic
matter storage in agricultural soils. New York: CRC Press. 213—62.

Hendrix, P.F, A.J. Franzluebbers and D.V. McCracken. 1998. Management effects on C
accumulation and loss in soils of the southern Appalachian Piedmont and of Georgia,
Soil Till. Res. 47: 245—251.

Huggins, D.R., G.A. Buyanovsky, G.H. Wagner, J .R. Brown, R.G. Darmody, T.R. Peck,
G.W. Lesoing, M.B. Vanotti and LG. Bundy. 1998. Soil organic C in the tallgrass
prairie-derived region of the corn belt: effects of long-term crop management. Soil
Tillage Res. 47: 219—234.

Janzen, H.H., C.A.Campbell, R.C. Izaurralde, B.H. Ellert, N. Jurna, W.B. McGill and
RP. Zentner. 1998. Management effects on soil C storage on the Canadian prairies.
Soil Till.Res. 47(3-4): 181-195.

Jenkinson, D.S., D.E. Adams and A. Wild. 1991. Model estimates of C02 emissions from
soil in response to global warming. Nature. 351: 304—306

Jenkinson, D.S., 1988. Soil organic matter and its dynamics. In: Wild, A. (Ed.), Soil
Conditions and Plant Growth, 11th ed. Longman, New York, pp. 564—607.

136

 

Jones, C., C. McConnell, K.Coleman, P.Cox, P.Falloon, and D. Jenkinson. 2005. Global
climate change and soil carbon stocks; predictions from two contrasting models for
the turnover of organic carbon in soil. Global Change Biology 11: 154—166.

Kellogg Biological Station. 2007. Long-term ecological research in row-crop agriculture.
Available at www.kbs.msu.edu/lter . KBS, Hickory Comers, MI.

Kirschbaum, M.U.F. 1995. The temperature dependence of soil organic matter
decomposition and the effect of global warming on soil organic carbon storage. Soil
Biology and Biochemistry, 27, 753— 760.

Kirschbaum, M.U.F. 2000. Will changes in soil organic matter act as a positive or
negative feedback on global warming? Biogeochemistry, 48, 21— 51.

Kirschbaum, M.U.F. 2004. Soil respiration under prolonged soil warming: are rate
reductions caused by acclimation or substrate loss ? Global change biology 10: 1870-
1877.

Knorr, W, I.C.Prentice, J.I.House and EA. Holland. 2005. Long-term sensitivity of soil
carbon turnover to warming. Nature 433: 298—301

Kortleven, J. 1963. Quantitative aspects of humus accumulation and humus
decomposition. Versl. Landbouwk. 0nderz. 69.1. 109

Krull, E.S., J.A. Baldock, and J .O. Skjemstad. 2003. Importance of mechanisms and
processes of the stabilisation of soil organic matter for modelling carbon turnover.
Functional Plant Biology 30:207—222.

Lal, R. 2004. Soil carbon sequestration impacts on global climate change and food
security. Science 304: 1623—1627.

Lark, R.M., P.H.Bellamy, and G.J.D.Kirk. 2006. baseline values and change in the soil,
and implications for monitoring. Eur. J. Soil Sci. 57: 916-921.

Luo. Y, S.Wan, H. Dafeng, and L.L.Wallace 2001. Acclimatization of soil respiration to
warming in a tall grass prairie Nature 413, 622-625

Marriott, EB, and M.M. Wander. 2006. Total and labile soil organic matter in organic
and conventional farming systems. Soil Sci. Soc. Am. J. 70:950—959.

Milliken, G.A., and DE. Johnson. 1992. Analysis of messy data: Designed experiments.
CRC Press, Boca Raton, F L.

Munos, J .D., R. Gehl, S. Snapp, A.N. Kravchenko. 2008. Identifying the factors affecting
cover crop performance in row crops. Presented at 9th International conference on
precision agriculture, Denver, Colarado, USA July 20-23.

Murty, D., M.U.F. Kirschbaum, E.M.Ross, and M.Heather. 2002. Does conversion of
forest to agricultural land change soil carbon and nitrogen? a review of the literature
Global Change Biology 8 2 , 105—123

137

Needelman, B.A., M.M. Wander, G.A. Bollero, C.W. Boast, G.K. Sims, and D.G.
Bullock. 1999. Interaction of tillage and soil texture: Biologically active soil organic
matter in Illinois. Soil Sci. Soc. Am. J. 63:1326—1334.

Nelson, D.W., and LE. Sommers. 1982. Total carbon, organic carbon, and organic
matter. In: Page AL, Miller RH, Keeney DR (eds) Methods of Soil Analysis Part 2:
Chemical and Microbiological Properties, Sec. Ed. American Society of Agronomy,
Madison, Wisconsin, USA. PP 539-580

Paustian, K., G.P.Robertson and ET. Elliott. 1995. Management impacts on carbon
storage and gas fluxes (C02, CH4) in mid-latitude cropland and grassland
ecosystems. In: Lal R, Kimble J, Levine E & Stewart BA (Eds) Soil Management and
Greenhouse Effect. Advances in Soil Science (pp 69—84). CRC Press, Boca Raton,
FL, USA.

Paustian, K., H.P. Collins, and EA. Paul. 1997. Management controls on soil carbon. p.
15—49. In E.A. Paul et al. Cole (ed.) Soil organic matter in temperate agroecosystems:
Long-term experiments in North America. CRC Press, Boca Raton, FL.

Paustian, K., J .Six, E.T.Elliott and H.W.Hunk. 2000. Management options for reducing
CO2 emissions from agricultural soils. Biogeochemistry 482147-163

Plante, A.F., R.T. Conant, C.E. Stewart, K.Paustian and J. Six. 2006. Impact of soil
texture on the distribution of soil organic matter in physical and chemical fractions.
Soil Sci. Soc.Am. J. 70(1): 287-296.

Puget, P., C. Chenu, and J. Balesdent. 1995. Total and young organic carbon distributions
in aggregates of silty cultivated soils. Eur. J. Soil Sci. 46:449—459.

Puget, P., and R. Lal. 2005. Soil organic carbon and nitrogen in a Mollisol in central Ohio
as affected by tillage and land use. Soil Till. Res. 80: 201—213.

Pulleman, M.M., J. Bourna, E.A. van Essen, and E.W. Meijles. 2000. Soil organic matter
content as a function of different land use history. Soil Sci. Soc. Am. J. 64:689—693.

Richter, D.B.J., M. Hofmockel, M. A. Callaham, Jr., D. S. Powlson, and P. Smith. 2007.
Long-Term Soil Experiments: Keys to Managing Earth's Rapidly Changing
Ecosystems. Soil Sci. Soc. Am. J. 71: 266-279

Robertson, G.P., J .R. Crum, and BC. Ellis. 1993. The spatial variability of soil resources
following long-term disturbance. Oecologia 96:451—456.

Robertson, G.P., K.M. Klingensmith, M.J. Klug, E.A. Paul, J .C. Crum, and BC. Ellis.
1997. Soil resources, microbial activity, and primary production across an agricultural
ecosystem. Ecol. Appl. 7: 1 58—170.

Robertson, G.P., E.A. Paul, and RR. Harwood. 2000. Greenhouse gases in intensive
agriculture: Contributions of individual gases to the radiative forcing of the
atmosphere. Science 289: 1922—1925.

138

 

Rustad, L.E., J.L. Campbell, and GM. Marion. 2001. A meta-analysis of the response of
soil respiration, net nitrogen mineralization, and aboveground plant grth to
experimental ecosystem warming. Oecologia, 126, 543— 562

SAS Institute. 2001. SAS user's guide. Version 9.1. SAS Inst., Cary, NC.

Schimel, D.S., B.H. Braswell, E.A. Holland, R. McKeown, D.S. Ojima, T.H. Painter, W.
J. Parton, and A. R. Townsend. 1994. Climatic, edaphic, and biotic controls over
storage and turnover of carbon in soils. Global Biogeochem. Cycles 8: 279—293.

Schipper, L.A., W.T. Baisden, R.L. Parfitt, C.Ross, J.J. Claydon, and G.Amold. 2007.
Large losses of soil C and N from soil profiles under pasture in New Zealand during
the past 20 years. Global Change Biology 13: 1138—1144.

Six, J ., E.T. Elliott, and K. Paustian. 1999. Aggregate and soil organic matter dynamics
under conventional and no-tillage systems. Soil Sci. Soc. Am. J. 63:1350—1358.

Six, J., R.T. Contant, E.A. Paul, and K. Paustian. 2002. Stabilization mechanisms of
soilorganic matter: implications for C-saturation of soils. Plant and Soil. 241: 155-
176.

Stockdale, E.A., N.H. Lampkin, M. Hovi, R. Keatinge, E.K.M. Lennartsson, D.W.
Macdonald, S. Padel, F.H. Tattersall, M.S. Wolfe, and CA. Watson. 2001.
Agronomic and environmental implications of organic farming systems. Adv. Agron.
70:261—327.

Stockdale, E.A., M.A. Shepherd, S. Fortune, and SP. Cuttle. 2002. Soil fertility in
organic farming systems—Fundamentally different? Soil Use Manage. 18:301—308

Teasdale, J.R., C. B. Coffman, and R W. Mangurn 2007. Potential Long-Term Benefits of
No-Tillage and Organic Cropping Systems for Grain Production and Soil
Improvement. Agron. J. 2007 99: 1297-1305.

Tisdall, J .M., and J .M. Oades. 1982. Organic matter and water-stable aggregates in soils.
Soil Sci. 62:141—163

USEPA, 1997. Climate change and Michigan EPA 230-F—97-008v

VandenBygaart, A.J., X.M. Yang, B.D. Kay, and J .D. Aspinall. 2002. Variability in
carbon sequestration potential in no-till soil landscape of southern Ontario. Soil
Tillage Res. 65:231—241 .

Villamil, M.B., G.A. Bollero, R.G. Darmody, F .W. Simmons, and D.G. Bullock. 2006.
No-Till Com/Soybean Systems Including Winter Cover Crops: Effects on Soil
Properties. Soil Sci. Soc. Am. J. 70: 1936-1944.

Webster, R., and M. A. Oliver. 1990. Statistical Methods in Soil and Land Resource
Survey. Oxford University Press, Oxford, England

139

 

 

Chapter V
SUMMARY AND CONCLUSIONS

As expected, management practices substantially influenced the spatial variability
pattern in soil C distribution. Total C, N content and overall variance observed at 15 year
old poplar from the surface depth (0-5 cm) was found to be closely similar to that of NT
followed by CT-cover and substantially different from that of CT. With spatial variability
characteristics, variogram values near the origin, variogram values at 1.5 m distance to
sill ratios at poplar was greater compared to other agronomic treatments, which show that

C is highly variable at very shorter distance at poplar, compared to agronomic treatments.

 

Variability characteristics of the total C, N in the studied plots were strongly affected by
topographical gradients at CT-cover and NT and least at poplar system. Overall the
results indicated that poplar system did not differ substantially with no-till system in
terms of soil C, N content and in overall C and N variability pattern in a landscape.

Comparing the spatial variability pattern between surface (0-5 cm) and below
plowing (20-30 cm) depth, at CT, spatial correlation range, nugget and sill values at
surface depth is substantially greater compared to below plow depth indicating greater
variability in surface depth compared to below plowing depth. However at CT-cover,
variability in C distribution pattern was found to be greater at even at subsurface depth.
Topographical features have major effect controlling the spatial variability pattern at CT-
cover compared to CT at both the studied depth. Inclusion of spatial variability

information substantially improved the precision of the treatment estimates in this study.

140

Relationship between topography and soil C, N was found to vary with different
tillage and management practices. Management practices significantly interacted with
topographical features as represented by WI with both C and N. Topography had a
stronger effect on soil C and N with the organic based chisel-plow system (CT-cover)
compared to the chemical input based no-till (NT) and chisel plow (CT) systems. Greater
regression slopes was observed between WI and C in CT-cover than CT; and greater
differences in C content between CT-cover and CT in valleys than uphill. Among the
landform elements, C and N content was found to be greater at valley areas compared to
upperslope position in all the studied treatments. At 0-30 cm depth, CT-cover and NT
was found to accumulate 2.2 and 1.81 Mg more C per acre than CT at uphill position and
3.7 and 6.2 Mg per acre more C than CT in the depressions. In overall, CT-cover was
found to be equally efficient in maintaining C and N content as compared to NT
treatments. Largest C and N benefits of CT-cover in the landscape were observed in
valleys as compared to uphill position in the landscape. The results demonstrate that
organic management has greater potential for C sequestration at lower landscape
positions.

With respect to long term trend in soil C storage during the past 15-20 years at

Michigan, soil C at conventional chisel-plow management has decreased by 0.9 to 1.15 g
kg '1 of soil and the never tilled grassland soil has decreased on 7.4 g kg '1 of soil since

1986 levels. Within conventional tillage system, C loss was found to be greater in
unfertilized compared to fertilized treatments. Soil C under no-till and organic based
chisel plow system tends to show no change in C content from the baseline C values. No-

till and organic tillage system might have prevented soil C loss as compared to

141

conventional management system due to lesser soil disturbance at NT and additional C
inputs in CT-cover but they did not lead to absolute gain in C. Baseline C was found to
have negative relationship with changes in C content, and C loss was found to increase
with increase in baseline values. Increase in silt and clay content leads to bring positive
changes (gain in C) in C content in no-till system. Rising winter temperature might be

one of the major factors that cause soil C loss at Michigan. In summary, none of the
studied treatment showed increase in C content from the baseline values. With the
increasing global warming potential, adoption of NT or inclusion of cover crops in the
crop rotation may be one of the prerequisite to sustain the present C content in the soils of

the temperate regions.

142

 

 

 

3 1293 03063 2073