ArticlePDF Available

Abstract and Figures

Temporal changes in soil C content vary as a result of complex interactions among different factors including climate, baseline soil C levels, soil texture, and agricultural management practices. The study objectives were: to estimate the changes in soil total C contents that occurred in the past 18 to 21 yr in soils under agricultural management and in never-tilled grassland in southwest Michigan; to explore the relationships between these changes and soil properties, such as baseline C levels and soil texture; and to simulate C changes using a system approach model (SALUS). The data were collected from two long-term experiments established in 1986 and 1988. Georeferenced samples were collected from both experiments before establishment and then were resampled in 2006 and 2007. The studied agricultural treatments included the conventional chisel-plow and no-till management systems with and without N fertilization and the organic chisel-plow management with cover crops. Total C was either lost in the conventional chisel-plowed systems or was only maintained at the 1980s levels by the conservation management systems. The largest loss in the agricultural treatments was 4.5 Mg ha(-1) total C observed in the chisel-plow system without N fertilization. A loss of 17.3 Mg ha(-1) occurred in the virgin grassland sod. Changes in C content tended to be negatively related to baseline C levels. Under no-till, changes in C were positively related to silt + clay contents. The SALUS predictions of soil C changes were in excellent agreement with the observed data for most of the agricultural treatments and for the virgin soil.
Content may be subject to copyright.
2078 SSSAJ: Volume 73: Number 6 • November–December 2009
SOIL & WATER MANAGEMENT & CONSERVATION
Soil Sci. Soc. Am. J. 73:2078-2086
doi:10.2136/sssaj2009.0044
Received 29 Jan. 2009.
*Corresponding author (kravche1@msu.edu).
© Soil Science Society of America
677 S. Segoe Rd. Madison WI 53711 USA
All rights reserved. No part of this periodical may be reproduced or
transmitted in any form or by any means, electronic or mechanical,
including photocopying, recording, or any information storage and
retrieval system, without permission in writing from the publisher.
Permission for printing and for reprinting the material contained
herein has been obtained by the publisher.
T
he soil is considered to be one of the major Earth reservoirs
of C. It holds about 1550 Pg as organic C and 950 Pg as in-
organic C (Lal, 2004). Rising global temperatures might reduce
soil C stocks by enhancing the release of stored soil organic C
as atmospheric CO
2
, still further intensifying global warming
(Jenkinson et al., 1991; Schimel et al., 1994; Bellamy et al., 2005;
Jones et al., 2005).
Recently, soil C losses were observed in several regions
across the world (Bellamy et al., 2005; Schipper et al., 2007;
Stevens and van Wesemael, 2008). In a study of more than 2000
sites across England and Wales, Bellamy et al. (2005) reported
C losses of about 0.12 kg C m
−2
yr
−1
with a net C loss of about
2.9 kg C m
−2
during the last 25 yr. Similarly, Schipper et al. (2007)
reported soil C losses of about 2.1 kg C m
−2
during the last 20 yr
in the pasturelands of New Zealand. Stevens and van Wesemael
(2008) observed an approximately 1.0 kg C m
−2
decrease from
1950–1960 levels in the soil C of temperate forest soils in the
Belgian Ardennes. Smaller scale studies conducted at several lo-
cations in the United States and Canada during the last couple
of decades also have reported soil C losses (VandenBygaart et al.,
2002; Olson et al., 2005; Varvel, 2006; Khan et al., 2007). Even
though some additional factors, such as reduction in the use of
manure, an increased tillage depth, or measurement inaccura-
cies, have been proposed as alternative explanations for reported
losses (Smith et al., 2007), on a global scale, soil C losses appear
to have been observed for a wide range of soils, land uses, and
management practices and seem to have accelerated during past
couple of decades, pointing to the recent increases in global tem-
peratures as a potential universal cause.
An increase in global temperature is likely to a ect soil C by
in uencing soil organic matter decomposition and mineraliza-
tion rates (Davidson et al., 2000; Knorr et al., 2005; Conant et
al., 2008) as well as soil and root respiration (Kirschbaum, 2004;
Jones et al., 2005). A review of data from several soil warming
experiments showed that a 3°C increase in soil temperature
increased the soil respiration rate by 20% (Rustad et al., 2001;
Richter et al., 2007).
Global climatic e ects are likely to interact with in uences
operating at smaller scales, including locally varying land use and
land management practices. Among land management practice
characteristics that greatly a ect soil C are tillage intensity and
plant residue inputs. Conventional tillage systems disrupt soil
structure and contribute to a rapid turnover of soil aggregates
(Balesdent et al., 2000), while minimal soil disturbance under
S. Senthilkumar
Dep. of Crop and Soil Sciences
Michigan State Univ.
East Lansing, MI 48824
B. Basso
Facultá Agraria
Univ. of Basilicata
Potenza, Italy
A. N. Kravchenko*
Dep. of Crop and Soil Sciences
Michigan State Univ.
East Lansing, MI 48824
G. P. Robertson
W.K. Kellogg Biological Station
Michigan State Univ.
Hickory Corners, MI 49060
Contemporary Evidence of Soil
Carbon Loss in the U.S. Corn Belt
Temporal changes in soil C content vary as a result of complex interactions among di erent
factors including climate, baseline soil C levels, soil texture, and agricultural management
practices.  e study objectives were: to estimate the changes in soil total C contents that
occurred in the past 18 to 21 yr in soils under agricultural management and in never-tilled
grassland in southwest Michigan; to explore the relationships between these changes and
soil properties, such as baseline C levels and soil texture; and to simulate C changes using a
system approach model (SALUS).  e data were collected from two long-term experiments
established in 1986 and 1988. Georeferenced samples were collected from both experiments
before establishment and then were resampled in 2006 and 2007.  e studied agricultural
treatments included the conventional chisel-plow and no-till management systems with and
without N fertilization and the organic chisel-plow management with cover crops. Total C
was either lost in the conventional chisel-plowed systems or was only maintained at the 1980s
levels by the conservation management systems.  e largest loss in the agricultural treatments
was 4.5 Mg ha
−1
total C observed in the chisel-plow system without N fertilization. A loss
of 17.3 Mg ha
−1
occurred in the virgin grassland soil. Changes in C content tended to be
negatively related to baseline C levels. Under no-till, changes in C were positively related to silt
+ clay contents.  e SALUS predictions of soil C changes were in excellent agreement with the
observed data for most of the agricultural treatments and for the virgin soil.
Abbreviations: CT, conventional tillage; CT-cover, conventional tillage with cover crops; CT-F, conventional
tillage with fertilizers; CT-NF, conventional tillage without fertilizers; LTER, Long Term Ecological
Research; NT, no-till; NT-F, no-till with fertilizers; NT-NF, no-till without fertilizers; POM, particulate
organic matter; RTM, regression to the mean; SALUS, System Approach to Land Use Sustainability.
SSSAJ: Volume 73: Number 6 • November –December 2009 2079
no-till (NT) systems is known to provide an overall positive ef-
fect on soil C storage (Franzluebbers, 2004; Puget and Lal, 2005).
Several long-term studies reported greater soil C in systems with
greater C inputs (e.g., Paustian et al., 1997), while removal of
crop residues has been o en found to reduce soil organic mat-
ter content (Huggins et al., 1998). Organic management systems
that rely on greater crop residue inputs have been reported to
increase soil organic matter content as compared to conven-
tional management systems (Pulleman et al., 2000; Robertson
et al., 2000; Stockdale et al., 2001; Marriott and Wander, 2006;
Teasdale et al., 2007).
Even though the overall e ects of land management can
be expected to be broad, their magnitudes are likely to di er in
response to variations in inherent soil characteristics, includ-
ing, among others, mineralogy (Rasmussen et al., 2006, 2008),
baseline C contents (Bellamy et al., 2005), and soil texture (Hao
and Kravchenko, 2007). A negative relationship between base-
line C contents and the change in C due to conversion to NT
management has been observed in a number of individual  eld
experiments and experiment reviews (e.g., VandenBygaart et al.,
2002; Tan et al., 2006). A relative loss in soil C in the topsoil
has also been found to increase with an increase in baseline C
values (Bellamy et al., 2005). Campbell et al. (1996) reported a
stronger relationship between soil texture (clay content) and C
content under NT than CT management at three sites in west-
ern Canada. Hao and Kravchenko (2007) observed a stronger
relationship between total C and the silt + clay content under
NT than under conventionally tilled management practices in
southwest Michigan.
e historic di culty in addressing the interactions between
rising global temperatures, land use and management practice
e ects, and intrinsic soil properties originates from the small
scale of most of the individual experiments and from di cul-
ties in meta-analyses due to the immense variability among the
individual studies in terms of management details, experiment
durations, and other factors.  e commonly encountered lack
of baseline C measurements further contributes to di culties
in comparing individual experiment results. In most of the  eld
studies, the di erences in soil C between management systems
have been examined at speci c time
points a er initiation of the experi-
ment.  e changes in soil C are thus
assessed on a relative scale via com-
parison with a certain control treat-
ment. Reliance only on comparisons
among the treatments can lead to
erroneous conclusions about soil C
changes (Khan et al., 2007). Not only
relative but also absolute changes in
soil C with time must be assessed to
monitor trends in soil C dynamics re-
lated to temporal climate change.
Modeling is a necessary tool for
spatially expanding the  ndings on C
changes obtained from individual ex-
periments to larger areas and regions,
as well as temporally to assess future
changes (e.g., Paustian et al.,1995).
For realistic predictions, however, be-
sides the soil characteristics, suitable models need to take into
account a number of management-, climate-, and plant-related
parameters. Moreover, before expanding the model predic-
tions spatially or temporally, the model performance must be
evaluated using existing data.  e System Approach to Land Use
Sustainability (SALUS) model was designed to simulate con-
tinuous crop, soil, water, and nutrient conditions under di erent
management strategies for multiple years (Basso, 2000; Basso et
al., 2006). It is equipped with functions allowing a comprehen-
sive holistic treatment of key details in plant and soil processes
under di erent land uses and management regimes in response
to daily weather variations (Fig. 1).
e objectives of this study were to: (i) examine the changes
that have taken place in soil C under di erent agricultural man-
agement practices and in virgin grassland during the past two
decades in southwest Michigan; (ii) analyze the relationships be-
tween the changes in soil C with baseline soil C and soil texture
under di erent tillage and management systems and in virgin
grassland; and (iii) assess the performance of SALUS in model-
ing the observed 20-yr trends under di erent management prac-
tices and varying baseline C levels and textures.
MATERIALS AND METHODS
e study was conducted at two long-term experiments at the
Kellogg Biological Stations Long-Term Ecological Research (LTER)
site in southwest Michigan (85°24' W, 42°24' N), referred to as the
Main Site and Interaction Site experiments. Soils at the LTER are Typic
Hapludalfs of the Kalamazoo ( ne-loamy, mixed, semiactive, mesic)
and Oshtemo (coarse-loamy, mixed, active, mesic) series (Mokma and
Doolittle, 1993).  e climate is temperate with cool, moist winters and
warm, humid summers, with approximately 90 cm of precipitation an-
nually, about half of which is snow.  e mean annual temperature is 9°C
(Grandy and Robertson, 2007).  e tilled portions of both studied sites
have been conventionally managed in row crop agriculture from at least
1950; before that, the land was under native vegetation (oak–hickory
[Quercus–Carya spp.] forest) (Robertson et al., 1993).
Fig. 1. Diagram of the components of the System Approach to Land Use Sustainability (SALUS) model.
2080 SSSAJ: Volume 73: Number 6 • November–December 2009
Main Site Experiment
e rst studied experiment is the LTER Main Site, which is a one-
factor randomized complete block design, with seven treatments and six
replications.  e experimental plots are 80 by 100 m in size.  e LTER
management systems used in this study are conventional tillage (chisel
plow) (CT-F) and no-till (NT-F) systems with conventional chemical
inputs, and a certi ed organic, chisel-plowed system with a winter legu-
minous cover crop and zero chemical inputs (CT-cover).  ese agro-
nomic treatments are in a corn (Zea mays L.)–soybean [Glycine max (L.)
Merr.]–wheat (Triticum aestivum L.) rotation. A detailed description
of the studied LTER treatments can be found at the Kellogg Biological
Station LTER website (Kellogg Biological Station, 2007).
e experiment was started in 1988. Before the experiment initia-
tion, in May of 1988, a set of 417 georeferenced soil samples was collect-
ed at the 0- to 15-cm depth from the area that then became Replicates 1
through 5 of the Main Site experiment (Robertson et al., 1997). During
the year before the experiment initiation, the entire site was planted
with corn, followed by fall-planted rye (Lolium sp.).  e 1988 data are
referred to here as the Main Site baseline samples.
In May of 2006, 50 of the locations sampled in 1988 were found
using a global positioning system (Trimble Receiver Type Pro XRS
Model 33302-51, Trimble Navigation Ltd., Sunnyvale, CA) and resam-
pled.  e sites chosen for resampling were those located within central
portions of the plots of the present LTER experiment. Overall, 11, 22,
and 17 locations suitable for resampling were found in the CT-F, NT-F,
and CT-cover treatments, respectively, with approximately three to four
locations per each plot.
At each sampling location, a hydraulic soil core unit from
Geoprobe Systems (Salina, KS) with 4.2-cm-diameter cylinder was
used to collect an undisturbed soil core of 0- to 40-cm depth.  e cores
were segmented into 0- to 15-, 15- to 20-, 20- to 30-, and 30- to 40-cm
depths.  e measurements from the cores that were used in this study
are total C, soil texture, and bulk density data for the 0- to 15-cm depth.
e 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 (Shatterbox
Model 8530, SPEX CertiPrep, Metuchen, NJ) to pass through a 250-
μm sieve. Total C was measured using an automatic Carlo-Erba CN ana-
lyzer (Carlo Erba Instruments, Milan, Italy). Soil texture analyses were
performed using the hydrometer method (Gee and Bauder, 1986).  e
average bulk density values in 2006 were 1.45, 1.50, and 1.40 g cm
−3
in
CT-F, NT-F, and CT-cover, respectively.  e di erence in bulk densi-
ties between the treatments was not statistically signi cant (P > 0.3).
Very high variability in bulk density measurements was observed among
the individual sites, which could be in part an artifact of soil compac-
tion during core sampling at some of the sites.  us, to estimate changes
in C stocks at the 0- to 15-cm depth, we used the average bulk den-
sity values. Bulk density in 1988 was assumed to be equal to that of the
CT-F treatment in 2006.
e 2006 sampling was performed during the same period of time
as the 1988 sampling (May), with corn being the previous years crop (as
in 1988).  is reduced errors associated with seasonal soil C variations
and di erences in crop residues. To minimize the errors associated with
di erences in laboratory procedures, the 1988 archived samples from all
50 locations were reanalyzed for total C using the same procedure that
was used for the analysis of the 2006 samples.  e changes in soil C are
reported based on the reanalyzed 1988 data.
Interaction Site Experiment
e LTER Interaction Site is a two-factor randomized complete
block design experiment with four replicates.  e two studied factors
are fertilization with two levels, fertilized and unfertilized, and tillage
with two levels, chisel plow tillage and NT. Each experimental plot is
27 by 40 m in size.  e experiment is in a corn–soybean–wheat rota-
tion. In the fertilized plots, urea is applied as a source of N (45 kg ha
−1
)
during the wheat years and liquid N fertilizer (28% active ingredient) is
applied during the corn years according to Michigan corn recommenda-
tions. No fertilizer is applied when soybean is grown. During the last
5 yr, neither P nor K were applied even on the fertilized plots.
In addition to the agronomic treatments, we studied two virgin
grassland sites adjacent to the Interaction Site experiment.  e only
management that has been done in the virgin grassland is mowing, con-
ducted every year since 1960 in late fall. A er the mowing, the plant
residues are le on the surface.
e Interaction Site experiment was established in 1986. In May
1986 before starting the experiment, 256 samples were collected from
the tilled portions and 65 samples from the virgin portions of the site
at the 0- to 20-cm depth (Robertson et al., 1993). Sampling locations
were georeferenced by taking the southwest corner of the Interaction
Site as a reference point and de ning the x and y coordinates of the
sampled locations as distances from the reference point. In the year
before soil sampling, the tilled portion of the site was planted to soy-
bean followed by a winter cover of annual ryegrass (Lolium multi orum
Lam.). Unfortunately, there were no archived samples available for the
Interaction Site experiment. In 1986, soil C measurements were con-
ducted using the Walkley–Black wet combustion technique (Nelson
and Sommers 1982).  e 1986 data are referred to here as baseline
samples for the Interaction Site.
In May of 2007, we collected a total of 57 samples from locations
sampled in 1986. Locations for resampling were selected from central
portions of the experimental plots. Approximately two to four samples
were collected in each plot. At each location, the sample was taken be-
tween the plant rows and was composited from three 2.5-cm-diameter
cores collected within a 0.2-m radius at the 0- to 20-cm depth. In addi-
tion, undisturbed samples of 4 cm in diameter and 7.6-cm height were
collected from each site for bulk density measurements.
In 2007, main e ects of tillage (P > 0.13), fertilizer (P > 0.30), or
interaction e ects (P > 0.80) on bulk density were not statistically sig-
ni cant. We used the average bulk density values of 1.37 and 1.42 g cm
−3
observed for conventionally tilled and NT treatments, respectively, for
estimating changes in C stocks on an areal basis at the 0- to 20-cm depth.
Bulk density in 1986 was assumed to be equal to that of the convention-
ally tilled treatments. For the virgin sites, the average bulk density mea-
sured in 2007 was 1.12 g cm
−3
; this value was assumed to be applicable
to the 1986 data as well.  e lack of bulk density measurements from
the 1980s is a drawback in the C stock calculations of this study that
reduces accuracy in the estimates of C stocks; however, it does not a ect
C concentration results.
As in the Main Site experiment, the sampling time (May) and the
previous year’s crop (soybean) were the same in both 1986 and in 2006.
e procedures of sample cleaning, processing, and analyses for total C
and soil texture were the same as those described for the Main Site above.
A summary of the studied treatments, numbers of samples, sam-
pling depths and timing, and preceding crops for both experiments are
shown in Table 1.
SSSAJ: Volume 73: Number 6 • November –December 2009 2081
SALUS Model
e SALUS models continuous crop, soil, water and nutrient
conditions under di erent management strategies for multiple years
(Basso, 2000; Basso et al., 2006).  e model is composed of three main
structural components: (i) a set of crop growth modules; (ii) a soil wa-
ter balance and temperature module; and (iii) a soil organic matter and
nutrient cycling module.  e model executes all components daily for
each management strategy being run.  e crop growth and develop-
ment component accounts for environmental conditions (particularly
temperature and light) when calculating potential plant growth rates.
is growth is then reduced based on water and N limitations.  e wa-
ter balance considers surface runo , in ltration, surface evaporation,
saturated and unsaturated soil water  ow, drainage, root water uptake,
soil evaporation, and transpiration.
e soil organic matter and nutrient module simulates organic
matter decomposition, N mineralization, and the formation of NH
3
and NO
3
, N immobilization, gaseous N losses, and three pools of
P. e soil organic matter (SOM) and N module is derived from the
Century model (Parton et al., 1988), with a number of modi cations in-
corporated.  e model simulates organic matter and N mineralization
and immobilization from three SOM pools (active, slow, and passive),
which vary in their turnover rates and characteristic C/N ratios.  ere
are two crop residue and fresh organic matter pools (structural and met-
abolic), for representing recalcitrant and easily decomposable residues,
based on residue lignin and N content. Decomposition and N min-
eralization rates for di erent pools are in uenced by soil temperature
and moisture, soil texture, and tillage intensity (as well as the pool C/N
ratio for N mineralization).  e main external inputs needed for the
soil process module are soil texture, bulk density, horizon depths, total
organic C and N, and initial mineral N content. Several modi cations
were made to adapt the model for use with daily-time-step crop growth
routines.  e original Century model operates on a monthly time step
and, therefore, rate constants were recalibrated to correct for the di er-
ence in integration interval (from monthly to daily). A surface active
SOM pool associated with the surface residue pools was added to better
represent conservation tillage systems and perennial crops. Soil organic
matter and litter pools were also added for up to 10 soil layers (vs. only
a single topsoil layer in Century).  e soil moisture control function
for decomposition was replaced to make decomposition a function of
water- lled pore space. Separate NH
3
and NO
3
pools were represented
with nitri cation rate calculations. An algorithm was developed that
determines the initial fraction of total organic matter C and N in each
of the three SOM pools (for model initialization) as a function of soil
texture, the type of original native vegetation, and time under cultiva-
tion, based on a steady-state analytical solution of the decomposition
equations (Paustian et al., 1992).
Data Analyses
General Statistical Analyses
Data analysis was conducted using SAS (SAS Institute, 2001).
e e ects of the studied factors and continuous covariates, includ-
ing baseline C and texture, were assessed using analysis of covariance
(ANCOVA) in PROC MIXED (Milliken and Johnson, 2002). For
the Main Site experiment, the statistical model included treatments as
a  xed factor, and blocks and plots nested within treatments as random
factors, with plots used as an error term for testing the treatment e ects.
For the Interaction Site, the statistical model included  xed e ects of
tillage, fertilization, and their interaction, and random e ects of blocks
and plots.
Mean baseline C data were varied among the treatments (Table 1).
To account for these initial variations when evaluating the changes in soil
C and making comparisons among the treatments, the baseline C values
were  rst standardized by subtracting the respective treatment means
and then used as a covariate in C change data analyses. Standardization
is the customary practice in ANCOVA for dealing with covariates that
are di erent among the studied treatments (Milliken and Johnson, 2002).
In both experiments, the assumptions of normality of the re-
siduals and homogeneity of variances were checked and analysis with
heterogeneous variances was conducted, if found necessary, using the
REPEATED/GROUP statement in PROC MIXED. When the  xed
e ects were found to be statistically signi cant, either at P < 0.05 or
at P < 0.10, the comparisons between the treatments were conducted
using t-tests.
Regression to the Mean Correction
Regression analysis of the relationships between the changes in
soil C and baseline C values was conducted with accounting for the
regression-to-the-mean (RTM) phenomenon.  e RTM occurs when
unusually large or small measurements obtained by chance during base-
line sampling are followed by measurements that are less extreme and
closer to the mean in the subsequent sampling.  us, even in the ab-
sence of the real relationship between the changes and the baseline lev-
els, larger change values are o en associated with more extreme baseline
values, leading to negative sample correlation (Barnett et al., 2005; Lark
Table 1. Summary of data collection for individual treatments at the Kellogg Biological Station Long-Term Ecological Research
Main Site (1988–2006) and Interaction Site (1986–2007) experiments.
Treatment†
N
Sampling depth Sampling time Previous crop Analysis method
Main Site
1988 2006 1988 2006 1988 2006 1988 2006
CT-F
11 0–15 0–15 May May corn corn
Carlo-Erba CN analyzer Carlo-Erba CN analyzer
NT-F
22 0–15 0–15 May May corn corn
CT-cover
17 0–15 0–15 May May corn corn
Interaction Site 1986 2007 1986 2007 1986 2007
1986 2007
CT-F
14 0–20 0–20 May May soybean soybean
Walkley-Black method Carlo-Erba CN analyzer
CT-NF
10 0–20 0–20 May May soybean soybean
NT-F
12 0–20 0–20 May May soybean soybean
NT-NF
11 0–20 0–20 May May soybean soybean
Virgin grassland 13 0–20 0–20 May May Walkley-Black method Carlo-Erba CN analyzer
The agronomic treatments are chisel-plow (CT) or no-till (NT) in a corn–soybean–wheat rotation either with (F) or without (NF) fertilization; CT-
cover is a certifi ed organic treatment with leguminous cover crops; the virgin grassland was mowed annually since 1960s.
‡ Number of samples co-located at each treatment.
2082 SSSAJ: Volume 73: Number 6 • November–December 2009
et al., 2006). To overcome the RTM e ect and to get unbiased estimates
of regression slopes and con dence intervals for regressions between
changes in soil C and baseline C values, we used a method proposed
by Blomqvist (1977) (Lark et al., 2006).  e sample regression slope
obtained from the ordinary simple linear regression between the change
and the baseline,
ˆ
β
, is adjusted as follows:
()
adjusted
ˆ
1
ˆ
V
V
β+
β=
[1]
where
adjusted
ˆ
β
is the adjusted regression slope obtained a er accounting
for the RTM e ect. e V component is calculated as
2
u
2
z
1
S
V
S
=−
where S
z
2
is the overall sample variance of the baseline C data and S
u
2
is the variance of the baseline data that is due to all the sources of un-
certainty in baseline measurements, estimated independently.  ese
include analytical errors and spatial variability due to imperfect identi-
cation of the site locations during resampling (Lark et al., 2006).  e
S
u
2
values were obtained using sample variograms of the baseline C val-
ues.  e variograms were calculated and  tted with variogram models
using the SAS procedures PROC VARIOGRAM and PROC NLIN.
e value of S
u
2
was set to be equal to the  tted variogram value at a
3-m lag distance.  is value was considered to be representative of the
spatial variability of the baseline C data within a 3-m area around the
sampled site.  is 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, S
u
2
accounts for baseline C variability
occurring at distances <3 m and includes the analytical errors as well.
Conclusions on whether the adjusted regression slopes are di erent
from zero have been reached based on examination of the con dence
intervals for the adjusted regression slopes calculated as described by
Blomqvist (1977). If the con dence intervals were found to include
zero, the regression slope was concluded to be not signi cantly dif-
ferent from zero.
e intercept for the RTM-adjusted regression equation,
0
ˆ
β
,
was obtained based on the value of
adjusted
ˆ
β
, the mean baseline C value,
x
, and the mean value for the C change,
y
:
0 adjusted
ˆˆ
yxβ= β
e relationships between the changes in soil C and baseline
values presented here are the results obtained a er accounting for
the RTM e ect.
RESULTS
In the Main Site, the mean C contents of contemporary
samples (2006) under NT-F (9.9 g C kg
−1
soil) and CT-cover
(9.8 g C kg
−1
soil) treatments were signi cantly greater than
that of CT-F (8.2 g C kg
−1
soil) (Table 2). In the Interaction
Site, the interaction between tillage and fertilization factors
was considered to be statistically signi cant (P = 0.07).  us
we report the results for the individual treatments.  e mean
contemporary C content of the CT-NF treatment tended to
be less than that of the CT-F treatment (P = 0.09) but there
were no di erences among the other treatments (Table 2).
e variability in the Interaction Site baseline (1986) C data
was found to be much higher than in 2007, which can be ex-
plained by di erent methods used for sample pretreatment
and analyses.  e soil C content of the virgin sites was much
higher than those of the agronomic treatments in both 1986 and
2007 (Table 2).
In none of the treatments of either the Main or Interaction
sites did C content increase from the baseline C levels. In the
Main Site, the mean changes in C from 1988 to 2006 adjusted
for the common level of baseline C values in the eNT-F and CT-
cover treatments were not signi cantly di erent from zero, in-
dicating no change in the C content over the baseline values. In
the CT-F treatment, the mean change was signi cantly less than
zero, indicating C loss.  e loss of C in the CT-F treatment was
signi cantly greater than that in the CT-cover treatment, while
the NT and CT-cover treatments were not signi cantly di erent
from each other (P < 0.1) (Table 2).
At the Interaction Site, the mean C change in the CT-NF
treatment was less than zero and less than those of the NT-F and
NT-NF treatments.  e mean C changes in the CT-F, NT-F,
and NT-NF treatments were not di erent from zero and not dif-
ferent from each other either. Carbon loss at the virgin site was
signi cantly greater than zero and greater than the losses of the
agronomic treatments (Table 2).
At the Main Site, even though all raw regression slopes for
the relationships between baseline C and the change in C were
negative (data not shown), a er RTM adjustment only the re-
gression slope for the CT-cover data was found to be less than
zero (Fig. 2a). At the Interaction Site, a signi cant negative rela-
tionship was observed between baseline C values and the change
in C values from 1986 to 2007 in all the studied treatments.  e
plots are shown in Fig. 2b, with fertilization treatments com-
bined for clarity.
Among the studied treatments, a signi cant positive rela-
tionship between changes in C content and the silt + clay con-
Table 2. Means and standard errors (in parentheses) for baseline
(1980s) and contemporary (2000s) soil C contents and for changes
in soil C content adjusted for standardized baseline C in the Main
Site and Interaction Site experiments; P values are from an ANCOVA
for testing the signifi cance of the factor effects.
Treatments† Soil C Change in soil C
——————— g kg
−1
of soil —————
Main site experiment
1988 2006 From 1988 to 2006
CT-F
9.3 (0.7) a 8.2 (0.3) a** −1.15 (0.27) a*
NT-F
10.6 (0.5) a 9.9 (0.5) b −0.71 (0.48) ab NS‡
CT-cover
10.0 (0.5) a 9.8 (0.3) b −0.16 (0.29) b
NS
P value
0.13 0.002 0.06
Interaction site experiment 1986 2007 From 1986 to 2007
CT-F
9.8 (1.4) a 9.5 (0.5) a −0.22 (0.46) ab
NS
CT-NF
9.8 (1.4) a 8.2 (0.5) b −1.54 (0.51) a
NT-F
7.4 (1.4) a 8.3 (0.5) ab 0.31 (0.48) b
NS
NT-NF
8.7 (1.4) a 8.9 (0.5) ab 0.19 (0.51) b
NS
P value
0.11 0.07 0.07
Virgin grassland 23.0 (0.9) 15.3 (0.6) −7.7 (1.0)
* Means within the same column within the experiment followed by the
same letter are not signifi cantly different at α = 0.1.
** Means within the same column within the experiment followed by the
same letter are not signifi cantly different at α = 0.05.
The agronomic treatments are chisel-plow (CT) or no-till (NT) in a corn–
soybean–wheat rotation either with (F) or without (NF) fertilization; CT-cover
is a certifi ed organic treatment with leguminous cover crops; the virgin
grassland was mowed annually since 1960s.
‡ NS, the mean value is not signifi cantly different from zero (P < 0.05).
SSSAJ: Volume 73: Number 6 • November –December 2009 2083
tent was observed only for the NT-F treatment at the Main Site,
with an R
2
value of 0.32, and for the NT-NF treatment at the
Interaction Site, with an R
2
value of 0.44 (Table 3).
DISCUSSION
e studied experiments used the same treatments for the
entire experiment duration and were located on land with a simi-
lar land use history.  is is a desirable characteristic for a study
addressing the temporal changes in soil C, since many factors,
especially historical land use, may a ect present soil C levels
and, potentially, soil C change rates (Stevens and van Wesemael,
2008). Another advantage of our study is the availability of geo-
referenced baseline data, which contributed to reducing errors
due to spatial variation. Yet another advantage is that at least in
one of the two experiments, we were able to use archived samples
reanalyzed with the same techniques as those used for contem-
porary samples.
Overall, the observed changes in C content varied with
the management system. At the Main Site, total C under CT
management has decreased by 12%. Under NT-F and CT-
cover treatments, there were no C losses but there were not any
C gains either. Note that in 2006 the total C values under the
NT-F treatment and the CT-cover treatment were higher than
those of the CT management (Table 2).  us, if only contem-
porary data were available, we could have concluded that NT-F
and CT-cover managements increased soil C.  e observed
contemporary di erences, however, are essentially re ecting the
C losses sustained by the soil under CT-F rather than gains in
NT-F and CT-cover. Our results indicate that while contempo-
rary comparisons among management practices provide valuable
information on the relative di erences generated due to human-
induced management, to obtain a comprehensive perspective on
soil C changes, the relative di erences must be examined along
with the absolute changes.
Similar to our study, Hendrix et al. (1998) observed a de-
cline in soil C content from the baseline status under CT at a 0-
to 20-cm depth in a 16-yr experiment performed in a  ne-loamy
soil in Georgia. Doran et al. (1998) observed C losses at 0 to 30
cm in a >20-yr-old experiment in a silt loam soil in Nebraska.
Both studies reported that the magnitude of C losses was greater
under CT than NT.  e lack of C gain under the NT treatment
observed in this study was consistent with the results of Puget
and Lal (2005) at the 0- to 30-cm depth in an 8-yr-old experi-
ment on a Mollisol of central Ohio; and of Eynard et al. (2005)
at the 0- to 20-cm depth in a 16-yr-old experiment on Ustolls
of South Dakota. In Mead, NE, on a Typic Argiudoll following
the  rst 8 yr (1984–1992) of gains in soil C in a corn–soybean
rotation, Varvel (2006) observed 3 to 5% losses from 1992 to
2002. VandenBygaart et al. (2002) observed C losses in Orthic
and Gleyic Luvisols in a majority of sampled sites at four agricul-
tural  elds in southern Ontario from 1985 to 2000.
e rate of change in C contents in any management sys-
tem depends on the balance between soil C inputs through plant
residues and C losses, which occur mainly through decomposi-
tion.  is balance exists for a particular environmental setting
that a ects decomposition rates, including air and soil tempera-
tures and soil moisture. Because the temperature sensitivity of
organic matter decomposition decreases with increasing tem-
perature, soil C losses can be expected to be higher in cold and
temperate regions than in the tropics (Kirschbaum, 1995, 2000).
An increase in global temperature by 1°C could lead to a loss of
>10% of soil C in regions with an annual mean temperature of
5°C (Kirschbaum, 1995). In Michigan, the mean air temperature
has increased during the last century from 8.1°C (1876–1905
average) to 8.7°C (1962–1991 average; USEPA, 1997). During
the next century, however, predictions of the Intergovernmental
Panel on Climate Change (IPCC) and the UK Hadley Centre’s
climate model (HadCM2) project an increase in mean air tem-
perature of 2.2°C (USEPA,1997). Data from the LTER weather
station adjacent to the studied experiments indeed indicate a re-
cent winter warming trend.  e number of days per year with
minimum daily temperature above freezing has been steadily
increasing during the past 15 yr (Fig. 3). Even though it can-
not be unequivocally concluded from this study that the recent
warming trends are the cause of the observed C losses, there is
undoubtedly an association between the two.  is is especially
true for the virgin grassland.  e warming trend appears to be
the only recently changed in uence on this treatment, which
otherwise remains managed in the same manner since the 1960s.
Fig. 2. Changes in total soil C plotted vs. baseline total soil C values
for (a) the Main Site experiment (1988–2006) and (b) the Interaction
Site experiment (1986–2007). Symbols represent observed data (CT,
chisel plow; NT, no-till; F, fertilizer added; CT-cover, chisel plow with
cover crops; Never-tilled, virgin grassland). The lines represents the
regression equation plots adjusted for the regression-to-the-mean
effect. *Regression equations where the slopes are signifi cantly
different from zero.
2084 SSSAJ: Volume 73: Number 6 • November–December 2009
e predictions of soil C changes obtained using SALUS
were consistent with the observed results (Fig. 4).  e model
predictions were in excellent agreement with the observed data
in the CT-F and NT-F treatments of both the Main Site and
Interaction Site experiments.  e model also predicted well the
observed large soil C loss in CT-NF treatment of the Interaction
Site experiment.  e model slightly overestimated C losses in the
CT-cover treatment. For the NT-NF treatment, the model pre-
dicted losses of C that were comparable with the prediction for
the CT-NF treatment; however, the observations for this treat-
ment indicated no statistically signi cant change. One of the rea-
sons for the discrepancy between the model and observed results
is that the NT-NF data from 1986 had a relatively large propor-
tion of sites with very low C. In 1986 in this treatment, four out
of a total of 11 observations had total C < 0.5%, while the other
treatments had one or at most two observations with total C <
0.5%. It is possible that SALUS did not correctly model soil C
related processes in such extreme settings of very low initial C
values with no fertilization and its further assessments at such
settings are necessary.
e observed results and model predictions of this study
agree well with the study of the potential e ects of global warm-
ing on soil C storage under CT and NT management practices
at the LTER by Paustian et al. (1995).  ey evaluated soil C stor-
age projections under the scenario of the mean annual tempera-
ture increasing by 2°C by 2050 based on IPCC estimates using
the Century model.  e simulation assumed that no changes
in management or in C input levels were made in the conven-
tional tillage system.  e model predicted a decline in C storage
from the baseline status under CT by taking baseline C values as
3 kg C m
−2
for 0 to 20 cm observed in the mid-1980s.
Baseline Carbon Levels
Negative relationships between baseline C and the change
in C content observed in this study were consistent with the
results of VandenBygaart et al. (2002) and Bellamy et al. (2005)
as well as with other studies that looked at the e ects of conser-
vation management practices in relation to baseline C (Tan et al.,
2006). Stevens and van Wesemael (2008) also observed greater
losses in soil C in the Belgian Ardennes associated with higher
initial C content values, and C gains at sites with lower initial C
contents. As in this study, their regression results were corrected
for the RTM phenomenon.  e virgin soils of this study that had
twice as much C as agricultural soils in 1986 lost 33% of their
initial C, compared with an average of 10% loss observed in the
conventionally tilled and conventionally managed agricultural
systems of this study.
One possible mechanism behind the negative relationship
between baseline C levels and C change might be related to a
soil’s saturation with C (Hassink, 1997; Six et al., 2002). Carbon
saturation capacity is de ned as the maximum amount of C that
can be sequestered by a soil under speci c climatic and manage-
ment 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 C content, the sites able to
protect soil C by physical and chemical mechanisms might be
undersaturated with C. In soils with higher baseline C contents,
most of the protective sites might be already saturated with C.
us, more C is being stored there in labile forms as light fractions
Table 3. Relationships between changes in soil C and soil silt +
clay contents.
Treatment† Regression model‡
R
2
Main site experiment
CT-F
ΔC = −0.11 − 0.00003(silt + clay) NS§
NT-F
ΔC = −0.68 + 0.01(silt + clay) 0.32
CT-cover
ΔC = −0.12 + 0.002(silt + clay) NS
Interaction site experiment
CT-F
ΔC = −1.8 + 0.03(silt + clay) NS
CT-NF
ΔC = 1.22 − 0.023(silt + clay) NS
NT-F
ΔC = −0.24 + 0.003(silt + clay) NS
NT-NF
ΔC = −4.3 + 0.08(silt + clay) 0.44
Virgin grassland ΔC = −0.49 + 0.004(silt + clay) NS
The agronomic treatments are chisel-plow (CT) or no-till (NT)
in a corn–soybean–wheat rotation either with (F) or without (NF)
fertilization; CT-cover is a certifi ed organic treatment with leguminous
cover crops; the virgin grassland was mowed annually since 1960s.
ΔC = C content in 2006 or 2007 minus C content in 1986 or 1988.
§ NS, regression slope is not signifi cantly greater than zero at α = 0.05.
Fig. 3. Number of days per year with minimum air temperatures
above 0°C recorded by the weather station at the Kellogg Biological
Station Long-Term Ecological Research site from 1988 to 2006.
Fig. 4. The observed changes in total C at the Main Site and Interaction
Site experiments and the predictions obtained using SALUS (CT,
chisel plow; NT, no-till; F, fertilizer added; NF, no fertilizer added; CT-
cover, chisel plow with cover crops; Never-tilled, virgin grassland).
At the Main Site, the changes were from 1988 to 2006 at the 0- to
15-cm depth, and at the Interaction Site the changes were from 1986
to 2007 at the 0- to 20-cm depth.
SSSAJ: Volume 73: Number 6 • November –December 2009 2085
and particulate organic matter (POM), the forms that could be
physically protected only within soil aggregates.  e amount of
unprotected C being held at equilibrium with the environment
is more strongly a ected by variations in environmental factors
a ecting 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 rela-
tively high baseline C in several experiments supports the satura-
tion capacity hypothesis (Hassink and Whitmore, 1997; Six et
al., 2002). Soils with both low and high baseline C values, when
subjected to an increase in temperature or heavy soil disturbance,
will lose C; however, high-baseline-C soils, which have more un-
protected C, will lose more C than lower baseline C soils.
In soils under agricultural management, soil C losses due to
an increase in the decomposition rate could be partly compen-
sated by increases in agricultural productivity due to improve-
ment in crop cultivars and the use of external chemical inputs.
In Michigan, annual agricultural productivity was reported to
increase by about 2.6% per year during 1960 to 1999 (www.ers.
usda.gov/Data/AgProductivity; veri ed 7 Sept. 2009).  ere
might not be such compensation at virgin sites, where the
productivity remained stable during the studied time period.
Paustian et al. (1995) predicted that baseline C can be sustained
or marginally increased if the agricultural productivity increases
annually by 1.4%, which might result in 40% more residue re-
turn compared with the present residue return rate.
Effect of Soil Texture
Positive correlations between changes in soil C and the
silt + clay contents in the NT treatments and no signi cant
relationships in the plowed treatments (CT and CT-cover) ob-
served in this study are consistent with literature reports of the
relationships between soil texture and soil total or organic C
(Needelman et al., 1999; Hao and Kravchenko, 2007). Besides
other mechanisms, greater silt + clay content contributes to
greater soil C storage also through enhancing the formation of
micro- and macroaggregates, which provide greater stability and
protection to soil organic matter.
e soil environment under NT is much more conducive
to aggregate formation than that of conventionally tilled soils.
Tillage limits the soils potential to protect C within aggregates,
since aggregate formation is greatly disrupted by plowing (Puget
et al., 1995; Paustian et al., 1997). For example, Beare et al.
(1994) observed that the pool of physically protected C in mac-
roaggregates accounted only for 10% of soil C stocks in a tilled
soil but 19% in NT soils.  e lack of relationship between C and
soil texture in the CT-cover treatment could be further due to
the fact that the greater C contents of organic-based systems are
o en mainly attributed to increases in POM. For example, an
organic-based system has been reported to have 30 to 40% more
POM than a conventional system (Marriott and Wander, 2006).
e POM content, however, typically is found not to be directly
related to soil texture (Plante et al., 2006; Franzluebbers and
Arshad, 1997; Puget and Lal, 2005).
CONCLUSIONS
e results of this study indicate that in the past 18 to
20 yr, the total C in soil under continuous, conventional, chis-
el-plowed management has declined. Carbon losses were the
greatest in conventionally tilled soil without N fertilization.
No C gains were observed in either of the two agricultural man-
agements that are regarded as conservational practices, i.e., NT
and organic management with cover crops.  at is, the conser-
vation management practices appeared only to have prevented
total C losses compared with conventional tillage management.
Greater losses were o en associated with greater baseline C val-
ues, while gains were more likely to be observed where baseline
C was low. Soil under virgin continuous grassland, which in
1986 had more than twice as much C as soils of the agricultural
part of the study, experienced the greatest loss. Changes in soil
C were positively correlated with the silt + clay content in NT,
but not in the CT treatments.
Associated with the observed C losses, a tendency for higher
winter temperatures was observed during the last two decades
of weather recorded by the weather station located in the study
area. An increase in temperature might be one of the factors that
contributed to the soil C losses. With the projected increase in
global temperature, the adoption of NT or the inclusion of cover
crops in the crop rotation may be one of the prerequisites to sus-
taining the present soil C levels.
REFERENCES
Balesdent, J., C. Chenu, and M. Balabane. 2000. Relationship of soil organic
matter dynamics to physical protection and tillage. Soil Tillage Res.
53:215–230.
Barnett, A.G., J.C. van der Pols, and A.J. Dobson. 2005. Regression to the mean:
What it is and how to deal with it. Int. J. Epidemiol. 34:215–220.
Basso, B. 2000. Digital terrain analysis and simulation modeling to assess spatial
variability of soil water balance and crop production. Ph.D. diss. Michigan
State Univ., East Lansing.
Basso, B., J.T. Ritchie, P.R. Grace, and L. Sartori. 2006. Simulation of tillage
systems impact on soil biophysical properties using the SALUS model. Int.
J. Agron. 4:677–688.
Beare, M.H., M.L. Cabrera, P.F. Hendrix, and D.C. Coleman. 1994. Aggregate-
protected and unprotected organic matter pools in conventional-tillage
and no-tillage soils. Soil Sci. Soc. Am. J. 58:787–795.
Bellamy, P.H., P.J. Loveland, R.I. Bradley, R.M. Lark, and G.J.D. Kirk. 2005.
Carbon losses from all soils across England and Wales. Nature 437:245–248.
Blomqvist, N. 1977. On the relation between change and initial value. J. Am.
Stat. Assoc. 72:746–749.
Campbell, C.A., B.G. McConkey, R.P. Zentner, F. Selles, and D. Curtin. 1996.
Long-term e ects 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. Global Change Biol. 14:868–877.
Davidson, E.A., S.E. Trumbore, 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 Tillage Res. 49:3–18.
Eynard, A., T.E. Schumacher, M.J. Lindstrom, and D.D. Malo. 2005. E ects of
agricultural management systems on soil organic carbon in aggregates of
Ustolls and Usterts. Soil Tillage Res. 81:253–263.
Franzluebbers, A.J. 2004. Tillage and residue management e ects on soil organic
matter. p. 227–268. In F. Magdo and R.R. Weil (ed.) Soil organic matter
in sustainable agriculture. CRC Press, Boca Raton, FL.
Franzluebbers, A.J., and M.A. Arshad. 1997. Particulate organic carbon content
and potential mineralization as a ected by tillage and texture. Soil Sci. Soc.
Am. J. 61:1382–1386.
Gee, G.W., and J.W. Bauder. 1986. Particle-size analysis. p. 383–412. In A. Klute
(ed.) Methods of soil analysis. Part 1. Physical and mineralogical methods.
2nd ed. Agron. Monogr. 9. ASA and SSSA, Madison, WI.
Grandy, A.S., and G.P. Robertson. 2007. Land use intensity e ects on soil C
accumulation rates and mechanisms. Ecosystems 10:59–74.
2086 SSSAJ: Volume 73: Number 6 • November–December 2009
Hao, X., and A.N. Kravchenko. 2007. Management practice e ects on surface
soil total carbon: Di erences along a textural gradient. Agron. J. 99:18–26.
Hassink, J. 1997.  e 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 A.P. Whitmore. 1997. A model of the physical protection of
organic matter in soils. Soil Sci. Soc. Am. J. 61:131–139.
Hendrix, P.F., A.J. Franzluebbers, and D.V. McCracken. 1998. Management
e ects on C accumulation and loss in soils of the southern Appalachian
Piedmont of Georgia. Soil Tillage 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 L.G. Bundy. 1998. Soil organic C
in the tallgrass prairie-derived region of the Corn Belt: E ects of long-term
crop management. Soil Tillage Res. 47:219–234.
Jenkinson, D.S., D.E. Adams, and A. Wild. 1991. Model estimates of CO
2
emissions from soil in response to global warming. Nature 351:304–306.
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 Biol. 11:154–166.
Kellogg Biological Station. 2007. 2007 Agronomic protocol: Long-term
ecological research (LTER) in row-crop agriculture. Available at houghton.
kbs.msu.edu/data/agronomic_protocol/2007AgronomicProtocol.pdf
(veri ed 6 Sept. 2009). KBS, Hickory Corners, MI.
Khan, S.A., R.L. Mulvaney, T.R. Ellsworth, and C.W. Boast. 2007.  e myth
of nitrogen fertilization for soil carbon sequestration. J. Environ. Qual.
36:1821–1832.
Kirschbaum, M.U.F. 1995.  e temperature dependence of soil organic matter
decomposition and the e ect of global warming on soil organic carbon
storage. Soil Biol. Biochem. 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
Biol. 10:1870–1877.
Knorr, W., I.C. Prentice, J.I. House, and E.A. Holland. 2005. Long-term
sensitivity of soil carbon turnover to warming. Nature 433:298–301.
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.
Marriott, E.E., 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 D.E. Johnson. 2002. Analysis of messy data: Designed
experiments. CRC Press, Boca Raton, FL.
Mokma, D.L., and J.A. Doolittle. 1993. Mapping some loamy Al sols in
southwestern Michigan using ground-penetrating radar. Soil Surv. Horiz.
31:71–78.
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 L.E. Sommers. 1982. Total carbon, organic carbon, and
organic matter. p. 539–580. In A.L. Page et al. (ed.) Methods of soil
analysis. Part 2. Chemical and microbiological properties. 2nd ed. Agron.
Monogr. 9. ASA and SSSA, Madison, WI.
Olson, K.R., J.M. Lang, and S.A. Ebelhar. 2005. Soil organic carbon changes a er
12 years of no-tillage and tillage of Grantsburg soils in southern Illinois.
Soil Tillage Res. 81:217–225.
Parton, W.J., J.W.B. Stewart, and C.V. Cole. 1988. Dynamics of C, N, P and S in
grassland soils: A model. Biogeochemistry 5:109–131.
Paustian, K., H.P. Collins, and E.A. Paul. 1997. Management controls on soil
carbon. p. 15–49. In E.A. Paul (ed.) Soil organic matter in temperate
agroecosystems: Long-term experiments in North America. CRC Press,
Boca Raton, FL.
Paustian, K., W.J. Parton, and J. Persson. 1992. Modeling soil organic matter in
organic-amended and nitrogen-fertilized long-term plots. Soil Sci. Soc.
Am. J. 56:476–488.
Paustian, K., G.P. Robertson, and E.T. Elliott. 1995. Management impacts on
carbon storage and gas  uxes (CO
2
, CH
4
) in mid-latitude cropland and
grassland ecosystems. p. 69–84. In R. Lal et al. (ed.) Soil management and
Greenhouse E ect. Adv. Soil Sci. CRC Press, Boca Raton, FL.
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. 70:287–296.
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 a ected by tillage and land use. Soil Tillage Res. 80:201–213.
Pulleman, M.M., J. Bouma, E.A. van Essen, and E.W. Meijles. 2000. Soil organic
matter content as a function of di erent land use history. Soil Sci. Soc. Am.
J. 64:689–693.
Rasmussen, C., R.J. Southard, and W.R. Horwath. 2006. Mineral control of
organic carbon mineralization in a range of temperate conifer forests.
Global Change Biol. 12:834–847.
Rasmussen, C., R.J. Southard, and W.R. Horwath. 2008. Litter type and soil
minerals control temperate forest soil carbon response to climate change.
Global Change Biol. 14:2064–2080.
Richter, D.B.J., M. Hofmockel, M.A. Callaham, Jr., D.S. Powlson, and P. Smith.
2007. Long-term soil experiments: Keys to managing Earths rapidly
changing ecosystems. Soil Sci. Soc. Am. J. 71:266–279.
Robertson, G.P., J.R. Crum, and B.G. Ellis. 1993.  e 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 B.G.
Ellis. 1997. Soil resources, microbial activity, and primary production
across an agricultural ecosystem. Ecol. Appl. 7:158–170.
Robertson, G.P., E.A. Paul, and R.R. Harwood. 2000. Greenhouse gases in
intensive agriculture: Contributions of individual gases to the radiative
forcing of the atmosphere. Science 289:1922–1925.
Rustad, L.E., J.L. Campbell, and G.M. Marion. 2001. A meta-analysis of the
response of soil respiration, net nitrogen mineralization, and aboveground
plant growth 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. Par tt, C. Ross, J.J. Claydon, and G. Arnold.
2007. Large losses of soil C and N from soil pro les under pasture in New
Zealand during the past 20 years. Global Change Biol. 13:1138–1144.
Six, J., R.T. Contant, E.A. Paul, and K. Paustian. 2002. Stabilization mechanisms
of soil organic matter: Implications for C-saturation of soils. Plant Soil
241:155–176.
Smith, P., S.J. Chapman, W.A. Scott, H.I.J. Black, M. Wattenbach, R. Milne, et
al. 2007. Climate change cannot be entirely responsible for soil carbon
loss observed in England and Wales, 1978–2003. Global Change Biol.
13:2605–2609.
Stevens, A., and B. van Wesemael. 2008. Soil organic carbon stock in the Belgian
Ardennes as a ected by a orestation and deforestation from 1868 to 2005.
For. Ecol. Manage. 256:1527–1539.
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 C.A. Watson.
2001. Agronomic and environmental implications of organic farming
systems. Adv. Agron. 70:261–327.
Tan, Z., R. Lal, and S. Liu. 2006. Using experimental and geospatial data to
estimate regional carbon sequestration potential under no-till management.
Soil Sci. 171:950–959.
Teasdale, J.R., C.B. Co man, and R.W. Mangum. 2007. Potential long-term
bene ts of no-tillage and organic cropping systems for grain production
and soil improvement. Agron. J. 99:1297–1305.
USEPA. 1997. Climate change and Michigan. Rep. EPA 230-F-97-008v. USEPA,
Washington, DC.
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.
Varvel, G.E. 2006. Soil organic carbon changes in diversi ed rotations of the
western Corn Belt. Soil Sci. Soc. Am. J. 70:426–433.
... These surveys initially indicated that SOC-rich soils were losing carbon, while SOC-poor soils were not (Bellamy et al., 2005), but this pattern was later partly attributed to regression to the mean (Lark et al., 2006;Potts et al., 2009). Several studies have followed the suggestion of Lark et al. (2006) by correcting for regression to the mean or at least estimating its effect size (Callesen et al., 2015;Hong et al., 2020;Saby et al., 2008;Senthilkumar et al., 2009 in paired control plots. These studies include analyses focused on agricultural practices and land-use change (Arndt et al., 2022;Berhane et al., 2020;Deng et al., 2016;Fujisaki et al., 2018;Haddaway et al., 2017;Hübner et al., 2021;Iwasaki et al., 2017;Sun et al., 2010) but also purely observational studies (Capriel, 2013;Hanegraaf et al., 2009), and meta-analyses of warming experiments (Crowther et al., 2016;van Gestel et al., 2018). ...
... Disentangling the effect of regression to the mean from the true underlying relationship between ∆SOC and SOC initial is possible given the right statistical approach. In fact, several studies have found persistent negative relationships between ∆SOC and SOC initial after applying a statistical correction (Hong et al., 2020;Lark et al., 2006;Senthilkumar et al., 2009). One correction approach relies on calculating the regression line between the change value (final − initial) and the initial value, and then correcting the slope derived from this regression (̂ ′ ) using variance estimates to generate an unbiased estimate (̂ ) (Blomqvist, 1977): ...
Article
Full-text available
Changes in soil organic carbon (SOC) storage have the potential to affect global climate; hence identifying environments with a high capacity to gain or lose SOC is of broad interest. Many cross-site studies have found that SOC-poor soils tend to gain or retain carbon more readily than SOC-rich soils. While this pattern may partly reflect reality, here we argue that it can also be created by a pair of statistical artifacts. First, soils that appear SOC-poor purely due to random variation will tend to yield more moderate SOC estimates upon resampling, and hence will appear to accrue or retain more SOC than SOC-rich soils. This phenomenon is an example of regression to the mean. Second, normalized metrics of SOC change-such as relative rates and response ratios-will by definition show larger changes in SOC at lower initial SOC levels, even when the absolute change in SOC does not depend on initial SOC. These two artifacts create an exaggerated impression that initial SOC stocks are a major control on SOC dynamics. To address this problem, we recommend applying statistical corrections to eliminate the effect of regression to the mean, and avoiding normalized metrics when testing relationships between SOC change and initial SOC. Careful consideration of these issues in future cross-site studies will support clearer scientific inference that can better inform environmental management.
... Soil organic carbon (SOC) is an important index of soil fertility and increasing SOC is regarded as an efficient way to alleviate climate change and meet the demands for food (Diacono and Montemurro 2010;Gomiero 2016;Mekuria et al. 2016). The intensive use of conventional tillage has considerably reduced the SOC contents of agricultural lands, relative to that occupied by natural vegetation (Senthilkumar et al. 2009, Chatterjee et al. 2018, which has led to dramatic declines in soil quality over the last century (Lal and Kimble 1997). Conventional tillage coupled with climate change accelerates SOC decomposition and restricts the SOC sequestration process (Dash et al. 2019;Ise and Moorcroft 2006).. No-till is one of the most widely promoted management practices for sequestering C in arable lands because it increases SOC through increased C inputs from high biomass productivity and reduces C losses, leading to a net transfer of C from the atmosphere to the soil (estimated to be 0.79 ~ 1.54 Gt C yr −1 globally), which helps mitigate climate change (Fuss et al. 2018). ...
Article
Full-text available
AimsNo tillage (No-till, NT), a form of conservation agricultural management, can improve food security and alleviate world climate change by influencing crop yield and soil organic carbon (SOC), but the effects of NT on crop yields and SOC and the major factors which influence these effects remain poorly understood.Methods We conducted a global meta-analysis using 5230 paired observations from 446 studies to evaluate the effects of NT on upland crop yield and SOC and to explore the associated influencing factors.ResultsGenerally, NT decreased upland crop yields by 5% (P < 0.05) and increased SOC sequestration by 9.9% (P < 0.05). Specifically, wheat and maize yields decreased by 7.7% and 2.3%, respectively. Our analyses indicate that NT combined with crop residue return and crop rotation reduced the negative effect of NT on crop yield from 5% to 2.44% and increased SOC up to 12.77%. The effects of NT on yield and SOC varied depending on the regulating factors. No-till increased yields in relatively arid areas, but reduced yields in more humid areas, whereas SOC is more likely to increase in humid regions. Soil organic carbon sequestration increased with temperature, but yield losses also increased in warmer regions.Conclusions Our study suggests that NT combined with crop residue return and crop rotation could enhance SOC sequestration under moist and warm conditions without compromising crop yield. It may also improve soil quality and help alleviate global food security and climate change.
... remained relatively static over the course of the growing season and did not decrease, even following frequent mechanical disturbance ( Figure 1). These results are consistent with others that have found larger C pools in the biologically based systems relative to the conventional system (62,65,66). The greater C levels despite frequent disturbance may be due to continuous C inputs from the cover crops included in the rotation (62). ...
Article
Full-text available
Soil carbon (C) is a major driver of soil health, yet little is known regarding how sensitive measures of soil C shift temporally within a single growing season in response to short-term weather perturbations. Our study aimed to i) Examine how long-term management impacts soil C cycling and stability across a management intensity and plant biodiversity gradient and ii) Assess how sensitive soil health indicators change temporally over the course of a single growing season in response to recent weather patterns. Here we quantify a variety of sensitive soil C measures at four time points across the 2021 growing season at the W.K. Kellogg Biological Station’s Long Term Ecological Research Trial (LTER) located in southwest Michigan, USA. The eight systems sampled included four annual soybean (Glycine max) systems that ranged in management intensity (conventional, no-till, reduced input, and biologically-based), two perennial biofuel cropping systems (switchgrass (Panicum virgatum) and hybrid poplars (Populus nigra x P.maximowiczii)), and two unmanaged systems (early successional system and a mown but never tilled grassland). We found that unmanaged systems with increased perenniality enhanced mineralizable C (Min C) and permanganate oxidizable C (POXC) values. Additionally, all soil health indicators were found to be sensitive to changes in short-term weather perturbations over the course of the growing season. The implications of this study are threefold. First, this study assess indicators of labile and stable C pools over the course of the growing season and reflects the stability of soil C in different systems. Second, POXC, Min C, and ß-glucosidase (GLU) activity are sensitive soil health indicators that fluctuate temporally, which means that these soil health indicators could help elucidate the impact that weather patterns have on soil C dynamics. Lastly, for effective monitoring of soil C, sampling time and frequency should be considered for a comprehensive understanding of soil C cycling within a system.
... Other studies have found that continuous loss of SOC with cultivation (e.g. Senthilkumar et al., 2009), but that diverse rotations (Ladha et al., 2011), N fertilizer (King and Hofmockel, 2017;Salter and Green, 1933), and inclusion of legumes (Heenan et al., 2004;Poffenbarger et al., 2020) slow the loss of SOC by 3-50%. ...
Article
Full-text available
Semi-arid regions support the majority of global wheat production, but they are threatened by continued soil erosion and increased risk of drought and extreme temperatures from climate change. Enhancing soil organic matter (SOM) is a key adaptation strategy for sustaining agriculture in these regions because it can improve water retention and nutrient cycling. When seeking to increase SOM, growers face a dilemma, known as the ‘soil carbon (C) dilemma’ (Janzen, 2006): How can SOM be increased, while also increasing the release of nutrients that accompanies decomposition? Our observational study examined whether growers could address this dilemma through intensification with diverse grain crops. We specifically tested whether incorporating legumes into a continuous rotation influences the form and amount of SOM as well as productivity in farms of the central Great Plains region of the U.S. by contrasting three, no-till rotation systems: 1) conventional wheat-fallow; 2) continuous grain-only rotations, and 3) continuous grain rotations that incorporate a legume crop. We sampled on-farm fields and experimental agricultural research station plots that had received one of these rotations for at least eight years. We compared yield, soil organic C (SOC), and nitrogen (N) stocks in bulk soil to 1 m, as well as SOC and N in particulate organic matter (POM) and mineral-associated organic matter (MAOM) in surface soils (0–10 cm) among the different rotations. We also estimated the long-term effects of the rotations on SOC fractions through a simulation of each rotation with the MEMS 1.0 model, verified with measured data. We found that intensifying the rotation with continuous grains led to 1.5-fold increase in aggregate size but did not change SOC stocks. Incorporating a legume to the continuous grain rotation resulted in 1 Mg C ha⁻¹ more SOC on average in surface soil compared to wheat-fallow rotations. No significant soil changes among rotations were observed at depth. Longer-term simulations of these cropping systems suggest that including legumes could allow for 10% greater SOC gains over time compared to wheat-fallow. Since legume incorporation also increased crop yields across the full rotation cycle, our results support our hypothesis that growers can solve the ‘soil C dilemma’ by simultaneously increasing the quantity and N-content of soil input through continuous cropping with both grains and legumes.
... Current research shows that NT may impact SOC storage in soil with high silt and clay contents (finer-textured) more dramatically than sandy soils (coarser-textured) because of their greater physical protection of SOC (Senthilkumar et al. 2009a, Hao and Kravchenko 2007, Hassink 1994, and Angers et al. 2007). Likewise, Needelman et al. (1999) detected a stronger impact of NT on SOC storage under lower rather than higher sand contents. ...
Thesis
http://deepblue.lib.umich.edu/bitstream/2027.42/107742/1/kkiluk.pdf
... indicates that winter climate changes may lead to lower soil levels C and isolation of the C ecosystem (Senthilkumar et al., 2009). Global warming can have adverse effects on plant growth. ...
Article
Full-text available
From a studied to collect data for temperature degrees during 1973 until 2019 there is clear evidence that show climate change is happening in our region. The collected data showed that the average annual temperature rate has raised by almost 1.3°C, the average of warmest year for this period was recorded in 2010 and it was 21.55°C. In 2016, average temperature degree was high 20.3°C as the global worm rising the lowest temperature in this period was 16.2°C recorded in 1992. Since 1973 until 1985 the average temperature was 18.39°C and since 1986 until 1998 the average temperature raised to 18.79°C, then since 1999 until 2010 the average temperature raised again to 19.93°C but in period 2011 to 2018 the average temperature raised to 20.09°C. The difference between warmest monthly average temperatures and the coldest month of the year or annual range in 1988 was the highest and valued 31.9°C while in year 1973 was the lowest (19.6°C) but the rate for the period 1973 to 2018 was 27.9°C. The absolute annual temperature range or differences between the maximum absolute temperature and the minimum absolute temperature during a year was obvious like in 1973 was the highest (53°C) and in 1987 was the lowest (35°C). Diurnal temperature variation or the daily difference between the maximum and minimum temperatures was limited in winter months almost recorded 6 - 9°C, but in summer months the ranges is bigger almost reported 15 - 18°C. Over these years specially from (2008 -2019) soils that affected with plant pathogenic nematodes increased dramatically in the study area witch was within 10 km2 because of rising temperatures degrees and need to use greenhouses instead of open field cultivation, these greenhouses numbers influencing on agroecosystem structure for a long terms. From this investigation, and to control plant pathogenic nematodes in the infested greenhouses, farmers used kinds of chemical pesticides that cause damages to the soils and the yields as well as changing agroecosystem structure. Agroecosystem structure for the study area has been changed relatively because of increasing the numbers of greenhouses that got almost 7000 greenhouse causing imbalance in agroecosystem by using a huge amount of water (125meter cubic/greenhouse).
... In our IAM, yield responses to P fertilizer and soil P availability are modeled with the SALUS model, which has been thoroughly tested against field measurements for soil carbon dynamics (e.g., Senthilkumar et al. 2009), crop yield (e.g., Basso et al. 2007;Asseng et al. 2013;Rosenzweig et al. 2013;Dzotsi, Basso, and Jones 2013), plant nitrogen uptake and phenology (e.g., Basso et al. 2010;Basso, Ritchie, et al. 2011;Basso, Kendall, and Hyndman 2013), nitrate leaching (Giola et al. 2012;Syswerda et al. 2012), water use efficiency (Basso and Ritchie 2012), and P dynamics (Daroub et al. 2003). SALUS is derived from the validated CERES models with added ability to quantify the impact of management strategies and their interactions with the soil-plant-atmosphere system on yield along with carbon, nitrogen, and P dynamics. ...
Article
This paper describes our efforts to integrate economic and biophysical models to evaluate the effects agri-environmental policies have on the value of freshwater ecosystem services. We are developing an integrated assessment model (IAM) that links changes in phosphorus-related management practices on farm fields to changes in the value of key freshwater ecosystem services, including biological condition, water clarity, species-specific fish biomasses, and beach algae. Our IAM approach enables examination of the effects of policies and conservation programs on ecosystem services and values. Results will help policy makers allocate conservation dollars to improve water quality, enhance ecosystem services, and promote more sustainable agricultural production. (JEL Q24, Q51) © 2020. by the Board of Regents of the University of Wisconsin System. All Rights Reserved.
Article
Full-text available
Plant diseases have caused serious challenges in the production of food globally. This has led to hunger and food insecurity. Different solutions for crop diseases have been introduced in the recent past that include cultural control using farm management methods, chemical control, resistant cultivars, and recently, biocontrol. Each of these solutions has varied weaknesses. Faced with the changes in climate and the recurrences of crop diseases, new strategies incorporating preventive measures would be important by reducing risks to crop production from crop diseases, thus alleviating food insecurity. Strategies for the prevention of these diseases and/or forecasting favorable environmental conditions for disease development have not been fully employed as preventive measures. The use of crop modeling has been used to advise farmers on planting procedures that would bring maximum yields using different management procedures at the farm level. Little is known about the use of crop models in crop disease control. In order to increase the use of crop models for these objectives, this review provides the current status quo and will help to stimulate more research in this regard.
Chapter
Full-text available
Rapid population growth, climate change, and limited natural resources have widened the gap between food production and consumption, contributing to global hunger. Improving cereal crop production is a critical hot spot challenge for closing this gap and ensuring global food security and nutrition. Previous data and findings from published literature demonstrated that cereal crop models have been applied and developed globally over the last 30 years under a wide range of climate, soil, genotype, and management conditions. However, when the models are applied to pests, diseases, phosphorus fertilization, potassium fertilization, iron, and zinc, further improvements are required. Furthermore, the integration of genotypes and phenotypes is critical for food security, necessitating careful consideration in crop models. We examined about 31 cereal crop models for increasing crop production and ensuring food and nutrition security. Furthermore, we discussed the current limitations in crop model application, as well as the critical need to integrate with other cutting-edge sciences, such as remote sensing, machine learning, and deep learning. This will undoubtedly improve crop model accuracy and reduce uncertainty, assisting agronomists and decision makers in ensuring food and nutrition security. In this chapter, we discussed the current and further improvements of cereal crop models in assisting breeders, researchers, agronomists, and policy makers in addressing current and future challenges related to global food security and nutrition
Article
Placement and fate of photo-assimilated carbon (C) newly added to the soil are important contributors of soil health. Soil pores control the movement of gasses, water, and microorganisms, thus potentially influencing new photo-assimilated C gains and losses. The objective of this study was to explore the associations between soil pores and additions and losses of root-derived C. Young cereal rye (Secale cereale L.) plants were grown in soil with inherent pore architecture destroyed by sieving and in soil with intact pore architecture, with each rye planted container having a section inaccessible to plant roots. Plants were pulse labeled with ¹³CO2, followed by sampling of intact soil cores and subjecting them to X-ray computed micro-tomography (µCT) scanning, some immediately after collection and some after a 21-day incubation. From the scanned cores we obtained soil micro-samples in specific locations corresponding to µCT images. The µCT images were used to quantify pore size distributions of the micro-sample soils, while soil δ¹³C signatures provided a quantitative measure of the presence of root-derived C before and after the incubation. In the intact soils, δ¹³C was positively associated with > 90 µm Ø pores, likely reflecting preferential rye root growth into legacy root channels. In soils where existing pore architecture was destroyed, ¹³C was preferentially added to 15–90 μm Ø pores when the soil was accessible to roots and to 7–40 μm Ø pores when the soil was accessible only to fungi, yet after the incubation the associations between ¹³C and pores were lost. The results identified the pore sizes associated with root-derived C additions to the soil via root and fungal routes and highlighted the importance of inherent pore architecture on the placement and persistence of such additions.
Article
Full-text available
The degree to which soil resource availability is linked to patterns of microbial activity and plant productivity within ecosystems has important consequences for our understanding of how ecosystems are structured and for the management of systems for agricultural production. We studied this linkage in a 48-ha site in southwest Michigan, USA, that had been cultivated and planted to row crops for decades. Prior to seeding the site to genetically identical soybean plants (Glycine max) in early spring, we removed soil samples from approximate to 600 locations; plant biomass was harvested from these same locations later in the season. Soil samples were analyzed for physical properties (texture, bulk density), chemical properties (moisture, pH, total C, total N, inorganic N), and biological attributes (microbial biomass, microbial population size, respiration potential, and nitrification and N-mineralization potentials). Plant analyses included biomass and C and N contents. Soil resource variability across this long-cultivated site was remarkably high, as was variability in microbial activity and primary productivity. In almost all cases variability exhibited a strong spatially explicit structure: for most properties and processes >50% of sample variance was spatially dependent at a scare of 5-60 m. Exceptions included microtopography, soil pH, and inorganic P, which were spatially dependent across the entire 1-1200 m range of separation distances examined in this study, and the culturable-bacteria population, which was not spatially autocorrelated at any scale examined. Both topographic relief and soil pH exhibited strongly nested structures, with autocorrelation occurring within two (topography) or more (pH) distinct ranges. Multiple regression analysis showed surprisingly little correlation between biological processes (soybean productivity, soil N turn-over, soil respiration), and static soil properties. The best predictor of soybean biomass at late reproductive stages (r(2) = 0.42) was a combination of nitrate N, bulk density, inorganic P, N-mineralization rates, and pH. Overall, results suggest a remarkable degree of spatial variability for a pedogenically homogeneous site that has been plowed and cropped mostly as a single field for >100 yr. Such variability is likely to be generic to most ecosystems and should be carefully evaluated when making inferences about ecological relationships in these systems and when considering alternative sampling and management strategies.
Article
Given a series of consecutive measurements on a random sample of individuals, it is often of interest to investigate whether there exists a relationship between the rate of change and the initial value. Assuming that the observations deviate in a random manner from the true values, straightforward regression computations will yield biased results. It is shown that, in the case of the normal distribution, the maximum likelihood (ML) estimates of the second-order moments of the true slope and the true initial value are obtained by simple adjustments of the corresponding moments of the estimated quantities. An asymptotic formula for the standard error of the regression coefficient of slope on initial value is derived, and the methods are applied to longitudinal blood pressure data. The case with concomitant variables is discussed briefly.
Article
Given a series of consecutive measurements on a random sample of individuals, it is often of interest to investigate whether there exists a relationship between the rate of change and the initial value. Assuming that the observations deviate in a random manner from the true values, straightforward regression computations will yield biased results. It is shown that, in the case of the normal distribution, the maximum likelihood (ML) estimates of the second-order moments of the true slope and the true initial value are obtained by simple adjustments of the corresponding moments of the estimated quantities. An asymptotic formula for the standard error of the regression coefficient of slope on initial value is derived, and the methods are applied to longitudinal blood pressure data. The case with concomitant variables is discussed briefly.