Relationship of Soil Respiration to Crop and Landscape in the Walnut Creek W atershed
T.B. Parkin1*, T.C. Kaspar1 and Z. Senwo2, J. H. Prueger,1 and J.L. Hatfield1
1 2
USDA -Agricultural R esearch Service, N ational Soil Tilth L aboratory , Ames, IA Department of Plant and Soil Science, Alabama A & M University, Normal, AL
*
Corresponding Author USDA-ARS National Soil Tilth Laboratory Ames, IA 50011
Ph: (515)-294-6888 Fax: (515)-294-8125 Email:
[email protected]
Acknowledgments: The authors would like to acknowledge the technical assistance and contributions of Otis Smith Jr., Tim Hart, Forrest Goodman, Shannon Kulisky and Irenus Tazisong. 1
1
ABSTRACT
2
Soil respiration is an important component of the carbon dynamics of terrestrial ecosystems.
3
Many factors exert controls on soil respiration, including, temperature, soil water content,
4
organic matter, soil texture, and plant root activity. This study was conducted to quantify soil
5
respiration in the Walnut Creek Watershed in Central Iowa, and to investigate the factors
6
controlling this process. Six agricultural fields were identified for this investigation; three of the
7
fields were cropped with soybean (Glycine max (L.) Merr.) and three cropped with corn (Zea
8
mays L.). Within each field, soil respiration was measured at 9 locations, with each location
9
corresponding to one of three general landscape positions (summit, side-slope, and depression).
10
Soil respiration was measured using a portable vented chamber connected to an infra-red gas
11
analyzer. Soil samples were collected at each location for measurement of soil water content,
12
pH, texture, microbial biomass, and respiration potential. Field respiration rates did not show a
13
significant landscape effect. However, there was a significant crop effect, with respiration from
14
cornfields averaging 37.5 g CO2 m-2 d-1 vs. an average respiration of 13.1 g CO2 m-2 d-1 in
15
soybean fields. In contrast, laboratory measurements of soil respiration potential, which did not
16
include plant roots, showed a significant landscape effect and an insignificant cropping system
17
effect. Similar relationships were observed for soil organic C and microbial biomass.
18
Additional analyses indicate that corn roots may be more important that soybean roots in their
19
contribution to surface CO2 flux, and that root respiration masked landscape effects on total soil
20
respiration. Also, failure to account for soil respiration may lead to biased estimates of net
21
primary production measured by eddy covariance.
2
1 2
1. Introduction Concern over global climate change has fostered an interest in documenting soil C
3
sequestration in agricultural systems as a strategy to offset atmospheric CO2 increases. This has
4
resulted in a greater effort to understand the factors affecting soil C storage, as well as to assess
5
soil C budgets (Lal et al., 1995). Carbon dioxide flux from soil to the atmosphere is the primary
6
mechanism of C loss from soils and a major component of terrestrial C budgets. Quantification
7
of C losses, relative to inputs, may be a valuable technique for estimating the rate of change of
8
soil C pools and for evaluating the impact of management practices on C sequestration in
9
agricultural systems (Buyanovsky et al., 1985; Curtin et al., 2000; Duiker and Lal, 2000;
10 11
Paustian et al., 1997). Micrometeorological methods such as Bowen ratio and eddy covariance offer
12
opportunities for determination of annual net ecosystem exchange of CO2 Within the past
13
decade, several flux monitoring networks have been established to assess net CO2 exchange from
14
terrestrial systems (Baldocchi, et al., 2001; Svejcar et al., 1997). However, a major focus of
15
previous work has been on forest and grassland systems, with relatively few measurements made
16
on cropped systems.
17
The fluxes measured by micrometeorological techniques can provide a direct estimate of
18
net ecosystem exchange of CO2. It has been demonstrated that indirect estimates of gross
19
primary production and ecosystem respiration can also be obtained (Falge et al., 2002). In
20
studies of seasonal CO2 flux from a variety of ecosystems, these investigators concluded that
21
differences in the temporal patterns of assimilatory and respiratory processes is an important
22
reason for characterizing these different components of net ecosystem exchange. Indeed, Falge 3
1
et al., (2002) stated, “Hence these drivers (temperature, moisture, light) will affect net ecosystem
2
carbon exchange (FNEE) differently as they force FGPP (gross primary production) and FRE
3
(ecosystem respiration) differently...
4
factors that influence seasonality of the component fluxes, FRE and FGPP, and govern the seasonal
5
patterns of net fluxes.” [italics added].
6
and heterotrophic respiration, similar logic dictates that driving factors may differentially control
7
the temporal dynamics of these two processes. This argues for detailed characterizations of
8
these heterotrophic and autotrophic respiratory processes to provide a better understanding of the
9
feedbacks on net ecosystem C exchange.
Consequently, we need to understand primarily the
Because ecosystem respiration is the sum of autotrophic
10
Respiration in terrestrial systems is the result of plant shoot respiration and soil
11
respiration that includes both root respiration and heterotrophic respiration (micro- and macro-
12
fauna). Nighttime CO2 fluxes quantified by micrometeorological techniques do not distinguish
13
between above and below ground sources. Partitioning respiration between above ground and
14
below ground sources may require a combination of techniques (Rochette et al., 1996). Soil
15
chambers have been widely used to quantify CO2 flux due to plant roots and soil heterotrophic
16
processes. Thus, soil chambers used in conjunction with micrometeorological methods may
17
yield greater insights into temporal dynamics of CO2 exchange, as argued for by Falge et al.
18
(2002).
19
Estimation of cumulative CO2 flux from the soil surface over the time periods required to
20
evaluate agricultural management practices remains problematic. Temporal and spatial
21
variability may often mask differences in CO2 flux arising from management changes (Duiker
22
and Lal, 2000; Hutchinson et al., 2000). The spatial variability of soil CO2 flux has been 4
1
characterized in several studies (Davidson et al., 2002; Rayment and Jarvis, 2000; Rochette et
2
al., 1991; Savage and Davidson, 2003), where coefficients of variation ranged from 25% to 85%.
3
Despite these general reports of the magnitude of the spatial variability associated with soil
4
respiration, there is little information on spatial variability in relation to landscape.
5
Development of a more complete understanding of the factors controlling in situ soil respiration
6
requires that the interactions associated with landscape position be assessed.
7
To provide additional insights into controls of soil respiration and to evaluate landscape
8
influences, this study was conducted as part of the SMEX02 soil moisture experiment in the
9
Walnut Creek watershed in central Iowa. The objectives of this experiment were to: i)
10
characterize the factors controlling spatial variations of soil respiration, and ii) relate soil CO2
11
flux measurements to net ecosystem CO2 flux determined by eddy covariance.
5
1
2. Materials and Methods
2
a. Site Description and Soil Characteristics
3
This study was conducted in the Walnut Creek watershed in central Iowa located 5 km
4
south of Ames, IA (41o75'N, 93o 41W’) as part of the multi-agency 2002 Soil Moisture
5
Experiment (SMEX02) (Kustas et al., 2003; Kustas et al., this issue). The Walnut Creek
6
watershed is approximately 5100 ha in size and consists predominantly of privately owned land
7
in corn/soybean production. The fields of this study ranged from 64 ha to 130 ha in size, and
8
have been in 2 y corn-soybean rotations since at least 1991. The typical tillage regime consists
9
of fall chisel plow following corn, and spring disking of both corn and soybean fields (Hatfield
10 11
et al., 1999). Soils within the Walnut Creek watershed were formed during the last glacial advance
12
into Iowa during the late Wisconsinan period between 14000 and 12000 years ago (Hallberg and
13
Kemmis, 1986). The area is characterized by low relief swell and swale topography, and surface
14
drainage is inherently poor. As a result, subsurface drainage has been installed across much of
15
the area. Numerous closed depressional areas, commonly called potholes, exist and have
16
accumulated organic matter from surrounding sideslopes. Soils are predominantly the Clarion-
17
Nicollet-Canisteo soil association. This association consists of well-drained Clarion soils located
18
on higher or sloping areas, somewhat poorly drained Nicollet soils located on side slopes,
19
Canisteo soils on poorly drained low areas, and very poorly drained Okoboji and Harps soils in
20
the depressional areas. The soils that have been mapped are Canisteo silty clay loam (fine-
21
loamy, mixed [calcareous], mesic Typic Haplaquoll), Clarion loam (fine-loamy, mixed, mesic
22
Typic Hapludoll), Harps loam (fine-loamy, mesic Typic Calciaquoll); Nicollet loam (fine-loamy, 6
1
mixed, mesic Aquic Hapludoll), and Okoboji mucky silt loam (fine, montmorillonitic, mesic
2
Cumulic Haplaquoll) (Soil Conservation Service, 1981).
3
b. Soil CO2 Flux
4
Soil CO2 fluxes were measured in 6 fields within the Walnut Creek Watershed during the
5
period of June 25 - July 3, 2002. Three of the fields were cropped to corn (WC06, WC13a,
6
WC15) and three fields were cropped to soybeans (WC03, WC14, WC16). Field WC13a was
7
not included in the overall SMEX02 sampling design, but was located directly east of, and
8
adjacent to field WC13. In each field a stratified sampling scheme was employed whereby three
9
replicate locations of each of 3 general landscape elements were identified (Summit, Side-slope,
10
and Depression), resulting in 9 locations per field. At each landscape position within each field
11
soil CO2 flux, soil temperature (surface and 5 cm) and soil water content (surface 6 cm) was
12
measured at 3 separate locations located approximately 1 meter apart. Soil water contents were
13
performed using a soil moisture meter (model TH2O, Dynamax, Houston, TX)1 An empirical
14
algorithm was used to correct for the slight temperature effect on soil moisture probe response
15
(Parkin and Kaspar, 2003). Soil samples were also collected for laboratory analysis of soil
16
microbial biomass, respiration, pH, organic C, and soil texture (described in section d, below).
17
Soil CO2 fluxes were measured using vented chambers (27 cm diameter) placed on the
18
soil surface (between plant rows) for 3 minutes. A polyethylene skirt attached to the outside of
19
the chambers extended in a concentric ring approximately 30 cm out from the chamber wall and 1
Reference to a trade or company name is for specific information only and does not
imply approval or recommendation of the company or product by the USDA to the exclusion of others that may be suitable. 7
1
was held on the soil with a length of chain to seal the chambers at the soil surface (Denmead,
2
1979).
3
(model CI-301, CID, Inc., Vancouver, WA) to quantify CO2 concentrations. Carbon dioxide
4
fluxes were calculated by performing linear regression on the CO2-concentration vs. time data or
5
by the nonlinear procedure of Hutchinson and Mosier (1981). Each site was sampled only once
6
during the 9 d period. To estimate daily soil respiration from these point-in-time measurements,
7
the 5 cm soil temperature values were used along with average daily 5 cm soil temperature in a
8
mathematical algorithm relating temperature and soil CO2 flux (Eq. 1).
9
Daily Average CO2 Flux = Rt * Q(DAT - T)/10
Headspace gas of the chambers was circulated through a portable infra-red analyzer
[Eq. 1]
10
where: Rt is the measured CO2 flux from a given soil chamber at time t, T is the soil temperature
11
at time t, DAT is the daily average temperature over the 9 d sampling period, Q is the Q10 factor,
12
and Daily Average CO2 Flux is the resulting estimated daily average flux average based on the
13
point-in-time chamber measurement (Parkin et al., 1996). For these corrections a Q10 factor of
14
1.22 was calculated from temporal-intensive soil CO2 flux measurements made with automated
15
soil chambers (described below).
16
Intensive temporal sampling of soil CO2 flux was conducted in a single corn field within
17
the watershed (field WC15). On DOY 170 ( 5 d prior to the sampling period) 6 automated
18
chambers were placed in this field at 3 different landscape elements; 2 chambers each at a
19
summit, side-slope, and depression location. Soil respiration was measured using automated
20
chambers similar to the design of Parkin and Kaspar (2003). Chambers were 30 cm x 30 cm x
21
15 cm stainless steel open ended boxes pressed into the soil approximately 5 cm. Chambers
22
were placed in the inter-row areas. The top of each steel box was fit with a wooden framework 8
1
that supported a sliding cover. The covers were supported by casters riding on steel tracks
2
attached to the sides of the chambers. Linear actuators driven by gear motors attached to the
3
frames served to open and close the covers at hourly intervals. Soil respiration was measured
4
every hour by sliding the cover over the chamber top to allow CO2 to accumulate in the chamber
5
headspace. Carbon dioxide was measured during a 6 minute period by pumping the chamber
6
gas through an infrared gas analyzer (model LI-800, LiCor Inc., Lincoln, NE). After 6 minutes
7
the chamber lids were withdrawn and the chambers were left open to the atmosphere for the
8
succeeding 54 m. To account for the nonlinear buildup of headspace CO2 concentrations within
9
the chambers, we calculated respiration rates using the algorithm of Hutchinson and Mosier
10
(1981). Each chamber was instrumented with thermocouples to measure air temperature, surface
11
soil temperature and soil temperature at 5 cm. The average difference between air temperature
12
measured outside the chambers at a height of 1.2 meters above the soil surface (in the canopy)
13
and the air temperature within the chambers (0.10 m above the soil surface) during the time they
14
were closed was 1.41 degrees C. We do not know to what extent this difference was a spatial
15
effect (temperatures taken at different heights above the soil surface); however, average air
16
temperature increases within the chamber during the 6 min that they were closed averaged only
17
0.11 C. The small increases in chamber air temperature were likely due to the fact that the
18
chambers were shaded in the corn canopy. Soil moisture probes, placed in the chambers, were
19
used to measure soil water content (0-6 cm), and a small fan was placed in each chamber to
20
recirculate air during the respiration measurements. Data from the automated chambers were
21
used to determine diurnal responses of soil CO2 flux in order to account for temperature / time-
22
of-day influences associated with the CO2 fluxes determined with the portable chambers at the 9
1
remote fields as described above.
2
c. Eddy Covariance CO2 Flux
3
Fifteen eddy covariance towers were spatially distributed across the Walnut Creek
4
Watershed representing multiple soil types and landscape position for both corn and soybean
5
production fields (Kustas et al., this issue). Two of these towers were in field WC15 where the
6
automated soil CO2 flux chambers were installed. Turbulent fluxes of sensible heat (H), Latent
7
heat (LE) and CO2 were measured using eddy covariance (Prueger et al., 2004). Instrumentation
8
consisted of a three dimensional sonic anemometer (CSAT3 Campbell Scientific Inc., Logan,
9
UT) and a fast response water vapor and CO2 density open path infrared gas analyzer (LI7500
10
LiCor Inc., Lincoln, NE). The EC instruments were mounted on 10 m towers at 3 m above the
11
ground surface. Ancillary instrumentation on each tower site included a net radiometer (Q*7.1
12
Radiation Energy Balance Systems, REBS, Seattle, WA), a high precision infrared radometric
13
temperature sensor (IRT, 15o fov; Apogee, Logan, UT) and an air temperature / relative humidity
14
sensor (Vaisala, HMP-35, Campbell Scientific Inc., Logan, UT). Two soil heat flux plates
15
(REBS HFT-3) were buried 0.06 m below the soil surface; one within the plant row and the
16
second in the inter-row space. Soil thermocouples (Cu-Co Type T) were placed 0.02 and 0.04 m
17
below the surface and above each soil heat flux plate.
18
d. Laboratory Measurements
19
Surface soil (0-6 cm) was sampled at each respiration site. In the laboratory, samples
20
were weighed and sieved (2 mm). Subsamples of field moist soil were collected for water
21
content and soil microbial biomass, and the remaining soil was air dried. Water content was
22
determined by oven drying 10 g field moist soil at 105 oC overnight. Soil microbial biomass C 10
1
was measured by fumigation and direct extraction with 0.5 M K2SO4 on duplicate 25 g field-
2
moist samples (Tate et al., 1988). Organic C in the fumigated and nonfumigated extracts was
3
measured using a Dohrmann DC-180 carbon analyzer (Rosemount Analytical Services, Santa
4
Clara, CA). Air dried samples were ground with a roller mill for organic C and N determination
5
by dry combustion with a Carlo-Erba NA 1500 CHN elemental analyzer (Haakes Buchler
6
Instruments, Paterson, NJ) after removal of carbonates (Nelson and Summers, 1996). pH was
7
measured in 1:1 distilled water:soil slurries. Soil texture analyses were performed by Midwest
8
Laboratories, Inc. (Omaha, NE). Physical and chemical properties of the soils are shown in
9
Table 1. Soil respiration potential was performed by placing 5 g air dried soil in 120 ml amber
10
tubes equipped with screw tops fit with butyl rubber septa. De-ionized water was added to each
11
sample to bring the water content to 60% water filled pore space in order to optimize microbial
12
activity (Linn and Doran, 1984). Carbon dioxide concentrations were measured over a 30 d
13
period with an infrared analyzer (Model 2200, Automated Custom Systems, Orange, CA). After
14
each CO2 reading, the headspace of each tube was flushed with compressed air. Samples were
15
run in triplicate, and mean respiration rates for each soil sample was calculated from the
16
cumulative amount of CO2 produced over the 30 d incubation period.
17
e. Statistical Analyses
18
Landscape and cropping effects on soil variables were assessed using 2 way ANOVA and
19
differences assessed by Tukey’s pairwise comparison. Correlations between variables were
20
assessed using Pearson Product Moment Correlation. All tests were performed with SigmaStat
21
software (SigmaStat v. 2.03, SPSS, Inc., Chicago, IL.).
11
1
3. Results
2
a. Soil Respiration and Landscape Interactions
3
Soil respiration rates measured in the field ranged from 10.6 g CO2 m-2 d-1 to 93.3 g CO2
4
m-2 d-1 across fields and landscape positions within the Walnut Creek Watershed (Table 2).
5
Highest rates were observed in field WC15, which was planted to corn, and lowest fluxes were
6
observed in the soybean fields (WC03, WC14, WC16). Average fluxes for the corn and soybean
7
fields were 37.5 and 13.1 g CO2 m-2 d-1, respectively. Analysis of variance showed that there
8
was a crop effect, with the corn fields having significantly higher (P
