Grazing Effects on Carbon Dynamics in the Northern Mixed-Grass Prairie.

Abstract

The role of rangelands in the regulation of atmospheric CO2 concentrations is a critical issue in global climate change research. Rangelands are complex ecosystems that occupy about 50% of the land area in the world and USA. We studied the effects of seasonal grazing on CO2 flux on small plots located on a silty range site in the northern mixed-grass prairie with an Eapa fine loam soil. Treatments were no grazing or short-duration intensive grazing during mid-May or mid-July in 1996, 1997, and 1998. Data were collected from mid-April to mid-October at about 30-day intervals to estimate standing crop, leaf area, soil organic C, root mass to a 30-cm soil depth, and diurnal variation of CO2 flux and soil respiration (at 08:00, 12:00, 16:00, and 24:00 hr) in closed chambers. Uptake of CO2 was greatest during spring and early summer, peak periods of precipitation and green biomass. Grazing removed an average of about 70% of the green standing crop with a subsequent reduction in CO2 uptake of 175% in May and 109% in July. Grazing in May reduced CO2 uptake for 30 days in two of the three years, whereas, grazing in July reduced CO2 flux only in 1998. Residual effects of grazing, however, declined in late summer and autumn with the onset of plant maturation. The potential C sink in the mixed-grass prairie of the Northern Great Plains appears to be small and will vary through time with intensity and timing of grazing as it interacts with climatic conditions.

Full text

Page 1

Grazing Effects on Carbon Dynamics in the
Northern Mixed-Grass Prairie
MARSHALL R. HAFERKAMP
M. D. MACNEIL
USDA-Agricultural Research Service
Fort Keogh Livestock and Range Research Laboratory
243 Fort Keogh Road
Miles City, Montana 59301, USA
ABSTRACT / The role of rangelands in the regulation of atmo-
spheric CO2concentrations is a critical issue in global climate
change research. Rangelands are complex ecosystems that
occupy about 50% of the land area in the world and USA. We
studied the effects of seasonal grazing on CO2flux on small
plots located on a silty range site in the northern mixed-grass
prairie with an Eapa fine loam soil. Treatments were no graz-
ing or short-duration intensive grazing during mid-May or mid-
July in 1996, 1997, and 1998. Data were collected from mid-
April to mid-October at about 30-day intervals to estimate
standing crop, leaf area, soil organic C, root mass to a 30-cm
soil depth, and diurnal variation of CO2flux and soil respiration
(at 08:00, 12:00, 16:00, and 24:00 hr) in closed chambers.
Uptake of CO2was greatest during spring and early summer,
peak periods of precipitation and green biomass. Grazing re-
moved an average of about 70% of the green standing crop
with a subsequent reduction in CO2uptake of 175% in May
and 109% in July. Grazing in May reduced CO2uptake for 30
days in two of the three years, whereas, grazing in July re-
duced CO2flux only in 1998. Residual effects of grazing, how-
ever, declined in late summer and autumn with the onset of
plant maturation. The potential C sink in the mixed-grass prai-
rie of the Northern Great Plains appears to be small and will
vary through time with intensity and timing of grazing as it in-
teracts with climatic conditions.
Rangelands
deserts, and tundra) occupy about 50% of the world’s
land area and contain more than 33% of aboveground
and belowground terrestrial C reserves (Allen-Diaz
1996). Follett and others (2001) suggest that, given the
size of the C pool in US rangeland, a better understand-
ing is needed of the current and potential effects of
management practices on C storage. Since 1990 re-
searchers have been actively studying the potential role
of grasslands for C sequestration with research in
seeded pastures (Franzluebbers and others 2000); tall
grass prairie (Verma and others 1989, 1992, Kim and
others 1992, Ham and Knapp 1998, Mielnick and Du-
gas 2000, Dugas and others 1999, Rice and Owensby
2001, Suyker and Verma 2001); mixed-grass prairie
(Dormaar and others 1995, Frank and others 1995,
2001, Manley and others 1995, Schuman and others
1999, 2001, 2002, LeCain and others 2000, Meyers
2001, Frank and Dugas 2001, Sims and Bradford 2001,
Frank 2002, Reeder and Schuman 2002; shortgrass
(includinggrasslands,shrublands,
steppe (Reeder and Schuman 2002; Reeder and others
this issue); and sagebrush steppe (Angell and others
2001). Inherent in all rangeland ecosystems are both
diurnal and seasonal variation in CO2flux, with the
direction of flux controlled by the balance between
photosynthesis and respiration during the growing and
dormant seasons (Norman and others 1992, Bremer
and others 1998, Dugas and others 1999, Mielnick and
Dugas 2000, Frank and Dugas 2001, Frank 2002).
Rangelands are also resource limited, particularly for
nitrogen and water (Willms and Jefferson 1993). Sever-
ity, duration, and timing of grazing and drought in
relation to plant phenology are important in determin-
ing the status of prairie vegetation as a C source or sink.
Inconsistent responses of soil organic C to grazing
have been reported on rangelands. Many of these dif-
ferences appear to be the result of variations in climate,
soil properties, landscape position, plant community
composition, grazing management practices, soil or-
ganic matter, depth of soil profile, and depth of the soil
profile sampled (Johnston and others 1971, Smoliak
and others 1972, Dormaar and others 1977, Bauer and
others 1987, Milchunas and Lauenroth 1993, Frank and
others 1995, Manley and others 1995, Biondini and
Manske 1996, Derner and others 1997, Povirk and oth-
ers 2001, Schuman and others 1999, Wienhold and
others 2001, Reeder and Schuman 2002). Some have
suggested that grazing leads to an increase in soil mi-
crobial biomass that is related to changes in root exu-
KEY WORDS: Northern Great Plains; Grazing; Carbon sequestration;
Grasslands; CO2flux; Montana
The paper is a contribution from the USDA-ARS and Montana Agri-
cultural Experiment Station, Miles City, Montana, USA.
Mention of any trade name or proprietary product does not constitute
a guarantee or warranty by the authors or USDA-ARS nor does it imply
the approval of these products to the exclusion of others.
Published online January 20, 2004.
DOI: 10.1007/s00267-003-9154-x
Environmental Management Vol. 33, Supplement 1, pp. S462–S474
© 2004 Springer-Verlag New York, LLC

Page 2

dation patterns following defoliation (Guitian and
Bardgett 2000, Hamilton and Frank 2001). Grazing can
also affect the redistribution of C within the system
(Reeder and Schuman 2002) through changes in rate
of recycling aboveground plant C into the soil, changes
in plant species composition, and the rate of incorpo-
ration of litter into the soil (Naeth and others 1991).
Excluding grazing allows litter to accumulate, which
may reduce rate of C incorporation into the soil. Litter
accumulation will also affect soil temperature (LeCain
and others 2000) and soil water content, which in turn
will affect soil organic matter decomposition rates, and
C and N cycling (Reeder and others 2001). Lateral and
vertical spatial variability on rangelands associated with
soil organic matter, soil aggregation, and other re-
sources poses challenges and opportunities for manag-
ers trying to enhance C storage (Bird and others 2001).
The objective of this research was to determine the
impact of short-duration intensive grazing in May or
July on CO2flux and C dynamics in the Northern Great
Plains. We hypothesized that intensive short-duration
grazing in May or July would have no effect on CO2flux
or C dynamics relative to environmental stress.
Study Area
Research was conducted at the Fort Keogh Livestock
and Range Research Laboratory (46°22’N 105°5'W)
near Miles City, Montana, USA. Regional topography
varies from rolling hills to broken badlands with small
intersecting ephemeral streams flowing into rivers in
broad, nearly level valleys. The area is representative of
the semi arid mixed-grass prairie of the Northern Great
Plains. Indigenous vegetation on the 22,500-ha re-
search station is a grama–needlegrass–wheatgrass (Bou-
teloua–Stipa–Agropyron) mix (Ku ¨chler 1964). Annual
precipitation averages 343 mm, with about 60% re-
ceived from April through September (Figure 1). Daily
temperatures range from ? 38°C during summer to
below ?40°C during winter. The average frost-free
growing season is 150 days.
This 1-hectare study site was a nearly level silty range
site, which is a common range site in the Northern
Great Plains. Soil was an Eapa (fine-loamy, mixed, su-
peractive, frigid Aridic Argiustoll). Vegetation was dom-
inated by a C3 annual grass, Japanese brome (B. japoni-
cus Thunb.); C3 perennial grasses, prairie Junegrass
[Koeleria macrantha (Ledeb.) J.A.], needleandthread
[Hesperostipa comata (Trin. & Rupr.) Barkworth], and
western wheatgrass [Pascopyrum smithii (Rydb.) A.
Love]; a C3 sedge, thread-leaf sedge (Carex filifolia
Nutt.); C4 perennial grasses, blue grama [Bouteloua
gracilis (Willd. ex Kunth) Lag. ex Griffiths] and red
threeawn [Aristida purpurea Nutt. var. longiseta (Steud.)
Vasey]; a succulent, prickly pear (Opuntia spp.); half-
shrubs, fringed sagewort (Artemisia frigida Willd), com-
mon sagewort (A. campestris L.), and phlox (Phlox hoodii
Richards); and forbs, western salsify (Tragopogon dubius
Scop.) and hairy goldaster [Heterotheca villosa (Pursh.)
Shinners var. villosa]. Perennial cool-season grasses and
sedges generally made up greater than 60% of the
green standing crop during April through July each
year. Perennial warm-season grasses and sageworts be-
Figure 1. Climate diagram developed for a
96-year period for Miles City, Montana.
Mean monthly temperature (°C) and precip-
itation (mm) indicate mesic spring, early
summer, and autumn periods interrupted by
late summer and early autumn drought
(stippling). Winter precipitation occurs as
snow. Months shaded in black have average
minimum temperatures ? 0°C. Those with
diagonal lines have absolute minimum tem-
peratures ? 0°C. Figure follows standard
form of Walter (1985).
Grazing Effects on CO2in Mixed-Grass Prairie
S463

Page 3

gin to increase in dominance during July. Elevation at
the study site was 719 m.
Methods
Environment
Long-term and current precipitation and tempera-
ture records were obtained from a weather station lo-
cated at Frank Wiley Field, Miles City, MT, 12.5 km
from the study site. Precipitation data for 1996 and
early 1997 were also obtained from a study area located
6 km from the study site. These supplemental data were
used because malfunction of the nearby Bowen Ratio
system prevented continuous precipitation measure-
ments during 1996 and 1997. Precipitation was moni-
tored on site during the remainder of 1997 and 1998.
Soil water was estimated during measurement periods
on an adjacent site using gravimetric methods for the 0-
to 8-cm depth and a dielectric soil water probe for the
15-, 30-, 60-, and 90-cm depths. Gravimetric samples
were oven-dried at 105°C for 48 hr.
Treatments
Three treatments were imposed on replicated plots.
Treatments were: no grazing (control), intensively
grazed by sheep in mid-May (May), and intensively
grazed by sheep in mid-July (July). Treatments were
arranged in a randomized complete-block design with
four blocks of three treatment plots (15 m ? 15 m).
The study was conducted during 1996, 1997, and 1998.
Sheep (28–35 head) were used to impose grazing treat-
ments. Grazing events were normally about 3 hr in
duration beginning early morning (05:00 to 08:00 hr).
The desired level of grazing (60–70% removal) was
generally attained in one grazing bout, and only rarely
were the plots grazed again the next morning. Two of
the four blocks were grazed during the same week, thus
requiring two weeks to complete the treatment appli-
cation in May and July each year.
Measurements
Seasonal and diurnal effects of grazing and environ-
ment were measured from mid-April to mid-October
(weather permitting) at about 30-day intervals each
year. As with the application of grazing treatments, two
blocks were sampled during the same week during each
30-day period.
CO2flux. Net CO2flux between the soil surface and
atmosphere was measured over one randomly located
1-m2sampling quadrat per replication. The 1-m2area
was located at the beginning of the study in early spring
of 1996, and the same plot was used each year. The plot
was delineated by pressing a 1-m2angle iron frame into
the soil. At 08:00, 12:00, 16:00, and 24:00 hr on desig-
nated days, we sampled CO2flux with a portable CO2
measurement system (LI-COR 6200, LI-COR Inc., Lin-
coln, NE, USA) connected to a 1-m3chamber placed
over each plot (Angell and Svejcar 1999, Angell and
others 2001). We also measured CO2flux at 12:00 hr on
control and grazed plots the day of grazing. The cham-
ber was constructed and operated according to meth-
ods presented in Angell and Svejcar (1999). An air-tight
seal was maintained by setting the chamber on a gasket
of closed cell foam resting on the angle iron frame and
sealing any gaps around the metal angle iron frame by
tamping loose soil. Before obtaining a measurement,
the chamber was placed on the foam gasket, the cham-
ber fan was turned on, and ventilation doors were left
open to maintain near-ambient conditions within the
chamber until data logging was initiated.
To obtain a measurement, the doors were closed
and after a 30-sec mixing interval, the measurement
period began, and CO2concentration was measured
for three 20-sec or two 30-sec periods. During measure-
ments of CO2concentration, we monitored leaf and air
temperature with thermocouples and photosynthetic
active solar radiation (PAR) with a quantum sensor.
Measurement periods were brief to minimize chamber
effects, and the chamber was covered when moving
between plots to reduce heating. Net CO2fluxes above
the canopy (positive flux values indicating movement
of CO2from the atmosphere to the terrestrial surface)
were calculatedfromrate
concentration.
Soil CO2flux was measured just before the large
chamber measurements using a 1-liter LI-COR 6000-09
soil respiration chamber (Norman and others 1992)
connected to a LI-COR 6200 infrared gas analyzer.
Measurements were taken in two soil respiration rings
(polyvinyl chloride 10.4-cm diameter, 5-cm depth) per
replication. These were located near the 1-m2frames.
After CO2was drawn below ambient concentrations
within the chamber, the CO2concentration was mea-
sured for four 15-sec or three 20-sec periods. Rings were
pressed into the soil in a new location each spring to a
depth of 2.5 cm, and then left in place during the
entire April–October sampling period. Emerging seed-
lings were plucked from inside each ring at least 48 hr
before taking measurements. Soil temperature was
measured using a temperature probe placed 8 cm deep
at the time of soil flux measurements. Soil samples were
collected to a 4-cm depth from each replication for
gravimetric soil water determinations on each sample
date. Samples were oven-dried at 105°C for 48 hr.
ofchangein CO2
S464
M. R. Haferkamp and M. D. MacNeil

Page 4

Aboveground biomass. On each sample date, standing
crop was clipped to ground level and sorted by plant
species, within one 0.25-m2quadrat (50 ? 50 cm)
randomly located within each replication. Litter was
also collected from this same plot. The standing crop
sample was refrigerated and sorted into live, current
dead (i.e., this year’s senesced biomass), and dead (i.e.,
previous year’s senesced biomass) tissue. Leaf area of
the green standing crop was then determined with a
LI-COR 3050-A leaf area meter for determination of
leaf area index (LAI). Biomass samples were dried at
60°C for 48 hr and then ground before laboratory
analyses. During 1996, quadrats were located randomly
throughout the entire study area prior to grazing. Sam-
pling began in the individual May and July plots after
grazing. The assumption was made that these samples
collected before grazing were representative of the bio-
mass in plots. During 1997 and 1998, samples were
collected within each plot on all dates before and after
grazing.
Belowground biomass. On each sample date two soil
cores were obtained (4.2-cm diameter, 30-cm depth)
within each 0.25-m2frame to estimate root mass. Addi-
tional soil cores were obtained to a 60-cm depth within
each 0.25-m2frame in July each year to determine bulk
density, root biomass, and soil C. Bulk density was
determined by the core method (Blake 1965). Roots
were washed from each soil core using a hydropneu-
matic root washer, dried at 60°C for 48 hr, and ground
prior to laboratory analysis. Soil samples for C determi-
nations were stored in a freezer until they were pro-
cessed by removal of roots from soil by gentle grinding
and sieving of samples through a 1-mm screen. Small
root particles passing through the screen were removed
by hand.
Laboratory analyses of soils and biomass. All samples
were removed from storage and dried at 60°C for 16 hr
before they were weighed for chemical analyses. Soil
and plant samples were analyzed for total C with a
Carlo/Erba automated dry combustion analyzer, and
inorganic soil C was determined by the modified pres-
sure-calcimeter method (Sherrod and others 2002).
Root samples were ashed, and weights corrected for
organic matter content. Soil C values were corrected for
bulk density. Net primary production of total C was
determined for aboveground biomass by summing in-
creases in C in live and current dead biomass.
Data Summary and Analysis
The fundamental design of the experiment was a
randomized complete block with repeated observations
collected through time. Data were analyzed using anal-
ysis of variance models in which block, treatment, and
block ? treatment interactions comprised the whole
plot. The significance of treatment effects was tested
using the mean square of the block ? treatment inter-
action. The subplot varied according to the sampling
strategy for the particular dependent variable, as de-
scribed previously. In general, the subplot contained a
set of nested effects describing the time course of data
collection (e.g., year, date within year, time within date
within year for those data collected more than once per
day), each of which interacted with the treatment ef-
fects.
Results
Impact of Environment
The amount and distribution of precipitation varied
during the three years of our study (Figure 2). Precip-
itation was above average at Frank Wiley Airfield during
January, March, May, October, and November in 1996;
April and July in 1997; and March, June, July, August,
and October in 1998. Although May 1996 was very wet,
precipitation was below average thereafter until the
following October. Precipitation was below average in
May and June in 1997 and from August 1997 through
February 1998. April and May were dry in 1998, but
rainfall was above average the remainder of the sum-
mer. Average monthly temperatures were near normal
during all three years of our study. Soil water varied by
sample date (Figure 3) in response to wet and dry
periods.
Impact of Grazing
Grazing immediately reduced leaf area index (LAI),
aboveground biomass, and net CO2flux. Leaf area
index was reduced from 0.29 ? 0.03 before grazing to
0.06 ? 0.03 after grazing. Live aboveground biomass
was reduced (P ? 0.05) with grazing in all three years
whether grazing occurred in May or July (Table 1).
Current dead was similar before and after grazing in
May but was reduced by grazing in July. Dead biomass
was reduced (P ? 0.05) from 26 ? 2 to 9 ? 2 g/m2
averaged across years. Grazing did not affect (P ? 0.05)
litter.
CO2flux measurements, taken before and after graz-
ing events on both May and July plots and the corre-
sponding control plot at 12:00 hr, revealed variation
among years (Table 2). There were, however, no signif-
icant (P ? 0.05) differences in CO2flux between con-
trol plots and May or July treatment plots before graz-
ing, and no differences between control plots measured
before or after grazing (Table 2). In contrast, CO2flux
was significantly reduced by grazing in May and July.
Grazing Effects on CO2in Mixed-Grass Prairie
S465

Page 5

Seasonal and Diurnal CO2Fluxes
Net. April fluxes, measured at 12:00 and 24:00 hr in
1998, averaged 0.8 ? 0.2 and ?0.6 ? 0.2 ?m CO2/m2/
sec, with no differences among treatments. In May
1996, fluxes that were measured only at 12:00 hr were
greater than fluxes measured at 12:00 hr in 1997 or
1998 (Figure 4). Grazing in May reduced (P ? 0.05)
CO2fluxes for at least 30 days in 1996 and 1998, but not
in 1997. Grazing in July only reduced (P ? 0.05) CO2
flux for at least 30 days in 1998. Fluxes were only
greater in grazed than control plots in July 1997. Sea-
sonal fluxes were generally greatest in May and June,
intermediate in July, and near or below zero in August,
September, and October. Diurnal fluxes (Figure 5)
were generally greatest at 08:00 hr, intermediate at
12:00 and 16:00 hr, and least at 24:00 hr. The greatest
difference in diurnal measurements occurred during
May, June, and July, whereas diurnal measurements
were similar during August, September, and October.
Soil. Seasonal soil flux varied significantly with treat-
ment ? date-within-year (P ? 0.05) and time-within-
date and year (P ? 0.05). April fluxes at 12:00 and
24:00 hr in 1998 averaged ?0.9 ? 0.1 and ?0.6 ? 0.1
?m CO2/m2/sec, with no differences among treat-
ments. Some treatment differences occurred during
May, June, and July, when soil flux was less on grazed
than nongrazed plots in June and July 1997 and 1998.
Although the differences were significant (P ? 0.05),
they were small, averaging ?2.6 vs ?2.2 ?m CO2/m2/
sec in June 1997, ?2.8 vs ?2.6 ?m CO2/m2/sec in July
1997, and ?1.6 vs ?1.4 ?m CO2/m2/sec in both June
and July 1998. Variation among times was related to
season of the year (Figure 6). During spring and au-
tumn, highest rates were generally during the warmer
periods at 12:00 and 16:00 hours, but during summer
no clear trends were found. The significant (P ? 0.05)
date-within-year interaction accounted for 67% of the
total sum of squares. Respiration was generally greatest
in June and July. Although the relationships between
soil flux and soil temperature and soil water were sig-
nificant (P ? 0.05), less than 20% of the variation was
explained by either variable.
Leaf area index (LAI). Average LAI was reduced (P ?
0.05) from 0.18 ? 0.01 with no grazing to 0.14 ? 0.01
with May grazing and 0.12 ? 0.01 with July grazing.
Leaf area index varied significantly (P ? 0.05) among
dates within years (Figure 7). Maximum values were
generally recorded during May, June, and July. During
1998, however, LAI in June was significantly lower than
in May or July.
Carbon Dynamics
Aboveground C. The live component of ungrazed and
July-grazed treatments peaked in May, June, and July
(Figure 8). Current dead (CDED) increased as the
growing season progressed in nongrazed and May-
grazed plots, but increased early and then decreased
abruptly after grazing in July. The dead component was
greatest early in the season, then gradually decreased
through time without grazing, and decreased abruptly
after grazing events in May or July.
Grazing reduced the live component for at least 60
days after grazing in May and 90 days in July (Figure 8).
Compared to nongrazed control plots, grazing in May
and July reduced current dead for the remainder of the
season. July-grazed plots contained less dead than con-
Figure 2. A combination of precipitation data recorded at
the Frank Wiley Airfield (National Oceanic and Atmospheric
Administration 1996–1998) located about 12.5 km from the
study site, on another study area located 6 km from the study
site, and onsite during 1996, 1997, and 1998. Horizontal lines
above the precipitation bars denote time of sampling, and
asterisks denote time of grazing.
S466
M. R. Haferkamp and M. D. MacNeil