| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a USDA-ARS, U.S. Water Conservation Lab., 4331 East Broadway Rd., Phoenix, AZ 85040
b Dean A. Martens (deceased), USDA-ARS, Southwest Watershed Research Center, 2000 East Allen Rd., Tucson, AZ 85719
* Corresponding author (jmclain{at}uswcl.ars.ag.gov)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: GHG, green house gas • IPCC, Intergovernmental Panel on Climate Change
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Increasing atmospheric concentrations of the trace gases CO2, N2O, and CH4 are a major focus of current studies on global climate change. Atmospheric CO2 has risen from 280 to 370 µL L–1 since the beginning of the Industrial Revolution and in the same time period atmospheric CH4 has risen from 0.75 to 1.8 µL L–1 (Khalil, 1999). Atmospheric CH4 has a lifetime of 12 to 17 yr and is responsible for approximately 25% of anticipated global warming (Intergovernmental Panel on Climate Change IPCC, 1995). Preindustrial N2O levels in the atmosphere were 285 nL L–1, and have risen to 311 nL L–1 today (Khalil, 1999). With an atmospheric lifetime of 120 yr, the increase in atmospheric concentrations of N2O could have significant implications for future climate change (Schlesinger, 1997).
Atmospheric concentrations of CO2, N2O, and CH4 are each affected by biological transformations of C and N in soils. Carbon dioxide efflux is strongly correlated with SM, litter quality, and T in semiarid soils, where low rainfall acts as a strong constraint on biological activities (Conant et al., 2000, 2004). Studies have also shown that N2O production (Virginia et al., 1982) and CH4 consumption (Striegl et al., 1992) in arid soils may be concentrated in brief periods following wetting events, but the mechanisms influencing fluxes of these two trace gases are not well defined. Nitrous oxide production in semiarid soils is reported to result from respiratory denitrification, a process whereby nitrate (NO3–) is used as a terminal electron acceptor by denitrifying bacteria in anoxic soils (Virginia et al., 1982). However, recent studies by McLain and Martens found that N2O fluxes from semiarid soils were strongly correlated to soil C mineralization rates (McLain and Martens, 2005a) and that N2O fluxes were significantly reduced by additions of an inhibitor of eukaryotic activity (cycloheximide) (McLain and Martens, 2005b). These results, together with the low soil C content and lack of persistent low moisture potentials common to semiarid soils, suggest that respiratory denitrification is limited in these life zones, and indicate that much of the N2O production in semiarid soils may result from the activity of aerobic heterotrophic fungi.
Consumption of atmospheric CH4 by methanotrophic microorganisms is ubiquitous in aerobic soils of temperate, tropical, boreal, grassland, and forest systems (Mosier et al., 1991; Keller et al., 1993; Topp and Pattey, 1997). The CH4 sink in arid soils has been reported to develop in response to SM inputs (Striegl et al., 1992) but is thought to be of short duration because methanotrophs have not been found to be xerotolerant in laboratory incubations (Schnell and King, 1996).
At present, Southwestern semiarid soils receive precipitation in a bimodal distribution consisting of intense, highly variable convective precipitation during the mid-summer months with longer duration, low intensity precipitation during the winter. Some climate change models, while controversial, suggest that global warming may shift the region's precipitation distribution (currently 60% summer and 40% winter) to higher winter rainfall percentages (Magaña et al., 1997). Because organic C mineralization increases with higher soil T (Peterjohn et al., 1994), shifts in precipitation to one dominated by winter moisture may be expected to reduce annual CO2 emissions. However, semiarid monitoring studies often limit sampling periods only to warm seasons and long-term studies of trace gas fluxes across different seasons are largely lacking (Potter et al., 1996).
Here we report the results of a continuous, 15-mo study conducted in three vegetation communities in southeastern Arizona, where fluxes of CO2, N2O, and CH4, as well as SM and T at the soil surface and at several depths were monitored. The first year of the study (2002) was characterized by an extremely dry winter and spring, followed by heavy monsoon rainfalls in July through September. Winter rainfall was more abundant during the cool season of the study (fall 2002–winter 2003), but the 2003 monsoon rainfall was 60% lower than 2002, allowing us to examine microbial responses to a range in monsoon precipitation. The environmental data collected enabled us to develop predictive models to quantify the impacts of SM and T on fluxes of CO2, N2O, and CH4 across a range of soils and vegetation types.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
120 000 ha of the total land area of the southwestern USA (Hendricks, 1985), they are significant because of their landscape importance and recreational use, and their value as wildlife habitats. The SPRNCA is thus the subject of considerable scientific study, and represents a unique system to quantify the effects of ecosystem desertification and vegetation change on trace gas fluxes.
Precipitation in the SPRNCA averages 350 mm per year and at present is bimodal in pattern with 58% of annual rainfall historically occurring during the monsoon season of July to September (National Oceanic and Atmospheric Administration—National Weather Service NOAA, 2005). Winter rains (October through March) account for 35% of the total annual precipitation, while dry conditions (7% of total annual precipitation) prevail in April through June. Measurement plots were located in three sites within a 100-m transect along an alluvial terrace; the first (mesquite site) is dominated by velvet mesquite (Prosopis velutina), a C3 leguminous tree. The second site (open site) is dominated by a mix of C3 and C4 annual herbaceous dicots, including peppergrass (Lepidium thurberi), Fremont's goosefoot (Chenopodium fremontii), and toothleaf goldeneye (Viguiera dentata), and the third (sacaton site) is populated by sacaton (Sporobolus wrightii), a C4 perennial bunchgrass. The open and sacaton sites were located > 25 m from mesquite trees to minimize mesquite impacts on the soils.
Trace Gas Flux Measurements
Trace gas monitoring in the SPRNCA began in July 2002 before the first monsoon precipitation. Monitoring was done once or twice weekly during the summer monsoon period, and monthly or bimonthly in the fall and winter. Gas fluxes were measured by the static chamber technique using 22-cm diam. PVC chambers permanently installed at the soil surface. On sampling dates, lids were firmly affixed to the chamber surface and subsamples of the chamber atmosphere were removed, using gas tight syringes, every 15 min for 1 h. Gas samples were analyzed in the laboratory using a Shimadzu GC14-A Gas Chromatograph (Shimadzu Scientific Instruments, Columbia, MD) fitted with flame ionization and electron capture detectors for measurement of CH4 and N2O concentrations, respectively. The column was an 80/100 mesh HayeSep-Q, 2 m x 3 mm ID (Supelco, Inc., Bellefonte, PA) set at 45°C, using N2 as a carrier gas (flow rate, 40 mL min–1). The injector and detector were maintained at 110 and 250°C, respectively. Certified CH4 and N2O standards (Praxair Technology, Sa Ramon, CA) were used for calibration. Net fluxes were calculated from the exponential regression of the time series of trace gas concentrations.
Carbon dioxide efflux from soils beneath each flux chamber was determined following removal of the final (1 h) samples for CH4 and N2O analysis. Headspace gas from each chamber was pulled through an S151 Infra-Red Gas Analyzer (IRGA) at 500 mL min–1 (Qubit Systems, Kingston, ON) while columns containing soda lime, affixed to a second port on the chamber lid, prevented introduction of ambient CO2. The peak gas concentrations from the IRGA were used to calculate net fluxes of CO2 from the soil surface. After passage through the IRGA, headspace gas from each chamber was bubbled into a flask containing a mixture of 2 mL of saturated SrCl2 (Sigma Chemical Co., St. Louis, MO) and 20 mL of 250 mM NaOH, trapping CO2 as SrCO3 (Harris et al., 1997). Following repeated washings of the precipitate with distilled water to remove excess NaOH, dried carbonate samples were analyzed for 13C isotope composition by dry combustion with a Europa Hydra 20/20 IRMS (Europa Scientific, Crewe, UK).
Plant and Soil Analyses
Soils at the SPRNCA sites are well-drained Typic Torrifluvents of the Pima series (fine-silty, mixed, superactive, calcareous, thermic Typic Torrifluvents). Before the start of the study, samples from the 0- to 45-cm soil depth were collected using a 5-cm soil core sampler. In the laboratory, soils were sifted through a 2-mm sieve and refrigerated at 4°C until analysis. Soil texture was determined using the hydrometer method (Gee and Bauder, 1986) and pH was measured in 1:5 (soil/water) dilutions using the using the method of Van Lierop (1990) with an Orion Model 310 m (Thermo Orion, Beverly, MA) and an Orion electrode. C and N content and isotope composition of both the vegetation collected during the growing season and the soils (O-horizon to 45 cm) were determined, following acid pretreatment to eliminate carbonates, by a dry combustion analyzer interfaced with a Europa Hydra 20/20 IRMS.
Soil NO3– concentrations were determined by deionized water (5 mL) extraction of 1.0 g of air-dried soil (shaken for 30 min), followed by centrifugation at 2000 x g (RSF) for 10 min. The supernatant was decanted into a clean vial, diluted to 10 mL, and analyzed for anions using a Dionex DX-500 ion chromatograph (Dionex Corp., Sunnyvale, CA) equipped with an AS-11 anion exchange column. Anions were separated with an isocratic 10 mM NaOH gradient and detected with a Dionex ED-40 electrochemical detector set in the conductivity mode.
Measurement of Environmental Variables
During the first 10 mo of monitoring, environmental variables known to impact trace gas fluxes were measured on each sampling date. Soil T, at the surface and at 5 cm below the surface, was measured using a digital stem thermometer (Indoor Health Products, Layton, UT). Soil moisture was measured at 0 to 5 and 5 to 10 cm using permanently installed gypsum soil blocks and a KS-D1 Soil Moisture Tester (Delmhorst Instrument Co., Towaco, NJ). Rainfall data was collected using a TR-525 Rain Gauge (Texas Electronics, Dallas, TX).
In May 2003, collection of environmental variables was expanded using weather stations equipped with data loggers installed at each of the monitoring sites. Every 5 min, these stations measured air and dew point T at 60 cm above the soil surface using HOBO ProTemp/RH Data Loggers (Onset Computer Corp., Pacasset, MA); soil T at the 15- and 30-cm depths using TMC6-HA Wide Range Temperature Sensors (Onset Computer Corp.); SM at 5 to 10 cm using a Watermark Irrometer Soil Moisture Sensor (Spectrum Technologies, Inc., Plainfield, IL); and ambient CO2 concentrations at 15 cm above the soil surface using Vernier CO2-BTA CO2 Gas Sensors (Vernier Software and Technology, Beaverton, OR). All data were stored in a HOBO H8 4-Channel External Data Logger (Onset Computer Corp.). On sampling days, the data were downloaded into a HOBO data shuttle while SM values at 15 and 30 cm were collected from the gypsum blocks. Rainfall at the each site was measured on each sampling date with Tru-Chek Direct Reading Rain Gauges (Edwards Manufacturing Co., Albert Lea, MN) that contained an overlying film of mineral oil to minimize evaporation.
Laboratory Incubations
Laboratory incubations to quantify CH4 consumption were performed using open area soils collected in 5-cm increments to 45 cm. Twenty grams of soil from each depth were placed in duplicate 250-mL flasks, sealed, and incubated under near-optimum conditions for methanotroph activity (T = 20°C; SM = –34 kPa) for 5 d. Gas samples were collected from the incubation flasks every 24 h and analyzed for CH4 concentrations using GC/FID. Following each gas sampling, flasks were flushed with ambient air and resealed. Daily CH4 consumption was calculated from the decrease in headspace CH4 over the 24-h incubation period.
Calculations and Statistical Analyses
On sampling dates, the average CO2, N2O, and CH4 flux for each vegetation zone was calculated from fluxes measured at two or three chambers permanently installed at each site. Seasonal fluxes of CO2, N2O, and CH4 were calculated from the daily average of all samplings during that season, multiplied by the number of days in the season. Global warming potentials (GWPs) were calculated by adding the total seasonal emission of N2O, normalized to a CO2 equivalent (g N2O x 310), to the total seasonal emission of CO2 (g CO2), and subtracting the total seasonal CH4 oxidation, normalized to a CO2 equivalent (g CH4 x 26) (IPCC, 1995).
Stepwise linear regression models were used to determine impacts of environmental variables on fluxes of trace gas. Models that revealed strong controls on trace gas fluxes by SM and T allowed the estimation of total annual trace gas fluxes using environmental data collected by other researchers in periods during which the above-described monitoring equipment was not installed (Jan–May 2002 and Oct–Dec 2003). Comparisons between vegetation sites across multiple measurement dates were performed using repeated-measures balanced analysis of variance. All statistical analyses were done using Minitab 13.32 Statistical Software (Minitab, Inc., State College, PA). A significance level of 
= 0.05 was used for all statistical tests.
Isotope composition of the SrCO3 collected from headspace gas and concentrations of respired CO2 in the chamber headspace were used to calculate the contribution of C3 vs. C4 plants to net CO2 respired. These calculations used a mixing equation (Gearing, 1991):
 |
13Cmix = isotope ratio of total headspace CO2; 13C3 = isotope ratio of CO2 derived of C3 plants (–27
); 13C4 = isotope ratio of CO2 derived from C4 plants (–13). |
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The results of the C and N analyses of the vegetation, surface litter, and soils from each vegetation zone are presented in Table 1. The vegetation isotopic signatures clearly defined the C4 sacaton grass (13C = –13.5) and the C3 mesquite (13C = –27.1), while the signature of the open/forb vegetation (13C = –19.9) indicated C input from both C3 and C4 plants (Table 1). The surface soil C isotopic values were similar to the vegetation signatures in the mesquite and open sites, but the sacaton soil value (–16.0, 0- to 5-cm depth) was slightly more depleted than the sacaton grass litter (Table 1).
|
Environmental Variables
Throughout the study, soil surface Ts were significantly (
< 0.001) lower under the shaded mesquite canopy, averaging 23.4°C over 15 mo compared with 26.9°C in the open sites and 28.5°C in the sacaton sites (Fig. 1
). Daily air Ts during the 2003 monsoon season revealed the same pattern, averaging 25.5°C for the mesquite sites and 1.9°C warmer outside the mesquite community (average of 27.4°C for the open and sacaton sites, Fig. 2b
and 2d). Belowground soil Ts during the 2003 monsoon season were also significantly lower ( < 0.001) in the mesquite sites (23.4°C at 15 cm and 22.7°C at 30 cm) compared with the open and sacaton sites (28.2°C at 15 cm and 28.2°C at 30 cm), but significant differences ( = 0.480) were not found between the open and sacaton sites (Fig. 2b and 2d).
|
|
No differences in total precipitation were measured during the 2003 monsoon between the open and sacaton sites on opposite sides of the mesquite community, but during the same time the total rainfall in the mesquite sites was lower by an average of 36%. Despite the reduced rainfall input, SM potentials (0–10 cm) averaged over the 15-mo monitoring period were significantly ( = 0.015) higher in the mesquite sites, averaging –732.9 kPa, compared with –866.4 (open) and –1114.6 kPa (sacaton).
Carbon Dioxide
Measured soil CO2 emissions were minimal from all sites (average: 32.1 ± 14.6 mg CO2 m–2 h–1) during the premonsoon of early July 2002. Subsequent monsoon rainfall (> 50 mm over 7 d) stimulated plant and microbial respiration, and CO2 production averaged 383.4 ± 51.8 mg CO2 m–2 h–1 across all sites on July 26. Over the 15-mo monitoring period, CO2 efflux closely followed rainfall patterns, with the lowest fluxes measured following periods of limited precipitation during the warm season (Fig. 3
). During the 2002 monsoon, no significant differences ( = 0.125) in CO2 production were found between sites, as efflux averaged 242.0 ± 64.3 (mesquite), 240.6 ± 51.8 (open), and 224.9 ± 62.9 mg CO2 m–2 h–1 (sacaton). In the fall (2002) and winter (2003), average CO2 production fell to 34.1 ± 13.4 mg CO2 m–2 h–1 in the mesquite sites, efflux statistically equal ( = 0.205) to the sacaton sites (28.1 ± 4.4 mg CO2 m–2 h–1) but significantly ( = 0.020) greater than the open sites (22.2 ± 4.7 mg CO2 m–2 h–1). During the 2003 monsoon, the mesquite sites showed the highest average CO2 flux (180.6 ± 80.6 mg CO2 m–2 h–1), significantly greater than production in the open (124.1 ± 63.8 mg CO2 m–2 h–1, = 0.003) and sacaton (153.4 ± 70.0 mg CO2 m–2 h–1, = 0.023) sites.
|
The isotopic signature of respired CO2 reflected the variations in the vegetation and soils (Table 1). Throughout the monsoon, fall, and winter of 2002–2003, CO2 respired from the C3 mesquite site was significantly ( < 0.001) more depleted in 13C than CO2 arising from the open (C3 and C4 mix) and sacaton (C4) sites (Fig. 4
). The differences in isotopic composition were smallest in the 2002 premonsoon when mesquite 13CO2 averaged –21.5 while the open/sacaton 13CO2 averaged –18.8. As the monsoon progressed, the differences between the mesquite and open/sacaton sites increased, and by early September of 2002, 13CO2 values averaged –24.3 in the mesquite and –16.7 in the open/sacaton sites. Thereafter, the CO2 isotopic signature from the mesquite sites averaged –20.3 during the drier fall periods and –22.6 when soils were moist, while the 13CO2 average in the open and sacaton sites ranged from –16.7 (dry soils) to –18.3 (following nonmonsoon precipitation).
|
= 0.008), open (r2 = 54.11, = 0.009), and sacaton (r2 = 70.80, = 0.001) sites from July 2002 through April 2003. During the monsoon (2002) period, SM alone was the strongest predictor of CO2 production in the mesquite (r2 = 67.83, = 0.004) and sacaton areas (r2 = 59.18, = 0.009). Although soil T was the strongest predictor of CO2 efflux in the open areas (r2 = 26.92), T alone did not explain a significant portion of variability in net respiration ( = 0.087). The installation of data loggers to collect daily averages in 2003 improved the ability of regression models to predict CO2 flux in the open and mesquite sites (Table 2).
|
Decreased precipitation during the 2003 monsoon corresponded to > 60% reductions in average N2O efflux from the mesquite sites compared with the monsoon 2002, to 8.5 ± 5.2 µg N2O m–2 h–1, but this decrease was not significant ( = 0.123) due to the large spatial variability in N2O production. Decreased monsoon precipitation in 2003 had the opposite effect in the open sites, where N2O production increased significantly ( = 0.050) in response to decreases in SM, to 5.9 ± 4.9 µg N2O m–2 h–1. A similar, nonsignificant ( = 0.879) trend was seen in the sacaton sites, where N2O efflux was 4.0 ± 3.5 µg N2O m–2 h–1 during the 2003 monsoon.
Stepwise linear regression models were developed to predict N2O production during the monsoon of 2002 using the T, SM, and rainfall data. The best model, using data collected in the mesquite sites, accounted for 23.2% of the variability in N2O efflux, but was not significant ( = 0.165) in its ability to predict N2O production. The expanded data collection in 2003 increased the ability of regression models to predict N2O production (Table 2), as SM, above-soil CO2 concentration, and soil T at 30 cm significantly predicted N2O production in all sites from June through September 2003 (Table 2).
Methane
Premonsoon flux measurements in July 2002 following 100 rain-free days revealed CH4 consumption rates close to zero (Fig. 3c). Monsoon precipitation induced the development of a sizeable CH4 sink that continued through the winter 2002 and spring 2003, averaging 25.7 ± 6.8 (mesquite), 32.6 ± 8.8 (open), and 20.1 ± 6.8 µg CH4 m–2 h–1 (sacaton) from July 2002 through April 2003 (Fig. 3c). The total 2002 monsoon CH4 consumption was highest in the open sites, totaling 61.1 mg CH4 m–2, whereas the seasonal totals for the sacaton and mesquite sites were 29.6 and 45.6 mg CH4 m–2, respectively, while the total nonmonsoon season CH4 consumption ranged from 118.8 mg (sacaton) to 159.0 mg (mesquite) to 160.4 mg CH4 m–2 (open).
Lack of rain in May and June 2003 reduced the soil CH4 sink strength by 25 to 55% at all sites, but CH4 consumption was restored with the onset of monsoon rainfall in July 2003 and thereafter averaged 26.4 ± 6.8 and 30.1 ± 10.8 µg CH4 m–2 h–1 in the mesquite and open sites through the end of the monsoon season. During the same time period, CH4 fluxes in the sacaton sites indicated a shift from a soil sink to a source, as early July measurements showed substantial CH4 production (> 85 µg CH4 m–2 h–1). Monsoon precipitation input decreased the net CH4 efflux but the CH4 production in the sacaton continued through July (55.9 ± 22.5 µg CH4 m–2 h–1) and August (6.9 ± 2.7 µg CH4 m–2 h–1).
Laboratory incubations of open area soils collected in July 2002 showed that under near-optimum environmental conditions (–34 kPa, 20°C), large potential CH4 consumption rates were measured to a depth of 45 cm. Maximum potential oxidation rates were found at a 10- to 15-cm soil depth (Fig. 5 ).
|
= 0.140). Models also showed that near-surface (5–10 cm) SM accounted for only 4.9 (mesquite) to 31.4% (open) of the variability in CH4 consumption from July 2002 through May 2003. The expanded data collection during the second monsoon season increased the ability of regression models to predict CH4 flux (Table 2). Soil moisture had the strongest influence on CH4 flux in the sacaton site, while SM and dew point accounted for CH4 flux variability in the open site. In the mesquite site, air T, dew point T, and SM interacted to significantly predict CH4 consumption from June through September 2003.
Greenhouse Gas Emissions under Different Seasonal Rainfall Patterns
This 15-mo study period encompassed two monsoon seasons with widely different rainfall patterns. Monitoring revealed net GHG emissions for the 2002 monsoon of 310.1 (mesquite), 298.0 (open), and 308.7 g m–2 (sacaton). During the post monsoon season (October 2002 through July 2003), average daily GHG emissions were lower for all three sites, but the length of the post-monsoon period vs. the monsoon period (> 300 d vs. 57 d) increased the overall GHG emissions to 450.0 (mesquite), 330.5 (open), and 345.8 g–1 m–2 (sacaton) for the non-monsoon seasons. The 60% decrease in monsoon rainfall in 2003 compared with 2002 decreased GHG emissions to 219.1 (mesquite), 150.0 g m–2 (open), and 186.5 g–1 m–2 (sacaton). The CH4 emissions in 2003 did not significantly affect the overall GWP of the sacaton sites, as total GHG emissions without the added CH4 production totaled 185.5 g m–2. Taken together, surface GHG emissions in July through September 2002 averaged 303.1 ± 4.7 g m–2 across all sites, but fell significantly ( = 0.009) to 183.2 ± 33.9 g m–2 during the same period in 2003, a decrease of more than 40%.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Moisture Controls on Carbon Dioxide Flux
Data encompassing 15 mo of monitoring showed significant correlations of CO2 efflux and SM in all three vegetation systems during all seasons, as might be expected from results of previous work showing strong SM influences on soil respiration in semiarid systems (Amundson et al., 1989; Conant et al., 2004). During the monsoon of 2003, respiration rates in the SPRNCA also responded significantly to soil T (Table 2), in agreement with studies performed in a variety of ecosystems (Rey et al., 2002; Conant et al., 2004). Rey et al. (2002) investigated controls on CO2 efflux at a Mediterranean site and found that soil respiration was controlled by T when soils were moist (volumetric > 20%) but during dry periods, respiration was limited by SM. Taken together, these results indicate that higher soil respiration rates in response to T increases may be constrained by low SM in semiarid soils.
During the wet monsoon of 2002, SM exerted the strongest control on CO2 efflux in the higher C content soils (mesquite and sacaton), while soil T was the strongest controller of CO2 efflux from the low C content (open) site. In the presence of adequate SM, both C pool size and C/N have been shown to be strong controllers of net soil respiration in semiarid soils (Conant et al., 2000), and the strong response to moisture input measured in the higher C/N soils (mesquite and sacaton) indicates that respiration in these sites is not limited by C availability. Shading and higher organic C content of mesquite and sacaton soils may have slowed moisture loss following precipitation input, but lack of shading and low C availability would have increased evaporative drying in the open area in response to increased Ts, reducing the net CO2 flux. Thus, the T control on CO2 efflux from the open sites may also reflect SM influences.
During the monsoon of 2003, CO2 sensors recorded extreme responses to precipitation in the open and sacaton sites, but not in the mesquite sites (Fig. 2a, 2c). The spikes in ambient CO2 concentration were likely due to environmental factors present in the open and sacaton sampling sites. These sites were in exposed areas and subject to more wind movement, which would lower the soil atmospheric pressure and increase out-gassing of porespace CO2 during periods of turbulent monsoon activity. The exposed nature of the open and sacaton sites may also have led to increased soil disturbance and mixing of the surface layers during precipitation. The 35% decrease in rainfall reaching the mesquite soil surface, and the thick layer of organic matter under the mesquite canopy, would both reduce precipitation impacts on CO2 evolution.
The variations in isotopic composition between the mesquite, open, and sacaton soils may reflect historic shifts in the SPRNCA vegetation. The average biomass 13C values of C4 plants (–13 ) are distinct from C3 plants (–26 ) (Smith and Epstein, 1971). The deepest soils of all three vegetation sites bear a strong isotopic signature of C4 plants, suggesting that the SPRNCA was dominated by grasslands until recent geologic time. Indeed, mesquite encroachment into grasslands, thought to result from overgrazing, fire suppression, and climate change, has been well documented in the southwest USA (Archer, 1995; Martin et al., 2003). Only the upper soils (0–5 cm) of the SPRNCA bear a strong isotopic signature of the current dominant vegetation.
Through the summer and fall of 2002, weekly precipitation (total rainfall in 7 d preceding the measurement date) was significantly ( < 0.005) correlated with 13C of the respired CO2 of the mesquite (r = –0.87), open (r = –0.75), and sacaton (r = –0.83) sites. Mixing models showed that nearly 98% of the CO2 respired from the mesquite soils during the 2002 monsoon was C3–C. Following the monsoon, the CO2 isotopic values decreased to 63% C3–C during the dry periods and to 80% C3–C after precipitation events. Carbon dioxide respired from the open and sacaton soils averaged 52.3 ± 0.6% C4–C during the monsoon. Thereafter, CO2 respired from the open and sacaton sites was enriched (62.3 ± 3.9%) in C4–C during dry periods and depleted (52.9 ± 5.7%) in C4–C following precipitation events. These differences in isotopic composition of CO2 produced in the mesquite soils versus the open and sacaton sites are striking, but simply reflect that the respired CO2 bore isotopic signatures closely resembling that of the mineralizing plant litter (Ludlow et al., 1976). After July, a distinct pattern emerged whereby the isotopic signature of respired CO2 became more enriched in C4–C during dry periods and less C4–enriched following rainfall. The C4–enriched CO2 more closely reflects the isotopic signature of deeper soils, suggesting that microbial activity increases downward where moisture availability may be higher during dry periods. Upon soil rewetting, surface heterotrophic activity increases and the isotopic signature of the CO2 respired more closely resembles that of the surface organic C. This scenario is quite plausible, given the very sandy texture (75–83% sand, 0–10 cm) of the SPRNCA surface soils. These coarse-textured upper soils allow diffusion of atmospheric gases from the surface, preventing the development of O2 limitations on deep soil microbial respiration.
Moisture Controls on Nitrous Oxide Flux
The distribution of nutrients in the SPRNCA vegetation systems is spatially heterogeneous, with excess C and N located under the established mesquite. Mesquite, like many savanna plant species, biologically fix N and have been found to contain leaf N three times higher than non-N fixing plants (Shearer and Kohl, 1989). Tiedemann and Klemmedson (1973) found that N availability was up to 15 times higher in soils underneath mesquite trees than in nearby grassy spaces. The N2O fluxes measured in the SPRNCA ecosystem reflect these vegetation differences (15-mo averages: 11.5 [mesquite], 3.7 [open], and 3.2 µg N2O m–2 h–1 [sacaton]) and were within the range reported in other arid life zones. Martin et al. (2003) reported peak fluxes of 6.8 ± 1.7 µg N2O m–2 h–1 under mesquite canopies in North Texas rangelands, while Peterjohn and Schlesinger (1991) found denitrification N losses of 12.9 µg N2O m–2 h–1 in a Chihuahuan Desert system and Johansson et al. (1988) reported fluxes averaging 4.0 µg N2O m–2 h–1 from a tropical savanna during the dry season.
Model predictions of significant correlations between both SM and CO2 concentrations and N2O efflux such as those found in the mesquite and open sites may be expected, as C and N mineralization rates generally increase following the rewetting of a dry soil (Cui and Caldwell, 1997; Franzluebbers et al., 2000). Rapid changes in soil water potential associated with rewetting cause microbes to undergo osmotic shock, inducing cell lysis and/or a release of intracellular solutes (Halverson et al., 2000) and yielding pulses of C and N as the remaining microbes mineralize the labile cell contents (Kieft et al., 1987).
The wide variation in soil and seasonal weather conditions at the SPRNCA suggest that several microbial processes may be responsible for N2O production in the three vegetation systems. The high organic C availability in the mesquite soils, in combination with extended periods of SM levels greater than field capacity, could potentially induce the activity of denitrifiers (Peterjohn and Schlesinger, 1991). However, SM levels higher than 80% field capacity were measured only three times during the summer of 2002 (following rainfalls of 30 mm or greater), and were never measured during the summer of 2003, suggesting that the SM conditions necessary for denitrifier activity were only transient in SPRNCA soils. In systems where SM potentials do not favor denitrification, autotrophic nitrifying microbes have been shown to account for a large portion of N2O efflux (Parton et al., 1988; Davidson et al., 1993), but McLain and Martens (2005b) found that mineralization of added ammonium and nitrite to SPRNCA soils was extremely slow, suggesting that autotrophic nitrification is not a dominant process in this system. Aerobic heterotrophic denitrification by soil fungi has been shown to dominate N2O production in grasslands (Laughlin and Stevens, 2002), and McLain and Martens (2005b) reported that applications of cycloheximide, an inhibitor of fungal activity, suppressed N2O production while additions of streptomycin, a bacterial inhibitor, stimulated N2O emissions in laboratory incubations of SPRNCA soils. Taken together, these data suggest that the activity of soil fungi may account for much of the N2O production in this system. Cooke and Rayner (1984) report that fungal populations are not restricted to soil water films as are bacterial populations and thus, soil fungi are able to maintain growth and activity even under dry conditions. The ability to tolerate conditions of low soil moisture may introduce an obvious advantage for fungal populations in semiarid soils, and may account in part for the increased N2O production in open area soils during the drier monsoon of 2003. In the low C open soils, heterotrophic fungi may have had to compete with heterotrophic bacteria for C substrates in 2002. Reduced competition for C in the drier 2003 monsoon may have increased fungal activity and net N2O efflux.
Moisture Controls on Methane Flux
Increases in CH4 consumption after precipitation have been reported in other arid systems. In Mojave Desert soils, CH4 consumption following a 15-mm rainfall increased by 250%, from 17.1 to 42.9 µg CH4 m–2 h–1 (Striegl et al., 1992). The SPRNCA CH4 flux increased from 3.3 ± 4.7 µg CH4 m–2 h–1 in early July of 2002 (after more than 100 d without precipitation) to 26.7 ± 11.4 µg CH4 m–2 h–1 in mid-July. A similar increase was measured in 2003, when CH4 uptake increased from 16.6 ± 11.4 µg CH4 m–2 h–1 premonsoon to 29.9 ± 11.4 µg CH4 m–2 h–1 following moisture input.
Although the response of methanotroph activity to precipitation agrees with published findings, the sizeable CH4 consumption by SPRNCA soils nearly year-round was surprising, as studies have indicated that the rapid drying common to surface soils of semiarid ecosystems may inhibit CH4 oxidation (Schnell and King, 1996). Schnell and King (1996) found methanotrophic activity to be substantially inhibited at water potentials below –500 kPa and yet, SPRNCA soils showed near-maximum in situ CH4 consumption rates through the winter and spring of 2003, when SM potentials in the top 20 cm of soils layer averaged –800 kPa. Porespace CH4 concentration data from a semiarid rangeland (McLain and Martens, 2004) indicate that extended drying common to Southwestern soils promotes a shift in CH4 oxidation activity to deep soil layers where moisture is retained, allowing methanotrophs to remain active during periods of extended surface soil dryness, and laboratory incubations showed that under optimum SM and T conditions, SPRNCA CH4 consumption occurred to 45 cm, confirming the presence of methanotroph activity through the soil column. Such a vertical shift in maximum CH4–oxidizing activity with moisture input may be unique to semiarid and arid ecosystems. In many temperate zone soils, methanotrophic bacteria are confined to near-surface layers as diffusive transport of CH4 and O2 from the atmosphere limits their activity (Born et al., 1990). However, as evidenced by the CO2 isotopic signature, the high sand content of SPRNCA soils reduces limitations on soil gas transport, allowing substrates critical to CH4 oxidation activity to diffuse to methanotroph populations at deeper depths.
The high porosity of the SPRNCA soils makes the monsoon 2003 CH4 production in the sacaton sites all the more anomalous. Methane production is a strictly anaerobic process and yet, significant CH4 efflux from the sacaton was measured through 2 mo of the monsoon season. One plausible explanation is that this large, localized belowground CH4 source results from termite activity. Termites, which contain methanogenic bacteria in their hindgut, produce 20–150 Tg of CH4 per year on a global scale (Bachelet and Neue, 1993; Madigan et al., 1997). Subterranean termites can consume over 50% of arid land net primary production, and thus they are the prominent invertebrate cycling C and other nutrients in desert grasslands (Silva et al., 1985). Peters and Conrad (1995) tested well-aerated soils adjacent to termite colonies in South Africa and reported as many as 17 600 methanogens g–1 soil.
Global Change and Trace Gas Fluxes in Semiarid Soils
At present, it is unclear how global change will alter precipitation and water availability in semiarid areas of the southwestern USA. Some scenarios suggest increased winter precipitation and decreased summer precipitation (Magaña et al., 1997), similar to the rainfall patterns observed in the current study in the winter and monsoon of 2003. The 60% decrease in monsoon rainfall in 2003 compared with 2002 decreased net soil CO2 efflux by 40% and decreased N2O production by more than 30% across all vegetation sites. Methane consumption fell by nearly 65% in 2003 compared with 2002, a decrease attributable solely to CH4 production in the sacaton plots, while soil CH4 consumption in the mesquite and open plots was statistically equal in 2002 and 2003.
By adding the monsoon net soil GHG production to the non-monsoon production, it is estimated that the SPRNCA produced 681.0 ± 70.0 g m–2 of net GHG in 2002 and 560.6 ± 97.4 g m–2 in 2003. Taken together, these data indicate that a rainfall shift from one dominated by summer rainfall to increased winter precipitation may decrease the net GWP of the SRPNCA soils by as much as 18%. Of course, definitive conclusions on the net (soil plus vegetation) GWP represented by the SPRNCA in current and future climates cannot be made without taking into account the vegetation, which may represent a significant CO2 sink. However, because arid and semiarid soils represent more than 30% of the terrestrial surface of the earth and are increasing in area, projected rainfall shifts may represent a negative feedback on global warming.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
Received for publication April 4, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Haile-Mariam, H. P. Collins, and S. S. Higgins Greenhouse Gas Fluxes from an Irrigated Sweet Corn (Zea mays L.)-Potato (Solanum tuberosum L.) Rotation J. Environ. Qual., May 1, 2008; 37(3): 759 - 771. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||
