Density fractionation and 13C reveal changes in soil carbon following woody encroachment in a desert ecosystem
Woody encroachment has dramatically changed land cover patterns in arid and semiarid systems (drylands) worldwide over the past 150 years. This change is known to influence bulk soil carbon (C) pools, but the implications for dynamics and stability of these pools are not well understood. Working in a Chihuahuan Desert C4 grassland encroached by C3 creosote bush (Larrea tridentata), we used two density fractionation techniques (2 and 7 pool density fractionations) and isotopic analysis to quantify changes in C pools and dynamics among vegetation microsites typical of an encroachment scenario (remnant intact grassland, shrub subcanopies, and in shrub intercanopy spaces within a shrub-encroached area). The C concentration of bulk soils varied with microsite, with almost twice the C in shrub subcanopies as in intercanopy spaces or remnant grasslands. Estimated SOC accumulation rates from Larrea encroachment (4.79 g C m−2 year−1 under canopies and 1.75 g C m−2 year−1 when intercanopy losses were taken into account) were lower than reported for higher productivity Prosopis systems, but still represent a potentially large regional C sink. The composition of soil C varied among microsites, with the shrub subcanopy C composed of proportionally more light fraction C (<1.85 g cm−3) and C that was soluble in sodium polytungstate. Grassland soils contained very little carbonate C compared to shrub subcanopies or shrub intercanopy spaces. Stable isotope analyses indicate that inputs from C3 shrubs were incorporated into all density fractions, even in heavy fractions in which shrub inputs did not change overall C concentration. The seven density fractionation yielded unexpected δ13C patterns, where the two heaviest fractions were strongly depleted in 13C, indicating strong fractionation following organic matter inputs. These results suggest that the utility of isotope mixing models for determining input sources may be limited in systems with similar fractionation patterns. We propose a five pool model for dryland soil C that includes a relatively dynamic light fraction, aggregate and heavy fractions that are stable in size but that reflect dynamic inputs and outputs, a potentially large and seasonally dynamic pool of soluble C, and a large pool of carbonate C. Combined, these results suggest that dryland soil C pools may be more dynamic than previously recognized.
KeywordsLarrea Bouteloua Soil organic matter Dryland Carbonates Soil carbon stabilization Woody encroachment
Soil organic carbon (SOC) is a major component of the global carbon (C) cycle, accounting for more C than the terrestrial biomass and atmospheric pools combined (Amundson 2001). The annual CO2 flux from soils to the atmosphere is > 10 times that from fossil fuel combustion (Schlesinger 1997), so small changes in SOC pools have the potential to greatly influence the atmospheric CO2 concentration and subsequently affect Earth’s climate. Arid and semi-arid ecosystems (hereafter ‘drylands’) cover 40 % of the terrestrial land surface and account for 30–35 % of terrestrial net primary productivity (Bailey 1996; Field et al. 1998), making them a major component of the global C cycle. While SOC concentrations in dryland soils are typically low, extensive land cover change and human disturbances in these areas alter the fluxes between atmospheric and SOC pools. Integrated over the large geographical extent of drylands, these flux changes may represent a significant source or sink of atmospheric C (Lal 2004). Woody encroachment, an increase in woody plant cover in formerly grass-dominated ecosystems, may currently be dramatically altering the sink strength of dryland soils. Modeling and field-based inventories suggest that geographically extensive woody proliferation in drylands accounts for a potentially large, although highly uncertain, portion of the North American C sink. For example, Pacala et al. (2001) estimate that woody encroachment in drylands accounts for 18–40 % of the C sink in the continental United States. Understanding the current and future dynamics of this terrestrial sink requires understanding both how vegetation change affects SOC pools and controls over the stability of these pools.
Woody encroachment has fundamentally changed many of the world’s drylands over the past 150 years (Archer 1994). While the causes of this land cover change are subject to considerable debate, changes in land management practices (particularly increased grazing pressure and decreased fire frequency) appear to be major factors (Archer et al. 2001). Climate change, exotic species, N deposition, and atmospheric CO2 enrichment may also play a role (Archer 1994, 1995; Van Auken 2000). Several recent syntheses suggest that aboveground C typically increases with woody encroachment. The magnitude of this change is positively related to mean annual precipitation, with net losses in aboveground C possible in very dry sites or with increased fire frequency (Barger et al. 2011; Eldridge et al. 2011; Knapp et al. 2008). A synthetic understanding of belowground C responses to woody encroachment has proven more elusive; the magnitude, and even direction (positive, negative, or neutral) of belowground C change in response to woody encroachment remains difficult to generalize (reviewed in Wessman et al. 2004; Barger et al. 2011). The prevailing dogma for dryland woody encroachment has been that woody plants create “islands of fertility” where SOC and soil nutrients accumulate due to changes in litter input quality and quantity (e.g., Garcia-Moya and McKell 1970; Schlesinger et al. 1990). Woody plants may also promote nutrient accumulation in subcanopy zones by translocating nutrients from intercanopy areas, enhancing deposition from aeolian, fluvial or animal transport processes (Belsky et al. 1989; Okin and Gillette 2001; Weltz et al. 1998; Lajtha and Schlesinger 1986) or slowing organic matter losses through depressed subcanopy decomposition (Throop and Archer 2007). The distribution of inputs may also shift, with greater root biomass in shrub subcanopy patches than intact grasslands (Hibbard et al. 2001), and a deeper distribution of woody roots than herbaceous roots (Jackson et al. 1996). Improving generalizations about the consequence of woody encroachment to SOC pools and dynamics is crucial to reducing uncertainty about ecosystem C storage responses to woody encroachment.
One challenge to understanding and generalizing woody encroachment impacts on SOC is that the majority of SOC analyses in drylands to date has focused on bulk SOC pools. SOC exists in various states of chemical, physical, or biological stabilization. As such, SOC consists of pools that vary in turnover rates, from highly labile pools (e.g., surface detritus that often persists for <1 year) to those that persist for decades to millennia (e.g., C in highly stabilized organo-mineral complexes; Trumbore 2000). The controls over the formation and destabilization of organo-mineral complexes are still not well understood, particularly in the context of soil disturbance or changing vegetation or climate (e.g., Diochon and Kellman 2009; Kane et al. 2005). Data on bulk (whole soil) SOC provide a short-term snapshot of C pools, but provide limited insight into long-term C dynamics (Trumbore 2000; Trumbore et al. 1995). In drylands, SOC may accumulate for many decades to centuries following initial woody encroachment (Liao et al. 2006b; Throop and Archer 2008; Wheeler et al. 2007) and SOC may persist following woody plant removal for >4 decades (McClaran et al. 2008). Understanding controls over SOC stability is crucial for understanding under what conditions SOC will persist, and whether changes in bulk pools are likely to translate into long-term storage. Furthermore, understanding controls over SOC stabilization is crucial for developing and parameterizing models that can accurately predict SOC dynamics under future climate, vegetation, and land management scenarios (e.g., Hatton et al. 2012; Parton et al. 1993, 1988).
A variety of physical, chemical, and biological fractionation techniques have been used to separate out meaningful fractions of SOC that differ in relative stability (Christensen 2002; Crow et al. 2007; McLauchlan and Hobbie 2004). While no single universally accepted method exists, separation of soils into two or more density fractions via floatation provides functionally-relevant information on both associations between organic material and mineral particles and their stability (Baisden et al. 2002; Crow et al. 2007; Sollins et al. 2006). The density of soil particles varies as a function of organic matter concentration, mineral particle density, and particle porosity (Sollins et al. 2009). Previous work in mesic systems suggests that sequential density fractionation (SDF) divides soil into fractions that differ in both mineralogy and organic matter composition (Sollins et al. 2009, 2006). Organic matter concentration typically decreases with increasing particle density, and C:N ratios, δ13C, and other biochemical indices suggest that organic matter associated with denser particles has undergone greater microbial processing than that associated with less dense particles. Furthermore, radiocarbon dating indicates that C associated with denser particles typically has a greater mean residence time (Sollins et al. 2009). To date, SDF has been performed in relatively mesic systems, and it is unclear whether the patterns observed in those systems would hold under the different climate and soil composition of drylands. However, SDF procedures are expensive and time intensive, and thus are typically performed on unreplicated soil samples (e.g., Sollins et al. 2009), limiting their relevance for understanding ecological patterns that differ in time and space. However, SDF procedures could be combined with other less time intensive and expensive measurements (e.g., two pool density fractionation or incubation techniques), thus allowing inferences from SDF to be extended through time or space.
We coupled density fraction techniques and stable isotope analyses to explore changes in SOC composition and input sources following woody encroachment by creosote bush, Larrea tridentata, in a former grassland in the northern Chihuahuan Desert in New Mexico, USA. Larrea is one of the prominent woody encroachers on shallow soils in lowland deserts throughout the southwestern United States. In contrast to mesquite (Prosopis spp.), which has encroached into deeper soils, Larrea is not capable of symbiotic N2 fixation. We performed a two pool density separation on a replicated suite of soils from three microsite types typical of an encroachment scenario (remnant intact grassland, shrub subcanopies, and in shrub intercanopy spaces within a shrub-encroached area; n = 3 microsites × 10 replicates = 30), separating soils into commonly-used light fraction (“LF”; particle density <1.85 g cm−3) and heavy fraction (“HF”; >1.85 g cm−3) pools. Soils were sampled to 10 cm depth under the entire subcanopy area (or 1 × 1 m patches for shrub intercanopy spaces and grassland), thus providing an integrated representation of the subcanopy area and embracing expected patterns of sub-canopy spatial variability (Throop and Archer 2008). To further explore how C was distributed with mineral particle density, we fractionated soils from two of the vegetation microsites (intact grassland and shrub subcanopy) into seven different density fractions, ranging in particle density from <1.65 to >2.65 g cm−3 (Sollins et al. 2009). The two density fractionation allowed us to conduct a replicated study quantifying differences in C distribution among microsites, while the more costly and labor intensive seven density fractionation provided more detail about C distribution among fractions differing in particle density. For both fractionation techniques, we used stable isotope analyses to assess the contribution of C3 and C4 pathways to soil C pools. Our goal was to explore C pools and potential stabilization processes following land cover change in a desert grassland ecosystem.
Soil samples were collected from the Chihuahuan Desert Rangeland Research Center in the northern Chihuahuan Desert (Doña Ana County, New Mexico, USA; 32°30′N, 106°47′W), where shrub establishment into former grassland has occurred over the past 150 years (Gibbens et al. 2005). The study site forms part of the National Science Foundation’s Jornada Basin Long Term Ecological Research site. The climate is arid (MAP = 245 mm, MAT = 14.7 °C), with the majority (>60 %) of precipitation occurring as monsoonal thunderstorms between 1 June and 1 October (Wainwright 2006). Soils were collected from the upper bajada of Mt. Summerford, where the historically dominant C4 perennial bunchgrasss (black grama, Bouteloua eriopoda) has been largely replaced by the evergreen C3 shrub creosote bush (Larrea tridentata) during the past century. Remnant patches of intact black grama grasslands selected for sampling were located slightly upslope, but in close proximity to, areas encroached by creosote bush (Larrea tridentata) that were selected for sampling. In areas dominated by creosote bush, shrub interspaces are largely devoid of perennial plants, with the exception of patches of the invasive African grass Eragrostis lehmanniana. Areas invaded by E. lehmanniana were avoided in the sampling campaigns. Soils at all sites selected for sampling are igneous-derived coarse-loamy alluvium in the Onite-Pajarito association (coarse-loamy mixed superactive thermic Typic Calciargids and Typic Haplocambids). Although historical livestock grazing at the site has been attributed as a primary cause of the grassland to shrubland transition, grazing has been absent from the site since 1982 (Havstad and Schlesinger 2006). Soil sampling focused on surface soils (0–10 and 0–15 cm) as these depths account for a large fraction of soil profile C in deserts (Jobbagy and Jackson 2000). Furthermore, surface soils (<10 cm depth) typically exhibit the most dynamic change in SOC as a response to woody plant establishment or removal in drylands and reported woody encroachment impacts on SOC appear to be typically restricted to <20 cm depths (Barth and Klemmedson 1978; Boutton et al. 2009; McClaran et al. 2008; Tiedemann and Klemmedson 1986).
A sampling campaign was conducted in November 2010 to characterize the SOC pool difference in response to Larrea encroachment. We selected 10 replicates for each of three vegetation microsites (n = 30). Larrea subcanopy (hereafter ‘subcanopy’) samples were collected under individual Larrea shrubs that were representative of the shrubs present at the site: approximately 2.0 m canopy diameter, at least one canopy diameter from the nearest neighboring Larrea or other woody plants, and no evidence of recent surface erosion. Shrub subcanopy soils may have strong bole to dripline gradients in SOC (Throop and Archer 2008). To integrate these spatial patterns in SOC across the entire subcanopy area, we collected all the subcanopy soil to 10 cm depth, homogenized it, and retained a 250 g subsample. For the Larrea intercanopy (hereafter ‘intercanopy’) and intact black grama grassland (hereafter ‘grassland’) microsites, we obtained all the soil from a 1 m × 1 m patch, homogenized it, and retained a 250 g subsample. Selected microsite patches were at least one canopy diameter from neighboring Larrea canopies and had no evidence of recent surface erosional processes or small mammal activity. Prior to bulk sample collection at each microsite replicate, a volumetric soil core to 10 cm depth was obtained for bulk density calculation. Soils were oven dried at 60 °C, passed through a 2 mm sieve, and stored dry until density fractionation. Bulk density was calculated as the mass of the fine earth (<2 mm) fraction divided by the volume of the entire core, thus accounting for volume displaced by >2 mm fragments (Throop et al. 2012).
Soils were separated into two density fractions using sodium polytungstate (SPT) following the procedures of Sollins et al. (2006), using an initial mass of 40 g soil for each microsite sample. Samples were gently shaken in 250 ml centrifuge tubes for 3 h and spun at 1,285 rpm for 25 min. The floating material and supernatant (hereafter ‘LF’ for light fraction) were aspirated off. Both LF and the remaining heavy fraction (HF) were rinsed multiple times with deionized water and collected on glass fiber filters (Whatman GF/F). An SPT density of 1.85 g cm−3 was used because results from the seven density fractionation (below) suggested that this density provided a clear separation between free organic matter and that bound to mineral particles. We refer to this density separation as the “two density fractionation”.
To explore the density distribution of C in more detail, two soil samples from contrasting microsites (grassland and shrub subcanopy, n = 1 sample/microsite) were subjected to a more detailed SDF procedure that yielded seven density fractions. The shrub subcanopy soil was collected from the edge of a Larrea canopy. Approximately 1 kg of soil was collected with a trowel from the top 15 cm of soil profile at each sampling location. An initial mass of 180 g soil for each sample was shaken (1–3 h) with SPT in a 1 L tube, centrifuged at 3,000 rpm, and the floating material and supernatant was aspirated off and rinsed. This procedure was repeated with sequentially increasing SPT target densities of 1.65, 1.85, 2.00, 2.20, 2.45, and 2.65 g cm−3 (actual values ± 0.03 g cm−3 of target densities); for expediency we refer to the resulting density fractions by the maximum particle density (e.g. 1.65, 1.85 g cm−3). We refer to this sequential density separation as the “seven density fractionation”.
For both density fractionation procedures, samples were dried at 60 °C, weighed, and ground using a ball mill. Subsamples of each density fraction and the bulk soil were analyzed in triplicate for C and N concentration on an elemental analyzer (ECS 4010, Costech Analytical, Valencia, CA) at New Mexico State University. A second set of replicate subsamples were acid fumigated prior to CN analysis to remove any carbonates (Harris et al. 2001); carbonate concentration of the soils was calculated from the difference in C concentration between fumigated and non-fumigated samples. Analysis of internal standards indicated an analytical error of <3 % for C. In addition to mass-based concentration of C and N of individual fractions (e.g., mg C g−1 of a fraction), C and N concentration of individual fractions relative to bulk soil were calculated from fraction C or N concentration and total fraction mass data (e.g., mg C g−1 bulk soil). Samples were analyzed for δ13C with a coupled continuous-flow elemental analyzer-isotope ratio mass spectrometer (EA-IRMS) system with a Carlo-Erba model 1108 EA interfaced to a Thermo-Finnigan Delta Plus XP IRMS at the Light Stable Isotope Facility of the University of California, Santa Cruz. 13C data are reported relative to the Pee Dee Belemnite (PDB) standard. Precision of in-house standards, which had been calibrated using international standards, was typically better that 0.2 per mil for δ13C. One standard was run for every ten unknowns, and two blanks and conditioning and calibration standards were included at the beginning and end of each run. Samples were run in duplicate and were always within the range of the standards.
Statistical analyses on the two density fractionation samples were performed in JMP 7.0.1 (SAS Institute, Cary, NC). One-way ANOVA procedures were used to test for among-vegetation microsite differences in response variables. LF, HF, and bulk pools were analyzed separately. Significance of differences among means were assessed with a post hoc Student–Newman–Kuels test. No statistical analyses were performed on the seven density fractionation samples due to lack of replication.
Carbon and nitrogen pools
Grass and shrub contributions to C pools
Increases in SOC under shrubs at our study site are congruent with the well-described island of fertility effect (Garcia-Moya and McKell 1970) that sometimes includes losses of C from intercanopy soils following shrub encroachment (Schlesinger et al. 1990). While we report our results primarily on a per mass basis, greater bulk density in the shrub microsites magnify the positive impact of shrubs on areally-based C estimates. Both density fractionation techniques suggest that increases in C under shrubs are driven by light fractions composed of plant debris and entrained mineral material. In contrast, C in the heavy fractions remained essentially unchanged between the subcanopy and grassland vegetation microsites. Although P. glandulosa encroachment led to increases in particulate organic matter C within a few decades of encroachment in south Texas, accumulation of C in more stabilized macroaggregates increased dramatically, but only in stands <80 years old (Liao et al. 2006b). Incorporation of C into heavy fractions may likewise occur in the future following continued growth and organic inputs from Larrea. Our stable isotope data suggest that while the heavy fraction C concentration remains quite static relative to vegetation change, the isotopic composition of all fractions, including the heavy ones, show shifts in isotopic signature consistent with C3 inputs into these pools. Thus, even the heavy fractions, which are thought to be most stabilized (Sollins et al. 2009), appear to have dynamic inputs and outputs of recently-derived organic material.
Although the concentration of C in our dryland study system is quite low, the magnitude and rate of change in SOC in response to woody encroachment were quite high. Shrub subcanopy microsites averaged 335 g C m−2 more than grassland sites, while shrub intercanopy sites averaged 90 g C m−2 less than grassland sites. Assuming that encroachment at this site occurred in the last 70 years (Gibbens et al. 2005) and linear C accumulation under shrubs through time, we estimate a subcanopy accumulation rate of 4.8 g C m−2 year−1 (to 10 cm depth). If Larrea canopies account for 50 % of the land area (Rango et al. 2005), landscape-level accumulation (subtracting out losses in intercanopy areas) would be 1.75 g C m−2 year−1. These values are much lower than values reported for N-fixing Prosopisglandulosa in south Texas (10–30 g C m−2 year−1 to 15 cm depth; Liao et al. 2006b) or P. velutina in the Sonoran Desert (12 g m−2 year−1 to 20 cm depth; Wheeler et al. 2007). However, Throop and Archer (2008) point out that single-point sampling under shrub canopies can drastically overestimate rates of C accumulation, and suggest that C accumulates under P. velutina in the Sonoran Desert at only 2.6 g C m−2 year−1 to 20 cm depth when spatial patterns are taken into account. Our method of sampling all the subcanopy soil integrates across the entire canopy area, and thus is most equivalent to the Throop and Archer (2008) estimate. This would translate into a substantial flux of C if similar patterns are found in the >19 million ha of southwestern North America where Larrea is the dominant shrub (Van Auken 2000).
Stable isotope analyses can be used to characterize the contribution of plants with C3 and C4 photosynthetic pathways to soil C pools, and woody expansion in southwestern US drylands have been particularly amenable to these analyses due to C3 shrub encroachment into former C4 grasslands (Boutton et al. 1998). Woody encroachment at our study site leads to an approximate −2 ‰ shift in δ13C for bulk soils, similar to that caused by small Prosopis velutina shrubs encroaching into Sonoran Desert grasslands (Wheeler et al. 2007) but smaller than the −3 ‰ to −6 ‰ shift following P. glandulosa encroachment in Texas (Liao et al. 2006a). However, although our soil collection locations in the remnant C4 grasslands were devoid of C3 shrubs in close proximity, δ13C of bulk grassland soils (−18.2 ‰) was depleted in 13C relative to C4 grass inputs (−14.2 ‰). This 13C depletion can not be explained by isotopic fractionation during SOM formation, as SOM is typically 13C enriched due to preferential respiration of light 12C (Ehleringer et al. 2000). Scattered C3 shrubs, annual C3 forbs present in high favorable precipitation years, or erosional inputs from C3 shrubs found upslope (e.g., Juniperous and Quercus) may account for this. Alternatively, low 13δC values may reflect disproportionate retention of organic compounds such as lignin that are typically depleted in 13C (Benner et al. 1987).
Increases in δ13C with increasing density fraction (up to the 2.45 g cm−3 fraction) are consistent with previous observations that have been interpreted as fractionation during microbial processing, although the dramatic drop in δ13C in the two heaviest fractions has not been reported in other studies (Dijkstra et al. 2006; Sollins et al. 2009). Residual CaCO3 following acid fumigation is not an explanation, as the isotopic signature of CaCO3 is relatively enriched in 13C (Monger et al. 2009). The low δ13C could reflect a disproportionate retention of material with low δ13C or an unknown fractionation source that preferentially removed 13C. With respect to the first possibility, the two heaviest pools may be enriched in lignin, which is generally depleted in 13C relative to bulk litter (Benner et al. 1987). However, while lignin is relatively resistant to decay in surface litters (e.g., Meentemeyer 1978), recent studies suggest that it is not preferentially preserved in soils and exerts little influence on the stability of mineral-associated soil fractions (reviewed in Thevenot et al. 2010). Furthermore, C stabilized onto mineral fractions generally has an isotopic signature more similar to microbially processed organic matter than to plant compounds (Grandy and Neff 2008; Kiem and Kögel-Knabner 2003).
Isotope mixing models are used on bulk soils and, increasingly, following physical or density separation techniques, to assess the relative contribution of C3 and C4 sources to SOC pools, but our results reveal some difficulties in model interpretation posed by fractionation during SOM formation. Ideally we would have a pure C4 grassland soil and a C3 shrub soil with no previous C4 history to serve as mixing model end-members. Such pure end-members can be obtained in controlled cropping systems, yet are rare in natural systems, and thus plant isotopic signatures are generally used as at least one, if not both, end-members in mixing models. However, SOM is generally 2–5 ‰ δ13C isotopically heavier than plant inputs due to fractionation during decomposition (Krull et al. 2002; Stevenson et al. 2005); isotopic differences among our seven density fractions suggest such fractionation. To estimate the C3 and C4 contribution to SOC, we used the lightest fraction material from the grassland and shrub subcanopy. This end-member mixing model suggests that roughly 50 % shrub subcanopy C is derived from C3 inputs. Calculating contributions to stable heavier fraction C is more complex. The large clay-rich fraction (2.45 g cm−3) in the grassland was enriched by 1.6 ‰ compared to the lightest fraction. Assuming that the same fractionation would occur for a pure shrub signal, we calculated that heavy fraction C in the shrub subcanopy reflects a 70 % contribution of residual C4-C. These rough calculations suggest input of C3 material into both labile and more stable C pools.
Comparison with soils from more mesic environments
Dryland soluble C pools
Many density fraction studies using SPT have noted losses of particulate or soluble C during the fractionation procedure. Kramer et al. (2009) reported losses of under 8 % for several soils during an overnight soaking experiment in SPT, perhaps due to particulate loss during the fractionation procedure. Indeed, Crow et al. (2007) reported larger losses of C (mean = 20 %) from the same Pacific Northwest coniferous soils after a larger-scale density fractionation, perhaps due to the greater number of steps involved during that experiment. However, patterns of C loss in our two density fractionation with SPT appear to be too great to be attributable to laboratory particulate losses during a single density separation, and the microsite pattern of soluble C loss suggests an alternate, more ecologically meaningful explanation. Because the largest soluble C losses occurred in shrub subcanopy soils (35 % of total C on a per mass basis versus 2–4 % in grass or intercanopy soils), we hypothesize that microbial activity and root exudation during periods where soils are drying, but still wet enough for biotic activity, lead to the build-up of soluble organic acids in surface soils. With the onset of rain, these products are likely mobilized or respired. Results of the Crow et al. (2007) study might also reflect stores of soluble C, as these soils were collected during the prolonged summer dry season in the Pacific Northwest. An alternative explanation, that C losses were composed of residual CaCO3, is unlikely due to both isotopic results (reported δ13C of CaCO3 is −12 to +2 ‰ at our study site; Monger et al. 2009) and the disparity between the distribution of CaCO3 and SPT soluble C among vegetation microsites. Support for our hypothesis of a soluble, perhaps seasonal, C pool comes from studies of streamwater chemistry in ecosystems with marked seasonality of precipitation where increased DOC flushes occur during rains following dry periods (Brooks et al. 2007; van Verseveld et al. 2009; Vanderbilt et al. 2003), also likely due to a build up of small organic molecules as soils are drying. The existence of a pool of potentially mobilized SOC in drylands has significant implications for our understanding of mechanisms of SOC stabilization.
Dryland carbonate C pools
Carbonates are a significant C pool both locally and globally, with soil carbonates accounting for the third largest pool in the global C cycle (Schlesinger 1997). Results from this study of surface soils indicate microsite differences in carbonates, with shrub intercanopy soils having greater carbonate concentration than grasslands or shrub subcanopies. Given the same parent material for all soils, we hypothesize that greater carbonates in intercanopy soils is related to erosion of surface horizons (Wilcox et al. 1996) rather than to microsite differences in carbonate formation. Our estimates of bulk SOC changes suggest that accumulation rates under Larrea canopies (4.79 g C m−2 year−1) are greater than carbonate accumulation rates (0.012–1.3 g C m−2 year−1; Monger and Martinez-Rios 2000), and SOC accumulation rates are likely greater than carbonates even when intercanopy SOC losses are accounted for (1.75 g C m−2 year−1).
Conceptual model of dryland soil carbon
Models of SOM dynamics have used different numbers of pools, defined in different ways, to represent fractions of organic matter that are chemically or biophysically distinct and have different turnover rates, even when it is generally acknowledged that turnover times and pools are likely continuous, not discrete. For example, the CENTURY model has been widely used to represent soil C pools based on three compartments for SOM (active, slow and passive) with different potential decomposition rates defined on a theoretical basis rather than experimentally (Parton et al. 1993, 1988). Hatton et al. (2012) built on this conceptual framework, proposing a similar model for forest SOM based on density fractions. Their model contains a light fraction composed of plant fragments not associated with minerals (equivalent to our LF), aggregates or organic matter associated with light clays (1.85–2.4 g cm−3), and primary mineral particles coated with pedogenic oxides and patches of organic matter (>2.4 g cm−3). This model provides a testable and simple framework for understanding soil C pools and their dynamics.
In our study, the soluble pool and the light fraction pools both increased with woody encroachment, suggesting that woody encroachment at in this system has significant impacts on short-term C dynamics. Increases in carbonate-C with woody encroachment may reflect increased erosion of surface soils with the loss of grasses, while changes in the isotopic composition of the heavy fraction indicate that this fraction reflects dynamic inputs and outputs. Understanding the processes and rates of C stabilization in drylands are crucial for estimating the current and future dynamics of C pools in globally extensive drylands, particularly in light of dramatic land cover changes due to woody encroachment and the current uncertainty of the role of this land cover change as a terrestrial C sink.
We greatly appreciate advice, ideas, and laboratory expertise from P. Sollins and assistance from C. Laney, D. Hewins, H. Lee, J. Smith, J. Throop, and J. Wigg. Thoughtful comments from two reviewers improved this manuscript. Financial support for this work was provided by the National Science Foundation (DEB 0817064 to KL, DEB 0953864 to HT) and USDA-NRI Managed Ecosystems (2005-35101-15408 to S. Archer, HT, and M. McClaran). The Jornada Basin LTER is supported by NSF DEB 0618210.
- Archer S (1994) Woody plant encroachment into southwestern grasslands and savannas: rates, patterns and proximate causes. In: Vavra M, Laycock WA, Pieper RD (eds) Ecological implications of livestock herbivory in the West. Soc Range Manag, Denver, CO, pp 13–68Google Scholar
- Archer S (1995) Tree-grass dynamics in a Prosopis-thornscrub savanna parkland: Reconstructing the past and predicting the future. Ecoscience 2:83–99Google Scholar
- Barger NN, Archer SR, Campbell JL, Huang CH, Morton JA, Knapp AK (2011) Woody plant proliferation in North American drylands: a synthesis of impacts on ecosystem carbon balance. J Geophys Res Biogeo. doi:10.1029/2010JG001506
- Boutton TW, Liao JD, Filley TR, Archer SR (2009) Belowground carbon storage and dynamics accompanying woody plant encroachment in a subtropical savanna. In: Lal R, Follett R (eds) Soil carbon sequestration and the greenhouse effect. 2nd edn. Soil Science Society of America, Madison, pp 181–205Google Scholar
- Brooks PD, Haas PA, Huth AK (2007) Seasonal variability in the concentration and flux of organic matter and inorganic nitrogen in a semiarid catchment, San Pedro River, Arizona. J Geophys Res 112:G03S04. doi:10.1029/2006jg000275
- Havstad KM, Schlesinger WH (2006) Introduction. In: Havstad KM, Huenneke LF, Schlesinger WH (eds) Structure and function of a Chihuahuan Desert ecosystem: the Jornada Basin long-term ecological research site. Oxford University Press, New York, pp 3–14Google Scholar
- Knapp AK, Briggs JM, Collins SL, Archer SR, Bret-Harte MS, Ewers BE, Peters DP, Young DR, Shaver GR, Pendall E, Cleary MB (2008) Shrub encroachment in North American grasslands: shifts in growth form dominance rapidly alters control of ecosystem carbon inputs. Glob Change Biol 14:615–623. doi:10.1111/j.1365-2486.2007.01512.x CrossRefGoogle Scholar
- Krull ES, Bestland EA, Gates WP (2002) Soil organic matter decomposition and turnover in a tropical ultisol: evidence from δ13C, δ15N and geochemistry. Radiocarbon 44:93–112Google Scholar
- Monger HC, Martinez-Rios JJ (2000) Inorganic carbon sequestration in grazing lands. In: Kimble JM, Lal R (eds) Follett RF. The potential of U.S. grazing lands to sequester carbon and mitigate the greenhouse effect. Lewis Publishers, Boca Raton, pp 87–118Google Scholar
- Okin GS, Gillette DA (2001) Distribution of vegetation in wind-dominated landscapes: implications for wind erosion modeling and landscape processes. J Geophys Res 106(D9):9673–9684Google Scholar
- Pacala SW, Hurtt GC, Baker D, Peylin P, Houghton RA, Birdsey RA, Heath L, Sundquist ET, Stallard RF, Ciais P, Moorcroft P, Caspersen JP, Shevliakova E, Moore B, Kohlmaier G, Holland E, Gloor M, Harmon ME, Fan S-M, Sarmiento JL, Goodale CL, Schimel D, Field CB (2001) Consistent land- and atmosphere-based U.S. carbon sink estimates. Science 292:2316–2320CrossRefGoogle Scholar
- Parton WJ, Scurlock JMO, Ojima DS, Gilmanov TG, Scholes RJ, Schimel DS, Kirchner T, Menaut J-C, Seastedt T, Gardia Moya E, Kamnalrut A, Kinymario JI (1993) Observations and modeling of biomass and soil organic matter dynamics for the grassland biome worldwide. Global Biogeochem Cy 7:785–809CrossRefGoogle Scholar
- Schlesinger WH (1997) Biogeochemistry: an analysis of global change, 2nd edn. Academic Press, San DiegoGoogle Scholar
- Sollins P, Kramer MG, Swanston C, Lajtha K, Filley T, Aufdenkampe AK, Wagai R, Bowden R (2009) Sequential density fractionation across soils of contrasting mineralogy: evidence for both microbial- and mineral-controlled soil organic matter stabilization. Biogeochemistry 96:209–231CrossRefGoogle Scholar
- Wainwright J (2006) Climate and climatological variations in the Jornada Basin. In: Havstad KM, Huenneke LF, Schlesinger WH (eds) Structure and function of a Chihuahuan desert ecosystem. Oxford University Press, Oxford, pp 44–80Google Scholar
- Wessman C, Archer S, Johnson L, Asner G (2004) Woodland expansion in US grasslands: assessing land-cover change and biogeochemical impacts. In: Gutman G, Janetos AC, Justice CO et al (eds) Land change science: observing, monitoring and understanding trajectories of change on the Earth’s surface. Kluwer Academic Publishers, Dordrecht, pp 185–208Google Scholar
- Wilcox BP, Pitlick J, Allen CD, Davenport DW (1996) Runoff and erosion from a rapidly eroding pinyon-juniper hillslope, vol 1. In: Anderson MG, Brooks SM (eds) Advances in hillslope processes. Wiley, New York, pp 61–77Google Scholar