SEARCH

SEARCH BY CITATION

Keywords:

  • LiDAR;
  • drylands;
  • fire;
  • land degradation;
  • vegetation patterns

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[1] Woody plant encroachment, a worldwide phenomenon, is a major driver of land degradation in desert grasslands. Woody plant encroachment by shrub functional types ultimately leads to the formation of a patchy landscape with fertile shrub patches interspaced with nutrient-depleted bare soil patches. This is considered to be an irreversible process of land and soil degradation. Recent studies have indicated that in the early stages of shrub encroachment, when there is sufficient herbaceous connectivity, fires (prescribed or natural) might provide some reversibility to the shrub encroachment process by negatively affecting shrub demography and homogenizing soil resources across patches within weeks to months after burning. A comprehensive understanding of longer term changes in microtopography and spatial patterning of soil properties following fire in shrub-encroached grasslands is desirable. Here, we investigate the changes in microtopography with LiDAR (light detection and ranging), vegetation recovery, and spatial pattering of soil properties in replicated burned, clipped, and control areas in a shrub-grass transition zone in the northern Chihuahuan Desert four years after prescribed fire or clipping. Results indicate a greater homogeneity in soil, microtopography, and vegetation patterning on burned relative to clipped and control treatments. Findings provide further evidence that disturbance by prescribed fire may allow for reversal of the shrub encroachment process, if the event occurs in the early stages of the vegetation shift. Improved understanding of longer-term effects of fire and associated changes in soil patterning can inform the use and role of fire in the context of changing disturbance regimes and climate.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[2] Land degradation in drylands, one of the major environmental issues of the twenty-first century, has implications for the world's food security and environmental quality. The rapid progression of land degradation, associated loss of biodiversity, and impact on ecosystem services, in response to climatic changes and anthropogenic disturbances, directly or indirectly impacts the livelihood of over 2 million human inhabitants, mostly in the developing world [Millennium Ecosystem Assessment, 2005]. Across the world, drylands are undergoing vegetation shifts such as shrub encroachment and exotic grass invasion in response to complex interactions between climate and anthropogenic factors [Archer, 1999; Bradley et al., 2006; Ravi et al., 2010]. On the other hand, many of these regions are also experiencing more frequent and intense impacts from fires and droughts [Millennium Ecosystem Assessment, 2005; Westerling et al., 2006; Bowman et al., 2009; Spracklen et al., 2009]. The combined effects of these drivers may initiate, accelerate, or maintain the land degradation processes in many dryland systems with ecohydrologic, biogeochemical, and socioeconomic implications [Huxman et al., 2005; Turnbull et al., 2012; D'Odorico et al., 2011, 2012]. Hence, understanding the drivers and feedbacks of dryland degradation is motivated by increasing need to estimate long-term changes in soil productivity, designing novel solutions for sustainable management solutions, and analyzing the effect of changing climate and disturbances regimes on land degradation.

[3] Shrub encroachment in the world-wide distribution of desert grassland, a common type of dryland, is a major driver of land degradation [Schlesinger et al., 1996; Archer, 1999; Van Auken, 2000; Ravi et al., 2009]. Alteration of desert grasslands changes the spatial pattern of soil properties from a more even distribution with very fine scale spatial heterogeneity, to a pattern where nutrients, organic matter, and particle-size distributions vary at a coarser spatial scale, typically consisting of raised mounds beneath shrub canopies and bare interspaces between shrubs [Schlesinger et al., 1996]. In shrub-encroached desert grassland, the degree of spatial heterogeneity in soil patterns has been identified as an indicator of desertification [Schlesinger et al., 1996; Turnbull et al., 2012].

[4] Ravi et al. [2009]observed that the redistribution of soil resources immediately subsequent to fire or vegetation removal (clipping) in shrub-encroached desert grasslands can promote the homogenization of microtopography and associated soil properties. In the interval of 2 weeks to 4 months after treatment,Ravi et al. [2009]observed a loss of soil and associated reduction in surface elevation of disturbed shrub (creosote -Larrea tridentata) microsites coupled with deposition of sediment and increase or no change in elevation of burned shrub-interspaces. Effects of fire were more pronounced than clipping; though redistribution of sediment by wind and water from previously vegetated mounds to grass patches within shrub-interspaces occurred after both types of disturbance [Ravi et al., 2009]. Fire, in particular, was hypothesized to counter the increase in soil resource heterogeneity associated with land degradation and shrub invasion of grasslands [Ravi et al., 2009; White, 2011]. Insight into the longer-term effects for fire and clipping, in environments where a heterogeneous patterning of soil resources existed due to shrub encroachment into grassland ecosystems, is key to understanding the mechanistic processes underlying the potential reversibility of woody plant encroachment. A comprehensive, mechanistic understanding of the changes in spatial patterning of soil resources accompanying these vegetation changes and disturbances are needed to design sustainable management and resource conservation strategies in drylands, especially within the context of expected magnitudes of global change [Herrick et al., 2010].

[5] Here, we investigate the changes in soil microtopography, vegetation recovery and spatial pattering of several soil properties in replicated burned, clipped, and control areas in a shrub-grass transition zone in the Northern Chihuahuan desert four years after a prescribed fire. We examined whether the properties we measured varied by treatment and considered the extent that treatment effects differed from those observed during days to months after prescribed fire or clipping [Ravi et al., 2009]. Microtopographic variability has traditionally been characterized with measurements of the relative elevation of the soil surface over a specified distance (length scale) and/or time interval. Such measurements have typically been based on the relative elevation of a level plane between metal pins that are inserted into the ground [e.g., White, 2011]. More recently, microtopographic variability has been determined from LiDAR remote sensing, most commonly using a metric of surface roughness, such as a measure of the variance in LiDAR point elevations [Davenport et al., 2004; Eitel et al., 2011; Guarnieri et al., 2009; Haubrock et al., 2009; Rango et al., 2000; Sankey et al., 2011]. Analysis of LiDAR data has been identified as a useful tool for quantifying the impacts of dryland disturbances of fire and erosion, in general, and LiDAR measurements of surface roughness have specifically been identified as useful indicators of the susceptibility of soil surfaces to these disturbances [Eitel et al., 2011; Sankey et al., 2010, 2011].

[6] We hypothesized that soil surface roughness would be lowest on the burned and clipped treatments and greatest in the control treatment, based on observations of the reduction in microtopography attributed to post-fire erosion and deposition at this [Ravi et al., 2009; White, unpublished manuscript] and nearby study sites [White et al., 2006], as well as LiDAR analysis of fire and erosion effects on surface roughness in other shrub deserts [Sankey et al., 2010]. We hypothesized that differences between subcanopy and shrub-interspace microsites in soil total N, total, organic, and inorganic C would be lowest on the burned and clipped treatments and greatest in the control treatment, based on the assertion that homogenization of microtopography is accompanied by a redistribution of soil nutrients during post-fire erosion and deposition processes [Ravi et al., 2009]. We also expected that variability in particle-size distributions would exist between microsites and treatments, and specifically hypothesized that burned and clipped treatments would have coarser particle-size distributions relative to the control given that wind erosion following vegetation reduction can deplete sites of fine (dust) soil particles [Sankey et al., 2011, 2012].

2. Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

2.1. Study Site

[7] The field sites were located in the Sevilleta National Wildlife Refuge of the U.S. Fish and Wildlife Service, located in the northern Chihuahuan desert, 30 km north of Socorro, NM (Figure 1). This semiarid region has an average annual precipitation of about 250 mm, mostly occurring during the summer monsoon season from June through September. The soil is classified as a Typic Calciorthid with sandy loam texture (http://websoilsurvey.nrcs.usda.gov, 2012). The field experiments were conducted in a shrub-grass ecotone (34.338056 N, 106.7175 W). The dominant grass and shrub species are black grama (Bouteloua eriopoda) and creosote bush (Larrea tridentata). The ecotone represented a transition between Chihuahuan desert grassland and desert scrub habitats [e.g., Báez and Collins, 2008; Ravi et al., 2009; He et al., 2010] providing the ideal setting to investigate the biophysical feedbacks of shrub encroachment.

image

Figure 1. Maps illustrate location of (a) New Mexico relative to North America, and (b) the Sevilleta National Wildlife Refuge (SNWR) relative to New Mexico; with photos of (c) creosote shrubs in (d) ecotone of creosote-encroached grassland at the SNWR.

Download figure to PowerPoint

[8] The field sites are in a relatively level landscape with dense grass patches, scattered shrub patches and bare (soil) interspaces. The grass cover was minimal at the shrub base but provided enough connectivity among shrubs to allow for the spread of fires in the presence of strong winds. The three treatments of burned, clipped (i.e., all vegetation mechanically removed), and control plots (circular plots, 6 m in diameter) replicated 3 times were established in the spring of 2007. The plots captured the heterogeneous nature of this landscape with at least three shrub patches in each plot prior to treatment. Each set of replicated plots was approximately 50 m apart and the treatments within a replicated set were about 20 m apart. For the clipped treatment, a system of platforms was used that permitted removal of all of the vegetation without touching or disturbing the soil. Detailed description of the experimental setup is provided in Ravi et al. [2009].

2.2. LiDAR-Derived Roughness

[9] In summer 2011, four years after establishing the replicated plots, a 3 × 3 m plot containing 2 shrubs and 2 interspaces within each replicate plot was scanned with an OPTECH ILRIS 3D Terrestrial Laser Scanner (TLS) mounted on a tripod from a single scan position 3 m above ground. The TLS had an average point spacing of 1.4 mm at 5 m range. The corners of each plot were marked to ensure that the plot boundaries could be identified in the scan image. Prior to performing the LiDAR scans, we carefully removed vegetation within each plot without touching or disturbing the soil surface using the system of platforms. The mean and standard error (SE) point density of the TLS point clouds was 379,923 (17,785) points/m2.

[10] Each TLS point cloud was subset to the 3 × 3 m square centered on the plot. Each subset point cloud was processed using previously developed and described methods and LiDAR tools [http://bcal.geology.isu.edu/tools-2/envi-tools; Glenn et al., 2006; Streutker and Glenn, 2006]. In this study, surface roughness is defined as the standard deviation of all LiDAR point elevations within an area (raster cell) of specified dimensions after the local macrotopographic slope has been removed (detrended). Surface roughness was determined for each subset at 0.03-, 0.05-, 0.1-, 0.3-, 0.5-, and 1.0–m raster cell resolution. A single surface roughness value was estimated for each raster cell.

2.3. Soil and Vegetation

[11] All of the aboveground vegetation was completely removed from each plot, sorted by functional group (shrubs and herbs), dried for 3 days at 65°C, and weighed. Soil samples were collected from the surface of subcanopy and shrub-interspace microsites on each plot (2 microsites × 3 plots × 3 treatments = 18 samples). Each sample was the aggregate of 3 subsamples collected from different locations on the respective microsite. Total C and N were determined with a CNS analyzer (model NA 1500, Nitrogen/Carbon/Sulfur analyzer, Carlo Erba Instruments, Milan, Italy) after soil samples had first been ground with a ball mill. Organic C was determined for ground soil samples by treatment with phosphoric acid, drying at 105°C; the remaining C content was measured with the CNS analyzer [Artiola, 1990]. Inorganic C was the difference between total and organic C. Clay, silt, and sand were determined by hydrometer method, and sand fractions were determined by dry sieving [Soil Survey Staff, 2009].

2.4. Statistical Analysis

[12] Data for soil and biomass were tested for normality prior to analysis and transformations were applied as appropriate. The natural log of biomass was compared among treatments using an ANOVA with treatment (burn, clipped, control), functional group (shrub, herb), and their interaction as fixed factors [IBM-SPSS software version 19]. Soil C, N, SOC, IC, sand, silt and clay were compared among the treatments and microsites using an ANOVA with treatment, microsite (subcanopy and shrub-interspace), and their interaction as fixed factors. Surface roughness for each raster cell resolution was compared among the treatments using an ANOVA with treatment specified as the fixed factor. Post hoc tests with Bonferroni corrections were used to identify significant differences (P < 0.05).

[13] Omnidirectional experimental semivariograms or simply variograms by convention, of surface roughness were constructed for each plot [ENVI software version 4.8]. The elevation data were translated to a raster layer of surface roughness with a resolution of 0.03 m, and semivariance values were calculated for lag distances from 1 pixel to 30 pixels (i.e., separation distances of ∼0.03–1.00 m). The mean variogram was then calculated for each treatment as the average of all variograms for all plots of that treatment. The variograms were then fitted with theoretical variograms [variogramfit, MATLAB software version 7.10.0]. The theoretical variograms quantify aspects of spatial autocorrelation as the parameters of “nugget,” “range,” and “sill,” which relate to the amount of random variability, the distance over which data are correlated, and the total variability of the data, respectively [Woodcock et al., 1988a, 1988b].

3. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

3.1. Surface Roughness

[14] The data indicate that soil surface roughness was greater on the control treatment, and lower for the clipped and burned treatments (Figure 2), and ANOVA results confirmed that surface roughness was significantly greater on the control treatment at each raster cell resolution examined with some exceptions. These exceptions were at 0.03 m, where differences between the clipped and control treatments were less strongly significant (p = 0.07); 0.30 m, where differences between the burned and control treatments were less strongly significant (p = 0.08); and 1.0 m, where no significant treatment effect was observed. ANOVA results indicated that surface roughness did not differ significantly between burned and clipped treatments at any of the raster cell resolutions examined.

image

Figure 2. Mean surface roughness on burned, clipped, and control treatments determined at raster cell resolutions of 0.03, 0.05, 0.10, 0.30, 0.50, and 1.0 m.

Download figure to PowerPoint

[15] The variograms reveal distinct differences in the spatial autocorrelation of surface roughness found in the control, clipped, and burned treatments (Figure 3). The semivariance at each separation distance increased along the continuum of burn-clipped-control treatments (Figure 3), suggesting that surface roughness values are most homogeneous for the burned treatment, intermediate for the clipped treatment, and most heterogeneous for the control. The range of autocorrelation was approximately 50% smaller on the burned relative to clipped treatment and control, suggesting that spatial patterns of surface roughness in the landscape occurred within microtopographic units that were smallest for the burn, and larger for the other treatments.

image

Figure 3. Autocorrelation structure of surface roughness, as depicted by mean omnidirectional semivariograms with 0.03-m lag and ∼1.00-m maximum separation distance. Semivariograms depict the spatial dependence (semivariance:y axis) of samples as a function of separation distance (distance: x axis). Smaller semivariance indicates greater relative spatial dependence (autocorrelation).

Download figure to PowerPoint

3.2. C and N

[16] In the burned treatment, soil N and C (TC, OC, and IC) concentrations did not differ significantly between the two microsites (Figure 4). On the clipped treatment, soil TC and IC were significantly greater on shrub relative to interspace microsites, but N and OC did not differ between microsites (Figure 4). Soil N and C (TC, OC, and IC) concentrations were significantly greater on shrub relative to shrub-interspace microsites on the control treatment (Figure 4).

image

Figure 4. Soil total nitrogen (TN), total, organic, and inorganic carbon (C), for burned, clipped, and control treatments. Means with different letters indicate statistically significant differences (p < 0.05) among treatments and microsites determined by ANOVA. Error bars indicate standard errors.

Download figure to PowerPoint

3.3. Particle-Size Distribution

[17] Burned and clipped treatments had coarser particle-size distributions relative to the control treatment in data plots for both microsite types (Figure 5). ANOVA results (Tables 1 and 2) confirmed that sand content was significantly greater on burned and clipped treatments relative to the control, and sand content was significantly greater on shrub relative to interspace microsites. Silt and clay content were significantly less on burned and clipped treatments relative to the control. Silt and clay content were significantly less on shrub relative to interspace microsites.

image

Figure 5. Mean particle-size distributions for subcanopy and shrub-interspace microsites on burned, clipped, and control treatments.

Download figure to PowerPoint

Table 1. Mean (SE) % (by Mass) Particle Size Content by Treatmenta
 Sand %Silt %Clay %
  • a

    Means with different superscript letters indicate statistically significant differences (p < 0.05) among treatments by particle size class determined by ANOVA.

Control74 (0.9)a21 (0.8)a5 (0.2)a
Clipped81 (0.9)b16 (0.8)b3 (0.2)b
Burn81 (0.9)b16 (0.8)b3 (0.2)b
Table 2. Mean (SE) % (by Mass) Particle Size Content by Micrositea
 Sand %Silt %Clay %
  • a

    Means with different superscript letters indicate statistically significant differences (p < 0.05) between microsites by particle size class determined by ANOVA.

Shrub83 (0.7)a14 (0.7)a3 (0.2)a
Interspace74 (0.7)b21 (0.7)b5 (0.2)b

3.4. Biomass

[18] Shrub biomass was 8–20 times greater on control relative to clipped and burned treatments after 4 years (Table 3). Herb biomass did not differ significantly between burned and control treatments and was significantly less by approximately 60–70% on the clipped treatment (Table 3). On the control plots, shrub biomass was 13 times greater than herb biomass. On clipped plots, shrub biomass was 5 times greater than herb biomass. On burned plots, shrub and herb biomass did not differ significantly (Table 3).

Table 3. Mean (SE) Shrub and Herbaceous Dry Biomass (in Grams) for Burned, Clipped, and Control Treatmentsa
 Shrub Biomass (g) (SE)Herbaceous Biomass (g) (SE)
  • a

    Means with different superscript letters indicate statistically significant differences (p < 0.05) among treatments and functional group determined by ANOVA.

Control14745 (8540)d1122 (318)bc
Clipped1644 (157)c336 (115)a
Burn620 (48)ab994 (122)bc

4. Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[19] In this study, we examined the spatial patterning of soil properties in a creosote-encroached grassland in the context of disturbances of shrub and herbaceous vegetation by fire and clipping, in order to better understand mechanistically the processes that control or constrain vegetation conversion. Fire and clipping have been proposed over relatively short time scales (weeks to months post-disturbance) to oppose the increase in soil resource heterogeneity associated with shrub invasion of grasslands [Ravi et al., 2009]. We observed that a legacy of fire and clipping effects persisted 4 years after the disturbances. The soil resource heterogeneity that occurs in shrub-encroached desert grasslands appeared to be especially impacted by fire, potentially alleviating a constraint for the reestablishment of grassland vegetation structure. We observed that fire and consequent shrub-mortality resulted in the depletion of larger shrub islands relative to smaller grass-dominated islands in the burned plots (Figures 3 and 6). Redistribution of microtopography (islands) may impact a number of processes that affect vegetation structure, such as altering connectivity, changing the abundance, stature and patterning of roughness elements (vegetation and microtopographic features), and impacting the flow of fire, wind, water, sediment, and nutrients on the landscape [Okin et al., 2009].

image

Figure 6. Example raster images of soil surface roughness determined from LiDAR at 0.03 m resolution for a control and a burned plot.

Download figure to PowerPoint

[20] The land degradation effects of shrub encroachment can be explained in terms of changes in the length of connected pathways (interconnected bare soil patches) or “functional connectivity” in the landscape, which determine rates of soil erosion. Short pathways existing in desert grasslands are known to be beneficial to the landscape, such as in increasing the availability of moisture and nutrients to vegetated patches [Ludwig et al., 2005]. However, once the path lengths exceed a certain threshold as a result of overgrazing and shrub encroachment, accelerated soil erosion processes can lead to irreversible loss of resources [Turnbull et al., 2008; Ravi et al., 2010]. Long-term studies in the Chihuahuan desert have demonstrated the reinforcement of shrub islands and accelerated soil erosion following grass removal [Li et al., 2008]. Our study highlights the role of fire in constraining either the growth or reestablishment of shrub islands and allowing for grass regrowth in the interspaces, thereby decreasing the length of connected pathways. This effect of fire in the early stages of shrub encroachment might provide some form of reversibility to shrub encroachment and promote grass recovery.

[21] Less microtopographic variability was evident on the burned and the clipped treatment relative to the control measured by surface roughness and determined over a range of spatial resolutions. The effects of fire on soil surface roughness at length scales of 0.03–1.00 m in this study were comparable to a previous study that examined height variability of microtopography with erosion bridges (metal pins inserted into the ground) over similar time periods, post-fire in the Chihuahuan Desert [White, 2011]. Differences in the spatial autocorrelation structure of microtopographic variability were also evident among the three treatments. Specifically, spatial variability of surface roughness occurred within microtopographic units that were smaller for the burned treatment relative to clipped and control treatments. The patch size of shrub and herb vegetation in creosote-encroached grasslands in this environment can decrease with burning, fragmenting into smaller patches that then regrow, eventually coalescing back into larger patches in the absence of additional fires [Parmenter, 2008]. The evolution and size of microtopographic features can closely correspond to growth and death of vegetation over time [Li et al., 2008; Ursino and Rulli, 2010]. The range of autocorrelation of surface roughness, which varied from 0.21 to 0.42 m among the treatments, provides an indication of the average microsite dimensions (e.g., diameter) (Figure 3). The smaller range of autocorrelation on burned plots might indicate that a reduction in the size of microtopographic features occurred in conjunction with a reduction in patch size induced by disturbance, and highlight the way that biotic and abiotic forces control the many surface dynamics that control resource redistribution. The smaller range of autocorrelation on burned relative to clipped plots might specifically indicate that a greater reduction in microtopography occurred after burning. This could be indicative of the greater mortality that might be attributed to fire compared to clipping [Gosz and Gosz, 1996]. This also might be indicative of the greater redistribution of sediment that occurred due to erosion processes after fire relative to clipping [Ravi et al., 2009]. Fire can result in soil compaction due to combustion of organic matter and roots, alteration of porosity and soil structure, and increased soil bulk density, which might provide further explanation for the greater reduction in microtopography observed after burning [Shakesby, 2011].

[22] Ravi et al. [2009]and White [unpublished manuscript] observed that erosion occurred after fire or clipping on the study plots. Particle-size distributions that were finer on the control imply that silt and clay particles might have been lost to a greater extent from the burned and clipped treatments during the 4 years after the treatments were performed. Because the plots were relatively flat, it is more likely that sediment removed from the surfaces occurred mostly by aeolian transport. Wind erosion can deplete soil of fine particles [Leys and McTainsh, 1996; Li et al., 2009b; Lyles and Tatarko, 1986]. Many previous studies have illustrated the potential for increased wind erosion [see review in Ravi et al., 2011] and specifically increased dust emission [Sankey et al., 2011, 2012] after fire. It is not certain how the loss of fine sediment following burning or clipping may impact landscape degradation over longer time intervals, as future changes in particle size distribution, for example dust capture by vegetation [Field et al., 2012], could potentially occur with increased time since disturbance.

[23] Erosion and deposition of sediment after disturbances alters the spatial patterning of soil nutrients in deserts [Li et al., 2007, 2008, 2009a]. We observed that the decreased spatial heterogeneity of soil C and N between microsites persisted 4 years after disturbance and was more evident on burned relative to clipped treatments. Soil C and N combust and volatilize with burning, reducing concentrations of the elements in compounds near the soil surface [Houghton et al., 2000; Knicker, 2007; Prieto-Fernández et al., 2004]. Homogenization of patterning of nitrogen (NH4+ and NO3−) has been reported immediately after wild and prescribed fires in the Chihuahuan Desert in juniper-encroached grasslands [White et al., 2006]. The homogenization that we observed was evident by the lower concentrations of TN, TC, and OC on burned shrub microsites, but similar concentrations were measured on interspaces among the three treatments. TN, TC, and OC concentrations that were significantly greatest on the shrub microsites in the control treatment, in conjunction with the smaller microsite dimensions associated with burning that were apparent from the LiDAR analysis, suggest that some loss of nutrients might have occurred with burning at the plot scale (i.e., among microsites). N and C concentrations in all treatments and microsites were nonetheless larger than described for nearby grassland soils in the study area (TN = 0.05%; TC = 0.50%) [Zeglin et al., 2007]. Concentrations of N forms in soil can vary seasonally and interannually with precipitation and drought at the scale of vegetation microsites [Kieft et al., 1998]. N concentrations influence C storage in soil, which varies by soil characteristics and microbial activity [Zeglin et al., 2007]. The effect of contemporary environmental variability in N on biomass production has been suggested to be secondary to effects of drought and fire in grasslands of the Chihuahuan Desert [Ladwig et al., 2011]. Growing season precipitation was at or below long-term averages for most years during the time period of our study (2007–2011) at the SNWR [Vargas et al., 2012].

[24] We observed an homogenization of vegetation patterning in the form of greater similarity in shrub and herb biomass between microsite types, in particular on burned treatments, suggesting that fire resulted in conditions that were more characteristic of grassland relative to the other treatments. Recovery of herb biomass and cover to pre-fire levels can occur within 1–2 years of burning in the Chihuahuan Desert [Gosz and Gosz, 1996; White et al., 2006; Parmenter, 2008]. However, herb biomass recovery is dependent on adequate summer precipitation and can vary dramatically by plant species; for example, post-fire recovery rates ofBouteloua eripoda (black grama) and Bouteloua gracilis (blue grama), the common grasses at the SNWR, can differ by an order of magnitude [Gosz and Gosz, 1996; Parmenter, 2008]. Creosote shrubs, which were observed to recover via sprouts at the ground level of burned and clipped stumps, can have very low mortality with burning, and the height and canopy diameter of creosote shrubs that survive prescribed fire can regrow to ∼50% of pre-fire levels after 4 years, and to ∼100% pre-fire level in 12 years [Parmenter, 2008].

[25] Ravi et al. [2009] illustrated the negative feedback of resource redistribution for desertification immediately after fire. We show here that the effects of the negative feedback in the form of spatial patterns of soil properties were still evident 4 years later; which is potentially a successional stage when ∼50% creosote recovery might have occurred [Parmenter et al., 2008]. Previous literature suggests that a fire interval of 5–10 years can be required to maintain many desert grasslands and counter creosote shrub expansion [Wright and Bailey, 1982; White et al., 2006]. Our results with those of White [2011]imply that a minimum fire interval of 4–5 y might be required to maintain the spatial patterning of soil properties that is characteristic of a landscape that has not been degraded by shrub encroachment. Perhaps increased fire frequency (either natural or prescribed) resulting in sequential fires that occur within the timeframe of regrowth of creosote to pre-fire levels (e.g., fire intervals of 5–10 y) might counter shrub expansion and promote the homogenization of soil patterning that is more characteristic of the desert grassland system [Wright and Bailey, 1982; Parmenter, 2008; White et al., 2006; White, 2011].

5. Conclusion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[26] Grasslands and rangelands comprise 65% of the world's drylands, provide fundamental ecosystem services, and are highly susceptible to degradation by woody encroachment by shrubs. Even though several studies have demonstrated the beneficial effects of prescribed fires in grassland ecosystems, very few studies have quantified the long-term changes in soil properties and their patterning in shrub-grass ecotones, or evaluated experimental contexts that allow for a partitioning of the mechanisms involved in constraining or facilitating vegetation conversion. Our results indicate that prescribed fire in the early stage of shrub encroachment may result in homogenization of soil and vegetation patterning on burned areas, providing some form of reversibility to shrub encroachment. We observed, using a combination of LiDAR derived microtopographic measurements, soil biogeochemical analyses, and vegetation biomass measurements, the diminishment of previously shrub vegetated microsites and the formation of numerous, small grass-dominated microsites several years following the prescribed fire, indicating the progression of the system toward a state more characteristic of grassland. Such changes are likely to impact landscape connectivity, and potentially alter the flow of fire, wind, water, sediment, and nutrients on the landscape with implications for resource redistribution. Importantly, vegetation removal alone (i.e., without fire) only impacted the evolution of some of the surface characteristics associated with redistribution of soil resources.

[27] Long-term studies on post-fire changes in resource redistribution are relevant for evaluating the use of prescribed fire as a management tool, as well as in light of the dramatic increases in the frequency and intensity of wildland fires in arid and semi arid regions of the world that are occurring as a function of climatic changes [Westerling et al., 2006; Bowman et al., 2009]. A comprehensive, mechanistic understanding of post-fire landscape responses will aid in the development of sustainable management solutions, in the context of changing climate and disturbance regimes.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[28] The authors gratefully acknowledge the contributions of Scott Collins and Jennifer Johnson (Sevilleta LTER, New Mexico, USA) for providing access to field and laboratory facilities and technical guidance. This research was supported by a U.S. Geological Survey Mendenhall Fellowship (Joel Sankey), NSF Award DEB-0620482 (University of New Mexico for Long-term Ecological Research) and Philecology Foundation of Fort Worth Texas (UA Biosphere 2). We thank Chris Soulard for comments on an earlier version of this manuscript. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Geological Survey.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
jgrg945-sup-0001-t01.txtplain text document0KTab-delimited Table 1.
jgrg945-sup-0002-t02.txtplain text document0KTab-delimited Table 2.
jgrg945-sup-0003-t03.txtplain text document0KTab-delimited Table 3.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.