Quantifying how landscapes have responded and will respond to vegetation changes is an essential goal of geomorphology. The Walnut Gulch Experimental Watershed (WGEW) offers a unique opportunity to quantify the impact of vegetation changes on landscape evolution over geologic timescales. The WGEW is dominated by grasslands at high elevations and shrublands at low elevations. Paleovegetation data suggest that portions of WGEW higher than approximately 1430 m a.s.l. have been grasslands and/or woodlands throughout the late Quaternary, while elevations lower than 1430 m a.s.l. changed from a grassland/woodland to a shrubland ca. 2–4 ka. Elevations below 1430 m a.s.l. have decadal timescale erosion rates approximately 10 times higher, drainage densities approximately 3 times higher, and hillslope-scale relief approximately 3 times lower than elevations above 1430 m. We leverage the abundant geomorphic data collected at WGEW over the past several decades to calibrate a mathematical model that predicts the equilibrium drainage density in shrublands and grasslands/woodlands at WGEW. We use this model to test the hypothesis that the difference in drainage density between the shrublands and grassland/woodlands at WGEW is partly the result of a late Holocene vegetation change in the lower elevations of WGEW, using the upper elevations as a control. Model predictions for the increase in drainage density associated with the shift from grasslands/woodlands to shrublands are consistent with measured values. Using modern erosion rates and the magnitude of relief reduction associated with the transition from grasslands/woodlands to shrublands, we estimate the timing of the grassland-to-shrubland transition in the lower elevations of WGEW to be approximately 3 ka, i.e., broadly consistent with paleovegetation studies. Our results provide support for the hypothesis that common vegetation changes in semi-arid environments (e.g., from grassland to shrubland) can change erosion rates by more than an order of magnitude, with important consequences for landscape morphology.
Understanding how climate change controls landscape evolution is a central problem in geomorphology. Climate changes are multifaceted, with changes in temperature (mean and variability), precipitation (mean and variability), and vegetation cover (type and density) often occurring simultaneously. The multifaceted nature of climatic changes can make it difficult to identify which aspects of climate change are most important in driving landscape modification in specific cases. However, given the accelerated climatic changes expected to occur in the coming decades, understanding how landforms are likely to respond to specific climatic drivers, acting alone or in concert, is critically important to society (e.g., Pelletier et al., 2015).
In the southwestern USA the existence of an extensive, regionally correlative fan and valley floor depositional unit (the Q3a unit of Bull, 1991) suggests that the Pleistocene-to-Holocene transition was a major perturbation in what are now semi-arid shrubland-dominated landscapes but which were predominantly pinyon–juniper woodlands at the Last Glacial Maximum (LGM). Where rates of aggradation in modern shrubland drainage basins have been measured, the Pleistocene-to-Holocene transition was associated with more than an order-of-magnitude increase above either LGM or mid-to-late Holocene rates (e.g., Fig. 3.11 of Weldon, 1986). The retreat of grasslands and woodlands to higher elevations and their replacement by shrublands likely exposed elevations of the southwestern USA between approximately 800 m and at least 1400 m a.s.l. to significant increases in percent bare ground, thus modifying the rainfall–runoff partitioning of hillslopes and their resistance to fluvial/slope-wash erosion. In especially arid areas of the southwestern USA such as the central Mojave Desert, the range of elevations affected by late Quaternary conversions of grasslands/woodlands to shrublands extends to elevations as high as 1800 m a.s.l. (Pelletier, 2014).
Given the strong correlation between percent bare ground and drainage density in the southwestern USA (Melton, 1957), it has been hypothesized that modern shrublands sufficiently high in elevation to have been grasslands or woodlands at LGM underwent large increases in drainage density during the Pleistocene-to-Holocene transition. Such an expansion of the fluvial drainage network could have mobilized hillslope deposits stored as colluvium during the last glacial epoch, mobilizing large volumes of sediment into the fluvial system during the transition to the present interglacial (Bull, 1991; Pelletier, 2014). This hypothesis is consistent with measured ages of the onset of aggradation in valley floor and alluvial fan depositional zones in the central Mojave Desert, in which aggradation occurs earliest (ca. 15 ka) in depositional zones fed by source regions with relatively low elevations in the 800–1800 m a.s.l. range and progressively later in areas fed by eroding regions at higher elevations (Pelletier, 2014). Alternative explanations for the punctuated nature of late Quaternary aggradation in the southwestern USA invoke changes in the frequency and/or magnitude of floods (Antinao and McDonald, 2013b) and argue that hillslope vegetation changes played a limited role (Antinao and McDonald, 2013a). To date, tests of the paleovegetation change hypothesis in the southwestern USA have been limited to studies of the timing of deposition on valley-floor channels and alluvial fans. Erosion of the source drainage basins themselves has been relatively understudied.
The Walnut Gulch Experimental Watershed (WGEW) provides an excellent
opportunity to test the paleovegetation change hypothesis in a drainage
basin that has been extensively monitored for decades. The western,
low-elevation portion of WGEW is currently a shrubland while the eastern
portion is predominantly a grassland (a small area of woodland occupies the
highest elevations). Paleovegetation data, however, indicate that all of
WGEW was a grassland or woodland until just a few thousand years ago.
Scientists working at WGEW have measured rates of sediment export from
watersheds using sediment samplers (e.g., Nichols et al., 2008) and rates of
sediment redistribution within watersheds using anthropogenic radionuclides
(e.g.,
The WGEW is part of the USA
Department of Agriculture (USDA) Agricultural Research Service's (ARS)
Southwest Watershed Research Center (SWRC). WGEW is located at the boundary
between the Chihuahuan and Sonoran deserts and elevations of between
approximately 1300 and 1600 m a.s.l. The approximately 150 km
The bedrock geology of WGEW includes sedimentary, plutonic, and volcanic rocks of Precambrian to late Cenozoic age (Fig. 1). Due to the complex nature of the rock types exposed in the southern portion of the watershed, we focus this study on the northern portion of the watershed, which is dominated by the Gleeson Road Conglomerate (GRC).
Maps of the bedrock geology and geomorphology of Walnut Gulch
Experimental Watershed (WGEW).
Relationships among landscape morphology and vegetation cover
in the study area.
The GRC was derived primarily from erosion of the Dragoon Mountains to the
east of WGEW and is estimated to be Plio-Pleistocene in age by Osterkamp (2008), who noted that “The upper part of the Gleeson Road Conglomerate is
probably equivalent both stratigraphically and in age to the
Plio-Pleistocene upper basin fill of Brown et al. (1966). To the west and
northwest, along the axis of the San Pedro River Valley, the upper part of
the Gleeson Road Conglomerate grades into fine-grained fluviatile and
lacustrine beds of the Plio-Pleistocene St. David Formation (Gray, 1965;
Melton, 1965).” The GRC dips gently (5–8
The northern portion of WGEW is composed of the whetstone pediment of Bryan (1926) and is divided into two parts: the Dissected Whetstone Pediment (DWP) at elevations below approximately 1430 m a.s.l. and the Upper Whetstone Pediment (UWP) at higher elevations (Fig. 1). The DWP is distinguished from the UWP by its lower relief and less well-developed soils. Differences between the DWP and UWP have been attributed primarily to headward extension of tributaries resulting from late Quaternary incision of the San Pedro River (Cooley, 1968) as well as renewed river and tributary incision following livestock grazing beginning ca. 1880 (Renard et al., 1993; Osterkamp, 2008). However, the boundary between the DWP and UWP coincides with the transition from the modern shrubland to the grassland (Fig. 2). As such, we hypothesize that vegetation cover and its changes over geologic timescales have also contributed to differences between the DWP and UWP.
Deep sandy gravel loams of the Blacktail–Elgin–Stronghold–McAllister–Bernardino Group occur in areas of the UWP. In the lower, western part of the watershed, soils are in the Luckyhills–McNeal Group. Soils of the Luckyhills–McNeal Group tend to be sandy and gravelly loams that are immature compared to soils of the Blacktail–Elgin–Stronghold–McAllister–Bernardino Group. The A horizon of the Luckyhills–McNeal Group is typically absent, having been removed by late Quaternary erosion (Breckenfeld, 1995). The boundary between these two soil groups coincides with the boundary between the DWP and UWP and the transition from the modern shrubland to the grassland.
Mean annual temperature at Tombstone (located in the western portion of WGEW
at an elevation of 1384 m a.s.l.) is 17.6
Modern vegetation cover in the USA–Mexico borderland region closely
follows elevation, with desert scrub occurring below approximately 1500 m,
grasslands from 1400 to 1700 m, lower encinal (
Paleovegetation records from the borderland region suggest that the western
portions of WGEW have transitioned from a grasslands/woodland to a shrubland
over the past few thousand years, while the eastern half of the watershed
has been a grassland/woodland for at least the past 40 000 years and likely
since the penultimate interglacial period. Low stalagmite
WGEW is home to two intensive monitoring sites, one in the shrubland of the DWP and the other in the grassland of the UWP. These sites provide the data necessary, in conjunction with the topographic analyses presented here, to calibrate a mathematical model that predicts the equilibrium drainage density as a function of vegetation cover and to test the hypothesis that differences in landscape morphology and erosion rates between the northwestern and northeastern portions of WGEW are partly the result of a transition from grassland/woodland to shrubland within the past few thousand years in the northwestern portion of the drainage basin that did not occur in the northeastern portion.
Watersheds 102, 103, 104, 105, and 106 are located in shrublands at an
elevation of approximately 1370 m a.s.l. in what is referred to as “Lucky
Hills,” which has been the site of a variety of intensive scientific studies
since the 1960s. Cover during the rainy season at Lucky Hills is
approximately 25 % bare soil, 25 % canopy, and 50 % erosion
pavement (rocks) (Nearing et al., 2007). Dominant vegetation includes
Creosote (
Sediment yield data and watershed characteristics. The erosion rate
calculation assumes a soil bulk density of 1500 kg m
Watershed 112 is located in the Kendall grassland site at WGEW, approximately
10 km east of Lucky Hills and at an elevation of approximately 1525 m.
Kendall drains to the west and has roughly equal areas of north- and
south-facing hillslopes. The site is predominantly covered by grass and forbs
with some shrubs and succulents with a combined canopy cover of approximately
35 %. Ground cover during the rainy season has been measured at 28 %
rock, 42 % litter, and 14 % plant basal cover (Nearing et al., 2007).
Compared to the 25 % bare soil at Lucky Hills, the bare soil exposed at
Kendall is negligible (i.e., a few percent or less). Historically, the
dominant desert grassland bunchgrasses at the site have been black grama
(
Nearing et al. (2005) used spatially distributed
Drainage density is higher in shrubland areas than in
grassland areas of the study site. Shaded relief images of representative
1.6 km
In this section we describe the methods used to quantify the similarities
and differences in landscape morphology between the modern grassland and
shrubland sites, with an eye toward providing the data necessary to
calibrate the mathematical model as described in Sect. 3.2. In all of the
topographic analyses described in this section we used a 1 m pixel
The drainage density in the shrubland areas appears to be much higher than
in the grassland areas (Fig. 3). To quantify this difference we used the
method developed by Pelletier (2013) to identify the network of valley
bottoms (i.e., where water flow is localized and fluvial processes are
responsible for the majority of erosion) in the vicinity of Lucky Hills and
Kendall. In this method, the DEM is filtered using the OWF, the contour
curvature is computed at every pixel, and valley heads are identified as the
areas closest to the divides where the contour curvature exceeds a
user-defined threshold value. In Pelletier (2013) and in this paper a
threshold contour curvature of 0.1 m
Relief was mapped as the difference between the highest elevation upstream from each pixel along flow lines. The results of the drainage network identification procedure described above were used to limit this analysis to the hillslope pixels only. We then computed the mean hillslope relief within 10 m elevation bins from 1320 to 1550 m a.s.l. The resulting graph quantifies how the mean hillslope relief varies with elevation across the shrubland-to-grassland transition.
We also computed the mean value of the along-channel slope gradient and curvature (i.e., the Laplacian) as functions of contributing area. Differences in mean curvature as a function of contributing area provide a quantitative signature of how late Holocene vegetation changes have modified the landscape morphology in the vicinity of the hillslope-to-valley-bottom transition.
In this section we describe the mathematical model used to quantify erosion over geologic timescales and its dependence on landscape morphology at WGEW. The mathematical model is used to predict the equilibrium drainage density, quantified as the mean distance along flow lines from divides to valley bottoms, in both shrublands and grasslands, in order to test the hypothesis that the difference in drainage density observed between the shrublands and grasslands at WGEW can be attributed, in part, to late Holocene vegetation changes in the shrubland portion of the watershed.
Erosion at WGEW over geologic timescales can be approximated as the sum of
erosion due to colluvial and fluvial/slope-wash processes. Sediment
transport by colluvial processes increases approximately linearly with slope
based on short-term monitoring studies (Kirkby, 1967) and cosmogenic
radionuclide analyses (McKean et al., 1993) when slopes are uniformly
soil-mantled and topographic gradients are modest. Linear slope-dependent
transport, combined with conservation of mass, leads to a diffusion equation
for topography (Culling, 1960):
We assume that fluvial/slope-wash processes in WGEW can be approximated as transport limited. We use the term fluvial/slope-wash to refer to all sediment transport by flowing water, wherever it occurs along the continuum from hillslopes (i.e., as sheet and rill flow) to channels (i.e., fluvial erosion). A transport-limited condition applies to landscapes in which most of the sediment is transported as bed-material load and sediment is readily deposited when the shear stress by flowing water declines with increasing distance along flow pathways. In the alternative detachment-limited end-member model of fluvial/slope-wash erosion, the shear stress required to detach sediment is much larger than the shear stress required to transport it; hence sediment redeposition is rare or nonexistent once detachment/entrainment occurs. Pelletier (2012) addressed the geomorphic conditions under which transport-limited vs. detachment-limited conditions are likely to occur, taking into account data for the relative proportion of sediment transported as bed-material load vs. wash load, among other factors, using WGEW as a case study. He concluded that among these two end-member models, WGEW is most accurately considered to be transport limited.
The assumption of transport-limited conditions implies that
fluvial/slope-wash erosion,
Next, we further parameterize the three terms in Eq. (
Next, we introduce three power-law relationships that relate volumetric
sediment flux to contributing area, channel width to contributing area, and
contributing area to distance along flow lines from topographic divides.
Sediment flux is a power-law function of contributing area at WGEW:
Miller (1995) measured 222 cross-sectional channel profiles in the field at
WGEW and used those data to calibrate a power-law relationship between
channel width and contributing area:
To estimate the time required for low-order channels to grow headward in
response to hillslope vegetation changes, we modeled the channels as a
diffusive system (e.g., Begin, 1988). In diffusive systems, the timescale,
Figure 3 illustrates the results of the drainage network identification for
1.6 km
Detailed shaded relief images (locations shown in Fig. 3)
illustrating the presence of parallel hillslope rills and gullies in the
shrubland areas (shown in
Schematic figure of a valley head. The profile shown along
A–A' is the cross section of the valley head where colluvial sediment flux
from hillslopes of mean gradient
Mean hillslope relief increases substantially across the shrubland-to-grassland transition (Fig. 6). Between elevations of approximately 1320 and 1430 m a.s.l., mean relief is uniformly low (approx. 0.5–1 m). Above elevations of approximately 1450 m, hillslope relief increases abruptly and continues to increase with increasing elevation.
Plot of mean hillslope relief as a function of elevation, illustrating the marked increase in relief above elevations of approximately 1430 m a.s.l. in the study area. Each data point represents the mean hillslope relief in 10 m wide elevation bins.
Plots of contributing area vs. mean distance from the
divide for the shrubland (approximated as the portion of the study area
below 1430 m a.s.l.) and grassland areas (above 1430 m). The dashed line
plots the piecewise power-law relationship (Eq.
Plots of mean topographic curvature as a function of
contributing area (distance from divide also shown along
Contributing area follows a piecewise power-law function of distance along
flow lines from topographic divides (Fig. 7), with one set of values for
Plots of mean topographic curvature (i.e., the Laplacian of
In this section we use the results of Sect. 3.1., together with analyses
of the sediment yield reported by Nearing et al. (2007), to constrain the
terms in Eqs. (
Plot of the decadal timescale volumetric sediment fluxes in
shrublands measured at the five watersheds of Lucky Hills as a function
of contributing area. The straight line is the result of a least-squares
regression to the logarithms of both sides of Eq. (
List of model parameters and values.
Measured (from DEM analysis) and predicted values (from Eqs.
We used Eqs. (
Figure 10 plots the magnitude of the three terms for a range of possible
mean distances from divide to valley bottom. The fluvial erosion rate is
We used the mean curvature vs. contributing area data plotted in Fig. 8
to reconstruct an average longitudinal profile from divides to valley
bottoms in shrublands and grasslands in order to infer the approximate
relief reduction associated with the late Holocene shift from
grasslands/woodlands to shrublands in WGEW. Figure 11 illustrates the
results of this integration. Integrating the curvature vs. contributing
area data in Fig. 8 twice leads to two constants of integration: one is constrained by the requirement that the slope along flow pathways
at divides is zero and the other by using a constant reference elevation at
a contributing area of approximately 300 m
Plots of the total erosion rate,
Plots of the mean longitudinal profile of hillslopes and valley bottoms in shrubland and grassland areas, constructed by integrating the mean curvature data in Fig. 8.
In order to test the hypothesis that 2–4 kyr is sufficient time for drainage
density to have fully responded to the recent
grassland/woodland-to-shrubland transition, we used Eqs. (
Equations (
In particular, Kirchner et al. (2001) demonstrated the potential pitfall of
applying sediment yields or erosion rates measured at one timescale to
another timescale by demonstrating that rates over different timescales
can differ by orders of magnitude. We agree that this is a significant
potential concern. However, we believe that our decadal-scale sediment
yields are an appropriate estimate for millennial-scale sediment yields
based on three lines of argument. First, as noted above, published analyses
have shown that fluvial sediment transport in WGEW, while highly episodic,
is not dominated by just a few extreme events. The effective discharge (i.e.,
the discharge above which half of the total load is transported) occurs many
times within a 30-year record. In a study of sediment transport in the
1995–2005 period, for example, Nearing et al. (2007) addressed this issue as
follows: “For six of the seven watersheds, between 6 and 10 events produced
50 % of the total sediment yields over the 11-year period.” That is, the
effective discharge has a recurrence interval of between approximately 1 and
2 years. Second, sediment yields calculated from 1995 to 2005 by Nearing et al. (2007) closely match yields measured over approximately 50 years using
We also propose that the difference in mean curvatures between shrublands
and grasslands within the range of contributing areas from
Figure 5 demonstrates that mean hillslope relief increases substantially across the shrubland-to-grassland transition. We propose that some of this difference in hillslope relief is a consequence of the difference in fluvial/slope-wash erosion rates, i.e., that higher erosion rates in the lower-elevation shrublands have caused relief reduction in the past few thousand years relative to the higher-elevation grasslands, and that this difference in fluvial/slope-wash erosion rates is the result of a geologically recent increase in drainage density in the shrublands of WGEW. However, it is likely that a portion of the difference in mean hillslope relief across the study site also reflects variable uplift rates, i.e., the fact that the uplift of any piedmont or foothill region tends to increase towards the mountain range. Flexural-isostatic response to erosion (which has been proposed to be an important component of late Cenozoic landscape evolution in southern Arizona; Menges and Pearthree, 1989; Pelletier, 2010b) of the Dragoon Mountains can be expected to have caused eastward tilting of WGEW, i.e., higher uplift rates in the higher elevations of WGEW compared to the lower elevations. Tilting would not explain the abrupt increase in relief at elevations just above 1430 m a.s.l., however, since no faulting occurs in the vicinity of this contour. Therefore, it is likely that some of the difference in hillslope-scale relief across the shrubland-to-grassland transition at WGEW is a result of the difference in erosion rates between the shrublands and grasslands. While we can be certain of grassland-to-shrubland shift only during the present interglacial period, the Quaternary period has seen many interglacial periods broadly similar in climate to the current period; hence it is likely that the lower elevations of WGEW have seen grassland-to-shrubland conversions more than once over the past approximately 2 Myr. Each of these episodes could have contributed to relief reduction of the lower elevations of the study site relative to the higher elevations.
Previous studies at WGEW have emphasized the role of base-level lowering and vegetation changes within the past 130 years on the differences in erosion rate between Lucky Hills and Kendall (Nearing et al., 2007). However, recent paleovegetation studies have provided a new perspective. Specifically, Holmgren's (2005) documentation of shrubland species in the region several thousand years before present at an elevation less than 100 m lower than Lucky Hills suggests that the lower elevations of WGEW likely shifted from a grassland/woodland to a shrubland prior to the 1880s. While base-level lowering has steepened hillslopes and channels close to the main-stem channel of Walnut Gulch, hillslope-scale relief is clearly larger at Kendall than at Lucky Hills (Fig. 6), indicating that base-level lowering may be a dominant factor only for those areas within close proximity to the main channel. The magnitude of the differences in topography (i.e., drainage density and the magnitude of erosion than can be inferred from the change) is difficult to fit into a period as short as 130 years. Given erosion rates measured over the past 60 at Lucky Hills, approximately 2 cm of erosion can be expected to have occurred over the past 130 years. Figure 11 suggests that erosion associated with a recent increase in drainage density is likely at least 10 times this value and thus more consistent with a vegetation change that occurred several thousand years before present.
The model of this paper contributes to our broader understanding of the controls on drainage density and it provides a mathematical model for predicting drainage density that may be useful in other study sites.
Previous studies have demonstrated that drainage density is controlled by
relief (e.g., Montgomery and Dietrich, 1992; Tarboton, 1992; Tucker and Bras,
1998), climate (Melton, 1957; Abrahams and Ponczynski, 1984), parent
material (e.g., Ray and Fischer, 1960; Day, 1980), and time (e.g., Ruhe, 1952;
Dohrenwend et al., 1987). While many studies have demonstrated the
importance of individual factors on drainage density, we lack a
comprehensive model for drainage density that integrates all of these
factors. Equation (
Drainage density is most commonly found to be an inverse function of mean annual precipitation or effective precipitation. This finding is consistent with the conceptual model of this paper that vegetation cover is the predominant climate-related variable that influences drainage density and that vegetation cover and drainage density are inversely related. Melton (1957), for example, documented an inverse correlation between drainage density and the precipitation-effectiveness (P-E) index at over 80 sites in the southwestern USA, including arid (low-elevation) and humid (high-elevation) climates. A similar negative correlation between drainage density and MAP was found by Abrahams and Ponczynski (1984). Naively, one might expect more precipitation to result in greater channel incision and hence less contributing area between divides and valley heads (and hence a higher drainage density), all else being equal. Melton (1957), however, proposed that greater aridity results in a lower vegetation density and, hence, a reduction in the cohesive strength protecting soils on hillslopes, thus leading to a higher drainage density. One might also expect a lower vegetation density to increase the runoff-to-rainfall ratio and hence also increase drainage density, but runoff intensity varied by only a factor of two across Melton's sites while drainage density varied by nearly two orders of magnitude. Istanbulluoglu and Bras (2007) provided theoretical support for Melton's interpretation, illustrating that a lower vegetation densities can lead to higher drainage densities through the cohesive or anchoring effect of plant roots.
There has been an ongoing debate in the geomorphic literature regarding the
importance of climate (not limited to but often defined as MAP) on erosion rates. Given the significant correlation
between MAP and erosion rates in many studies within individual mountain
ranges (e.g., Reiners et al., 2003; Bookhagen and Strecker, 2012), it is
perhaps surprising how little correlation exists between MAP and erosion
rates in global compilation/synthesis studies (e.g., von Blanckenburg, 2005;
Portenga and Bierman, 2011). Even studies that emphasize the climatic
control on erosion rates note that
Recent work on the role of vegetation, and its changes through time, can provide a basis for understanding the relatively low correlation between erosion rates and MAP in unglaciated areas outside the dominant influence of periglacial processes and the complex relationship between erosion rates and climate in such areas more generally. For example, Torres Acosta et al. (2015) recently documented a negative correlation between erosion rates and both vegetation cover and MAP in Kenya. They proposed that the primary effect of more humid conditions is to increase vegetation cover on hillslopes, thereby reducing erosion rates on otherwise similar slope gradients. This concept is consistent with the classic Langbein and Schumm (1958) curve. Langbein and Schumm (1958) proposed that sediment yields are maximized in semi-arid climates (all else being equal) because such climates generate sufficient rainfall to detach and transport soil in overland/rill flow but insufficient vegetation cover to protect/anchor the soil. As MAP increases in this conceptual model, more precipitation is available to drive erosion, but this effect is more than offset by a decrease in the susceptibility of soil to erode due to the increased anchoring effect associated with greater plant cover/biomass. The results of this paper demonstrate further complexity in the erosion–climate relationship, i.e., that the change in climate (and hence of vegetation cover) can be as important or more important that its mean state. It is important to emphasize that the effect of vegetation on the rate of erosion by colluvial processes may be entirely different than its effect on fluvial/slope-wash processes. All else being equal, increased vegetation cover is likely to increase erosion rates by colluvial processes, since more plants can be expected to drive higher rates of bioturbation (e.g., Osterkamp et al., 2012). As such, it is crucial to consider colluvial and fluvial/slope-wash processes separately when considering the effects of vegetation on hillslope erosion.
In this study we leveraged all relevant data from a uniquely well-studied semi-arid watershed to test the hypothesis that late Holocene vegetation changes can modulate drainage density, hillslope-scale relief, and watershed-scale erosion rates. We documented that areas below 1430 m a.s.l. have decadal-scale erosion rates approximately 10 times higher, drainage densities approximately 3 times higher, and hillslope-scale relief approximately 3 times lower than elevations above 1430 m. We calibrated all the terms of a mathematical landscape evolution model and used the model to predict the equilibrium drainage density associated with shrublands and grasslands. Model predictions for the increase in drainage density associated with the shift from grasslands/woodlands to shrublands are broadly consistent with measured values. Using modern erosion rates and the magnitude of relief reduction associated with the transition from grasslands/woodlands to shrublands, we also estimated the timing of the grassland-to-shrubland transition in the lower elevations of WGEW to be approximately 3 ka, i.e., broadly consistent with constraints from paleovegetation studies. Our work provides a mathematical model for predicting equilibrium drainage density in transport-limited fluvial environments that may be applicable in other study sites.
The DEM used in this paper can be obtained upon request from the corresponding author. All other data used in the paper are in the published literature.
We wish to thank Tyson Swetnam for drafting Fig. 2c. Jon D. Pelletier was partially supported by NSF award no. 1331408. We thank two anonymous reviewers for thoughtful, constructive reviews that helped us improve the quality of the paper. Edited by: S. Mudd