Legacy earthen berms influence vegetation and hydrologic complexity in the Altar Valley, Arizona

Mry H.Nihols , Sr E.Duke , Chndr Holifield Collins , Luren Thompson

a Southwest Watershed Research Center, USDA-ARS, 2000 E.Allen Rd., Tucson, AZ, 85719, USA

b Southern Plains Area Research Center, USDA- ARS, SPARC, College Station, TX, 77845, USA

c University of Arizona, College of Science Graduate, Tucson, AZ, 85719, USA

Keywords:Legacy conservation structure Earthen bank Earthen berm Land degradation Rangeland

A B S T R A C T Across the American Southwest water development has played a critical role in managing rangelands.Earthen berms have been constructed throughout US rangelands to manage runoff and reduce erosion.The berms altered runoff patterns to increase soil moisture with positive local vegetative response.However, altered runoff patterns can be considered a disturbance that affects broader scale vegetation patterns.We hypothesized that the hydrologic impacts of earthen berms in semiarid rangelands will be reflected in contrasting upslope and downslope vegetation patterns.A supervised classification of grass,shrubs, and bare soil was performed using orthographic imagery taken in June 2016 to quantify the effects of 181 earthen berms in the uplands and floodplain of the Altar Valley in Southern Arizona, US.Intact berms blocked runoff,creating downslope runoff“shadows”within which the precipitation is the dominant water input.We documented more bare soil downslope of intact berms in comparison with upslope.Grass and shrub cover affected by berms were related to soil texture.Grass and shrub cover were not different above and below intact berms on fine textured soils,but on sites with coarser textured soil,grass cover was lower downslope of berms.Where breaches occurred on coarser textured soils, the up and downslope differences in grass cover diminished.This study points to the role of conservation structures in adding additional complexity to already heterogeneous landscapes by creating patchwork assemblages of vegetation and bare soil proximal to earthen runoff and erosion control berms.

1.Introduction

Throughout the world, the topography of agricultural landscapes has been altered to control the flow of water with both positive and negative impacts on ecohydrologic processes.In cultivated agricultural lands, hillslopes are commonly altered to form terraces that reduce slope lengths thereby positively affecting hydrologic processes such as infiltration.However, the failures of such terraces can lead to increases in hydrologic connectivity and concentrated runoff (Meerkerk et al., 2009).On floodplains,earthworks constructed to control runoff have been shown to alter hydrologic and vegetation patterns by reducing the lateral extent of surface flooding (Kingsford, 2000; Steinfeld & Kingsford, 2013).In drylands where agriculture relies on small dams and water diversions to ensure water security, earthen banks and dams can reduce the area that is hydrologically connected to the watershed outlet (Callow & Smettem, 2009).In addition, anthropogenic influences that support agricultural operations, such as roads, can alter overland flow and thus influence vegetation patterns in areas where soil moisture is derived from surface runoff (Schlesinger &Jones,1984).

In the western United States(US),the primary agricultural use of rangelands is livestock production (Starrs, 1998).Limited water supply and low annual vegetation production characterize many of these lands (Havstad et al., 2009).Rangelands are inherently heterogeneous landscapes associated with underlying geology, soils,landscape position,and topography that influence the distribution of soil moisture supplied by highly variable and uncertain precipitation (Sheppard et al., 2002) and runoff (Goodrich et al., 2008).The importance of such heterogeneity is increasingly recognized in the study and management of rangeland landscapes (Fuhlendorf et al., 2017).Both inherent heterogeneity and disturbances, such as grazing and fire, can interact to create process feedbacks that result in vegetation patterns that shift in both time and space at the landscape scale(Archibald et al.,2005;Fuhlendorf&Engle,2004).In rangelands managed for livestock production,controlling runoff by managing flood flows to enhance soil moisture and stabilize water supplies can reduce variability in soil moisture and increase the extent of specific desired plant communities (Fuhlendorf &Engle, 2001).Thus, water management is integral to rangeland management.

Throughout US rangelands, earthen berms have been constructed to control surface runoff.Berms may be constructed singly and are often constructed in series for multiple reasons including 1)to form dams that create stock tanks or farm ponds (Berg et al.,2016; Nichols, 2006), 2) to direct water into stock tanks, 3) to control overland flow on both uplands and in floodplains where they reduce runoff velocities and spread runoff laterally thus increasing infiltration time to increase soil moisture and enhance vegetation production (Rango & Havstad, 2011), and 4) to prevent runoff from reaching channel banks and gully heads thereby preventing scour and reducing erosion.If designed and constructed properly,berms for controlling overland flow and reducing erosion retain runoff on the upslope side and route excess runoff nonerosively around the berm end.Constructed berms have a finite lifespan and in the absence of maintenance two common modes of berm failure are flanking (lateral scour around the berm end) and breaching(perpendicular break through the berm).Both intact and unmaintained berms are a significant control on runoff pathways(Nichols et al., 2018; Frankl et al., 2021).

Altered topography and runoff interact with soil properties such as texture and depth to influence soil moisture that varies in both space and time thereby influencing plant responses (McAuliffe,1994, 2003; McAuliffe & Hamerlynck, 2010) and landscape vegetation distribution patterns (Hamerlynck et al., 2002).Perennial warm season grasses can take advantage of moisture in shallow soil layers and become dormant when these supplies are depleted.In contrast, woody plants that have both deeper and laterally more extensive root systems can both take advantage of shallow moisture and reach moisture that may be more persistent at depth(Jackson et al., 1996; McAuliffe, 2003).Recently, prolonged droughts in the Southwestern US have provided an opportunity to study how soil characteristics and soil hydrologic behavior affect desert plant mortality (Hamerlynck & McAuliffe, 2008).In the Mojave Desert,drought impacted plant canopy die-back was found to be greater in weakly developed soils,while those plants growing on older well-develop soils showed less plant mortality.Although much of the literature relating vegetation patterns to soil moisture distribution focuses on natural hydrologic variability, it is reasonable to expect that distinct vegetation patterns will also be found in landscapes where common rangeland management practices have altered topography and thus runoff patterns and soil moisture distribution.

The goal of this study is to explore the influence of runoff and erosion control berms on proximal vegetation patterns in semiarid rangelands in southern Arizona,USA.Our objectives are to:1)apply a remote sensing approach to quantify vegetation and bare soil upslope and downslope of earthen berms,and 2)to investigate the role of earthen berms on vegetation patterns.This study is based on observations of landscape scale patterns.As such, hypothesis testing is limited to identifying general spatial patterns and is not intended to determine the causal relations among environmental variables.We hypothesize that the hydrologic impacts of earthen berms in semiarid rangelands will be reflected in contrasting upslope and downslope vegetation patterns.Both intact and flanked berms are expected to retain runoff on the upslope side,although if flanked a portion of the runoff will be routed around berm ends.In both cases runoff does not reach the adjacent area downslope of the berms.Therefore,if the berms are intact or flanked,we expect there will be more vegetation and less bare soil on the upslope side of berms in comparison with the downslope side.If the berms are breached and the up and downslope sides are hydrologically connected,we expect no location dependent differences in vegetation and bare soil.

2.Methods

2.1.Study site

This study was conducted in the Altar Valley in southern Arizona,US(Fig.1).The semiarid watershed encompasses 247,000 ha that drain from south to north.There is no naturally occurring perennial surface water in the Altar Valley and runoff is generated primarily in response to intense and infrequent summer convective thunderstorms.More than half of the 415 mm of annual rainfall occurs from July-September during the North American Monsoon(Adams & Comrie,1997).

Fig.1.Location of sites within the Altar Valley,Arizona,UTM 12N,selected to quantify the influence of legacy earthen berms on vegetation and bare soil.Site A is located on the floodplain and sites B through E are upland sites.

Soils in the valley are heterogeneous consisting of aridisols and entisols (Beaudette & O’Geen, 2009; https://casoilresource.lawr.ucdavis.edu/gmap/).Alluvial slopes are covered with thin, stony soils and floodplain soils are typically deep sedimentary accumulations.Vegetation on floodplains historically was dominated by big sacaton (Sporobolus wrightii Munro ex Scribn.).Currently,mesquite (e.g., Prosopis velutina Wooton) and shrubs (e.g., yellow paloverde(Parkinsonia microphylla Torr.)and creosote bush(Larrea tridentata (DC.) Coville) occupy the floodplains and Johnsongrass(Sorghum halepense (L.) Pers.) has replaced much of the sacaton.Uplands support native perennial grasses including grama grasses(Bouteloua spp.),bush muhly(Muhlenbergia porteri Scribn.ex Beal),and tobosagrass(Pleuraphis mutica Buckley).However much of the upland area is now occupied by woody shrubs among which mesquite trees are dominant,as well as by annual grasses and nonnative perennial grasses such as Lehmann lovegrass (Eragrostis lehmanniana Nees).

Approximately 80% of the valley supports 19 agricultural operations, primarily ranches, ranging in size from 4000 to 28,000 ha.These ranches are made up of a mosaic of private,federal,state,and county owned lands.Water management is fundamental to ranching operations in the valley.Since the late 1880's, the landscape has been modified to build earthen stock tanks and berms to control surface water distribution and supply,as well as to mitigate erosion (Nichols & Degginger, 2021; Nichols et al., 2018).We identified areas within the Altar Valley where multiple earthen berms were constructed through the 1900s beginning in approximately the 1930s.We selected five areas where berm density was high.We have limited information on the management objectives of the berms,and exact construction dates are unknown.However,current land managers have indicated that most of the berms were constructed for runoff harvesting and/or erosion control.Four of the sites are located on alluvial fan uplands on the west side of the valley and one is in the floodplain (Fig.1).During prior research(Nichols & Degginger, 2021), berms were identified and classified as intact,breached,or flanked based on Google Earth Imagery and aerial LiDAR data (2011 and July 9-14, 2016) sourced from Pima County Regional Flood Control District.We do not have information on the dates of berm failures.For each berm, soil texture data consisting of percent sand, silt, and clay were obtained from the National Cooperative Soil Survey accessed through the online SoilWeb tool developed by USDA-NRCS in cooperation with the California Soil Resource Laboratory at the University of California,Davis (https://casoilresource.lawr.ucdavis.edu/gmap/).SoilWeb is based on the NRCS Soil Survey Geographic (SSURGO) database(available online at https://sdmdataaccess.sc.egov.usda.gov.Accessed 06/27/2022), within which soil variability is represented by map units that are used for ranch level land use planning.We were not able to collect soil samples in the field, thus the assignment of soil texture to each berm was determined at the map unit scale.

2.2.Cover classification

A supervised classification was performed using the image classification wizard within ArcMap(ArcGIS Desktop[version 10.4],ESRI, Redlands CA) to identify grass, shrubs, and bare soil within orthographic imagery collected in June 2016 sourced from the Pima Country Regional Flood Control District.The 3-band red,green,blue(RGB) imagery was collected at a resolution of 0.75m.The orthophoto set consists of individual 1609 by 1609 m tiles.Between three and 10 tiles covered each of the five study areas within which we analyzed the influence of berms.These tiles were used without any smoothing or resampling for both training and validation of classified pixels.Within the orthophoto tiles, for each berm, we manually delineated an upslope and a downslope polygon with sides approximately equal to the length of the berm.Within each polygon, we identified homogeneous pixels for each class.Individual pixels were classified visually as grass, shrub/tree, or bare soil based on RGB color.The visually classified pixels made up the training data sets that were used within ArcGIS to classify pixels across entire tiles.

After classifying all the pixels within each tile,50 pixels for each class were randomly selected within each tile and used to assess classification accuracy.The machine classified value for each randomly selected pixel was not revealed to the classifier until after the classifier independently assigned a class to the pixel.A confusion matrix was computed to compare the number of pixels assigned to each class relative to the actual class determined visually.Cohen's kappa coefficient was computed as an index of classification accuracy (Foody, 2002).Our overall target accuracy was 80%,with no individual class less than 70%(Thomlinson et al.,1999; Wulder et al., 2006).After assuring accurate classification based on our target accuracies,the percent shrubs/trees,grass,and bare soil were quantified within each polygon.

2.3.Statistical analysis

This observational dataset was explored for patterns of similarity among earthen berm structures for soil texture, vegetative cover,and amount of bare soil in the context of berm status(intact,flanked, or breached).Principal Components Analysis (PCA) based on correlations was used to find the dominant sources of variance in soil and surface cover variables.Soil properties and vegetation cover are typically correlated and thus contain redundant information and PCA can reduce the dimensionality of such data.Through PCA, the data were linearly transformed to rotate the original feature space to represent the data without correlation in the new component space.Additional un-supervised Cluster Analysis methods were used to validate the PCA and identified very similar groups based on soils and cover.PCA groups were used because it gives a measure of variability explained by this method,and data points are oriented within a meaningful continuous data space.The second PC axis dominated by cover is a multivariate axis of surface cover so the individual components(bare soil,grass,and shrub cover) cannot be differentiated using the PC2 values.Therefore, within each of the three dominant soil texture PC1 groups, bare soil and grass and shrub cover were assessed to test the hypothesis that upslope values are equal to downslope values for each structure condition(intact,breached,or flanked).We used a one-way ANOVA for the berm-status cover-position combined variables and made three specific comparisons of upslope vs downslope cover within each berm status.We chose this approach over a two-way ANOVA because a two-way ANOVA analysis fits main effects for berm-status but is not appropriate because we do not have knowledge of the age or time since failure which are factors that confound the interpretation.Further, the comparisons of interest lie within the interaction term which is equivalent to the one-way ANOVA we have used.For comparisons across the three berm statuses,we adjusted our p-value threshold for interpretation based on a simple Bonferroni adjustment (p = 0.05/3 = 0.017)(Westfall & Wolfinger, 1997).Statistical analyses were conducted using SAS JMP®software(JMP®,Version<16.1>.SAS Institute Inc.,Cary, NC,1989-2021).

3.Results

3.1.Berm characteristics and cover classification

A total of 181 berms were included in the analysis (Table 1).Approximately 89% of the berms were less than 100 m long, and 40%were less than 50 m long.Slightly more than half(52%)of the berms on clay loam and sandy loam soils remain intact,while 43%of the berms on loamy sand remain intact.Flanks, identified by visual scour pathways around the end of a berm, were morecommon than breaches.Fig.3 provides an example of the relation of the drainage network affected by flanked berms (indicated by flow paths that go around the end of a berm) and associated vegetation patterns in an upland area within the valley.Almost half(47%) of the berms on the loamy sand soils were flanked in comparison with 31% on the finer textured clay loam soils.However,breaches were more common on the clay loams soils(17%of berms)in comparison with those on loamy sand soils (9%).The overall cover classification accuracy ranged from 82 to 96% and kappa coefficients ranged from 81 to 93% (Table 2).

Table 1 Status dependent number and length of legacy earthen berms in the Altar Valley, Arizona, by Principal Component (PC) group.

3.2.Principal component analysis

The correlations between soil texture (Table 3) and cover type are shown in Table 4.We focus our analysis on the dominant two Principal Components that explain 93%of the variability in the data(Fig.2).Within the first Principal Component (PC1) three distinct groups separated by percent sand (+ values) vs clay and silt (-values)defined the first major axis of variability.Soil characteristics explain 57.9% of the variability in the data.The second Principal Component(PC2)explains 36.6%of the variability and was defined by cover type of percent bare soil(+values)and percent vegetation(- values) (Fig.2).

Table 3 Mean (standard deviation) soil texture of legacy earthen berms in the Altar Valley,Arizona by principal component 1 (PC1) group.

The classification of berms according to soil texture allowed us to identify three groups with homogeneous soil texture that broadly represent berms from different sites and structure conditions: Group1 (red) is dominated by structures from site E with some structures from A and B, Group2 (green) predominantly includes structures from site B, and Group3 (blue) includes all structures within site D and some structures from sites A and C.The berms on sites making up Group1 are located on clay loam soilssignificantly higher upslope of intact berms in comparison with downslope grass cover on both sandy loam(40 vs.30%)and loamy sands(51 vs 41%)(Table 5).On the clay loam soils,we also found no differences in upslope and downslope grass cover associated with berms that have failed through either flanking or breaching.On the coarser textured soils, we did find significantly higher grass cover upslope of flanked berms in comparison with downslope grass cover.However,we did not find differences in grass cover upslope of breached berms in comparison with downslope grass cover on characterized by lower sand(25%)in comparison with higher sand associated with berms on sandy loams of Group2 (66% sand) and loamy sands of Group3(83%sand)(Table 3).Within each soil group,we analyzed differences in bare soil, grass, and shrub cover based on spatial relation to the berms(upslope or downslope)for each of the berm conditions(intact,breach,or flanked).A limitation of this point-in-time study is that it is not possible to make direct comparisons among the status of the berms (intact, flanked, or breached)because we do not know when the berms were built or compromised.Thus, we do not know how much time the vegetation has had to adjust to altered surface runoff.As a result, our interpretations are focused on upslope and downslope comparisons for each of the three berm statuses.

Table 2 Confusion matrices and kappa coefficients for cover classification of vegetation associated with legacy earthen berms at each of five sites in the Altar Valley, Arizona.

Table 4 Correlations among vegetation cover classes associated with legacy earthen berms in the Altar Valley,Arizona.Bold indicates a significant difference(p-value<0.01)for the correlation.

Fig.2.Principal component analysis(PCA)model of soil and surface cover with legacy earthen berms in the Altar Valley,Arizona.The left panel displays all data points above and below all berms in the new PC data space.The right panel is a bi-plot of the first two principal components, showing the first PC axis dominated by soil textures and with three distinct groupings according to soil clearly differentiated.The second PC axis is dominated by soil cover variables.Bare soil cover shown as positive values and vegetation(grass and shrub) values fall in the negative direction.

3.3.Bare soil

The amount of bare soil is significantly higher downslope of intact berms located on all three soil types in comparison with upslope values(Table 5).On the finer textured clay loam soils,if the berms failed through flanking or breaching, we did not find differences in up and downslope bare soil.In contrast,on the coarser sandy loams and loamy sands, we found that bare soil was significantly higher downslope of intact berms as well as those that failed through flanking or breaching.

3.4.Grass cover

Overall, as expected, grass cover was higher upslope of intact berms on each soil type.However, on clay loam soils grass cover upslope of intact berms (44%) was not statistically different from downslope grass cover (40%).In contrast, grass cover was the coarser soils.Overall, both up and downslope grass cover associated with failed berms was less than that associated with intact berms along with a corresponding increase in bare soil(Table 5).

3.5.Shrub cover

Patterns in location dependent differences in shrub cover are more variable than patterns found in grass cover.On clay loam soils there was no difference in shrub cover up and downslope of berms that are intact, flanked, or breached.On coarser soils we found inconsistent shrub cover responses to berms depending on soil type and berm status.Although shrub cover was higher upslope than downslope of intact berms on sandy loams oils,there was no shrub cover difference upslope and downslope of intact berms on coarser loamy sands.It is notable that overall, the lowest shrub cover was found proximal to intact berms on loamy sands (15%upslope and 11% downslope).Although no difference was found between up and downslope shrub cover proximal to breached berms on sandy loam soils, we did document up and downslope differences in shrub cover associated with breached berms on loamy sand soils.In summary, these results document patterns in cover upslope and downslope of the berms that vary with soil characteristics.

4.Discussion

Rangeland heterogeneity is both inherent and associated with disturbances such as grazing and fire (Fuhlendorf et al., 2017).Although soil and water conservation structures can be effective tools for reducing erosion and increasing water use efficiency,theytoo can be considered a disturbance.Construction typically creates local site disturbances including excavating, piling, and packing soil, thus altering topography.Once constructed, runoff interacts with the altered topography to seek new pathways thereby changing the spatial distribution of soil moisture,as well as altering sediment transport and deposition dynamics (Stavi et al., 2020).This creates downslope runoff “shadows” within which berms block runoff creating proximal areas downslope where precipitation is the dominant water input.These changes can cause unintended consequences as topography and runoff dynamics continue to evolve (Nichols et al., 2018).For example, over time, as deposition occurs in response to reduced velocities,and as runoff backs up behind the berms, the accumulated sediment increases the possibility that runoff will route around the end of berms.Once concentrated flow paths are established around the ends of berms,continued erosion creates channels that can convey increasingly large volumes of runoff that no longer travels as sheet flow, thus limiting the potential to infiltrate and enhance soil moisture.Concentrated flow paths and distinct runoff shadows influenced by the presence of berms can be readily identified in imagery and high-resolution topographic data.

Table 5 Simple t-test results comparing upslope to downslope cover for%bare soil,%grass cover,and%shrub cover within each PC group by status of legacy earthen berms in the Altar Valley,Arizona.These are simple evaluations of evidence of difference and not causal effects per se.The evidence suggests a difference of>10%in cover classes is detectable as significant at alpha=0.017,which is a modest Bonferroni multiple comparison adjustment for making three comparisons(0.05/3=0.017)to minimize chance of Type-1 errors.

In semiarid areas,vegetation and hydrologic processes are both complex (Hutjes et al., 1998) and strongly linked (Ludwig &Tongway,1995; Ludwig et al., 2005; Wilcox et al., 2003a).Our understanding of the ecohydrologic processes interacting to create and sustain patches in rangelands is grounded in conceptual frameworks such as the trigger-transfer-reserve-pulse model(Ludwig & Tongway, 2000), which describe the interaction and feedbacks that unevenly distribute water and nutrients resulting in natural vegetation patches.Vegetation patchiness is characteristic of many drylands where precipitation and soil moisture are naturally variable (Noy-Meir,1973).In patchy rangelands, runoff from sparsely vegetated areas can concentrate in vegetation patches and initiate feedbacks whereby increased infiltration and soil moisture enhance vegetation which in turn captures more runoff (e.g., Wilcox et al., 2003b; Ludwig et al., 2005).In practice, water and nutrients can be concentrated by constructing berms to harvest runoff generated during occasional precipitation events thereby increasing soil moisture that is critical for vegetation germination,establishment, and growth (Rango & Havstad, 2011).However,retaining water on the upslope side of berms has consequences for lands downslope.This study points to the role of conservation structures in adding additional complexity to already heterogeneous landscapes by creating patchwork assemblages of vegetation and bare soil proximal to earthen runoff and erosion control berms.

Fig.3.A typical example of the relation of the drainage network (a) and vegetation patterns (b) influenced by legacy earthen berms in an upland area within the Altar Valley, Arizona.The blue boxes (a) indicate up and downslope areas that may be affected by earthen berms.In panel b, the berms are indicated in red, and more vegetation can be seen upslope (to the left) of the berms than downslope.

In the Altar Valley in southern Arizona, patchiness is shown to be affected by earthen berms that effectively retain runoff and sediment sufficient to create upslope and downslope differences in vegetative cover that vary among sites and with soil characteristics.In relatively fine textured, clay-enriched soils, moisture holding capacity can be high (McDonald et al., 1996).Precipitation input may provide sufficient moisture on these soils to support grasses and shrubs without dependence on run-on subsidies that may be important to vegetation on coarser textured soils.Infiltration and the depth to which water travels are controlled by soil characteristics such as texture (McAuliffe, 2003), which has been shown to play a substantial role in mediating hydroecologic dynamics in drylands (Hamerlynck & McAuliffe, 2008).In general, shrubs are better able to take advantage of both shallow and deep moisture(McAuliffe,2003).Our finding of mixed responses in shrub cover to berms on the coarser soils may be the result of altered moisture distribution or may simply reflect the amount of time that the vegetation has had to adjust.Determining the relative importance of the causal factors requires new field experiments.However,because we document that when the landscape is hydrologically re-connected through berm breaches, differences in up and downslope grass or shrub cover can diminish, our results suggest that run-on enhancement may be an important source of water for vegetation.

Berms are associated with increased bare soil in downslope runoff shadows.In combination with reduced vegetive cover,increased bare soil in the runoff shadows can be expected to promote degradation through erosion and soil loss (Ludwig et al.,2005; Parsons et al.,1996; Turnbull et al., 2008).The potential to reverse such degradation by re-establishing hydrologic connectivity is evidenced by upslope and downslope similarities in vegetative cover proximal to breached berms, although the temporal threshold for reversal is unknown.However, re-establishing hydrologic connectivity alone may not be sufficient to reduce degradation as there are other variables that could drive the response.Many of the berms on these soils were constructed to control erosion and are located in areas with complex topography that has developed through excessive erosion and gullying.It is possible that a vegetation limiting threshold has already been crossed and cannot be reversed by reestablishing hydrologic connectivity alone.

This study highlights the enduring legacy of historical land management practices and their persistent influence on ecohydrologic patterns.Clearly, a more complete understanding of hydrologic and ecologic process dynamics based on specific field experiments are required to understand the thresholds and feedbacks that will determine the ecological potential of future management actions in landscapes with berms.Additional research is needed to understand the spatial scale of offsite impacts as well as the impact of cumulative effects in areas where multiple berms have been constructed.Such cumulative effects have the potential to affect large watershed areas.For example, Berg et al.(2016)documented a decrease in downstream reservoir sedimentation associated with increasing stock pond density over a multidecadal period in central Texas.A more complete understanding of the history of berm construction in the Altar Valley is needed to understand such cumulative impacts and their impacts on the temporal dynamics of ecohydrologic and sedimentation processes.Although we attempted to interrogate aerial imagery to date berms,that effort was complicated by low resolution imagery prior to approximately the mid-1950s that is not sufficient to show many of the smaller berms.Local knowledge can fill some of these information gaps, however, even among on the ground land managers,critical knowledge of local legacy management practices (Rhodes et al., 2021) is often lost when lands change ownership.As programs such as the U.S.Geological Survey 3D Elevation Program(U.S.Geological Survey, 2021) continue to expand, data with sufficient resolution to represent small berm features increasingly will be available.Such spatially expansive, high-resolution data provide the opportunity for land managers to assess broad landscapes through time (West, 2003) and to build databases of rangeland earthworks that are needed to improve our understanding of human altered watersheds.However, currently, many unanswered questions remain.How long does it take for the area in a runoff shadow to cross an ecological threshold in response to altered surface runoff? Is precipitation in a runoff shadow sufficient to erode soil past a threshold of vegetation support and how long does it take? How can sites with high potential for positive vegetation response following berm removal be identified?

Even in the absence of a more complete understanding of the impact of berms on ecohydrologic processes, knowledge of the landscape patterns they induce can be important for conservation planning and informing management decisions(Bestelmeyer et al.,2011).For example,there may be potential to use temporary berms for enhancing vegetation over a time scale that does not cause the downslope land to cross a threshold.In some cases,the removal of legacy berms may have the potential to reestablish hydrologic connectivity in support of restoration efforts.Currently, Ecological Site Descriptions (ESDs) provide land managers with a framework for classifying rangeland soils and vegetation.They also describe general topography, climate, and hydrology.Recently, recommendations and methodologies for incorporating hydrologic data and ecohydrologic relationships have been proposed to increase the utility of ESDs to assess rangelands, target conservation and restoration practices, and predict ecosystem response to management (Williams et al., 2016).Incorporating information on past conservation practices and land management actions,including the location and impact of earthen berms,can further extend the utility of ESDs and also improve predictions by the simulation models that provide data for ESDs.However, expanding the scope of conservation planning to explicitly include structural land treatments requires adequate basic data.Geospatial databases of berms and other structures such as that developed for the Altar Valley to map more than 1000 runoff and erosion control structures (Nichols &Degginger, 2021; Nichols et al., 2018) are needed.Simple inventories and assessments could be incorporated into ongoing field activities.After the data are developed, the information could be incorporated in management decision-making tools such as ESDs.

5.Conclusions

Earthen berms that control rangeland runoff need to be managed to ensure they are operating as intended and with a good understanding of the role they play in altering and influencing runoff and vegetation patterns.Although berms designed and constructed in cooperation with the technical assistance of entities such as the US Natural Resources Conservation Service come with maintenance expectations,many berms are constructed ad-hoc.In such cases, maintenance can be non-existent.Given the extent of western US water development during the past century, earthen berms can be found throughout managed rangelands.Many are not maintained and even those that are maintained impart landscape scale disturbances.In some cases, removing obsolete or nonfunctional berms may be justifiable to meet management objectives.Ideally, funds for maintenance would be budgeted for as a specific component of projects that include berm construction.

The existence, condition, and influences of structures such as berms that alter topography and runoff pathways should be explicitly considered in rangeland and watershed management,as well as restoration planning.Physical controls on runoff that may occur up-and down-slope of land management boundaries should be an integral part of conservation planning that currently may not extend to watershed boundaries.In many cases, upslope runoff diversions located beyond ranch boundaries may constrain the potential of proposed practices to manipulate vegetation.Finally,management decisions should consider the influence of soil characteristics on vegetation dynamics affected by altered runoff pathways.

Acknowledgements

Lauren Thompson was supported by a University of Arizona/NASA Space Grant Undergraduate Research Internship.We thank the Pima County Regional Flood Control District for access to lidar data, Michelle Cavanaugh for GIS assistance, and ranchers who were willing to allow access to lands and shared historical knowledge.The Altar Valley Conservation Alliance hosted several workshops and meetings that established early thoughts on the study.This material is based upon work supported by the U.S.Department of Agriculture, Natural Resources Conservation Service, Conservation Effects Assessment Project-Grazing Lands component, under agreement number NR213A750023C013.