A recent drying trend that is expected to continue in the southwestern US underscores the need for site-specific and near real-time understanding of vegetation vulnerability so that land management actions can be implemented at the right time and place.
We related the annual integrated normalized difference vegetation index (iNDVI), a proxy for vegetation production, to water balance across landscapes of the Colorado Plateau. We determined how changes in production per unit of water (vegetation responses) and the water balance amounts at which production shifted from above to below average values (pivot points), varied across dominant vegetation and soil types.
Precipitation (PRCP), actual evapotranspiration (AET), water deficit (D), and soil moisture (SM) explained 13%–82% of variation in vegetation production. Along an increasing water availability gradient, vegetation responses to PRCP and AET increased, responses to SM decreased, and responses to D became more negative. We found trade-offs between vegetation responses and pivot points within and across all vegetation types that were mediated by soil properties.
Synthesis and applications. The water needed by native vegetation to maintain production depends on plant traits. The water available to vegetation depends on climate and soil properties that change along environmental gradients. Tracking this biologically relevant water availability in relation to water need provides an indicator of vegetation growth or stress that can help guide the time and place for management actions.

1 INTRODUCTION

Recent elevated temperatures and droughts have been linked to large scale reductions and mortality of vegetation in forests (Williams et al., 2010), grasslands (Moran et al., 2014), and shrublands (Munson, Sankey, Xian, Villarreal, & Homer, 2016) of the southwestern US. Further increases in aridity are projected for the region by mid-century (Seager et al., 2007), which will impact vegetation and have cascading effects on ecological and socioeconomic functions (Miller, Belote, Bowker, & Garman, 2011). Monitoring vegetation production in relation to water availability can reveal ecosystem vulnerability and provide indication of future landscape transformations under climate change (Higginbottom & Symeonakis, 2014). Forecasts of vulnerability can promote anticipatory and proactive forms of land management that move beyond “wait and see” or post hoc approaches (Bradford, Betancourt, Munson, & Wood, 2018).

In water-limited regions, vegetation persistence depends on a positive balance between growth and reproduction during wet conditions and resistance during dry conditions. Drought-resistant plant traits include small xylem diameter, low leaf surface area and stomatal density, deep rooting depth, and other attributes that conserve water but come with a cost that limits growth (Noy-Meir, 1973). Balancing carbon capture with water conservation has been framed in terms of “cost–income,” “safety–efficiency,” and “response–resistance” trade-offs for grass, herbaceous dicot, and woody plant species (Angert, Huxman, Chesson, & Venable, 2009; Meinzer, McCulloh, Lachenbruch, Woodruff, & Johnson, 2010; Munson, 2013; Ocheltree, Nippert, & Prasad, 2016; Orians, Gordon & Solbrig, 1977), but this principle has not been widely tested at landscape scales. Scaling up from plant species and associated traits to understand where different vegetation types fall along the response–resistance spectrum also has important land management implications for predicting landscape vulnerability under climate change.

Precipitation is the primary control on water availability, but heterogeneity in topography, soils, vegetation traits, and land use in drylands interact to affect the spatial distribution, timing, and use of water (Hamerlynck & McAuliffe, 2008; Munson et al., 2015; Sala, Gherardi, Reichmann, Jobbagy, & Peters, 2012). Accounting for precipitation that is biologically relevant, including water storage (soil moisture [SM]), water use (actual evapotranspiration [AET]), and water need (deficit [D]), can improve predictions of vegetation production because these variables are more closely linked to vegetation traits than precipitation (Stephenson, 1998). Water availability and production are affected locally by soil water holding capacity (Sala, Parton, Joyce, & Lauenroth, 1988; Sala et al., 2012) and factors that influence capacity, such as texture and profile depth (Bisigato, Hardtke, & Francisco del Valle, 2013).

We determined the relationships of vegetation production to water availability, water use, and need using a satellite-based vegetation index. We then used these relationships to define vegetation responses and drought resistances that describe the growth and water loss trade-offs of vegetation types that enable persistence. Our specific objectives were to (a) determine relationships between water balance and vegetation production of four important vegetation types on the Colorado Plateau; (b) define response to wet conditions and resistance to dry conditions; and (c) identify soil physical properties that mediate vegetation responses and drought resistances.

2 MATERIALS AND METHODS

2.1 Study area

Our study focused on four widespread vegetation types in seven protected national park units on the Colorado Plateau, hereafter identified by dominant species and functional type: grasslands, blackbrush (Coleogyne ramosissima) shrublands and big sagebrush (Artemisia tridentata), shrublands, and pinyon–juniper (Pinus edulis–Juniperus osteosperma) woodlands (Table 1; Figure 1). We selected polygons greater than 6.25 ha (the area of a single MODIS pixel) that were delineated along ecological site boundaries or in areas where ecological sites have not been mapped along soil or vegetation map unit boundaries (Witwicki, Munson, & Thoma, 2016). Ecological sites are unique combinations of vegetation and soil properties that provide an organizing framework for understanding management-relevant patterns in production at landscape scales (Herrick, Bestelmeyer, Archer, Tugel, & Brown, 2006; Munson, Duniway, & Johanson, 2016).

Table 1. Mean and (minimum and maximum) of soil physical properties and climate in target polygons

Vegetation type	Area (ha)	Polygon counta	Annual precipitation (mm)b	Annual temperature (°C)b	Sand (%)	Clay (%)	Water holding capacity (mm)	Soil profile depth (cm)
Blackbrush shrublands	11,790	168	214 (83, 543)	12 (8, 16)	77 (37, 97)	9 (1, 25)	28 (2, 106)	66 (5, 200)
Grasslands	11,672	282	218 (87, 421)	13 (9, 16)	71 (22, 95)	12 (1, 44)	60 (2, 150)	112 (10, 200)
Pinyon–juniper woodlands	3,661	42	353 (128, 1, 028)	10 (4, 16)	58 (34, 98)	17 (2, 32)	58 (19, 144)	79 (16, 200)
Sagebrush shrublands	681	27	307 (108, 622)	6 (3, 10)	42 (30, 82)	25 (8, 38)	115 (46, 171)	155 (36, 200)

Note.

^a Polygons that had significant relationship between production and one or more water balance variables.
^b POR: period of record evaluated in this study calculated as water year means October 1 to September 30.

Details are in the caption following the image — **Figure 1**
Open in figure viewer PowerPoint

National parks, monuments, and recreation areas in the northern Colorado Plateau and target polygon centroids used to relate vegetation greenness to water balance. This study is designed to augment upland vegetation monitoring in parks that are part of the Northern Colorado Plateau Inventory and Monitoring Network (https://science.nature.nps.gov/im/units/ncpn)

2.2 Climate and soils

Across all target polygons, mean annual precipitation (2000–2014) was 214–380 mm and mean annual temperature was 7–13°C (Table 1; Daymet: http://daymet.ornl.gov, Accessed 29 April 2015). Soils were generally residuum and eolian deposits, colluvium on steeper slopes, and deep sandy and gravelly alluvium in valley bottoms derived from sedimentary parent materials. Water holding capacity, percent sand and clay in the top 1 m, and profile depth for each polygon were determined as a spatially weighted average of soil properties that intersected each vegetation target polygon (Web Soil Survey available at http://websoilsurvey.nrcs.usda.gov, Accessed 24 May 2015).

2.3 Satellite imagery and site selection

We began with 987 target polygons where soil property data were available from published soil surveys (Supporting Information Table S1). From these, we selected 564 polygons that had not burned and had no or minimal presence of annual invasive exotic species. Because invasive exotic species have an earlier phenology in our study region than native species, we were able to determine invaded parts of the landscape by isolating polygons with consistent early season green-up (Kokaly, 2011). We approximated annual gross primary production with the normalized difference vegetation index (NDVI) derived from a time series of 16-day maximum value composite, 250-m resolution MODIS images (MOD13Q1; LPDAAC 2011) from 2000 to 2014. We removed low-quality pixels affected by clouds, aerosols, and snow, then gap filled and smoothed the time series to a monthly period (R Core Team, 2016; signal package). We minimized the soil reflectance contribution to NDVI by subtracting the 15-year minimum NDVI for each polygon from each monthly NDVI value before calculating water year iNDVI (Fensholt & Rasmussen, 2011):

$urn:x-wiley:00218901:media:jpe13250:jpe13250-math-0001$ (1)

where NDVI_m is monthly NDVI from October of the previous year to September the current year and NDVI_min is the minimum NDVI value from 2000 to 2014. Annual NDVI anomaly (∆iNDVI_yr) was calculated as

$urn:x-wiley:00218901:media:jpe13250:jpe13250-math-0002$ (2)

where iNDVI_yr is the integrated annual NDVI for the year of interest and iNDVI_avg the 15-year average integrated NDVI.

2.4 Water balance

We used mean monthly temperature and monthly precipitation sums as inputs to a Thornthwaite-type monthly water balance model (Lutz, van Wagtendonk, & Franklin, 2010). The model partitions precipitation (PRCP) into rain or snow; the latter of which accumulates until temperatures (T) are warm enough to melt. SM is the quantity of water stored in the top meter of soil at the end of each month (mm). Soil is treated as a single layer that has a maximum storage capacity defined by the water holding capacity. Potential evapotranspiration (PET) is the amount of water that could be evapotranspired with available energy if water availability was unlimited to a short grass. Actual evapotranspiration (AET) is the monthly loss of water from soil via transpiration and evaporation which is limited by SM. Water deficit (D) is the amount of additional water vegetation would use if it was available, calculated as the difference between monthly PET and AET (Stephenson, 1998). These variables were summed or averaged in the case of SM, for the biologically relevant water year October 1 to September 30.

2.5 Analysis

We fit linear regressions between water balance variables (PRCP, SM, AET, and D) and ∆iNDVI after we determined nonlinear models only improved fit in 6% of target polygons. We derived the vegetation response, which is the slope of the regression—or the change in iNDVI per unit change in water balance (Figure 2). We also derived the water balance pivot point, which is the point where the regression slope intersects the x-axis (x-intercept) and there is no net change in iNDVI. The water balance pivot point is the water balance value at which there is a transition from below to above average greenness. A water balance pivot point is an indicator of drought resistance because vegetation with a low PRCP pivot point or high D pivot point is able to maintain above average greenness with low water input and high water deficit respectively. When deficit exceeds a vegetation pivot point there is a reduced capacity for growth, which is generally reversible in subsequent years but may lead to a decline in abundance or mortality if water balance amounts beyond the pivot point are sustained across years.

2.6 Trade-offs and influence of soil properties

We determined the relationship between vegetation response and pivot point for each water balance variable using nonlinear quantile regression for polygons that showed a significant relationship (p < 0.1) between a water balance variable and production (Fensholt & Rasmussen, 2011). We relaxed the significance to capture a wider range of environmental conditions where the vegetation types occur (a larger land area) in northern Colorado Plateau parks before performing quantile regression. We used quantile regression to better understand the range and limits of vegetation responses and account for unmeasured factors that influence responses, relative to a pivot point (R Core Team, 2016; qreg package). We compared vegetation response–pivot point relationships at the 5th, 50th, and 95th quantiles using the Wald test. Unequal cross-quantile differences in slope suggest the relationship between vegetation responses and pivot points may be affected by plant interactions with non-water balance factors. When appropriate, we performed natural logarithm transformation of pivot points and vegetation responses prior to quantile regression to improve data normality. Finally, we explored whether soil physical properties influenced vegetation responses and drought resistance by regressing vegetation responses and pivot points against soil properties for precipitation, the only water balance variable not influenced by soil properties. All statistical analyses were conducted in r (R Core Team, 2016).

3 RESULTS

3.1 Water balance and production

We found significant relationships (p < 0.10) between annual production and water balance variables in 519 of 564 polygons that represent 97%, 97%, 81%, and 56% of blackbrush shrublands, grasslands, pinyon–juniper woodlands, and sagebrush shrublands respectively (Supporting Information; Thoma, Munson, & Witwicki, 2018). At the polygon level, water balance variables explained 13%–82% of variation in ∆iNDVI. When polygons were averaged to the vegetation type level, water balance variables explained 29%–39% of variation in ∆iNDVI (Table 2). At the annual time step, we found no indication that any one water balance variable was a superior indicator of vegetation change across all vegetation types.

Table 2. Average vegetation responses and pivot points by vegetation type and water balance variable in national parks on the Colorado Plateau for polygons that had a significant (p < 0.10) relationship between ∆iNDVI and water balance. Summary statistics averaged across polygons as shown in Figure 2 for a single polygon. SE is the sample SE of the mean pivot point and vegetation response

Vegetation type	Water balance variable	Vegetation response Mean ± SE (∆iNDVI × 1,000 mm⁻¹)	Pivot point mean ± SE (mm)	Minimum adjusted r²	Maximum adjusted r²	Mean adjusted r²
Blackbrush shrublands	D	−0.70 ± 0.03	584 ± 7	0.14	0.8	0.38
	AET	0.88 ± 0.03	204 ± 2	0.13	0.72	0.35
	PRCP	0.81 ± 0.03	215 ± 2	0.15	0.77	0.39
	SM	31.24 ± 2.33	3 ± 0.2	0.13	0.74	0.31
Grasslands	D	−0.73 ± 0.02	540 ± 7	0.13	0.82	0.36
	AET	0.96 ± 0.02	207 ± 2	0.13	0.74	0.36
	PRCP	0.90 ± 0.02	214 ± 2	0.13	0.73	0.37
	SM	15.57 ± 1.39	6 ± 0.3	0.13	0.71	0.32
Pinyon–juniper woodlands	D	−1.96 ± 0.14	320 ± 30	0.14	0.64	0.38
	AET	1.94 ± 0.13	292 ± 7	0.14	0.6	0.33
	PRCP	1.18 ± 0.08	350 ± 12	0.14	0.66	0.36
	SM	13.99 ± 2.31	17 ± 2	0.13	0.51	0.29
Sagebrush shrublands	D	−2.64 ± 0.2	197 ± 19	0.14	0.59	0.31
	AET	2.86 ± 0.23	287 ± 10	0.14	0.55	0.3
	PRCP	1.81 ± 0.21	303 ± 14	0.14	0.56	0.32
	SM	8.74 ± 1.18	38 ± 5	0.17	0.57	0.3

Note.

D: cumulative annual deficit; PRCP: cumulative annual precipitation; AET: cumulative annual actual evapotranspiration; SM: average annual soil moisture.

3.2 Vegetation responses and pivot points

All vegetation type responses were negative in relation to water stress (D) and positive in relation to water availability and use (PRCP, SM, and AET; Table 2). The balance between increase and decrease in greenness determined by the sum of AET and D vegetation responses was positive for blackbrush shrublands, grasslands, and sagebrush shrublands (0.18, 0.23, and 0.22 ∆iNDVI × 1,000 mm⁻¹, respectively) but neutral for pinyon–juniper woodlands (−0.02 ∆iNDVI × 1,000 mm⁻¹). Vegetation responses to AET were greater than responses to PRCP for all vegetation types and the difference between responses to PRCP and AET increased from the most xeric to the least xeric vegetation types (Table 2; Figure 3). Similarly, the responsiveness to D became more negative from most xeric to least xeric vegetation types. Vegetation response to SM was greatest in blackbrush shrublands (31.24 ± 2.33 ∆iNDVI × 1,000 mm⁻¹), followed by grasslands (15.57 ± 1.39 ∆iNDVI × 1,000 mm⁻¹), then pinyon–juniper woodlands (13.99 ± 2.31 ∆iNDVI × 1,000 mm⁻¹), and sagebrush shrublands (8.74 ± 1.18 ∆iNDVI × 1,000 mm⁻¹).

We found that blackbrush shrublands and grasslands (mean annual precipitation [MAP] < 300 mm = most xeric) had similar pivot points for AET (204 ± 2 and 207 ± 2 mm, respectively), PRCP (215 ± 2 and 214 ± 2 mm), and SM (3 ± 0.2 and 6 ± 0.3 mm), but the D pivot point in blackbrush shrublands was 44 mm greater than in grasslands (Table 2). Pinyon–juniper and sagebrush shrublands (MAP >300 mm = least xeric) had similar pivot points for AET (292 ± 7 and 287 ± 10 mm, respectively), but the PRCP and D pivot points in pinyon–juniper woodlands were 47 and 123 mm greater than and SM was 21 mm less than, sagebrush shrublands.

3.3 Trade-off relationships across vegetation types

When vegetation response was related to pivot point across all vegetation types, response increased exponentially with PRCP and AET pivot points, but decreased exponentially with SM pivot points (Figure 3, Supporting Information Table S2). As D pivot points decreased, a logarithmic negative response became increasingly negative. Pinyon–juniper woodlands and sagebrush shrublands were more responsive at higher AET and lower D pivot points than the other vegetation types. There was a considerable overlap in the relationships between vegetation responses and pivot points in blackbrush shrublands and grasslands.

3.4 Trade-off relationships within vegetation types

The slope between vegetation response and pivot point within each vegetation type was consistently positive with D and negative with SM (Figure 4; Supporting Information Table S3). Slopes with AET and PRCP changed from negative to positive in grasslands and blackbrush shrublands at a PRCP pivot point of 197 and 206 mm respectively (Supporting Information Table S4; model 2). Slopes with AET were negative in pinyon–juniper woodlands, but positive in sagebrush shrublands (Figure 4g,h), and slopes with PRCP were negative in pinyon–juniper woodlands and sagebrush shrublands (Figure 4g,h). Cross-quantile slopes differed in blackbrush shrublands and grasslands vegetation types associated with water availability (PRCP, AET, SM) and deficit (D), and in pinyon–juniper woodlands with water deficit (D). Cross-quantile differences in slopes were not significantly different in the other vegetation types.

3.5 Influence of soil properties

We found a weak positive effect of soil depth on blackbrush shrublands vegetation responses to PRCP (r² = 0.04; Figure 5a; Supporting Information Figure S1). Similarly, effects of soil texture, water holding capacity, and depth on grasslands vegetation responses were weak and only explained a very small proportion of variation (r² = 0.01–0.07). We found a negative relationship (r² = 0.26) between soil depth and vegetation response in pinyon–juniper woodlands. In sagebrush shrublands, we found a moderate negative vegetation response with clay content (r² = 0.41), and strong positive relationship with sand content (r² = 0.65).

Precipitation pivot points in blackbrush shrublands and grasslands decreased (drought resistance increased) with higher clay content, but a greater change in drought resistance and more of the variation in pivot points was explained for blackbrush shrublands (r² = 0.30) than for grasslands (r² = 0.03; Figure 5b; Supporting Information Figure S1). In both blackbrush shrublands and grasslands, higher sand content, greater water holding capacity and deeper soil depths decreased drought resistance, but little of the variation in drought resistance was explained by these soil properties (r² = 0.09–0.24). Pinyon–juniper woodlands drought resistance decreased with greater soil depth (r² = 0.32). In contrast to our findings in blackbrush shrublands and grasslands, sagebrush shrublands drought resistance decreased with increasing clay content (r² = 0.52) and increased with increasing sand content (r² = 0.24).

4 DISCUSSION

Our results demonstrate that vegetation responses to water balance and pivot points, previously determined using repeat measurements of plant species cover in relatively small (1–100 m²) permanent plots (Munson, 2013), can be scaled to landscapes of heterogeneous species assemblages and soil properties.

4.1 Water balance and production

We found that production at half to nearly all the polygons within a vegetation type was correlated with one or more water balance variables. Strong relationships were either due to a dominant species signal or a mixture of dominant and subdominant species that responded consistently across years at the annual time step. A lack of correlation at some sites may be explained by the annual time step in our analysis not accounting for antecedent or isolated seasonal conditions that may be influential in drylands (Robinson et al., 2013; Sala et al., 2012).

We expected SM to be an important determinant of vegetation production. However, like Gremer et al. (2015) who used a mechanistic soil water model, we found little difference between water balance variables in their relationship with annual vegetation production (r² range = 0.29–0.39). This is likely because dryland ecosystems have limited run-off and deep infiltration, which means almost all PRCP that enters the soil returns to the atmosphere via AET (Noy-Meir, 1973). In our study, 83%, 92%, 94%, and 97% of PRCP was returned to the atmosphere via AET in pinyon–juniper woodlands, sagebrush shrublands, blackbrush shrublands, and grasslands respectively. We likely overestimated PET because we used the evapotranspiration rate from a shortgrass reference crop. We made a coarse assessment of water holding capacity from published soil surveys, which may have altered the rate of soil drying and plant water availability and use. Furthermore, our “single bucket” water balance model did not account for SM variation with depth that interacts seasonally with different root morphologies. Some of these issues were also noted by Huber, Fensholt, and Rasmussen (2011) as reasons why SM relationships with production were not much stronger than PRCP in the African Sahel.

4.2 Vegetation responses and pivot points

We found xeric vegetation types were more responsive to SM, while less xeric vegetation types were more responsive to D (Figure 4c,d,m,n; Supporting Information Table S3). Using a precipitation manipulation experiment across Great Plains grasslands, Heisler-White, Blair, Kelly, Harmoney, and Knapp (2009) attribute a similar response to precipitation event size and dry period length. At their most xeric site, where vegetation was already adapted to long dry periods, creating fewer, larger events increased production by elevating SM. At their least xeric site, where vegetation was less well adapted to dry intervals, creating longer dry periods between events decreased production. Thus, consistent with Heisler-White et al. (2009), xeric-adapted vegetation in our study responded more strongly to elevated SM because it relieved plant water stress, whereas vegetation adapted to more consistent periods of high SM responded more strongly to changes in D, which increases with dry period length.

4.3 Trade-off relationships across and within vegetation types

The relationships between vegetation response and water balance pivot points in this study are consistent with two primary strategies of plants in dryland regions. At opposite ends of a continuum, one strategy relies on persistence through periods of water scarcity and the other relies on rapid response to water availability (Orians Gordon & Solbrig, 1977; Ocheltree et al., 2016) . Evidence of a net positive balance on the Colorado Plateau was found in the long-term neutral to slightly positive sum of AET and D vegetation responses for each vegetation type. This suggests that vegetation in our study area is adapted to recent climate conditions and provides evidence of conditions which may trigger potential losses in vegetation production in the future if these differences become negative.

We acknowledge that water balance pivot points only suggest that vegetation is drought resistant and further evaluation of plant structural and physiological characteristics are necessary to fully make this determination. While pivot points vary within vegetation types, they fall close to the average water balance conditions and are influenced by the collective greenness signal of dominant and subdominant species of a site. Additional evaluation of sites that receive the same amount of mean annual precipitation but vary in subdominant species can refine the application of pivot points. Our methods do not test limits of resistance (Evans, Byrne, Lauenroth, & Burke, 2011) but provide an indication of production potential and the need for site specific evaluation of local vegetation performance. Coupling remote sensing with field-based biomass harvests and expanding knowledge of how plant reproduction, dispersal, and adaptation capacity are influenced by climate, are also needed to gain perspective on long-term plant persistence.

Dryland vegetation types that achieve a positive cost–income balance do so within a performance range that constitutes their ecohydrological niche, which we described by annual vegetation response–resistance trade-off quantiles. Upper trade-off quantiles of water availability and use (PRCP, ET, and SM) define upper limits of vegetation performance in terms of income or response, which is the accrual of biomass per unit of water used with respect to its pivot point. Our findings illustrate a nonlinear relationship between pivot points and responses which agrees with Fensholt and Rasmussen (2011) who identified plant traits, composition, soil characteristics, and ecological memory effects as potentially confounding a “global” relationship between production and precipitation on an annual basis. However, for a given location, the uppermost quantile of this relationship is linked to vegetation physiological or energetic limits that define growth rates according to resource availability. Performance below the upper limit may be influenced by factors other than water availability that suppress performance such as competition, grazing or nutrient limitation, some of which can be managed to increase performance. Lower quantiles in drought (D) trade-offs defined production losses under a range of drought resistances. In the course of our study, production losses were highest for vegetation that received more PRCP and was relatively drought intolerant. Blackbrush shrublands likely had high drought resistance (high D and low PRCP and AET pivot points) because of the xerophytic leaves, woody biomass, and relatively deep root system of the dominant shrub. It is less intuitive why grasslands also had high drought resistance, but seasonal dormancy and low mean annual precipitation may partially explain this result (Moran et al., 2014; Noy-Meir, 1973; Witwicki et al., 2016). Conversely, pinyon–juniper woodlands and sagebrush shrublands were less drought resistant but more responsive to water availability estimated by PRCP and AET, reflecting plant adaptations to higher water availability and likely herbaceous subdominant species in their understories. Departures above the highest quantiles could result from incomplete screening of polygons that contained sufficient annual invasive exotic species that have a large early season response that inflates vegetation response relative to the native vegetation pivot point. Refining upper and lower boundaries, and understanding what ecological mechanisms underlie them, are worthwhile avenues for future research. Cross-quantile differences indicate non-water balance factors such as differences in subdominant species may affect trade-offs in blackbrush shrublands and grasslands associated with water availability (PRCP, AET, and SM) and deficit (D), and in pinyon–juniper woodlands with water deficit (D). Cross-quantile differences in slopes were not significantly different in the other vegetation types, which suggest water balance was the primary determinant of trade-off relations.

4.4 Influence of soil properties

Three vegetation response–resistance relationships within vegetation types diverged from those found across vegetation types and can be explained by the mediating effect of soil properties. The relationships in pinyon–juniper woodlands described by AET and PRCP, and sagebrush shrublands described by PRCP, were negative and contrasted with positive relationships expected based on response–resistance theory. In sagebrush shrublands, a negative relationship between clay content and vegetation response and a positive relationship between clay content and pivot points may partially explain observed vegetation responses less than expected relative to PRCP pivot points. Similarly, increasing soil depth in pinyon–juniper woodlands was negatively related to vegetation responses and likely suppressed vegetation responses as PRCP and AET pivot points increased.

A logarithmic relationship between vegetation responses and pivot points with respect to D and SM indicated these metrics were better measures of the trade-off niche than annual PRCP or AET across the moisture gradient in our study. While annual PRCP and AET are roughly equivalent measures of water availability in our dryland study area, D only accrues during the growing season when water needs are not met, and thus emphasizes water need during a critical part of the growing season. Small changes in average annual SM can be due to differences in annual precipitation or a few large precipitation events.

A “U” shaped pattern and cross-quantile differences in slopes in the quadratic vegetation response–resistance relationship of PRCP and AET (Figure 5f,i,j) also suggests that factors other than drought resistance were limiting vegetation responses in blackbrush shrublands and grasslands. We found that including soil properties in models of “U” shaped patterns improved fits in the trade-off relationships, but the soil property terms did not control the pattern (Supporting Information Table S4). Rather, a break point along aridity gradients near 200 mm annual PRCP in these two vegetation types where slopes change from negative to positive may be due to subdominant vegetation composition or another unmeasured factor.

We found evidence that soil texture can strongly mediate drought resistance, but the effect differs by vegetation type. For example, across the range of textures, increasing sand content was associated with 47 mm decrease in drought resistance in blackbrush shrublands and clay content was associated with 164 mm increase in drought resistance in sagebrush shrublands (Supporting Information Figure S1).

Texture also affected vegetation production response, most notably in sagebrush shrublands where increases in sand content doubled vegetation response across the range of textures. Across all vegetation types, we found that percent sand or clay content had a larger influence on drought resistance and vegetation response than water holding capacity and depth. Contrasting effects of clay and sand likely influenced vegetation performance by altering water movement, storage, and retention in the soil profile. In drylands, the inverse texture hypothesis suggests greater plant production in sandy soils due to deeper infiltration where moisture moves into the rooting zone and is protected from high rates of evaporation at the soil surface (Noy-Meir, 1973). Our results suggest that the effects sand and clay content have on pivot points in drylands may counter predictions of the inverse texture hypothesis at extreme ends of the soil texture gradient. For example, where sand content was very high, as in the most xeric blackbrush shrublands type (average sand content = 77%), an increase in clay content may have increased water holding capacity and plant drought resistance. Similarly, where a soil had high clay content, as in the less xeric sagebrush shrublands (average clay content = 25%), an increase in sand content likely promoted additional hydrological conductivity and made water more accessible to plant roots (Clark, 1990). While the importance of soil properties on productivity at landscape scales has been demonstrated (Farrar, Nicholson, & Lare, 1994), there have been few studies demonstrating how unique combinations of soil and vegetation properties define ecohydrological niches in dryland landscapes. Collectively, our findings suggest that in some vegetation types, soil properties need to be explicitly considered in climate vulnerability assessments.

4.5 Implications for land management

Land managers intuitively track weather as an indicator of vegetation condition. The accuracy of these assessments can be improved if the relationships between biologically relevant water and vegetation greenness are quantified through the pivot point approach. Tracking the seasonal evolution of water balance variables and adopting short-term forecasts of climate dynamics (e.g., http://www.cpc.ncep.noaa.gov/products/predictions; Accessed June 4, 2017), in relation to pivot points provides an early warning of changes in vegetation production. Shortfalls of water availability in relation to the pivot point indicate reduced vegetation growth, which can trigger drought mitigation and adaptation measures. These measures may include managing nonclimate stressors like human disturbances, livestock, or non-native species, to reduce competition for water in drought sensitive areas and preparing fire suppression. In contrast, excess water availability in relation to the pivot point can create an opportunity to capitalize on favourable conditions. These opportunities include restoration efforts, assisted migration, or land treatments that may require longer periods of recovery, including prescribed burning and brush management.

The water balance pivot point framework provides land managers with knowledge of when and where vegetation is likely to respond to changes in water availability. First, the framework conveys information on which aspects of water balance vegetation types are sensitive. Vegetation types such as blackbrush shrublands in our study may be more sensitive to interannual changes in SM than precipitation, which indicates the importance of event size and soil properties in mediating water availability. Second, defining vegetation responses and pivot points can help land managers discriminate among differences in sensitivity among vegetation types and develop triage approaches that target management actions for vegetation types that are vulnerable. Third, understanding how variability in vegetation responses and pivot points within a vegetation type is attributable to landscape and soil properties can help define refugia where vegetation is resilient to water shortages, such as sagebrush shrublands on sandy soils in our study.

During our study, water balance and vegetation production fluctuated, and we did not observe large-scale vegetation transitions (Supporting Information Figure S2). However, vegetation mortality and a critical transition to a new vegetation type may occur if prolonged water balance conditions exceed pivot points in the future. While this concept needs further testing, water balance pivot points can potentially serve as an early warning sign of increased risk of crossing an ecosystem threshold. Because vegetation is the foundation for wildlife habitat and carbon cycling, pivot points can guide conservation efforts to increase at-risk species and carbon storage capacity (Scott, Biederman, Hamerlynck, & Barron-Gafford, 2015). While our study focused on national parks, pivot points can be used to compare how vegetation responds to water balance in areas that experience different land-use activities. Employing a pivot point approach across the landscape allows land managers to pinpoint the water availability conditions at sites of interest that are required to maintain productivity for habitat and grazing.

ACKNOWLEDGEMENTS

Funding was provided by the National Park Service, DOI National Climate Adaptation Science Center, and U.S. Geological Survey Ecosystem Mission Area. We thank Erin Bunting for providing maps of early season invasive species used in our analysis and Alice Wondrak-Biel for figure editing. We also thank three anonymous reviewers for comments that helped improved this manuscript. The authors claim no conflict of interest. Any use of trade, product, or firm names in this paper is for descriptive purposes only and does not imply endorsement by the US Government.

AUTHORS’ CONTRIBUTIONS

S.M.M. and D.P.T. conceived the study; D.P.T. conducted analyses; D.P.T., S.M.M., and D.L.W. wrote the manuscript. All authors gave approval for publication.

DATA ACCESSIBILITY

Data available via the Dryad Digital Repository https://doi.org/10.5061/dryad.8h5h762 (Thoma et al., 2018).

Supporting Information

REFERENCES

Angert, A. L., Huxman, T. E., Chesson, P., & Venable, D. L. (2009). Functional tradeoffs determine species coexistence via the storage effect. Proceedings of the National Academy of Sciences of the United States of America, 106, 11641–11645. https://doi.org/10.1073/pnas.0904512106
10.1073/pnas.0904512106
CASPubMedWeb of Science®Google Scholar
Bisigato, A. J., Hardtke, L., & Francisco del Valle, H. (2013). Soil as a capacitor: Considering soil water content improves temporal models of productivity. Journal of Arid Environments, 98, 88–92. https://doi.org/10.1016/j.jaridenv.2013.08.004
10.1016/j.jaridenv.2013.08.004
Web of Science®Google Scholar
Bradford, J. B., Betancourt, J. L., Munson, S. M., & Wood, T. E. (2018). Anticipatory natural resource science and management for a dynamic future. Frontiers in Ecology and the Environment, 16(5), 295–303
10.1002/fee.1806
Web of Science®Google Scholar
Clark, G. A. (1990). Measurement of soil water potential. HortScience, 25(12), 1548–1551.
10.21273/HORTSCI.25.12.1548
Web of Science®Google Scholar
Evans, S. E., Byrne, K. M., Lauenroth, W. K., & Burke, I. C. (2011). Defining the limit to resistance in a drought-tolerant grassland: Long-term severe drought significantly reduces the dominant species and increases ruderals. Journal of Ecology, 99(6), 1500–1507. https://doi.org/10.1111/j.1365-2745.2011.01864.x
10.1111/j.1365-2745.2011.01864.x
Web of Science®Google Scholar
Farrar, T. J., Nicholson, S. E., & Lare, A. R. (1994). The influence of soil type on the relationship between NDVI, rainfall, and soil moisture in semiarid Botswana. II. NDVI response to soil moisture. Remote Sensing of the Environment, 50, 121–133. https://doi.org/10.1016/0034-4257(94)90039-6
10.1016/0034-4257(94)90039-6
Web of Science®Google Scholar
Fensholt, R., & Rasmussen, K. (2011). Analysis of trends in the Sahelian “rain-use efficiency” using GIMMS NDVI, RFE and GPCP rainfall data. Remote Sensing of Environment, 115(2), 438–451. https://doi.org/10.1016/j.rse.2010.09.014
10.1016/j.rse.2010.09.014
Web of Science®Google Scholar
Gremer, J. R., Bradford, J. B., Munson, S. M., Duniway, M. C., Survey, U. S. G., Biological, S., … Duniway, M. C. (2015). Desert grasslands responses to climate and soil moisture suggest divergent vulnerabilities across the southwestern United States. Global Change Biology, 21(11), 4049–4062. https://doi.org/10.1111/gcb.13043
10.1111/gcb.13043
PubMedWeb of Science®Google Scholar
Hamerlynck, E. P., & McAuliffe, J. R. (2008). Soil-dependent canopy die-back and plant mortality in two Mojave Desert shrubs. Journal of Arid Environments, 72, 1793–1802. https://doi.org/10.1016/j.jaridenv.2008.05.002
10.1016/j.jaridenv.2008.05.002
Web of Science®Google Scholar
Heisler-White, J. L., Blair, J. M., Kelly, E. F., Harmoney, K., & Knapp, A. K. (2009). Contingent productivity responses to more extreme rainfall regimes across a grasslands biome. Global Change Biology, 15(12), 2894–2904. https://doi.org/10.1111/j.1365-2486.2009.01961.x
10.1111/j.1365-2486.2009.01961.x
Web of Science®Google Scholar
Herrick, J. E., Bestelmeyer, B. T., Archer, S., Tugel, A. J., & Brown, J. R. (2006). An integrated framework for science-based arid land management. Journal of Arid Environments, 65, 319–335. https://doi.org/10.1016/j.jaridenv.2005.09.003
10.1016/j.jaridenv.2005.09.003
Web of Science®Google Scholar
Higginbottom, T. P., & Symeonakis, E. (2014). Assessing land degradation and desertification using vegetation index data: Current frameworks and future directions. Remote Sensing, 6(10), 9552–9575. https://doi.org/10.3390/rs6109552
10.3390/rs6109552
Web of Science®Google Scholar
Huber, S., Fensholt, R., & Rasmussen, K. (2011). Water availability as the driver of vegetation dynamics in the African Sahel from 1982 to 2007. Global and Planetary Change, 76(3–4), 186–195. https://doi.org/10.1016/j.gloplacha.2011.01.006
10.1016/j.gloplacha.2011.01.006
Web of Science®Google Scholar
Kokaly, R. F. (2011). DESI—Detection of early-season invasives (software-installation manual and user's guide version 1.0): U.S. Geological Survey Open-File Report 2010-1302, 39 pp.
Google Scholar
Lutz, J. A., van Wagtendonk, J. W., & Franklin, J. F. (2010). Climatic water deficit, tree species ranges, and climate change in Yosemite National Park. Journal of Biogeography, 37(5), 936–950. https://doi.org/10.1111/j.1365-2699.2009.02268.x
10.1111/j.1365-2699.2009.02268.x
Web of Science®Google Scholar
Meinzer, F. C., McCulloh, K. A., Lachenbruch, B., Woodruff, D. R., & Johnson, D. M. (2010). The blind men and the elephant: The impact of context and scale in evaluating conflicts between plant hydraulic safety and efficiency. Oecologia, 164, 287–296. https://doi.org/10.1007/s00442-010-1734-x
10.1007/s00442-010-1734-x
PubMedWeb of Science®Google Scholar
Miller, M. E., Belote, R. T., Bowker, M. A., & Garman, S. L. (2011). Alternative states of a semiarid grasslands ecosystem: Implications for ecosystem services. Ecosphere, 2, 1–18.
10.1890/ES11-00027.1
CASWeb of Science®Google Scholar
Moran, M. S., Ponce-Campos, G. E., Huete, A., McClaran, M. P., Zhang, Y., Hamerlynck, E. P., … Hernandez, M. (2014). Functional response of U.S. grasslands to the early 21st-century drought. Ecology, 95(8), 2121–2133. https://doi.org/10.1890/13-1687.1
10.1890/13-1687.1
PubMedWeb of Science®Google Scholar
Munson, S. M. (2013). Plant responses, climate pivot points, and trade-offs in water-limited ecosystems. Ecosphere, 4(9), 1–15.
10.1890/ES13-00132.1
Web of Science®Google Scholar
Munson, S. M., Duniway, M. C., & Johanson, J. K. (2016). Rangeland monitoring reveals long-term plant responses to precipitation and grazing at the landscape scale. Rangeland Ecology and Management, 69, 76–83. https://doi.org/10.1016/j.rama.2015.09.004
10.1016/j.rama.2015.09.004
Web of Science®Google Scholar
Munson, S. M., Sankey, T. T., Xian, G., Villarreal, M. L., & Homer, C. G. (2016). Decadal shifts in grass and woody plant cover are driven by prolonged drying and modified by topo-edaphic properties. Ecological Applications, 26(8), 2478–2492.
10.1002/eap.1389
PubMedWeb of Science®Google Scholar
Munson, S. M., Webb, R. H., Housman, D. C., Veblen, K. E., Nussear, K. E., Beever, E. A., … Tallent, N. G. (2015). Long-term plant responses to climate are moderated by biophysical attributes in a North American desert. Journal of Ecology, 103(3), 657–668. https://doi.org/10.1111/1365-2745.12381
10.1111/1365-2745.12381
Web of Science®Google Scholar
Noy-Meir, I. (1973). Desert ecosystems: Environment and producers. Annual Review of Ecology and Systematics, 4, 25–51. https://doi.org/10.1146/annurev.es.04.110173.000325
10.1146/annurev.es.04.110173.000325
Google Scholar
Ocheltree, T. W., Nippert, J. B., & Prasad, P. V. V. (2016). A safety vs efficiency trade-off identified in the hydraulic pathway of grass leaves is decoupled from photosynthesis, stomatal conductance and precipitation. New Phytologist, 210(1), 97–107. https://doi.org/10.1111/nph.13781
10.1111/nph.13781
CASPubMedWeb of Science®Google Scholar
Orians Gordon, H., & Solbrig, O. T. (1977). A cost-income model of leaves and roots with special reference to arid and semiarid areas. The American Naturalist, 111(980), 677–690. https://doi.org/10.1086/283199
10.1086/283199
Web of Science®Google Scholar
R Core Team. (2016). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org/
Google Scholar
Robinson, T. M. P., La Pierre, K. J., Vadeboncoeur, M. A., Byrne, K. M., Thomey, M. L., & Colby, S. E. (2013). Seasonal, not annual precipitation drives community productivity across ecosystems. Oikos, 122(5), 727–738. https://doi.org/10.1111/j.1600-0706.2012.20655.x
10.1111/j.1600-0706.2012.20655.x
Web of Science®Google Scholar
Sala, O. E., Gherardi, L. A., Reichmann, L., Jobbagy, E., & Peters, D. (2012). Legacies of precipitation fluctuations on primary production: Theory and data synthesis. Philosophical Transactions of the Royal Society B: Biological Sciences, 367(1606), 3135–3144. https://doi.org/10.1098/rstb.2011.0347
10.1098/rstb.2011.0347
PubMedWeb of Science®Google Scholar
Sala, O. E., Parton, W. J., Joyce, L. A., & Lauenroth, W. K. (1988). Primary production of the central grasslands region of the United States. Ecology, 69, 40–45. https://doi.org/10.2307/1943158
10.2307/1943158
Web of Science®Google Scholar
Scott, R. L., Biederman, J. A., Hamerlynck, E. P., & Barron-Gafford, G. A. (2015). The carbon balance pivot point of southwestern U.S. semiarid ecosystems: Insights from the 21st century drought. Journal of Geophysical Research G: Biogeosciences, 120(12), 2612–2624.
10.1002/2015JG003181
CASWeb of Science®Google Scholar
Seager, R., Ting, M., Held, I., Kushnir, Y., Lu, J., Vecchi, G., … Naik, N. (2007). Model projections of an imminent transition to a more arid climate in Southwestern North America. Science, 316(5828), 1181–1184. https://doi.org/10.1126/science.1139601
10.1126/science.1139601
CASPubMedWeb of Science®Google Scholar
Stephenson, N. (1998). Actual evapotranspiration and deficit: Biologically meaningful correlates of vegetation distribution across spatial scales. Journal of Biogeography, 25(5), 855–870. https://doi.org/10.1046/j.1365-2699.1998.00233.x
10.1046/j.1365-2699.1998.00233.x
Web of Science®Google Scholar
Thoma, D. P., Munson, S. M., & Witwicki, D. L. (2018). Data from: Landscape pivot points and responses to water balance in national parks of the southwest U.S. Dryad Digital Repository. https://doi.org/10.5061/dryad.8h5h762
Google Scholar
Williams, A. P., Allen, C. D., Millar, C. I., Swetnam, T. W., Michaelsen, J., Still, C. J., & Leavitt, S. W. (2010). Forest responses to increasing aridity and warmth in the southwestern United States. Proceedings of the National Academy of Sciences of the United States of America, 107(50), 21289–21294. https://doi.org/10.1073/pnas.0914211107
10.1073/pnas.0914211107
CASPubMedWeb of Science®Google Scholar
Witwicki, D. L., Munson, S. M., & Thoma, D. P. (2016). Effects of climate and water balance across grasslands of varying C 3 and C 4 grass cover. Ecosphere, 7, 1–19.
10.1002/ecs2.1577
Web of Science®Google Scholar

Citing Literature

Volume56, Issue1

January 2019

Pages 157-167

Journal list menu

Tools

Follow journal

Landscape pivot points and responses to water balance in national parks of the southwest US

Abstract

1 INTRODUCTION