Desert shrub responses to experimental modification of precipitation seasonality and soil depth: relationship to the two-layer hypothesis and ecohydrological niche



  1. Ecohydrological niches are important for understanding plant community responses to climate shifts, particularly in dry lands. According to the two-layer hypothesis, selective use of deep-soil water increases growth or persistence of woody species during warm and dry summer periods and thereby contributes to their coexistence with shallow-rooted herbs in dry ecosystems. The resource-pool hypothesis further suggests that shallow-soil water benefits growth of all plants while deep-soil water primarily enhances physiological maintenance and survival of woody species. Few studies have directly tested these by manipulating deep-soil water availability and observing the long-term outcomes.
  2. We predicted that factors promoting infiltration and storage of water in deep soils, specifically greater winter precipitation and soil depth, would enhance Artemisia tridentata (big sagebrush) in cold, winter-wet/summer-dry desert. Sagebrush responses to 20 years of winter irrigation were compared to summer- or no irrigation, on plots having relatively deep or shallow soils (2 m vs. 1 m depths).
  3. Winter irrigation increased sagebrush cover, and crown and canopy volumes, but not density (individuals/plot) compared to summer or no irrigation, on deep-soil plots. On shallow-soil plots, winter irrigation surprisingly decreased shrub cover and size, and summer irrigation had no effect. Furthermore, multiple regression suggested that the variations in growth were related (i) firstly to water in shallow soils (0–0.2 m) and secondly to deeper soils (> 1 m deep) and (ii) more by springtime than by midsummer soil water. Water-use efficiency increased considerably on shallow soils without irrigation and was lowest with winter irrigation.
  4. Synthesis. Sagebrush was more responsive to the seasonal timing of precipitation than to total annual precipitation. Factors that enhanced deep-water storage (deeper soils plus more winter precipitation) led to increases in Artemisia tridentata that were consistent with the two-layer hypothesis, and the contribution of shallow water to growth on these plots was consistent with the resource-pool hypothesis. However, shallow-soil water also had negative effects on sagebrush, suggesting an ecohydrological trade-off not considered in these or related theories. The interaction between precipitation timing and soil depth indicates that increased winter precipitation could lead to a mosaic of increases and decreases in A. tridentata across landscapes having variable soil depth.


Hydrological niches are proposed to be ‘potent’ aspects of species coexistence and thus diversity of plant communities (Silvertown et al. 1999). Species' hydrologic requirements and trade-offs should be core predictors of their landscape response to climate shifts, especially as they relate to the temporal patterning of precipitation in water-limited systems (Weltzin et al. 2003; Debinski et al. 2010; Schlaepfer, Lauenroth & Bradford 2011; Kulmatiski & Beard 2013). Many upland plant communities experience seasonal shifts from surplus to deficit in their water balance (Walter 1973), which causes factors such as soil water storage, saturation or deficit to select for species' ecohydrological attributes and particularly trade-offs (e.g. between tolerance of saturated vs. dry soils, Araya et al. 2011). The seasonal transition from hydrating to desiccating conditions is furthermore overlaid on vertical gradients in soil water content, and this spatiotemporal patterning of water availability is probably a critical aspect of many species' ecohydrological niche (Pañuelas, Terradas & Lloret 2011).

Evidence for the importance of spatiotemporal patterns of soil water to species' growth requirements typically comes from observations that species separate in their water-use patterns along hydrological gradients (i.e. inferred from their realized and not fundamental niches; Schlaepfer, Lauenroth & Bradford 2011) or from patterns of soil water use indicated by soil water balance or isotopic tracers (e.g. Kulmatiski, Beard & Stark 2006; Nippert & Knapp 2007). Aside from these observational findings, direct experimental evidence for the importance of ecohydrological niches is lacking. This knowledge gap could be addressed by altering the key spatiotemporal hydrologic attributes thought to affect a species and observing the population outcomes (i.e. population growth or indicators of it), preferably over long time frames. Long-term experimental manipulations of hydrological patterns are also valuable for testing predicted outcomes of climate changes, such as in the amount and timing of precipitation (Easterling et al. 2000; Mote & Salathé 2010).

In arid and semi-arid grassland and savanna ecosystems, the relative abundance of shrubs and herbs in a community relates to the ratio of winter to summer precipitation (Archer 1989; Brown, Valone & Curtin 1997; Gao & Reynolds 2003; Browning et al. 2008). An explanation for this is Walter's two-layer hypothesis, which states that coexistence of deeper-rooted woody species and shallower-rooted herbs is made possible by vertical separation of their water uptake, particularly in drier ecosystems (Walter 1973; reviewed in Ward, Wiegand & Getzin 2013). Although originally proposed for summer precipitation, the two-layer model appears more likely to apply to winter-wet/summer-dry systems, where precipitation falling in the dormant season can infiltrate to deeper-soil layers (below ~0.3–0.5 m) where it is stored and used preferentially by deep-rooted woody plants during warm and otherwise dry summer conditions (Ehleringer et al. 1991; Paruelo & Lauenroth 1996; Schenk & Jackson 2002; Ogle & Reynolds 2004). Uptake of deep-soil water by individual shrubs in semi-arid settings has been confirmed in numerous ecophysiological studies and has been related to increased instantaneous plant water status or photosynthesis, such as for the dominant shrubs of western North American deserts, Larrea tridentata, Artemisia tridentata NUTT (big sagebrush) and others (e.g. Sala et al. 1989; Donovan & Ehleringer 1994; Leffler et al. 2004). The presence of roots in shallow- or deeper-soil horizons has also been a favoured indicator of the applicability of the two-layer hypothesis to ecosystems codominated by woody and herbaceous species (Schenk & Jackson 2002; Ward, Wiegand & Getzin 2013). However, the detection of water originating from deep soils in xylem, or greater water status or photosynthesis or stem diameter, or even the presence of roots at a given depth can suggest – but not prove – that water in deep or shallow soils is important to the abundance of a species in its community.

At the population or greater level, use of deep-soil water by species such as A. tridentata is, on the one hand, confirmed by water balance (e.g. in > 12 studies reviewed in Wilcox et al. 2012) and appears consistent with the outcomes of ecohydrologically informed species-distribution modelling (Schlaepfer, Lauenroth & Bradford 2011). Experimental removal of A. tridentata ssp. tridentata (basin big sagebrush) combined with winter-rainout shelters revealed that its deep-soil water use has important effects on the surrounding plant community and confers community resistance to invasion by exotic, tap-rooted forbs (Prevéy, Germino & Huntly 2010). On the other hand, a number of studies did not detect population response to alterations in seasonal precipitation. Bates et al. (2006) reported no changes in cover or density in A. t. spp. wyomingensis (Wyoming big sagebrush) in response to 7 years of altered seasonal timing of precipitation under rainout shelters built on intact communities. Plot cover of A. t. ssp. vaseyana (mountain big sagebrush) also did not change in response to large and long-term increases or decreases in snowpack (Loik, Griffith & Alpert 2013). Species can respond to changes in precipitation seasonality and soil water storage at individual or population levels, and measurement of responses at both levels could bridge the inconsistent information on response of desert shrub species to altered precipitation amount and timing. For example, imposed winter droughts were associated with decreased growth of individuals of the deep-rooted warm-desert shrub Ceratoides lanata, despite no differences in photosynthesis (Schwinning, Starr & Ehleringer 2005). Moreover, opportunistic water-use strategies and use of multiple soil water reservoirs in species such as sagebrush (e.g. Donovan & Ehleringer 1994; Leffler et al. 2004) should affect population responses to altered precipitation. The inconsistent support for the two-layer hypothesis among observational or ecophysiological studies (reviewed in Ogle & Reynolds 2004) led to refinements such as Ryel et al.'s (2008) resource-pool hypothesis, which proposed that deep-soil water selectively benefits physiological maintenance and survival of deep-rooted species, while water from shallower, nutrient-rich soils water promotes growth.

Our objective was to determine whether responses of sagebrush to variations in key hydroclimate and soil factors that are postulated to constitute its ecohydrological niche would be consistent with the two-layer hypothesis. Leaf-to-canopy responses to 20 years of supplemental winter or summer irrigation on both shallow-soil (1-m soil) and deep-soil (2 m depth) plots were evaluated for sagebrush in a plant community context. Potential soil water storage increases across these precipitation × soil treatments, from shallow soils receiving no supplemental irrigation, to deep soils receiving summertime irrigation, to deep soils receiving irrigation in winter. From the two-layer and resource-pool hypotheses, we predicted that plant size and stand-level abundance would increase with soil water storage potential across the treatments (i.e. fixed effects of deeper soil and irrigation, especially when added in winter), but that regression models would nonetheless show shallow-soil water content across all treatments to be an important determinant of growth. Additionally, we predicted that physiological adjustments in water-use efficiency (WUE) would also occur, possibly providing a compensating, alternative response to changes in overall population abundance (e.g. shifts in WUE could enable cover to change less in response to soil water). Climate responses of big sagebrush are a major concern due to its decline as wildfire and invasive plant stressors increase, and because of its importance to many animal species (e.g. to Greater Sage Grouse, pygmy rabbits, mule deer, pronghorn antelope; Welch 2005). Notably, increases in winter precipitation, along with increased minimum temperatures, are considered to be substantive climate changes that will affect big sagebrush ecosystems (Abatzoglou & Kolden 2011).

Materials and methods

The study was conducted at an ongoing ecohydrological field study located in the Idaho National Laboratory, which is located in sagebrush steppe of the Snake River Plain (northern Great Basin; vegetation described in Anderson & Inouye 2001). Local climate is typical of sagebrush steppe communities: for daily air temperatures, the annual mean is 5.5 °C, the mean of the warmest month is 20 °C, the mean of the coldest month is −8.8 °C, and extreme maximum and minimums are 38.3 and −43.9°C, respectively (monthly statistics for 38 years, Clawson, Start & Ricks 1989). Annual precipitation is 220 mm, with > 60% falling during the winter and early spring months. The experimental plots were created in 1993 and consist of three replicated blocks of two artificially constructed soil profiles and three irrigation treatments used in the current analysis, along with additional levels of the soil-profile treatment and two plant community types not evaluated here. Plots were 8 × 8 m in area.

The two soil-profile designs used in this particular analysis were plots with 1-m (relatively shallow) or 2-m soil depths (relatively deep; hereafter referred to as ‘shallow’ or ‘deep’ plots, respectively). Deep plots are representative of soil horizons throughout much of the Great Basin and particularly the loess-derived soils of the upper Snake River Plain (deep, well-drained, with little horizonation beyond the top few cm of soil). Shallow plots represent soil horizons with less loess deposition (e.g. topographic ridges) or where top soil is underlain by impermeable petro-calcic layers (i.e. caliche), and both are conditions common in the Great Basin. Shallow plots were underlain by a flexible membrane layer and a 0.6-m-deep compacted clay layer (hydraulic conductivity < 1 × 10−7 cm s−1) beginning at 1 m depth, both sloped at 3% grade to transport excess water off of these plots. Both soil-cap configurations are underlain by a layer of gravel. All experimental plots were constructed from the same fill soil, a frigid xeric haplocalcid (mean texture 19% sand, 48% silt and 33% clay) obtained from the top 1–2 m of soil from a nearby site at the Idaho National Laboratory. Each plot was filled in 0.2-m increments, with each increment compacted to 1.29 g cm−3. Soil horizons were not reconstructed, but lateral patchiness (islands of fertility) and surface-enriched depth profiles of organic carbon, nitrogen, phosphorous and other soil properties reformed considerably as of the 7th and 17th years of the study, trending towards structure similar to soil of surrounding, undisturbed communities. Specifically, in non-irrigated, 2-m-deep plots, organic carbon (C) was 12 mg g−1 and phosphorus (P) was nearly 60 μg g−1 in the top centimetre of soil under sagebrush crowns, compared to generally < 8 mg g−1 organic C and < 30 μg g−1 P in bare-soil canopy gaps or in deeper soils (McGonigle, Chambers & Gregory 2005; comparing to meta-analysis of Sankey et al. 2012). In these same plots, soils at 15–20 cm compared to 95–100 cm had 1.5% compared to 1.3% total C, 0.5% compared to 0.2% organic C and 3.6 mg kg−1 compared to 1.5 mg kg−1 inorganic nitrogen, respectively (Sorensen, Germino & Feris 2013). Soil pits dug within experimentally constructed plots in 2012 revealed formation of argillic horizons due to leaching of clay to deeper layers, as evident in clay films on pedogenic CaCO3 mineral deposits and aggregates.

The three irrigation treatments were as follows: ambient precipitation with no supplemental irrigation, supplemental irrigation in summer and supplemental irrigation in winter. The summer irrigation treatment consisted of four applications of 50 mm of water at biweekly intervals beginning in mid-June (200 mm total), simulating large summer-monsoon rainfall events that wet soils down to ~0.4 m depth. The winter irrigation treatment applied 200 mm of water in a series of wetting events of variable size (weather and infiltration-time dependent), occurring over a 1- to 2-week period in October and/or early April each year, just before or after winter snowpack when plots were snow-free but plants were assumed to be dormant. The extra soil water from irrigation stored through winter or added just after winter combined with spring rains to cause deep recharge of soil water (> 2 m soil depth). While the 200 mm year−1 increases in both of the irrigation treatments are nearly a doubling of average annual precipitation, irrigations occurred during non-rainy conditions and so effective soil wetting was reduced. Moreover, we did not intend to mimic predictions of future precipitation. Rather, our questions centred on the effects of strong shifts in seasonal timing of added precipitation. Irrigation was applied to the plots via a drip line and emitter system located on the ground. Drip lines were constructed of 8 m lengths of tubing spaced approximately 0.5 m apart. Emitters were placed at 0.5-m intervals, which delivered water at about 17.6 L min−1 plot−1, which achieved dispersion yet avoided ponding.

Plant community cover was established with five shrub species (Artemisia tridentata ssp. tridentata, A. t. wyomingensis, Chrysothamnus viscidiflorus, Ericameria nauseosa and Krascheninnikovia lanata), five perennial grass species (Elymus elymoides, E. lanceolatus, Leymus cinereus, Achnatherum hymenoides and Hesperostipa comata) and two forb species (Linum perenne and Hedysarum boreal). The forbs were seeded and all other species transplanted from local sagebrush communities in mid-November 1993, with effort to extract as much of the root as possible, and some replacement of transplant failures and watering of all plots in the first summer (a drought year). Plots were planted with eight individuals of A. t. ssp tridentata and eight A. t. ssp wyomingensis, but only A. t. ssp tridentata (diploid) survived.

Soil Water Content

Soil volumetric water content (θ) was determined using a neutron moisture probe (CPN Model 503DR Hydroprobe, Concord, CA, USA). Measurements were made biweekly April–October from 2002–2007 (calendar years). One 2.2-m aluminium access tube was located approximately in the middle of each plot, and soil moisture was determined at 0.2, 0.4, 0.6, 0.8, 0.9, 1.1, 1.2, 1.4, 1.6 and 2.0 ± 0.1 m depths. Soil water potentials of −1 and −1.5 MPa corresponded to soil water volumetric contents of 19% and 14% m3/m−3, according to soil water retention curves for soils collected at 15–20 and 95–100 cm depths from each plot and evaluated using a WP4-T Dewpoint PotentiaMeter (Decagon Devices, Pullman, WA, USA).

Shrub Cover, Density, Volume and Stem Growth

Sagebrush cover was determined using infield and photometric point-intercept methods. Infield data (1995–2007) were collected using point-intercept frames (33 frames, 36 points/frame) at the end of June each year. Some of the sagebrush became excessively tall for point-intercept frames, so sagebrush cover in 2010 onward was estimated from high-resolution aerial photographs using an on-screen, computer point-intercept method using imagej software (grid plug-in; NIH, Bethesda, MD, USA), which we previously compared with field measures (Janzen 2009). Photographs were taken from a camera mounted on a 3 m pole in the middle of each plot in late June. Density (# individuals/area over whole plot) and crown size were determined in 2009 for all adult sagebrush in each plot (adults defined nominally as >0.3 m in height; 133 shrubs total). Volume of individual sagebrush crowns was determined in 2009 by measuring crown height and maximum width in two orthogonal directions and multiplying these values assuming that shrub crowns were ellipsoid (Vora 1988; Cleary, Pendall & Ewers 2008). Total sagebrush canopy volume within plots (‘total canopy volume’) was the sum of individual crown volumes. There was no variation across the three precipitation treatments on 2-m soils in either specific leaf area (123 ± 4–126 ± 5 g cm−2; in 2009) or leaf-area index under shrubs (4.0 ± 0.4–4.4 ± 0.4 m3 m−3; in 2011, Sunfleck PAR Ceptometer, Decagon Devices, Pullman, WA) with five subreplicate sagebrush sampled/plot.

Plant Carbon Isotope Ratios

The isotope ratio of plant-tissue carbon (12C/13C) varies with the concentration gradient of CO2 between air and leaves during the photosynthesis that acquired the carbon. This CO2 gradient affects intrinsic water-use efficiency, the amount of CO2 assimilated per unit water diffusing through stomata (Farquhar, O'Leary & Berry 1982). Leaf samples from the previous year's growth (three plants harvested as subsamples per replicate plot) were harvested in June 2009 and again in February 2014, dried as described previously, ground to powder and analysed for 13C/12C at the Idaho State University Interdisciplinary Laboratory for Elemental and Isotopic Analysis using a Costech ECS 4010 elemental analyzer interfaced to a Thermo Delta V Advantage continuous-flow isotope-ratio mass spectrometer (Thermo Fisher Scientific, Inc., Waltham, MA, USA) for 2009 samples. Leaf samples collected in 2014 (up to five subsample plants per plot, less when fewer than five plants/plot) were analysed with model 2020 CRDS laser spectrometer (Picarro Inc, Santa Clara, CA, USA) also interfaced with a Costech ECS 4010. Analytical precision, calculated from analysis of standards distributed throughout each run, was ≤ ± 0.2‰. Isotopic values are reported in the conventional δ-notation (δ = ([Rsample/Rstandard] – 1) * 1000, where R = 13C/12C) relative to the international standard Vienna Pee Dee Belemnite (VPDB) and expressed as per mil (‰). More negative values of δ13C result from a smaller gradient of CO2 between air and leaves and indicate less water-use efficiency assuming no differences in leaf-air water deficits (Farquhar, O'Leary & Berry 1982).

Statistical Analysis

Significance of effects of irrigation treatment and soil-profile type on shrub density (#shrubs/plot), and crown and canopy volumes were determined using a mixed model anova for a full-factorial, randomized complete block, split-plot design with n = 3 blocks (proc mixed, sas v. 9.2 software, SAS Institute, Cary, NC, USA) with log-transformation to meet anova assumptions and Sheffé's test for mean comparisons. Soil-profile type was the main fixed factor, irrigation treatment was the fixed subplot factor, and block and the block * soil type interaction were specified as random factors. Annual shrub cover and δ13C were also analysed as mentioned previously, but with repeated measures and Tukey's test for mean comparisons. To identify relationships between shrub crown and canopy volume and different soil water storage compartments (< 0.2 m, < 0.6 m, 0.6–1 m, < 1 m, or > 1 m; measured in spring, fall or over all dates), a priori Akaike information criterion (with correction for small data sets, AICc) model selection was performed using stepwise regression in jmp v. 8.0 (SAS Institute). Our AIC goal was to compare the effect size of explanatory variables (soil-depth compartments) on shrub growth, and not to develop robust predictive models.


Soil Water

Winter irrigation increased volumetric soil water content (θ, averaged over all depths and years from 2002–2007) twice as much compared to supplemental summer irrigation, relative to non-irrigated plots (increases of up to eight percentage points for winter irrigation compared to up to four percentage points for summer irrigation, respectively, averaged across all depths; Fig. 1). This corresponded to about a 23% increase in absolute soil moisture (m3 of water in soil of entire plot) in winter irrigation (19.7 m3) compared to summer irrigation (15.9 m3) plots. Winter irrigation increased θ at all soil depths, and the increased θ persisted through midsummer (through July; Fig. 1) and was associated with soil water potentials above −1 MPa. Summer irrigation increased θ only in the top 40 cm of the soil, and increases in soil water lasted only ~7–14 days after each application, before being lost via evapotranspiration.

Figure 1.

Volumetric soil water content (% m3/m3) in spring (April–May, maximum wetness), early-summer (June–July) and late summer (September, minimum wetness) between 2002 and 2007 for shallow (1 m-SOIL) or deep (2 m-SOIL) plots with ambient precipitation (AMB), summer irrigation (SUM) or winter irrigation (WIN). Error bars are SE, which are smaller than symbols.

Sagebrush Cover, Density and Volume

Sagebrush cover was similar between summer irrigation and non-irrigated treatments in both types of soil profiles through 2013. In the absence of irrigation, soil depth had no effects on sagebrush cover, density, or crown or canopy volume (Figs 2 and 3). Winter irrigation led to considerably greater sagebrush cover in deep-soil plots (< 0.0001, Tukey, for 1995–2013), due primarily to marked increases since 2006 (Table 1; bottom panel of Fig. 2). In contrast, sagebrush cover in shallow profiles (no roots below 1 m depth) was less in the winter-irrigated than non-irrigated plots, marginally over all 20 years (T2,260 = 2.8, Tukey, = 0.06), and more significantly for each year up to 2006 (Table 1; top panel of Fig. 2). Winter irrigation effects on shallow soils were significant for the 2010–2013 period only after rerunning the anova on the log of cover and without 3 plots that had no surviving sagebrush (T2,260 = 3.7, Tukey, < 0.01; Table 1, Fig. 2). By 2010, no living sagebrush remained in one plot of each of the following combinations: (i) shallow soil and no irrigation, (ii) shallow soil and winter irrigation and (iii) deep soil and no irrigation.

Table 1. Summary anova Table showing effects of irrigation and soil-profile type on foliar cover (‘Cover’), mean number of shrubs per plot (‘Density’), mean crown volume of individual shrubs (m3/shrub, ‘Crown volume’), mean total canopy volume of shrubs per plot (m3/plot, ‘Canopy volume’) and leaf δ13C (‰) of sagebrush
 SoilIrrigationSoil × irrigation
d.f. F P d.f. F P d.f. F P
Cover, 1995–20131, 235.90.032, 2603.750.032, 26027.7<0.001
Cover, 2010–20131, 259.80.022, 530.190.832, 538.61<0.001
Density1, 21.850.312, 80.320.142, 80.330.73
Crown volume1, 20.050.842, 82.860.122, 83.750.07
Canopy volume1, 21.380.362, 82.430.152, 82.840.12
δ13C, 2009 & 20141, 20.170.722, 2611.6<0.0012, 261.950.16
Figure 2.

Mean sagebrush cover (%) in experimental plots from 1995–2013 for shallow (1 m-SOIL) or deep (2 m-SOIL) plots with ambient precipitation, summer irrigation or winter irrigation. Significant differences in cover among irrigation treatments (winter or summer) compared to non-irrigated control (ambient) plots within each year and soil-depth level are indicated with a * (< 0.05). Not shown are additional significant differences (< 0.05) between winter-irrigated and non-irrigated plots on 1-m soils that still had sagebrush over the 2010 through 2013 years combined. Error bars are SE.

Figure 3.

Comparison of (a) mean shrub density (# shrubs/plot), (b) mean individual shrub crown volume and (c) mean total-shrub canopy volume per whole plot for sagebrush on shallow (1 m-SOIL) or deep (2 m-SOIL) soils with ambient precipitation, summer irrigation or winter irrigation. Letters show differences among irrigation treatments within each soil treatment level at = 0.05. There were no significant treatment effects in panel A. Error bars are SE.

There was no significant difference in sagebrush density (# plants/plot) due to irrigation regime in 2009 (Table 1 and Fig. 3a). Individual shrub crowns, as well as plot-level canopies, were about threefold greater in deep plots with winter irrigation compared to no irrigation, did not vary with irrigation on the shallow plots (Table 1 and Fig. 3b,c). There were no significant differences in mean individual-crown or total canopy sagebrush volumes between summer irrigation and non-irrigated plots, in either soil-profile type. The relative abundance of small sagebrush (< 1 m tall) to large shrubs (> 1 m tall) was greater in shallow plots (14% compared to 9%, respectively) than in deep plots (mean of 13% in both size categories; see Fig. S1 in Supporting Information).

AICc Modelling of Shrub Volume Relationships to Soil Water

Regression models that had multiple terms (i.e. multiple soil-depth compartments) showed that soil water content in 0–0.6 m and > 1.0 m depths best predicted sagebrush volume (Table 2). For models with single terms, the best predictor of shrub volume at the individual scale was springtime soil water content in the shallowest depths (0–0.2 m), while at the canopy-scale it was springtime soil water below 1 m depth (Table 2). For models with one or two terms, soil water content measured over all times and especially at seasonal maximum (April–May) was important predictors, whereas seasonal minimum (September) water became important only after adding additional variables to the models (Table 2).

Table 2. Results from AICc modelling analysis comparing individual shrub crown volume (‘Crown’) and whole-plot canopy volume (‘Canopy’) of sagebrush to soil water content in soil layers. The top three models for each number of allowed factors are shown. Factors are mean seasonal (April–September) soil water content within soil-depth intervals. Fifteen independent variables were evaluated as follows: five depth categories (< 0.2 m, < 0.6 m, 0.6–1 m, < 1 m or > 1 m) averaged over three periods; either over all of the water content sampling dates, or April–May only to represent maximum seasonal soil water content (indicated by *), or September only to represent minimum seasonal soil water content (indicated by )
Model factorsNumber R 2 RMSEAICc
  1. RMSE, root mean square error; AICc, Akaike information criterion with a second-order correction for small sample size.

< 0.2 m*10.222.76175.17
< 0.2 m10.182.82176.63
< 0.6 m10.132.91178.80
< 0.6 m*,> 1.0 m*20.352.57171.48
< 0.6 m,> 1.0 m*20.332.59172.10
< 1.0 m*,> 1.0 m*20.302.65173.69
> 1.0 m*,< 0.2 m,> 1.0 m30.402.49171.12
< 0.6 m,> 1.0 m*,> 1.0 m30.382.54172.34
< 0.6 m,< 0.2 m*,> 1.0 m*30.372.56172.88
< 0.6 m*,> 1.0 m*,< 0.2 m,< 0.6 m40.462.41170.58
< 0.2 m*,> 1.0 m*,< 0.2 m,> 1.0 m40.442.45171.72
< 0.6 m*,> 1.0 m*,< 0.2 m,> 1.0 m40.432.48172.59
< 0.6 m*,> 1.0 m*,< 0.2 m,< 0.6 m,> 1.0 m50.472.43172.98
< 1.0 m,< 0.6 m*,> 1.0 m*,< 0.2 m,< 0.6 m50.462.44173.36
0.6–1.0 m,< 0.6 m*,> 1.0 m*,< 0.2 m,< 0.6 m50.462.44173.39
> 1.0 m*10.3224.16326.96
< 0.2 m*10.2026.18332.60
< 0.2 m10.1327.41335.79
< 0.6 m*,> 1.0 m*20.4821.41319.99
< 0.6 m,> 1.0 m*20.4821.55320.46
> 1.0 m,> 1.0 m*20.4821.55320.46
> 1.0 m*,< 0.2 m,> 1.0 m30.5220.99320.25
< 0.6 m,> 1.0 m*,> 1.0 m30.5121.12320.67
< 0.6 m,> 1.0 m,> 1.0 m*30.5121.16320.81
< 0.6 m,> 1.0 m*,< 0.2 m,> 1.0 m40.5420.89321.68
< 0.6 m*,> 1.0 m*,< 0.2 m,> 1.0 m40.5420.91321.74
< 0.6 m*,> 1.0 m*,< 0.2 m,< 0.6 m40.5420.95321.87
< 0.6 m*,> 1.0 m*,< 0.2 m,< 0.6 m,> 1.0 m50.5520.95323.85
< 0.6 m,0.6–1 m*,> 1.0 m*,< 0.2 m,> 1.0 m50.5520.96323.87
< 0.6 m,< 1.0 m,> 1.0 m*,< 0.2 m,> 1.0 m50.5520.96323.88

Carbon Isotope Ratios

There was no significant difference in δ13C between the soil-profile treatments over all irrigation treatments (Table 1). Leaf δ13C was about 1 ‰ more negative in shrubs in winter irrigation compared to non-irrigated plots of both soil depths in the 2009 sampling (soil X irrigation for 2009; T2,8 = 4.5, Tukey, < 0.05; Table 1 and Fig. 4). In the 2014 sampling, δ13C varied 2.5 ‰ among irrigation treatments on the shallow soils, becoming considerably reduced on non-irrigated plots and greater with irrigation, particularly winter irrigation (soil × irrigation for 2014; T2,8 = 4.5, Tukey, < 0.05). Precipitation accumulated in the 12 months preceding the sampling was 250 mm in 2009 and 90 mm in 2014, which compare to the long-term average of 220 mm year−1. Thus, soil × irrigation effects had a stronger effect on δ13C following additional years of treatment application and an extended drought period.

Figure 4.

Variation in δ13C of leaves of sagebrush for shallow (1 m-SOIL) or deep (2 m-SOIL) plots with ambient precipitation, summer irrigation or winter irrigation, in 2009 and in 2014. Asterisks indicate significant differences at < 0.05 between summer and winter irrigation treatments compared to non-irrigated control plots (ambient) within each soil-depth level. δ13C was also significantly less on the shallow ambient plots compared to all other treatment combinations in 2014. Error bars are SE.


We expected that deeper-soil or winter irrigation treatments would promote sagebrush abundance, but that growth would nonetheless be related to shallow-soil moisture, based on the two-layer hypothesis and subsequent modifications of it (Walter 1973; Paruelo & Lauenroth 1996; Ogle & Reynolds 2004; Ryel et al. 2008). Variation in soil depth alone under ambient precipitation had few effects on big sagebrush size and cover, possibly due to a large compensatory increase in water-use efficiency on non-irrigated, shallow soils. However, soil depth interacted with amount and timing of precipitation to cause large differences in sagebrush abundance. Consistent with the modified hypotheses, winter but not summer irrigation led to substantial infiltration and greater sagebrush growth on deep-soil plots, and AIC modelling revealed that shallow-soil water in spring was a relatively strong predictor of sagebrush growth. The surprising and markedly less sagebrush abundance on shallow plots with winter irrigation was not predicted by the hypotheses. The overall lack of summer irrigation effects on sagebrush was not surprising given its small effect on soil moisture, which could have resulted from evaporation or grass transpiration following irrigation.

Factors Promoting Deep Infiltration Increase Growth, but the Mechanism Involves Increased Shallow-Soil Water

The positive response of sagebrush to winter irrigation on deep soils was likely attributable to increased availability of water throughout all soil layers, and to sagebrush's dimorphisms in having both deep and shallow roots as well as larger drought-deciduous leaves and smaller evergreen leaves. The winter irrigation enhancement of soil moisture persisted across all soil depths, from spring through midsummer (Fig. 1), and furthermore crossed a nominal hydrological threshold in comparison with non-irrigated plots (i.e. > 19% VWC or −1 MPa − the point at which root hydraulic conductance in sagebrush is reduced below 50%; Ryel et al. 2002). In late summer, winter irrigation plots still had more moisture in deeper soils. Winter irrigation may have extended the March–June period during which soil water is readily available, when mineralization and nitrogen diffusion can occur, and when vegetative stem growth is most vigorous in sagebrush (Caldwell 1985; Evans & Black 1993; Ryel et al. 2008), and furthermore may have increased photosynthetic opportunities later in summer while minimizing risks of hydraulic (cavitation) stress (Ryel et al. 2008, 2010). Sagebrush has been shown to use both shallow- and deep-soil water (Donovan & Ehleringer 1994; Leffler et al. 2004; Schlaepfer, Lauenroth & Bradford 2011) and thus would have benefited from the uniform distribution of soil water from winter irrigation. Summer irrigation, on the other hand, increased soil moisture amounts only in shallow soils (top 40 cm) and only during late June through July when reproductive but not vegetative stem growth typically occurs (Caldwell 1985; Evans & Black 1993), and no increases in sagebrush cover, density or size resulted. Thus, a treatment designed to increase deep infiltration and water storage for use by this tap-rooted species during otherwise hot and dry conditions (i.e. winter irrigation) (i) does indeed increase growth, and more so than addition of the same amount of precipitation added in a way that does not lead to appreciable infiltration or storage (i.e. summer irrigation), but (ii) the mechanism of enhancement involves use of the moisture from shallower horizons when water is abundant.

Density of sagebrush was not increased by winter precipitation on deep soils, and its increase appeared to result more from growth of individuals. The resource-pool hypothesis of Ryel et al. (2008) suggests that such a growth increase would result from greater nutrient supply in spring. Although we did not measure nutrient supply, available nutrient concentration data provide mixed support for a contribution of N to the growth differences we observed. Sorensen, Germino & Feris (2013) reported greater nutrients in shallower compared to deeper soils of the 2-m-deep plots of our experiment, but the only changes in soil nutrients among treatments were increases in response to summer irrigation. Thus, there is little correspondence between the sagebrush responses to treatments reported here and soil fertility (total C and N, or organic or inorganic forms). Also, leaf N content of the abundant sagebrush in winter irrigation + deep-soil plots was 1.76 ± 0.02%, which compared with 2.0 ± 0.1%N across all treatments (from data accompanying δ13C analysis, not shown).

Growth Response to Winter Precipitation in Shallow Soils: A Paradox?

The negative effect of winter irrigation on sagebrush in shallow-soil plots was somewhat unexpected given the hypotheses and the general paradigm that more moisture, particularly in winter, should increase abundance in a semi-arid species such as big sagebrush. The outcome may thus also indicate a trade-off in big sagebrush in which tolerance of dry soils accompanies vulnerability to wet soil conditions, consistent with Araya et al.'s (2011) finding that hydrological trade-offs exist and contribute to ecohydrological niche formation. The lack of sagebrush cover on shallow-soil and winter-irrigated plots was not likely due to water deficit, given the accompanying reduction in δ13C, but rather could have resulted from temporary soil saturation (springtime θ > 30%; Fig. 1) and possibly denitrification or anoxia. We also did not detect any differences in shoot N or C between the winter irrigation + shallow-soil plots compared to grand averages for all other treatments combined that would indicate N limitations. Specifically, leaves in winter irrigation + shallow-soil plots had 1.9 ± 0.07% N, 50.7 ± 0.18% C and 5.5 ± 0.23 ‰ δ15N compared to grand averages of 2.0 ± 0.1% N, 50.9 ± 0.18% C and 5.34 ± 0.55 ‰ δ15N, respectively (from data accompanying δ13C analysis in 2009). In favour of the anoxia explanation, Lunt, Letey & Clark (1973) found sagebrush root growth to be strongly reduced for prolonged periods following experimental soil O2 deprivation in a controlled laboratory study, and more so than in other species. Growth of sagebrush seedlings was inhibited by saturation (Leffler et al. 2004), and sagebrush is absent from topographic depressions in the closed basins of the Idaho National Laboratory region (e.g. Lost River terminus) that are typical of many watersheds and the broader Great Basin ecoregion and are periodically inundated (Ganskopp 1986; Shive et al. 2012).

Native and exotic bunch grass (i.e. Agropyron cristatum) cover increased in response to winter irrigation in shallow-soil plots during the study years (Janzen 2009) and may have exerted competitive pressure on sagebrush by pre-empting soil water or nutrients. Roots of bunch grasses and some forb species are relatively abundant above (shallower than) 0.5 and especially 0.4 m depth, but often also reach below 0.5 m or even 1 m depth (Caldwell 1985). Increased grass growth on shallow winter irrigation plots may result in greater below-ground competition with A. tridentata roots in shallow soils (less opportunity for spatial partitioning of soil water).

A monotonically increasing relationship between soil-moisture storage and a water dependency consistent with the two-layer hypothesis or alternative models is common inference made for the ecohydrological niche of basin big sagebrush, as well as for other woody species that occupy similar winter-wet/summer-dry hydroclimate and soil conditions (Schenk & Jackson 2002; Ogle & Reynolds 2004; Ryel et al. 2008, 2010; Schlaepfer, Lauenroth & Bradford 2011; Ward, Wiegand & Getzin 2013). Specifically, deep-rooted and semi-evergreen species such as big sagebrush are assumed to be promoted by increased winter precipitation or deeper soils, and their enhancement of soil water storage. These assumptions are only partially supported by the experimental results reported here. An implication of the opposing effects of winter irrigation on deep compared to shallow soils is that changes in the amount or seasonality of precipitation may not lead to uniform sagebrush responses across landscapes with varying soil depths, such as the 0.4 to > 2 m soil depths on undulating basalt bedrock in the region evaluated here. Increases in winter/summer precipitation or larger and more intense rain events with climate change may increase occurrence of both saturated patches where soils are shallower or finer textured, and infiltration patches where soils are deeper and coarser, leading to a landscape mosaic of respective decreases or increases in A. tridentata. These edaphic considerations and trade-offs in response to climate shifts are rarely considered in plant species' vulnerability assessments (e.g. Richardson et al. 2014), and they support the concept of hydro-topographic ‘points of sensitivity’ to climate shifts in upland landscapes (Debinski et al. 2010).


Both authors contributed equally. We thank Kathleen Lohse for the soil pit and horizon evaluation; Laura Bond for statistical advice; and Sue Phillips, Robert Jones and two reviewers for helpful comments. Jay Anderson designed and oversaw the original experimental set-up. We thank Amy Forman and Roger Blew at S.M. Stoller Corporation, and Sara Bachman, Jeremy Greth, Jackie Hafla, Brandy Janzen, Jeff Morgan, Jess VanderVeen, Bill Davidson, Xochi Campos, Liam Junk and Masa Takahashi for logistical support, sample or data collection and laboratory assistance. Bryce Richardson and Stewart Sanderson provided cytometer analyses. The research was funded by grants to MJG from Idaho EPSCoR (EPS 0814387) and US Geological Survey Northwest Climate Science Center. Any use of trade, product or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.