1. Intrinsic water use efficiency (WUEi) reflects the trade-off between photosynthetic carbon gain and water loss through stomatal conductance and is key for understanding dryland plant responses to climate change. Stipa tenacissima is a perennial tussock C₃ grass with an opportunistic, drought-avoiding water use strategy that dominates arid and semiarid steppes across the western Mediterranean region. However, its ecophysiological responses to aridification and woody shrub encroachment, a major land-use change in drylands worldwide, are not well understood.
2. We investigated the variations in leaf stable isotopes (δ¹⁸O, δ¹³C, δ¹⁵N), nutrient concentrations (N, P, K), and culm water content and isotopic composition (δ¹⁸O, δ²H) of paired pure-grass and shrub-encroached S. tenacissima steppes along a 350 km aridity gradient in Spain (10 sites, 160 individuals). 
3. Culm water isotopes revealed that S. tenacissima is a shallow-rooted grass that depends heavily on recent rainwater for water uptake, which may render it vulnerable to increasingly irregular rainfall combined with faster topsoil drying under climate warming and aridification. With increasing aridity, S. tenacissima enhanced leaf-level WUEi through more stringent stomatal regulation of plant water flux and carbon assimilation (higher δ¹³C and δ¹⁸O), reaching exceptionally high δ¹³C values (-23 to -21‰) at the most arid steppes. Foliar N concentration was remarkably low across sites regardless of woody shrub encroachment, evidencing severe water and N co-limitation of photosynthesis and productivity. Shrub encroachment decreased leaf P and K but did not affect S. tenacissima water status. Perennial grass cover decreased markedly with both declining winter rainfall and shrub encroachment suggesting population- rather than individual-level responses of S. tenacissima to these changes.
4. The fundamental physiological constraints of photosynthetic C₃ metabolism combined with low foliar N content may hamper the ability of S. tenacissima and other drought-avoider species with shallow roots to achieve further adaptive improvements in WUEi under increasing climatic stress. A drought-avoiding water use strategy based on early stomatal closure and photosynthesis suppression during prolonged rainless periods may thus compromise the capacity of S. tenacissima steppes to maintain perennial grass cover, sustain productivity and cope with ongoing climate aridification at the drier parts of their current distribution. 

We selected 10 locations along a ~350 km transect from Central to Southeastern Spain (Fig. S1) representing an increasing climatic aridity gradient encompassing much of the geographical distribution of S. tenacissima in the Iberian Peninsula, including areas with MAP ranging from 346 to 464 mm and mean annual temperature (MAT) from 12.6 to 16.3℃ (Supporting Information Table S1, S2). The climate is semiarid Mediterranean, with asynchronous wet and warm seasons, as maximum and minimum precipitation fall in winter and summer, respectively. The soils are classified as Calcisols and are weakly developed and thin, with sandy loam texture and high pH values (>7–8) (Maestre et al. 2009). Transpiration fluxes and primary productivity in semiarid S. tenacissima steppes are rather low, with peak plant ecophysiological activity occurring shortly after the irregular rainfall pulses (Domingo et al. 2011).

Using the geographical coordinates of each location, we obtained the elevation and available climatic information from Worldclim (http://www.worldclim.org; Hijmans et al. 2005), and measured the minimum distance to the Mediterranean Sea (DTMS) using Google Earth. Climate variables included temperature and precipitation (mean annual, average of the wettest and coldest quarter and month, average of the driest and warmest quarter and month). We calculated the Aridity for each location as A= 1- (annual precipitation/annual potential evapotranspiration). Mean annual vapor pressure deficit (VPD) for each location was obtained from TerraClimate data (Abatzoglou et al. 2018). Several geographic and climatic variables showed substantial multi-collinearity along the aridity gradient (Table S3).

Field surveys

Field sampling was conducted during May 2008, and no permissions were needed for fieldwork. At each sampling location, we selected two paired sites with contrasting woody shrub cover (i.e. pure S. tenacissima grass steppes vs. nearby shrub-encroached steppes with 26.4% shrub cover; see Maestre et al. 2009), totaling 20 sites. The encroaching shrub species were Quercus coccifera, Pistacia lentiscus, Rhamnus lycioides, Juniperus phoenicea and Juniperus oxycedrus. Plant cover at each site was measured using the point-intercept methodology in four 30-m transects separated 8 m from each other.

At each of the 20 sites, we sampled 8 adult S. tenacissima individuals (replicates) resulting in 160 individuals. The canopy size of each tussock individual was measured using two diameters (parallel and perpendicular to the steepest slope) and calculated using the formula for the area of an ellipse. At sites undergoing woody shrub encroachment, we sampled tussock individuals located at a fixed distance (5-m) from the canopy edge of woody shrubs to exclude direct, immediate competition effects between shrubs and S. tenacissima at close range. Our assessment of how woody shrub encroachment impacts on the water and nutrient status of S. tenacissima focuses on diffuse competitive effects of encroaching shrubs, and therefore excludes the effects of direct, interspecific shrub-grass interactions at close range.

Isotopic and nutrient analyses

Fully expanded and healthy-looking leaves with their corresponding basal culms were collected for isotopic and nutrient analyses. Green leaf samples were oven-dried (60℃, 24h) and finely ground using a ball mill. Leaf δ¹³C and δ¹⁵N were measured by continuous flow dual isotopic analysis using an IsoPrime 100 IRMS (IsoPrime, UK) interfaced to a CHNOS C/N Elemental Analyzer. Leaf δ¹⁸O was measured in continuous flow using an Elementar PYRO Cube interfaced to a Thermo Finnigan Delta V IRMS (Thermo Finnigan, Germany). The analytical precision was ±0.1‰ for leaf δ¹³C and ±0.2‰ for leaf δ¹⁸O and δ¹⁵N. Ground leaves were digested with HNO₃:HClO₄ (2:1, v:v) and P and K concentrations were measured in the digested solution by a Perkin Elmer Inductively Coupled Plasma 5500 atomic absorption spectrometer.

Time-integrated WUEi at leaf level was estimated based on foliar δ¹³C values (Cernusak et al., 2013) as determined by the atmospheric CO₂ concentration (385.97 ppm in 2008) and the ratio of atmospheric [CO₂] to intracellular [CO₂] (c_i/c_a) as follows:

WUEi = A/g_s = c_a [1-(c_i/c_a)]×0.625                           Eqn (1)

d¹³C_leaf = d¹³C_atm – a - (b-a)×(c_i/c_a)                           Eqn (2)

where δ¹³C_leaf is the foliar isotope ratio, δ¹³C_atm is isotopic ratio of atmospheric CO₂ (-8.321‰ in 2008) and a and b are isotopic fractionation factors associated with CO₂ diffusion (4.4‰) and Rubisco discrimination (27‰), respectively.

Non-transpiring basal culms (5-6) from the base of the perennial tussock grasses near ground level were collected from each target individual as a proxy of the isotopic signature of source water used by S. tenacissima at the peak of the growing season. The basal culms were quickly placed in capped glass vials, sealed with parafilm, transported in coolers, and stored in the freezer (-20℃) until water extraction. Water extractions were performed using cryogenic vacuum distillation and lasted 2-4 h at 105 ºC until culm water was extracted completely. Culm water content (CWC) was calculated as:

CWC=(W_wet-W_dry)/W_wet·100%                                Eqn (3)

where W_wet is the weight of fresh culm sample and W_dry is the weight of culm sample after water extraction. The culm water δ¹⁸O and δ²H values of half of the samples (n=80) were determined by continuous flow using a Thermo Gas Bench II interfaced to a Thermo Finnigan Delta plus XL IRMS (Thermo Finnigan, Germany). The analytical precision was ±0.12‰ for δ¹⁸O and ±0.6‰ for δ²H. Water extracted from non-transpiring tissues in grasses inevitably includes some phloem water mixed with xylem water, which may cause some isotopic fractionation issues (Barnard et al., 2006; Jiang et al., 2022).