Volume 230, Issue 4 p. 1378-1393
Full paper
Free Access

Vertical decoupling of soil nutrients and water under climate warming reduces plant cumulative nutrient uptake, water-use efficiency and productivity

José Ignacio Querejeta

Corresponding Author

José Ignacio Querejeta

Departamento de Conservación de Suelos y Agua, Centro de Edafología y Biología Aplicada del Segura – Consejo Superior de Investigaciones Científicas (CEBAS-CSIC), Murcia, 30100 Spain

Author for correspondence:

José Ignacio Querejeta

Email:[email protected]

Search for more papers by this author
Wei Ren

Wei Ren

Departamento de Conservación de Suelos y Agua, Centro de Edafología y Biología Aplicada del Segura – Consejo Superior de Investigaciones Científicas (CEBAS-CSIC), Murcia, 30100 Spain

Chongqing Key Laboratory of Karst Environment, School of Geographical Sciences, Southwest University, Chongqing, 400715 China

Search for more papers by this author
Iván Prieto

Iván Prieto

Departamento de Conservación de Suelos y Agua, Centro de Edafología y Biología Aplicada del Segura – Consejo Superior de Investigaciones Científicas (CEBAS-CSIC), Murcia, 30100 Spain

Search for more papers by this author
First published: 07 February 2021
Citations: 51

Summary

  • Warming-induced desiccation of the fertile topsoil layer could lead to decreased nutrient diffusion, mobility, mineralization and uptake by roots. Increased vertical decoupling between nutrients in topsoil and water availability in subsoil/bedrock layers under warming could thereby reduce cumulative nutrient uptake over the growing season.
  • We used a Mediterranean semiarid shrubland as model system to assess the impacts of warming-induced topsoil desiccation on plant water- and nutrient-use patterns. A 6 yr manipulative field experiment examined the effects of warming (2.5°C), rainfall reduction (30%) and their combination on soil resource utilization by Helianthemum squamatum shrubs.
  • A drier fertile topsoil (‘growth pool’) under warming led to greater proportional utilization of water from deeper, wetter, but less fertile subsoil/bedrock layers (‘maintenance pool’) by plants. This was linked to decreased cumulative nutrient uptake, increased nonstomatal (nutritional) limitation of photosynthesis and reduced water-use efficiency, above-ground biomass growth and drought survival.
  • Whereas a shift to greater utilization of water stored in deep subsoil/bedrock may buffer the negative impact of warming-induced topsoil desiccation on transpiration, this plastic response cannot compensate for the associated reduction in cumulative nutrient uptake and carbon assimilation, which may compromise the capacity of plants to adjust to a warmer and drier climate.

Introduction

The widespread drying of the land surface with global warming is the result of warming-induced increases in potential evapotranspiration driven by increasing vapour pressure deficit (VPD) and net radiation (Scheff & Frierson, 2014). As terrestrial ecosystems get warmer, climatic aridity over land will increase (Fu & Feng, 2014; Huang et al., 2016) and droughts will become more frequent and intense (Dai, 2013; Cook et al., 2014). The projected soil drying expands far beyond drylands or regions with decreasing precipitation trends, as the global increase in evapotranspiration also causes soil drying in more mesic regions (Zhao & Dai, 2015; Berg et al., 2017). However, projections of climate change impacts focus mostly on surface (upper 10 cm) soil water content, despite the fact that many vegetation types can easily access soil/bedrock water down to 2–3 m or even greater depths (Canadell et al., 1996; Schulze et al., 1996; Schenk & Jackson, 2002a). Recent modelling studies have predicted large decreases in soil moisture content near the surface, but much smaller decreases or even increases in moisture in deeper soil layers in many regions (Berg et al., 2016). Terrestrial ecosystems around the Mediterranean basin are among the most vulnerable to global warming (Sala et al., 2000) as predictions for this region include temperature increases of 2–5°C for the second half of the 21st century (Giorgi & Lionello, 2008). Global warming will affect global air circulation patterns and will change regional precipitation regimes, which for Mediterranean regions implies reduced rainfall amounts with more frequent extreme climatic events including prolonged droughts (Guiot & Cramer, 2016).

A better understanding of soil moisture responses to global warming is critical to anticipating impacts on vegetation. The depths at which plants extract soil water are dependent on soil moisture conditions and on plants having active roots in the different soil/bedrock layers (Jackson et al., 2000; Ryel et al., 2002; Ding et al., 2021). In dryland ecosystems, plants with a dimorphic root system are able to use water from both topsoil and subsoil/bedrock layers that have been recharged during previous seasons (Ehleringer et al., 1991; Pierret et al., 2016; Rempe & Dietrich, 2018; Carrière et al., 2020; Dawson et al., 2020; Germon et al., 2020; Schwinning, 2020; Nardini et al., 2021). Ryel et al. (2010) postulated that dryland plants have evolved to rapidly utilize the soil moisture available in shallow depths to maximize nutrient capture. The shallow soil ‘growth water pool’ is exploited during the period of high resource acquisition and growth, and is directly coupled to nutrient availability, whereas the ‘maintenance water pool’ in deeper subsoil/bedrock is largely decoupled from nutrient resources and is characterized by greater temporal stability (Ryel et al., 2008). Many plant species shift their depth of water uptake to deeper subsoil/bedrock layers during drought periods to sustain transpiration (Querejeta et al., 2007b; Voltas et al., 2015). However, greater reliance on deep water sources may limit their ability to obtain sufficient nutrients as these are typically most abundant in the topsoil layer (Jobbagy & Jackson, 2001; Barbeta et al., 2015). Vertical hydraulic redistribution (‘hydraulic lift’; Caldwell et al., 1998; Bauerle et al., 2008) and nutrient uplift by plant roots could buffer the negative impacts of water and nutrient decoupling during drought periods, especially in deep-rooted species (Jobbagy & Jackson, 2001, 2004; Sardans & Peñuelas, 2014).

The acquisition of nutrients by plants depends on the availability of soil water, as nutrients move towards absorptive roots by diffusion and mass flow and are taken up and transported through the plant as a result of water potential gradients and water flux between roots, xylem and leaves (Schlesinger et al., 2016). Projected changes in temperature and precipitation regimes with climate change in the Mediterranean area could significantly reduce topsoil moisture contents (Miranda et al., 2009; León-Sánchez et al., 2018) and thus soil nutrient availability for plants (Kreuzwieser & Gessler, 2010; Rouphael et al., 2012; He & Dijkstra, 2014; Gessler et al., 2017; Luo et al., 2018; Peñuelas et al., 2018). However, Grossiord et al. (2018) recently proposed that climate warming might also enhance soil nitrogen cycling rates and plant nutritional status during hotter droughts, thereby mitigating the impacts of climate aridification on plant performance. The effects of warming on soil nutrient availability and plant nutrient uptake may be heavily context-dependent and warrant further research.

We used Helianthemum squamatum shrublands as a model system to assess the impacts of simulated climate warming and drying on plant water- and nutrient-use patterns. Helianthemum squamatum is a small shrub species (20–40 cm) that grows at low elevations (40–900 m) on gypsum outcrops with semiarid climate, and it is one of the most representative and widespread perennial endemic gypsophytes on the Iberian Peninsula (Escudero et al., 1999). Helianthemum species serve as hosts of ectomycorrhizal fungi (Terfezia sp., Picoa sp.) that produce desert truffles of high commercial value (Marqués-Gálvez et al., 2020). Maximum rooting depth in this shrub species is 65–100 cm (Guerrero-Campo et al., 2006; Palacio et al., 2017), which makes it suitable for evaluating potential changes in soil resource utilization in response to climate manipulation. The root system of H. squamatum is dimorphic with shallow roots in the thin fertile topsoil horizon and deeper taproots with the capability to penetrate into hard subsoil gypsum layers (Escudero et al., 2014), thereby allowing exploitation of shallow or deep moisture pools depending on availability (Palacio et al., 2017). In a manipulative field experiment, we exposed H. squamatum shrubs to warming and/or rainfall reduction for 6 yr to simulate projected climate conditions during the second half of the 21st century. Warming reduces the mean annual moisture content of upper soil layers in this semiarid shrubland through faster evapotranspiration after rain pulses, which shortens the duration of moist topsoil conditions that favour nutrient uptake by roots. A lowered moisture content in the fertile upper layer may lead to decreased nutrient diffusion and mobility with nutrients becoming ‘locked up’ in a drier soil, slower litter decomposition and nutrient mineralization and cycling and lowered abundance or activity of fine roots and mycorrhizal fungi (León-Sánchez et al., 2018, 2020; Prieto et al., 2019).

We hypothesized that climate warming and drying will lead to more frequent and longer periods of dry topsoil conditions, and thus to enhanced vertical decoupling between nutrient and water availability in the soil/bedrock profile; that shrubs growing under chronically warmer and drier conditions will be forced to shift root water uptake from dry topsoil to deeper and wetter subsoil/bedrock layers; and that greater reliance on water from less fertile subsoil/bedrock layers will decrease cumulative plant nutrient uptake, thereby leading to reductions in plant photosynthesis and productivity.

Materials and Methods

Study site and experimental design

The study was carried out near Aranjuez, in central Spain (40°02′N–3°32′W, 495 m altitude). The study area has a continental Mediterranean climate, with a mean annual temperature of 15°C and an average rainfall of 358 mm (for the period 1977–2016) concentrated mainly in the spring and autumn months. During 2017, the experimental area received slightly lower rainfall than the historical average (310 mm). Soils are derived from gypsum (CaSO4·2H2O), have pH values of c. 7 and are classified as Gypsiric Leptosols (IUSS WRB, 2014). Soils are shallow (10–15 cm deep) overlying weathered gypsum subsoil/bedrock and show a very thin superficial organic horizon (1–2 cm thick). Vegetation is a native grassland and shrubland community with low plant cover (< 40%) and is dominated by the perennial tussock grass Macrochloa tenacissima (L.) Kunth and the gypsophilous shrub H. squamatum (L.) Dum. Cours.

In February 2011, we established a manipulative field experiment to examine the long-term effects of simulated climate change on vegetation and soil. Manipulated climate treatments were as follows: warming (W), rainfall reduction (RR), and their combination (W + RR). The warming treatment simulates the predictions for the second half of the 21st century in the western Mediterranean region (Giorgi & Lionello, 2008) and was achieved by installing open-top chambers (OTCs) made of transparent methacrylate that passively increase mean air and surface soil temperatures (Supporting Information Fig. S1). To simulate projected reductions in precipitation, we used passive rainout shelters made of the same material that intercepts and excludes c. 30% of the incoming rainfall from the plots. Finally, the combined W + RR treatment was achieved by installing both OTCs and rainout shelters over the same experimental plot (Fig. S1). The experiment includes 10 replicate plots per each climate manipulation treatment plus 30 control plots, making a total of 40 experimental plots of c. 1 m2 distributed across a 100 × 50 m area. These plots were randomly assigned to the different climate treatments and were at least 2 m from each other. The target shrub H. squamatum is the dominant (often the only) plant species present in the experimental plots.

Warming (W and W + RR) increases mean annual air and surface soil (0–5 cm) temperatures by 2.5°C. Mean ( ± SD) air temperature inside OTCs increased by 1.83 ± 1.16°C (winter), 2.54 ± 0.96°C (spring), 3.29 ± 1.03°C (summer) and 2.04 ± 1.21°C (autumn) for the 2011–2015 period (León-Sánchez et al., 2018). Warming treatment effects are greatest during late spring and summer when midday temperatures inside the OTCs increase up to 6–7°C on hot days. VPD was higher in plots exposed to warming (W and W + RR; 1.311 Pa) than in those exposed to ambient temperature (control and RR; 1.042 Pa); the differences in VPD were greatest during spring (1.488 vs 1.138 Pa, respectively) and summer (2.924 vs 2.330 Pa, respectively). Topsoil water content (0–5 cm depth) was decreased to the same extent by the OTCs and the rainout shelters across the three climate manipulation treatments (by 2–3% relative to the control). The rainout shelters did not affect air or soil temperatures or VPD. For further details on treatment effects on microclimatic conditions, see León-Sánchez et al. (2018).

Plant water uptake patterns

Measuring the isotopic composition of soil/bedrock and xylem water provides a powerful tool to assess the depth of water uptake in woody plants (Dawson et al., 2002). Little fractionation occurs during water uptake by roots, so xylem isotopic composition generally matches the average isotopic composition of the different water sources taken up by roots along the soil profile (but see Ellsworth & Williams, 2007 and Barbeta et al., 2020 for some exceptions in which hydrogen isotopic fractionation may occur, especially in halophytes and xerophytes). When the potential soil water sources available are isotopically different from each other, it is possible to distinguish their relative contribution to xylem water (Ehleringer & Dawson, 1992; Rothfuss & Javaux, 2017). In dryland ecosystems, intense evaporation in upper soil layers during rainless periods results in marked vertical gradients in the isotopic composition of soil water with depth (Allison et al., 1983), thereby allowing discrimination between plants tapping deep or superficial water sources (Querejeta et al., 2007b; Moreno-Gutiérrez et al., 2012; Palacio et al., 2017).

Six years after the initiation of climate manipulations, we assessed plant water-use patterns in H. squamatum shrubs growing inside (W, RR and W + RR) and outside the climate change treatments (control) during the peak growing season when plant physiological activity is highest (May 2017). Within each plot, one H. squamatum individual was selected and two to three 5- to 7-cm-long basal woody twigs (2–4 mm diameter) were collected during the early morning and were immediately placed together in one air-tight glass vial. Scraping off the bark from the twigs was impractical because of their small size, so some contamination of nonenriched xylem water with enriched phloem water may have occurred, although the high transpiration rate of H. squamatum should minimize this risk (Martín-Gómez et al., 2017). Topsoil samples (0–10 cm) were collected underneath each target shrub using a hand auger (AMS Inc, American Falls, ID, USA) and placed in an air-tight glass vial. Additionally, topsoil and subsoil/weathered bedrock samples were collected at 5 cm depth intervals from the surface to 65 cm deep from three different profiles located outside the climate change treatments. These three profiles were located only 2–3 m away from the experimental plots and are thus representative of the soil/bedrock conditions within the experimental plots. Vials containing plant and soil samples were immediately capped and wrapped with parafilm, transported to the laboratory in an ice-filled cooler and stored frozen. Xylem and soil water was extracted in a cryogenic vacuum distillation line at 100°C and 10 millitorr vacuum pressure for 2 h (Ehleringer & Osmond, 1989), and total soil water content was measured gravimetrically by weighting the soil samples before and after water extractions. Free soil water content was also measured gravimetrically by weighting soil subsamples before and after oven drying (50°C, 72 h). Crystallization water content (water bound within gypsum) was determined as the difference between total and free soil water content. Water δ2H was measured using a dual-inlet hot chromium reactor unit (H/Device™; Thermo Scientific, Waltham, MA, USA) interfaced with a Thermo Delta Plus XL mass spectrometer (Thermo Fisher Scientific) and water δ18O was determined by continuous flow using a Thermo Gas Bench II interfaced to a Thermo Delta Plus XL mass spectrometer. δ2H and δ18O are expressed in delta notation relative to the Vienna Standard Mean Ocean Water (VSMOW, ‰). Long-term external precision values for δ2H and δ18O are ±0.60‰ and ±0.12‰, respectively. Deuterium excess (d-excess, ‰) was calculated as the deviation from the Global Meteoric Water Line (d-excess = δ2H – 8 × δ18O; Dansgaard, 1964).

Leaf gas exchange, nutrient concentrations and isotopic composition

Net photosynthetic rate (A), stomatal conductance (gs) and transpiration rate (E) were measured in mid-May 2017 with a portable photosynthesis system (LI-6400, Li-Cor, Inc., Lincoln, NE, USA) equipped with a Li-Cor 6400-01 CO2 injector. Leaf gas exchange was measured on mature, sun-exposed leaves. The CO2 concentration in the cuvette was maintained at 400 μmol mol−1 CO2 and measurements were performed at a saturating light intensity of 1500 μmol m2 s−1 and at ambient air temperature and relative humidity. The air flow was set to 250 μmol s−1. The leaves used to determine gas exchange were collected and photographed with a Sony α200 DSRL camera (Sony Corp, Tokyo, Japan) and the resulting image was processed using Photoshop CS (Adobe Systems, San José, CA, USA) to determine leaf area and correct leaf gas exchange measurements. Intrinsic water-use efficiency (WUEi) was calculated as the A/gs ratio, whereas instantaneous WUE (WUEt) was calculated as the A/E ratio.

Leaf carbon (δ13C) and oxygen (δ18O) isotopic composition were measured in mature sun-exposed leaves that were collected in May 2017 from the same individuals used for plant water sources and leaf gas exchange measurements. Leaf samples were oven-dried at 60°C for 72 h and finely ground with a ball mill before being weighted and encapsulated into tin or silver capsules for analyses. Leaf δ13C and C and N concentrations were determined by continuous flow dual isotope analysis using a CHNOS Elemental Analyzer interfaced to an IsoPrime100 mass spectrometer. Leaf δ13C isotope composition is expressed in delta notation (‰) relative to the Vienna Pee Dee Belemnite standard (V-PDB). Leaf δ18O was determined in continuous flow using an Elementar PYRO Cube interfaced to a Thermo Delta V mass spectrometer. Leaf δ18O isotope composition is expressed in delta notation (‰) relative to VSMOW for δ18O. Long-term external precision values for leaf δ13C and δ18O determinations are ±0.10‰. and ±0.20‰, respectively. All stable isotope analyses were conducted at the Center for Stable Isotope Biogeochemistry, University of California-Berkeley (USA). Leaf δ13C can be used to assess variation in traits that covary with leaf gas exchange and water relations, including WUEi (Dawson et al., 2002). Leaf δ18O reflects the isotopic composition of the source water utilized by plants as well as leaf-level evaporative effects (Barbour, 2007). Plants using large proportions of deeper, less evaporated water sources with depleted δ18O values are expected to show the imprint of this distinct water source signal in their leaves (Sarris et al., 2013; Barbeta et al., 2015). Moreover, the oxygen isotope enrichment of leaves above source water (Δ18O = δ18Oplant − δ18Osource water) can help to identify interplant differences in gs, and E as Δ18O can provide a time-integrated indication of cumulative transpiration over the growing season (Barbour, 2007).

Leaf phosphorus (P), potassium (K), calcium (Ca), sulphur (S), iron (Fe), copper (Cu) and zinc (Zn) concentrations were measured by inductively coupled plasma optical emission spectrometry (ICP- OES; Thermo Elemental Iris Intrepid II XDL, Franklin, MA, USA) after a microwave-assisted digestion with HNO2 : H2O2 (4 : 1, v/v). Four leaves per plant were weighed after drying (60°C for 72 h) to determine their leaf DW, and leaf nutrient contents were calculated by multiplying nutrient concentration by leaf DW.

Shoot biomass production and post-summer survival

To measure shoot biomass production, one representative terminal shoot per target shrub (c. 10 cm length) was collected in May 2017. Shoots were oven-dried at 60°C for 72 h and weighed to obtain the total shoot DW, standardized per unit shoot length. We could not carry out a more extensive biomass sampling to avoid destruction of whole individuals in this long-term experiment. Finally, we measured post-summer plant survival in October 2017. Plant survival rate was estimated as the percentage of H. squamatum individuals present in spring in each experimental plot that were still alive at the end of the summer drought period. Individuals showing no green leaves and no resprouting at 2 months after summer drought termination (first autumn rainfalls) were considered dead.

Soil/subsoil organic matter and nutrient contents

Soil and subsoil/bedrock samples were air-dried, sieved (2 mm) and ground to a fine powder with a ball mill. The samples were digested in nitric-perchloric acid using a microwave, before determination of the total concentrations of P in an ICP-OES analyser. The total contents of organic C and N in soil/subsoil samples were determined by a Thermo Scientific CN Flash2000 analyser.

Statistical analyses

We analysed differences in the isotopic composition of soil/subsoil water (δ18O, δ2H, d-excess) and soil/subsoil total and free water contents with depth using general linear mixed models (LMMs) with depth as a fixed factor and soil/subsoil profile as a random factor. We used general linear models (LMs) to assess the effects of warming (W), rainfall reduction (RR) and their interaction on the isotopic composition of topsoil and xylem water, topsoil water content, leaf gas exchange, leaf nutrient concentrations and isotopic composition and shoot biomass production, using W (yes or no) and RR (yes or no) as factors. We performed a principal component analysis (PCA) with six leaf nutrient concentrations (N, P, K, Fe, Cu and Zn) to obtain a multidimensional overview of plant nutrient status. We then extracted the loadings of the first and second PCA axes (PCAAxis1 and PCAAxis2) and carried out a general LM with W and RR as factors. To test the effects of warming in nonstomatal limitations to carbon assimilation we performed slope tests between warmed (W and W + RR) and nonwarmed (control and RR) using standardized major axis regressions (SMA) for A : gs and A : E relationships. Additionally, we carried out a stepwise regression with A as the dependent variable and stomatal conductance (gs) and plant nutrient status (PCAAxis1, see later) as independent variables. We calculated the estimated proportion of the different water sources using Bayesian mixing models for stable isotope data in the package siar for R (Parnell et al., 2010). We used the xylem water δ18O and δ2H of each individual plant as mixing values. The isotopic composition of topsoil water (0–10 cm) was measured underneath shrubs in each of the climate treatments. We considered two isotopically distinct water pools (topsoil water in each climate treatment and the pooled data for the 10–65 cm depth interval (mean ± SD) as inputs for the model. We then used the function siarmcmcdirichletv4, in which the output is calculated on a population basis. We classified individual plants into the four climate treatments, setting the number of iterations, burn-in and thinning to 500 000, 50 000 and 10 000, respectively. This method considers population variability when estimating water source proportions. Post-summer plant survival was analysed using generalized linear models with a binomial distribution where survival was the dependent variable and W and RR were the predictor factors. All calculations and statistical analyses were performed with the R software (v.2.15.3; R Core Team, 2019) using the packages ade4 (Chessel et al., 2004), effects (Fox et al., 2014), hmisc (Harell Jr, 2015), lme4 (Bates et al., 2015), nlme (Pinheiro et al., 2014) and smatr (Warton et al., 2012). Data shown throughout the text are means ± SE.

Results

Nutrients, water content and isotopic composition in the soil/subsoil profile

Organic matter and N and P contents were highest in the topsoil layer but decreased steeply with depth (Fig. 1). Conversely, free water content was lowest near the surface and increased with depth (Fig. 2a). Soil/subsoil water isotopic values (δ18O and δ2H) were most enriched in the topsoil layer and became more negative with depth (Figs 2b, 3). Water d-excess showed very negative values in the topsoil layer, indicating strong evaporative isotopic fractionation, and became less negative with depth (Figs 2c, 3). Topsoil water content was lower in warmed than in nonwarmed plots (Fig. 4a; Table S1), indicating enhanced surface soil desiccation with warming. Differences in topsoil water content among the climate treatments were entirely attributable to intertreatment differences in free water content, given that gypsum crystallization water content was remarkably similar across climate treatments (Fig. 4b,c). The isotopic composition of topsoil water was similar across climate treatments (Figs 3, 4c, S2; Table S1).

Details are in the caption following the image
(a–c) Changes in mean organic matter content (SOM) (a) and total nitrogen (N) (b) and phosphorus (P) (c) concentrations with depth in the soil/bedrock profile. Values are means ± SE (n = 3) for each depth except for 65, 67.5 and 72.5 cm where n = 1. Soil contents in SOM, total N and P decreased significantly with depth (F = 4.606, P < 0.001; F = 7.415, P < 0.001; and F = 2.829, P < 0.01, respectively).
Details are in the caption following the image
(a–c) Changes in total (black dots) and free (grey dots) soil water content (a), mean soil water δ18O (grey) and δ2H (black) isotopic values (b) and deuterium excess values (d-excess) (c) with depth along the soil/bedrock profile. Values are means ± SE (n = 3). Evaporation is more intense near the surface, which causes evaporative isotopic fractionation and more negative d-excess values in the topsoil. Both total and free-soil water contents increased with depth (F = 2.4367, P < 0.05 and F = 1.6714, P < 0.09, respectively); soil water δ18O and δ2H values became more negative with depth (F = 19.951, P < 0.001, and F = 2.378, P < 0.001, respectively), whereas d-excess increased with depth (F = 5.147, P < 0.001).
Details are in the caption following the image
Isotopic composition (δ18O and δ2H) in topsoil and subsoil water (circles) and in Helianthemum squamatum xylem water (triangles) under ambient conditions (control, C) and under climate change conditions (W, warming; RR, rainfall reduction; W + RR, warming and rainfall reduction). The local meteoric water line (LMWL) for peninsular Spain (δ2H = 12.40 + 8.49× δ18O; years 2000–2006, solid black line; Díaz-Teijeiro et al., 2009) is shown. The two regression lines for plant xylem water samples (black dashed line) and topsoil/subsoil water samples (red dashed line) represent evaporation lines with flatter slopes than the LMWL (regression slopes of 4.74 and 3.35, respectively) indicating evaporative isotopic fractionation. Isotopic values for gypsum crystallization water (black circle) were obtained from Palacio et al. (2014).
Details are in the caption following the image
(a–d) Total topsoil (0–10 cm) water content (a), free water content in topsoil (b), gypsum crystallization water content (c) and deuterium excess values of topsoil water (d-excess, ‰) (d) under Helianthemum squamatum shrubs growing under ambient conditions in control plots (C) and under simulated climate change conditions (W, warming; RR, rainfall reduction; W + RR, warming and rainfall reduction). Data are means ± SE. Asterisks indicate significant effects of the experimental climate change factors (Warming, *, P < 0.05; ns, nonsignificant).

Impacts of climate manipulation on xylem water isotopic composition and plant water sources

Plants were using varying mixtures of evaporatively enriched water from the topsoil layer and (less enriched) water from subsoil layers during spring (Fig. 3). The isotopic composition of xylem water in H. squamatum was affected by both warming and rainfall reduction, which resulted in higher d-excess values in W, RR and W + RR plants than in control plants (Fig. 5; Table S2). Bayesian models revealed that plants from control plots used a larger proportion of topsoil water (80.6% from the top 10 cm) and relied less on water stored in deeper subsoil/bedrock layers (19.4% from 10–65 cm depth interval). By contrast, plants growing under climate change conditions used comparatively smaller proportions of water from the topsoil layer (59.6%, 66.3% and 61.7% for RR, W and W + RR, respectively) and greater proportions of water stored in deeper subsoil/bedrock layers (33–40%; Figs 6, S3).

Details are in the caption following the image
(a–c) Deuterium excess (d-excess, ‰) (a), oxygen (δ18O, ‰) (b) and hydrogen isotopic values (δ2H, ‰) (c) of xylem water in Helianthemum squamatum shrubs growing under ambient conditions in control plots (C) and under simulated climate change conditions (W, warming; RR, rainfall reduction; W + RR, warming and rainfall reduction). Data are means ± SE. Asterisks indicate significant effects of the experimental climate change factors (*, P < 0.05; **, P < 0.01).
Details are in the caption following the image
Results from Bayesian stable isotope mixing models showing the estimated contribution of topsoil water (0–10 cm) and subsoil/bedrock water (10–65 cm depth interval) to the xylem water of Helianthemum squamatum shrubs growing under ambient conditions in control plots or under simulated climate change conditions (W, warming; RR, rainfall reduction; W + RR, warming and rainfall reduction).

Impacts of climate manipulation on plant nutrient status, gas exchange, growth and survival

Warming (W and W + RR) reduced the foliar concentrations of multiple essential plant nutrients including N, P, K, Fe, Cu and Zn and increased leaf C : N and C : P ratios (Tables 1, S3). In contrast, leaf Ca and S concentrations were unaffected by warming. W + RR plants generally showed the lowest foliar nutrient concentrations, which translated into significant W × RR interactions for leaf P and Zn (Tables 1, S3). A PCA of leaf nutrient concentrations across climate treatments revealed two highly significant components. The first axis (PCAaxis1) accounted for 40.8% of the variance and was associated with N, P, Fe, Cu and Zn which had strong positive loadings on PCAaxis1 (ranging from 0.49 to 0.73; Table S4). The individual plant scores along PCAaxis1 thus provided an integrative measure of overall leaf nutrient status. The second component (PCAaxis2) accounted for 17.0% of the variance and was mainly associated with variations in K concentrations. Plant scores along PCAaxis1 were reduced by warming, indicating lower overall nutrient concentrations in warmed plants (especially in W + RR, as revealed by a significant W × RR interaction), whereas scores along PCAaxis2 were not significantly affected by climate manipulation (Tables 1, S3). The lower foliar nutrient concentrations combined with the lower leaf DWs observed under warming conditions (F1,24 = 4.41, P = 0.04) led to lower foliar nutrient contents in W and W + RR plants (Tables S5, S6).

Table 1. Mean (±SE) leaf nutrient concentrations nitrogen (N), phosphorus (P), potassium (K), iron (Fe), copper (Cu), zinc (Zn), calcium (Ca) and sulphur (S), stoichiometric ratios (carbon (C) : N, C : P) and nutrient status (axes 1 and 2 of the principal component analysis (PCA) with leaf nutrient concentrations; see the Materials and Methods section) and shoot biomass in Helianthemum squamatum shrubs growing under simulated climate change conditions (warming, W; rainfall reduction, RR; warming and rainfall reduction, W + RR) and under ambient conditions (control).
N (mg g−1) P (mg g−1) K (mg g−1) Fe (μg g−1) Cu (μg g−1) Zn (μg g−1) Ca (mg g−1) S (mg g−1)
Control 15.46 ± 0.07 0.513 ± 0.002 7.2 ± 0.1 73.3 ± 5.5 4.5 ± 0.5 7.7 ± 0.5 35.58 ± 1.41 37.43 ± 2.46
RR 16.09 ± 0.08 0.591 ± 0.004 6.6 ± 0.1 78.6 ± 4.7 4.5 ± 0.5 8.4 ± 0.6 41.55 ± 3.30 43.10 ± 4.08
W 13.17 ± 0.03 0.466 ± 0.002 6.0 ± 0.1 41.7 ± 3.3 3.4 ± 0.3 7.7 ± 0.6 43.83 ± 3.61 46.16 ± 4.06
W + RR 12.92 ± 0.02 0.439 ± 0.002 5.2 ± 0.1 36.0 ± 3.1 3.7 ± 0.4 5.3 ± 0.2 40.12 ± 1.98 46.50 ± 4.39
C : N C : P Leaf nutrient statusAxis1 Leaf nutrient statusAxis2 Shoot biomass (mg cm−1)
Control 24.28 ± 0.55 738.9 ± 41.08 0.742 ± 0.289 −0.181 ± 0.326 27.0 ± 2.1
RR 24.34 ± 1.40 674.5 ± 48.78 1.399 ± 0.432 0.341 ± 0.420 21.4 ± 2.2
W 28.87 ± 0.87 808.9 ± 42.40 −0.994 ± 0.287 −0.295 ± 0.309 21.4 ± 2.4
W + RR 29.45 ± 0.92 881.6 ± 60.44 −2.023 ± 0.281 0.300 ± 0.313 20.7 ± 2.6

Warming strongly decreased photosynthesis in the W and W + RR treatments, with 34.5% lower average A-values in warmed than in nonwarmed plants (Tables 2, S7). On the other hand, warming enhanced transpiration, with W and W + RR shrubs showing, on average, 50% higher gs and 56% higher E than shrubs growing under ambient temperature (Tables 2, S7). Simultaneous decreases in A and increases in gs and E under warming resulted in 60.3% lower WUEi and 62.9% lower WUEt in warmed than in nonwarmed plants. A lower WUEi was linked to lower shoot biomass production (Fig. S4). Moreover, the slopes of the A : gs and A : E regression lines were much lower in warmed plants (W and W + RR) than in plants growing under ambient temperature (control and RR), which revealed the existence of strong nonstomatal limitations to carbon assimilation (i.e. nutritional, biochemical) in warmed plants (Fig. 7).

Table 2. Mean (± SE) photosynthetic rate (A, μmol CO2 m2 s−1), stomatal conductance (gs, mol H2O m2 s−1), intrinsic water-use efficiency (WUEi, μmol CO2 mol H2O−1), transpiration rate (E, mmol H2O m2 s−1) and instantaneous water-use efficiency (WUEt, μmol CO2 mmol H2O−1) in Helianthemum squamatum shrubs growing under simulated climate change conditions (warming, W; rainfall reduction, RR; warming and rainfall reduction, W + RR) and under ambient conditions (control).
A gs WUEi E WUEt
Control 21.03 ± 3.14 0.18 ± 0.03 127.3 ± 11.2 4.71 ± 0.74 4.67 ± 0.31
RR 19.41 ± 2.82 0.16 ± 0.03 137.5 ± 14.8 4.66 ± 0.92 4.48 ± 0.45
W 15.81 ± 2.88 0.26 ± 0.04 63.9 ± 10.4 7.49 ± 1.18 2.19 ± 0.33
W + RR 9.91 ± 3.53 0.23 ± 0.06 37.2 ± 6.6 7.14 ± 1.81 1.21 ± 0.20
Details are in the caption following the image
(a, b) Linear regressions between net photosynthetic rate (A) and stomatal conductance (gs) (a) and net photosynthetic rates (A) and transpiration (E) (b) in Helianthemum squamatum shrubs growing in control (Cont), rainfall reduction (RR), warmed (W) and warmed + rainfall reduction (W + RR) conditions. Each point represents one single replicate. Lines represent fitted linear regressions with zero intercept for each treatment. Legends within graphs show regression slopes (a), percentage of the variance explained by the model (R2) and significance of the relationship (P). Regression slopes were significantly smaller in warming (W) and warming and rainfall reduction (W + RR) plants than in control and rainfall reduction (RR) plants (slope test likelihood ratio (LR) = 27.54, P < 0.001 and LR = 28.05, P < 0.001 for A : gs and A : E, respectively).

Leaf δ18O was lower in W, RR and W + RR plants than in control plants (Fig. 8; Table S8), which is consistent with their greater utilization of deeper water sources with more negative δ18O signature (as leaf δ18O correlated with xylem water δ18O across treatments; r = 0.31, P = 0.09). Moreover, leaf Δ18O was also decreased by warming (Fig. 8; Table S8), further indicating higher time-integrated stomatal conductance and cumulative transpiration over the growing season in warmed plants. Leaf δ13C showed a 1.1‰ lower average value in W + RR plants than in control plants (Fig. 8; Table S8). Foliar δ13C was positively correlated with leaf P concentration (Fig. S5), suggesting that greater P limitation was associated to lower time-integrated WUEi.

Details are in the caption following the image
(a–c) Leaf oxygen isotopic composition (δ18O, ‰) (a), oxygen isotopic enrichment above source water (Δ18O, ‰) (b) and leaf carbon isotopic composition (δ13C, ‰) (c) of Helianthemum squamatum shrubs growing under ambient conditions in control plots (C) and under simulated climate change conditions (W, warming; RR, rainfall reduction; W + RR, warming and rainfall reduction). Data are means ± SE. Asterisks indicate significant effects of the experimental climate change factors (*, P < 0.05; **, P < 0.01; ns, nonsignificant).

Shoot biomass production was reduced by both warming and rainfall reduction (by 21–27%; Tables 1, S4). Post-summer drought survival was also reduced by both warming and rainfall reduction (χ2 = 3.805, P = 0.05 for W and χ2 = 4.081, P = 0.04 for RR), with effects mainly driven by the strong negative impact of the W + RR treatment (33% lower survival than in control plots; Fig. S6).

Across treatments, xylem water δ2H and δ18O values correlated positively with leaf nutrient status (scores on PCAaxis1; Fig. 9). In particular, xylem water δ18O was positively associated with leaf P concentration (r = 0.45, P = 0.013) and strongly negatively associated with leaf C : P ratios (r = 0.63, P < 0.001; Fig. 9), indicating a lower uptake of limiting P from the fertile topsoil layer with increasing reliance on deep subsoil/bedrock water. Moreover, shrubs using a greater proportion of deep subsoil/bedrock water (higher xylem water d-excess values) also had lower WUEi and leaf δ13C and less shoot biomass production (Fig. 10). Xylem water d-excess correlated positively with leaf Ca and S concentrations (P = 0.024 and P = 0.059), indicating greater accumulation of these excess nutrients in plants that were using greater proportions of water from subsoil/bedrock layers.

Details are in the caption following the image
(a–c) Relationship between hydrogen (δ2H) (a) and oxygen (δ18O) (b) isotopic composition of xylem water and leaf nutrient status (axis 1 of the leaf nutrient principal component analysis), and between oxygen (δ18O) isotopic composition of xylem sap and leaf carbon : phosphorus (C : N) ratios (c) of Helianthemum squamatum shrubs growing under ambient conditions in control plots and under simulated climate change conditions (green, control; blue, rainfall reduction; orange, warming; pink, warming + rainfall reduction). Pearson correlation coefficients (r), significance of the correlations (P-value) and the model-predicted relationships (solid line) with 95% confidence intervals (dashed lines) are shown.
Details are in the caption following the image
(a–c) Relationship between deuterium-excess values (d-excess) in xylem water and intrinsic water-use efficiency (WUEi) (a), leaf δ13C isotopic composition (b) and shoot biomass per unit length (c) of Helianthemum squamatum shrubs growing under ambient conditions in control plots or under simulated climate change conditions (green, control; blue, rainfall reduction; orange, warming; pink, warming + rainfall reduction). More negative d-excess values indicate greater proportional uptake of evaporatively enriched water from the topsoil layer. Pearson correlation coefficients (r), significance of the correlations (P-value) and the model-predicted relationships (solid line) with 95% confidence intervals (dashed lines) are shown.

Evidence of nutrient limitation of plant photosynthesis and WUE

Across climate treatments, photosynthesis correlated negatively with foliar C : P and C : N ratios (r = −0.45, P = 0.010 and r = −0.36, P = 0.043, respectively), which supports significant nutrient limitation of plant carbon assimilation in this low-fertility dryland ecosystem. Stepwise multiple regression showed that foliar nutrient status (PCAaxis1) and stomatal conductance (gs) jointly explained 52% of interplant variability in photosynthesis. Whereas gs explained 32% of inter-plant variability in A, PCAaxis1 explained an additional 20% of the variability, thereby revealing that carbon assimilation was colimited by both water and nutrients at the peak of the growing season. WUEi and WUEt were also positively correlated with foliar N concentrations (r = 0.42, P = 0.018 and r = 0.38, P = 0.032, respectively) and foliar nutrient status (PCAaxis1 r = 0.59, P < 0.001 and 0.62, P < 0.001, respectively) and negatively correlated with leaf C : N ratios (r = −0.37, P = 0.043 and r = −0.34, P = 0.031, respectively). Overall, plants with lower leaf nutrient concentrations showed poorer carbon assimilation capacity and less efficient use of water.

Discussion

Changes in plant water use under climate warming and drying conditions

Control plants used a high proportion of water extracted from the topsoil layer, a pattern consistent with those of other dryland species (Ryel et al., 2010; Prieto et al., 2014; Prieto & Ryel, 2014; Barbeta et al., 2015). However, after long-term exposure to simulated climate change conditions (6 yr), shrubs shifted to greater water extraction from deeper subsoil/bedrock layers, with a decreased contribution of water from topsoil. Enhanced evapotranspiration rate in the W and W + RR treatments leads to faster depletion of topsoil moisture after rainfall pulses and thus lower mean annual topsoil moisture contents (León-Sánchez et al., 2018, 2020). Lower topsoil moisture availability was thus likely the key driver of the observed shift to deeper water uptake, a pattern that was more evident under warming (W and W + RR) than under rainfall reduction alone (RR) (Figs 4 and 5, Sarris et al., 2013). Drier topsoil conditions may have reduced the production and/or activity of fine roots and caused more frequent functional impairment or mortality of roots and mycorrhizal fungi during prolonged rainless periods, thereby hampering nutrient uptake from this fertile topsoil layer (León-Sánchez et al., 2018). However, the higher transpiration and lower leaf Δ18O of warmed plants revealed a plastic response that suggests greater root allocation and/or water uptake activity in deeper subsoil/bedrock layers to compensate for decreased moisture availability in topsoil (Schenk & Jackson, 2002b; Ansley et al., 2007).

Increased reliance on deeper and wetter subsoil layers may be needed in warmed shrubs to sustain higher transpiration rates aimed at meeting the higher evapotranspiration demands of warmer air with higher VPD (Will et al., 2013; Urban et al., 2017) and reducing leaf overheating (as maximum summer temperatures reached > 45°C within the OTCs). Warmed plants had a sparser canopy with fewer leaves per unit shoot length, which probably facilitated the concentration of the plant’s transpiration flux in fewer leaves (León-Sánchez et al., 2018). Moreover, the nutrient-deficient W and W + RR plants may have further increased transpiration rate to enhance the mass flow of soil nutrients to roots (Cramer et al., 2008). However, these plastic responses failed to fully compensate for the warming-induced desiccation of the fertile topsoil layer, as warmed shrubs still exhibited reduced leaf nutrient concentrations, carbon assimilation and shoot biomass production despite increased gs and E (Tables 1, 2).

Xylem water d-excess was negatively correlated with leaf P concentration and with scores in PCAaxis1, thereby confirming that greater utilization of less evaporated water sources stored in deeper and less fertile subsoil/bedrock layers was linked to a poorer leaf nutrient status. As hypothesized, climate change-induced shifts in water uptake depth in W and W + RR shrubs reduced their carbon assimilation rates and WUE through increased nutrient limitation of photosynthesis. Low nutrient concentrations in leaves are often linked to nutrient limitation of photosynthesis and poor A and WUE, which generally translates into lower plant biomass production (Reich et al., 1989; León-Sánchez et al., 2016, 2018, 2020; Salazar-Tortosa et al., 2018). The reductions in leaf nutrient concentrations, photosynthesis, WUE and growth found in W and W + RR plants in May 2017 were highly representative of the plant responses observed during the previous 5 yr period (León-Sánchez et al., 2018), which broadens the temporal scope of our findings.

Interestingly, the RR treatment under ambient temperature exerted few negative impacts on foliar nutrient concentrations. This unexpected finding suggests a specific role of topsoil warming ‘per se’ in reducing the abundance and/or nutrient uptake activity of fine roots and mycorrhizal fungi within upper soil layers in the W and W + RR treatments (Lansac et al., 1995; Kasai et al., 2000; Augé et al., 2001; Wilson et al., 2016; León-Sánchez et al., 2018). However, the RR treatment still reduced total plant nutrient uptake and contents through inhibition of above-ground biomass growth relative to the control treatment (Tables 1, S6). The high foliar nutrient concentrations found in the RR treatment may reflect decreased nutrient dilution effects linked to drought-stress inhibition of shoot growth (Skirycz & Inzé, 2010; Gao et al., 2019).

The crystallization water contained in gypsum mineral structure can become a relevant source of water for plants during dry and hot periods (Palacio et al., 2014, 2017). Extraction of water from soil/subsoil samples at 100°C can lead to partial gypsum dehydration (which occurs at temperatures > 50°C) and thus contamination of free soil/subsoil water with released crystallization water, thereby affecting its isotopic composition and leading to overestimation of free water content in soil/subsoil samples (Figs 2, 4). However, the d-excess values of xylem water in H. squamatum shrubs were much higher (−14‰ to −18‰; Fig. 4a) than the d-excess values typically found in gypsum crystallization water in the Iberian Peninsula (−60‰ to −80‰; Palacio et al., 2014, 2017). This large disparity (see Fig. 3) suggests that crystallization water held in the mineral structure of gypsum was not a significant water source for plants in our study. Relatively high free water contents in subsoil/bedrock and mild ambient temperatures at the time of sampling (spring) may explain why gypsum crystallization water was a negligible source of water for H. squamatum in our study.

Shoot biomass production and survival

The higher transpiration observed in W and W + RR plants relative to control plants at the peak of the growing season suggest that water storage in the whole rooting zone was not the most limiting resource for biomass production under warming despite a drier topsoil. The contrasting water-use patterns observed in the different climate treatments can be interpreted in light of the two-water-pool model proposed by Ryel et al. (2008, 2010). This model proposes the existence of two distinct water pools in soil/bedrock profiles: a shallow soil water pool (or ‘growth pool’) that is critically important for biomass production, and a deeper subsoil/bedrock water pool (or ‘maintenance pool’) that is mainly used during rainless periods when the upper soil is dry, but that is less effective for supporting growth. We found that faster warming-induced depletion of the topsoil ‘growth water pool’ after rainfall pulses may strongly constrain the availability and uptake of nutrients necessary for plant photosynthesis and growth. These negative effects of warming on plant nutrient status and productivity may be further exacerbated by longer-term negative plant–soil feedbacks that operate under elevated temperature, including decreased nutrient resorption before leaf senescence and decreased litter decomposition and nutrient cycling (Prieto et al., 2019; Prieto & Querejeta, 2020). In contrast to the W and W + RR treatments, RR plants showed relatively high foliar nutrient concentrations, photosynthesis and WUEi. This suggests that their poor above-ground biomass production may have been the consequence of increased below-ground carbon allocation and partitioning to increase root : shoot ratios (Brunner et al., 2015; Eziz et al., 2017), and/or direct drought-stress inhibition of above-ground growth that reduces nutrient dilution effects (Skirycz & Inzé, 2010). Slightly lower leaf δ13C and Δ18O values in RR plants than in control plants despite few differences between them in A, gs or E could be a result of accumulation of 13C- and 18O-depleted osmolites in RR plants in response to enhanced drought stress. Post-photosynthetic carbon isotopic fractionation and increased export of 13C-enriched sugars to roots may have contributed further to lower leaf δ13C in RR plants (Badeck et al., 2005).

Based on our findings, we propose a conceptual model of the long-term impacts of warming-induced topsoil desiccation on vegetation (Fig. 11). Faster surface soil drying after rainfall pulses with warming leads to chronically lower moisture contents in the fertile topsoil layer (i.e. ‘growth water pool’) and thus lower nutrient mineralization, mobility and availability to roots. As an adaptive plastic response to a drier topsoil, warmed plants shift to greater utilization of deeper water sources, although the much lower nutrient availability in subsoil/bedrock layers (i.e. ‘maintenance water pool’) eventually leads to lower cumulative nutrient uptake, which in turn exacerbates nutrient limitation of photosynthesis and impairs stomatal regulation. The combination of lower A with enhanced gs and E causes a drastic reduction in WUEi and WUEt that may lead to faster depletion of moisture storage within the entire rooting zone, which further contributes to reduce plant productivity through a shortening of the growing season. Moreover, when rainfall reduction aggravates the impacts of increased evapotranspiration and topsoil desiccation with warming, plants may become more vulnerable to mortality caused by hotter droughts, as observed in W + RR shrubs (Adams et al., 2009; Anderegg et al., 2012; Will et al., 2013). The combination of smaller rainfall inputs and higher evapotranspiration demand in W + RR plots may lead to lower recharge and earlier consumption and exhaustion of subsoil/bedrock water storage (‘maintenance water pool’), which might explain the higher drought mortality observed in this treatment during the summer when the topsoil is dry (León-Sánchez et al., 2018). This finding suggests that sufficient water storage in the whole rooting zone is critical for plant survival during the summer drought period.

Details are in the caption following the image
Conceptual model of the long-term impacts of climate warming-induced topsoil desiccation on plant nutrient status, gas exchange, growth and survival. Superscripts denote observed effects of warming and/or rainfall reduction in Prieto & Querejeta (2020)1, Prieto et al. (2019)2 and León-Sánchez et al. (2018)3. WUEi, intrinsic water-use efficiency.

The greater vertical decoupling of water and nutrient availability in soil/bedrock profiles under warming could decrease or offset the projected increases in plant photosynthesis and WUE in response to rising atmospheric CO2 concentration (Swann et al., 2016; Peñuelas et al., 2017, 2020; Marqués-Gálvez et al., 2020) through increases in nutritional limitation of carbon assimilation. Negative impacts could be expected even in nondryland ecosystems with irregular rainfall distribution and frequent occurrence of long rainless periods, in which warming-induced increases in evapotranspiration and accelerated topsoil drying (Berg et al., 2017) could also lead to longer and more frequent periods of spatial decoupling between topsoil nutrients and subsoil/bedrock water availability. We predict that shallow-rooted plants growing on thin or weakly developed soils (where nutrient availability peaks at the surface but declines sharply with depth) will be more prone to spatial decoupling between nutrients and water under a warmer and drier climate (e.g. leptosols and regosols occupying 1655 and 260 million ha worldwide, respectively; IUSS WRB 2014). Conversely, plants growing on deep and fertile soils with a more homogeneous vertical distribution of nutrients are expected to be less vulnerable to warming-induced topsoil drying (Rivas-Ubach et al., 2016). Hydraulic lift (HL) could buffer the impacts of warming-induced topsoil desiccation through nocturnal soil remoistening that fosters nutrient mineralization, solubilization and foraging by roots (Aanderud & Richards, 2009; Armas et al., 2012; Prieto et al., 2012a,b), and/or through direct transfer of HL water to mycorrhizal fungi to sustain their functionality in dry upper soil (Querejeta et al., 2003, 2007a; Egerton-Warburton et al., 2008).

In conclusion, this study provides some insights into how a dryland shrub species is responding to 6 yr of warming-induced reductions in water availability within upper soil layers under simulated climate change. A drier fertile topsoil under warming forces a greater proportional utilization of water from deeper, wetter, but less fertile subsoil/bedrock layers. This leads to decreased cumulative uptake of nutrients and increased nutrient limitation of photosynthesis and growth along with impaired WUEi, thereby amplifying and aggravating the negative impacts of warmer and drier conditions on plant performance. This study highlights how even a moderate reduction in the proportion of water taken up from the fertile upper soil layers under warming can have disproportionately strong negative impacts on plant nutrition and photosynthesis that are in part independent of impacts on plant water relations. A shift to greater utilization of deeper subsoil/bedrock water sources may buffer the potentially negative impact of warming-induced topsoil drying on stomatal conductance and cumulative transpiration, but this strategy cannot compensate for the associated reduction in cumulative plant nutrient uptake and carbon assimilation over the growing season. Increased vertical decoupling between nutrients and water in soil/bedrock profiles may thus compromise the capacity of dryland plants to adjust plastically to a warmer and drier climate.

Acknowledgements

We thank María José Espinosa for her help with laboratory work and Emilio Nicolás for help with leaf gas exchange measurements. Funding was provided by the Spanish Ministerio de Economía y Competitividad (project CGL2013-48753-R) and Fundación Séneca (projects 19477/PI/14 and 20654/JLI/18), both co-funded by European Union FEDER funds. WR acknowledges financial support from the China Scholarship Council.

    Author contributions

    JIQ conceived and designed the experiment; JIQ, IP and WR performed the research and collected the data; JIQ and IP analysed and interpreted the data; and JIQ and IP wrote the manuscript with feedback from WR.