The resilience of perennial grasses under two climate scenarios is correlated with carbohydrate metabolism in meristems

Abstract

Extreme climatic events (ECEs) such as droughts and heat waves affect ecosystem functioning and species turnover. This study investigated the effect of elevated CO₂ on species’ resilience to ECEs. Monoliths of intact soil and their plant communities from an upland grassland were exposed to 2050 climate scenarios with or without an ECE under ambient (390 ppm) or elevated (520 ppm) CO₂. Ecophysiological traits of two perennial grasses (Dactylis glomerata and Holcus lanatus) were measured before, during, and after ECE. At similar soil water content, leaf elongation was greater under elevated CO₂ for both species. The resilience of D. glomerata increased under enhanced CO₂ (+60%) whereas H. lanatus mostly died during ECE. D. glomerata accumulated 30% more fructans, which were more highly polymerized, and 4-fold less sucrose than H. lanatus. The fructan concentration in leaf meristems was significantly increased under elevated CO₂. Their relative abundance changed during the ECE, resulting in a more polymerized assemblage in H. lanatus and a more depolymerized assemblage in D. glomerata. The ratio of low degree of polymerization fructans to sucrose in leaf meristems was the best predictor of resilience across species. This study underlines the role of carbohydrate metabolism and the species-dependent effect of elevated CO₂ on the resilience of grasses to ECE.

The contrasting resilience of two perennial grasses under future climate scenarios was associated with carbohydrate metabolism, such as fructan polymerization in leaf meristems, that could contribute to plant dehydration tolerance.

Introduction

Grasslands are the most important agro-ecosystems worldwide; they deliver many ecosystem services, ranging from forage supply to soil carbon storage and biodiversity preservation (Pilgrim et al., 2010; Gaujour et al., 2012). However, climate change caused by increased greenhouse gas emissions in the atmosphere may jeopardize the stability and functions of many ecosystems (Intergovernmental Panel on Climate Change, 2014). Reduced precipitation coupled with increased temperatures will lead to more frequent and intense droughts (Tubiello et al., 2007; Dai, 2011) and severe heat waves (Trnka et al., 2011; Orlowsky and Seneviratne, 2012; Seneviratne et al., 2014). The interactions between these combined climatic factors result in complex responses that challenge our current understanding and will affect plant physiological processes in a way that cannot be predicted from single-factor treatments (Albert et al., 2011a; Dieleman et al., 2012; Mueller et al., 2016). An important issue is to explore to what extent extreme climatic events (ECEs, as defined by Smith, 2011a) will interact with the predicted rise in CO₂, in particular, regarding their impact on plant mortality and therefore on overall ecosystem resilience (Way, 2013; Niu et al., 2014; Felton and Smith, 2017).

Many studies have explored the role of moderate water deficits on plant water status and carbon uptake under elevated CO₂ concentrations (eCO₂) (Clark et al., 1999; Korner, 2000; Volk et al., 2000; Knapp et al., 2001; Morgan et al., 2004; Grant et al., 2014). Generally, eCO₂ enhanced plant biomass production and improved water relations under drought (Clark et al., 1999; Ghannoum et al., 2000; Robredo et al., 2011). In addition, depending on the warming intensity (Dieleman et al., 2012), eCO₂ alleviated the drought response in a range of grassland species (Naudts et al., 2013), and mitigated the impacts of climate extremes on the growth and production of Arabidopsis thaliana (Zinta et al., 2014). Conversely, the stimulatory effects of eCO₂ on growth and photosynthesis may decline due to increases in temperature stimulating photorespiration (Ziska and Bunce, 1994) or due to water and nitrogen limitations (Reich et al., 2014). However, the effects of combined severe climatic extremes and CO₂ concentration on the ability of perennial herbaceous plants to survive and recover from severe stress have been explored relatively little (Roy et al., 2016).

The accumulation of water-soluble carbohydrates (WSCs) under eCO₂ may contribute to increases in both drought resistance, that is, the maintenance of leaf growth under moderate water deficit, and also drought survival under severe water deficit (Roy et al., 2016; Volaire, 2018). Under severe drought, once complete leaf senescence is reached, the dehydration tolerance of meristematic tissues of shoots and roots ensures plant survival by maintaining cell integrity through cell membrane stabilization (Volaire et al., 2014; Zwicke et al., 2015). WSCs, particularly fructans (soluble sucrose-derived fructose polymers), contribute to dehydration tolerance (Volaire, 1995; Livingston et al., 2009) through cellular protection by membrane stabilization (Hincha et al., 2007) and reactive oxygen species (ROS) scavenging (Peshev et al., 2013). These WSC reserves also play a role during recovery, since they are hydrolysed to fuel regrowth after rehydration (Amiard et al., 2004). Under eCO₂, temperate grassland species generally accumulated more WSCs, but a high variability of responses between species has been reported (Casella and Soussana, 1997; Allard et al., 2003; Dumont et al., 2015). Moreover, climate extremes under eCO₂ also elicited larger increases in the quantities of WSCs including fructans in grasses (AbdElgawad et al., 2014). Since the protective properties of WSCs differ among compounds, in terms of both membrane stabilization (Hincha et al., 2007) and ROS scavenging (Morelli et al., 2003), the biochemical composition of the WSC pool should be considered. Indeed, temperature conditions can modify not only the amount of WSCs but also their composition (Abeynayake et al., 2015), such as the distribution of fructan polymers [i.e. the relative content of low and high degree of polymerization (DP) fructans].

An experimental approach seems best suited to explore the impacts of combined environmental factors on plant responses (Reyer et al., 2013). At the Montpellier (France) Ecotron, eCO₂ was shown to enhance both the resistance and the short-term overall resilience of an upland grassland community submitted to an ECE consisting of a severe edaphic drought and air warming (Roy et al., 2016). Nevertheless, plant responses to eCO₂ are species-specific (Reich et al., 2001; Roumet et al., 2002; Teyssonneyre et al., 2002b; Maestre et al., 2007). Exploring the responses of dominant species to climate extremes will be key for predicting ‘winner’ and ‘loser’ species (Dukes, 2007) and therefore future ecosystem dynamics and function (Hoover et al., 2014; De Boeck et al., 2018). Temperature regimes interact with eCO₂ to affect plant physiology and growth according to plant functional groups; for instance, C₃ plants are more responsive than C₄ species (Wang et al., 2012). Water stress and eCO₂ also interact to affect in different ways the growth of grasses that have contrasting drought strategies (Fernandez et al., 2002; Wullschleger et al., 2002). The botanical composition of temperate grasslands was shown to be modified under eCO₂ (Teyssonneyre et al., 2002b), and changes in species abundance could result from differential mortality of different species under ECE (Grant et al., 2014). Overall, in combination with an ECE, eCO₂ may differentially affect species or plant functional groups (Morgan et al., 2011) and/or have a mitigating effect, particularly on drought survival (Volaire, 2018) and growth recovery after drought, as was shown at the community level (Roy et al., 2016).

This study was designed within an experiment conducted at the ecosystem level (Roy et al., 2016) to analyse the effect of eCO₂ on the resistance and resilience to an ECE of two coexisting C₃ grass species (Dactylis glomerata and Holcus lanatus). We investigated traits associated with nitrogen, carbon, and water use, at shoot level (biomass), leaf level (water potential, growth, nitrogen content), and surviving leaf meristem level (water status, membrane stability, non-structural carbohydrate metabolism). The chosen species are known to respond positively to eCO₂, but with higher drought resistance and survival for Dactylis (Teyssonneyre et al., 2002a, b). The extent of the intraspecific variability in response to an ECE and eCO₂ was therefore tested according to the following hypotheses: (i) eCO₂ increases resistance to ECE (leaf growth maintenance during ECE) through higher water and WSC content; (ii) eCO₂ improves survival and biomass recovery and therefore resilience after ECE; and (iii) WSC metabolism and fructan composition in leaf meristems play a central role to support cell protection in the most severe stages of ECE, and contribute to plant resilience.

Materials and methods

Experimental design

The experimental design was previously described by Roy et al. (2016). The study tested the species’ response to future climate scenarios projected for the period of the 2050s (Ciais et al., 2005) For the representative year 2045, the projected annual means for air temperature and precipitation at Saint-Genès-Champanelle, France, were 10.9 °C and 770 mm (corresponding to +2.3 °C and –33 mm, respectively, compared with the 1990–2009 means). Therefore, the control treatment consisted of warmer and drier climatic conditions than the long-term climatic conditions occurring at Saint-Genès-Champanelle. The atmospheric CO₂ concentration projected for the 2050s under the A2 scenario is 520 ppm compared with the ambient CO₂ concentration (aCO₂) of 390 ppm measured at the Mauna Loa Observatory (Hawaii) in 2010.

Forty-eight monoliths (1 m² each) of intact soil and their resident plant communities were excavated in June 2009, down to the bedrock at 0.6 m depth, from an extensively managed upland semi-natural grassland site (Redon, 45°43′N, 03°01′E, 800 m a.s.l.) near Saint-Genès-Champanelle in the French Massif Central. The average botanical composition of plant communities was initially dominated by C₃ perennial grasses (60%), legumes (35%), and forbs (5%). The grasses D. glomerata and H. lanatus were present in all monoliths and constituted 37% and 14%, respectively, of the initial above-ground community biomass. The soil in this site is a Cambisol (59.5% sand, 19.2% silt, 21.4% clay, pH H₂O 5.9).

The excavated monoliths were brought to Clermont-Ferrand INRA research station (45°46′N, 03°08′E, 350 m a.s.l., long-term 1980–2010 mean annual temperature 11.4 °C) for temperature acclimation until September 2009, where they received ambient precipitation and irrigation to maintain soil water content (SWC) near 80% of field capacity. They were then transported to Montpellier, and four monoliths representative of the species composition of the grassland were inserted in each of the 12 Ecotron macrocosms, where they acclimated, at aCO₂, from April 2010 to early March 2011 to the climatic conditions of the representative year 2045. The Ecotron climate-regulation system tracked the hourly means of air temperature, air humidity, and daily precipitation projected by the model.

From mid-March 2011 to the end of the experiment, six macrocosms randomly selected out of the 12 were subjected to eCO₂ (520 ppm) while the other six were subjected to aCO₂ (390 ppm) (Fig. 1). ECE included a reduction in precipitation by 50% during 4 weeks in midsummer (25 June to 21 July: date D1) followed by 2 weeks (22 July to 4 August: date D2) with no irrigation and a concomitant heat wave (+3.4 °C compared with the 2050s average) (Fig. 1). This increase in air temperature corresponded to 7.1 °C above the 2000–2009 average for the same period, a value above the average of the 14 consecutive hottest days of the exceptional heat wave in summer 2003. This ECE, mimicking the strongest of such events projected by the downscaled ARPEGEv4 model over 2040–2060, was applied to three randomly selected macrocosms out of the six at each CO₂ concentration. After the ECE, from 5 to 31 August, irrigation was progressively increased in the treatment with ECE to obtain the same cumulative precipitation as in the control treatment (Con) referring to the 2050s climatic conditions without ECE. After the ECE and until the end of the experiment on 3 November, the macrocosms subjected to each CO₂ treatment had the same climatic conditions, replicating the downscaled model projections for the representative year 2045. Each of the four experimental treatments (390 Con, 390 ECE, 520 Con, 520 ECE) was replicated three times. For further details on the experimental conditions, see Roy et al. (2016).

Fig. 1.

Experimental design and time course of the experiment. After a 1-year acclimation period, the vegetation was cut at 5 cm height (14 March) and the monoliths were submitted to four climatic scenarios: ambient (aCO₂: 390 ppm, blue) or elevated (eCO₂: 520 ppm, red) atmospheric CO₂ concentration, combined with an extreme climatic event (390 ECE and 520 ECE) or without ECE (390 Con and 520 Con). The ECE included a reduction in precipitation by 50% during 4 weeks in midsummer (25 June to 21 July: D1) followed by 2 weeks (22 July to 4 August: D2) with no irrigation and a concomitant heat wave (+3.4 °C compared with the projected 2050s average). Thereafter, the precipitation was gradually increased. The vegetation was cut at 5 cm height again on 26 April, 9 June, and 3 November. Leaf meristematic tissues were sampled at D1 (21 July) and D2 (4 August).

Plant community-level measurements

SWC was continuously measured at three soil depths (7, 20, and 50 cm) with TDR probes (IMKO, Ettlingen, Germany) and averaged across soil depths to assess relative SWC (RSWC) at each sampling date. The field capacity of the soil (SWC_fc) was 32.90% and the minimum soil moisture (SWC_min) at the end of the ECE, when all plants were senescent, was 8.16%. RSWC was calculated as the fraction of soil moisture available for plants with this equation: RSWC=(SWC–SWC_min)/(SWC_fc–SWC_min). The vegetation was cut at 5 cm height on 14 March, 26 April, 9 June, and 3 November, to mimic mown grassland. After harvest, samples of the last three cuts were frozen at –18 °C, then defrosted, sorted by species, and oven-dried at 60 °C for 48 h to determine specific level above-ground biomass. The March cut was excluded from the analyses as it represented biomass production from the previous autumn, before the onset of CO₂ treatment. For each species and each of the four treatments, resilience to ECE was estimated by the biomass resilience index (RESIL), calculated as the ratio of above-ground biomass harvested after the ECE (November cut) to above-ground biomass harvested before the ECE (mean of April and June cuts). A RESIL value of 1 corresponds to full recovery of species biomass. This index was chosen from among others (Ingrisch and Bahn, 2018) because the biomass data harvested in November integrate ECE resistance and recovery periods. This index allows comparison of the treatments because it is calculated for each replicate macrocosm.

Plant traits

Before the onset of ECE (21 June) to 3 weeks after the ECE (22 September), lamina length was measured twice a week on the fastest-growing leaves of four tillers per species (D. glomerata and H. lanatus) and per macrocosm to analyse the dynamics of leaf elongation rates. The mean leaf elongation rates measured during the ECE (LER_ECE) and the maximum leaf elongation rates measured during the recovery period (LERr) were analysed. Pre-dawn leaf water potential was measured on three detached green laminae from each species in each monolith on 17 June, 21 July (D1), and 4 August (D2) (Scholander pressure chamber, model 1000, PMS Instrument Company, Corvallis, OR, USA). On the same days, the relative water content (RWC) of these laminae was measured by weighing lamina tissue before and after rehydration overnight at 4 °C and after drying at 80 °C for 24 h.

Leaf and leaf meristem traits

For each species, leaves collected during the cuts on 26 April, 9 June, and 3 November and those used for water potential measurements were oven-dried at 60 °C for 48 h and then ball milled (MM200, Retsch, Germany). Leaf nitrogen content was analysed in the Isotopic platform of INRA Nancy (Isoprime100, IsoPrime, Manchester, UK). At dates D1 and D2, measurements were carried out on the leaf meristematic tissues (three tillers for each macrocosm), which are the organs that survive the longest in perennial grasses. From the fraction containing the lowest 20 mm of the tillers, the enclosed bases of immature leaf meristems were dissected out and split out into two subsamples. One was immediately weighed and then dried at 80 °C for 24 h to determine its water content. The other fresh subsample was used to measure electrolyte leakage (Volaire and Thomas, 1995) from the cells (seed analyser G2000, Wavefront Inc., Ann Arbor, MI, USA) and calculate the coefficient of membrane stability (CMS).

For the WSC analysis, on the same dates, three other subsamples of fresh leaf meristems per species and monolith were quickly weighed, dropped into liquid nitrogen and stored at –80 °C before freeze-drying at –100 °C for 48 h. Finely powdered samples (30–50 mg) were extracted in 80% ethanol and purified in mini-columns (Mobicols, MoBITec, Göttingen, Germany) with ion-exchange resins (Amiard et al., 2003). WSCs (mg g⁻¹) were analysed by HPLC on a cation exchange column (Sugar-PAK, 300×6.5 mm, Waters Corporation, Milford, MA, USA), eluted at 0.5 ml min⁻¹ and 85 °C with 0.1 mM Ca-EDTA in water, and quantified using a refractive index detector (2410 Differential Refractometer, Waters Corporation, Milford, MA, USA). External standards used to quantify carbohydrates were glucose, fructose, sucrose, and Cichorium intybus inulin (Sigma-Aldrich, MO, USA).

Carbohydrates in the assay mixture were analysed by high-performance anion exchange chromatography and pulsed amperometric detection (HPAEC-PAD ICS-3000, Dionex, CA, USA) equipped with a CarboPac PA1 anion-exchange column (4×250 mm) with elution at 1 ml min⁻¹ with 150 mM NaOH and an increasing concentration of sodium acetate: 0 mM (0–6 min), 50 mM (6–12 min), 100 mM (12–18 min), 175 mM (18–19 min), 250 mM (19–30 min), 425 mM (30–60 min). Fructan type (levan versus inulin) and DP were identified by comparison with standards of commercial fructans (1-kestotriose, 1,1-kestotetraose; Megazyme, Ireland), purified fructans (6-kestotriose; kind gifts of Professor I. Izuka, Hyogo, Japan) and inulin extracted from Taraxacum spp. Comparison of standards with D. glomerata and H. lanatus extracts revealed that both species accumulated β2,6 linked fructan (levans), in accordance with the literature (Chatterton et al., 1993; Bonnett et al., 1997). Owing to the lack of commercially available analytical standards for fructans, absolute quantification of each peak was not possible. Instead, the relative amounts of the same DP were compared for each sample. The peak area of each DP of β2,6 fructan polymer (using the major peak when multiple peaks were present) from DP3 to DP65 was used to calculate the relative area of each peak as a percentage of the sum of DP3 to DP65 peak areas. As the amperometric response declines with increased chain length, it was not possible to compare the relative area of fructans differing in DP. The relative area of each peak (or group of peaks) expressed as a percentage of total peak area was used to analyse the treatment effects for each peak (or range of DPs). For analysis of the correlations between traits across species (Supplementary Table S1 at JXB online), the area percentage was weighted by the total fructan content in each sample.

Statistical analysis

Statistical analyses were carried out with R software (RStudio Team, 2015). Data were transformed when necessary (log, inverse, square, power 4, 5, or 6, boxcox) before analysis of variance (ANOVA) with the ‘lmtest’ package to conform to the assumptions of normality (Shapiro–Wilk test), homogeneity of variances (Bartlett test), and independence (Durbin–Watson test) or normality (Shapiro–Wilk test) of residues. For data for which the assumptions of normality, homogeneity of variances, and/or normality of residues were not obtained, analyses were carried out with the non-parametric Kruskal–Wallis test. For each trait, the factors species (D. glomerata, H. lanatus), CO₂ concentration (390, 520 ppm), climatic condition (Con, ECE), and date (D1, D2), as well as all of their interactions, were explored. Relationships between cell membrane stability and WSC, and relationships between resilience and plant traits, were analysed with non-parametric Spearman correlation tests.

Results

Responses of above-ground biomass

For both species, above-ground biomass in the spring before the ECE was not affected by the CO₂ concentration and was approximately 2-fold greater for H. lanatus than for D. glomerata (P=0.007; Table 1, Fig. 2A, B). In the autumn after the ECE (recovery period), no difference in biomass production was found between the species; the biomass of both species was slightly higher under eCO₂ (P=0.077), with a significant interaction between species and ECE (P=0.01; Table 1, Fig. 2A, B). As a result, the RESIL for control plants of D. glomerata was around 1.1 (±0.6) in both CO₂ treatments, and equal to 1.4 (±0.7) and 1.7 (±0.8) after the ECE under aCO₂ and eCO₂, respectively (Fig. 2C). By contrast, the RESIL was close to zero after the ECE for H. lanatus but it was 2-fold higher (0.75±0.47) for the control plants under eCO₂ than under aCO₂ (0.34±0.16) (Fig. 2D).

Table 1.

Effect of species (D. glomerata, H. lanatus), CO₂ concentration (390, 520 ppm), climatic condition (Clim) (Con, ECE), date (D1, D2) and their interactions, based on ANOVA, on above-ground biomass (BIOM), resilience index (RESIL), mean leaf elongation rate during ECE (LER_ECE), maximum leaf elongation rate during the recovery period (LERr), leaf pre-dawn water potential (POT), leaf relative water content (RWC), and leaf nitrogen content (LNC)

Period	Variable	Species	CO2	Clim	Species× CO2	Species× Clim	CO2× Clim	Species× CO2×Clim	Date	Date× Species	Date× CO2	Date× Clim
Before ECE	BIOM	0.007	NS		NS
During ECE	LERECE	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	0.0037	NS
	POT	<0.0001	NS	<0.0001	NS	<0.0001	0.0188	NS	<0.0001	NS	NS	<0.0001
	RWC	0.03	NS	<0.0001	NS	NS	0.019	NS	<0.0001	NS	0.006	<0.0001
	LNC	<0.0001	<0.0001	NS	0.032	NS	NS	NS	NS	NS	NS	NS
Recovery after ECE	LERr	<0.0001	<0.0001	<0.0001	<0.0002	<0.0001	0.0042	<0.0001
	LNC	0.0101	0.0100	<0.0001	NS	NS	NS	NS
	BIOM	NS	0.077	NS	NS	0.010	NS	NS
	RESIL	0.0047	NS	NS	NS	NS	NS	NS

P values are shown when <0.05. NS, Not significant. Three periods are considered: before ECE, during ECE, and during recovery after ECE (see Fig. 1)

Fig. 2.

Above-ground biomass (A, B) and resilience index (C, D) of two perennial grass species, D. glomerata (A, C) and H. lanatus (B, D) under four climatic scenarios: 390 (blue) or 520 (red) ppm atmospheric CO₂ concentration, combined with the ECE or without ECE (Con). Above-ground biomass was the biomass harvested above 5 cm height before the ECE (B_ef; mean of two cuts on 26 April and 9 June) and after the ECE at the end of the recovery period (R_ecov; 3 November cut). For the recovery period, the absence of a bar for H. lanatus grown at 390 ECE coincides with mortality. The resilience index was calculated by the R_ecov to B_ef above-ground biomass ratio; A value of 1 corresponds to full recovery of species biomass. Mean ±SD are shown (n=3).

Responses at leaf level

During the ECE, the mean leaf elongation rate (LER_ECE) was significantly (P<0.0001) affected by all fixed factors (species, CO₂, ECE) and with significant interactions between factors (Table 1). Across all treatments, LER_ECE was 3- to 5-fold higher for D. glomerata than for H. lanatus, for which the variability of data was also greater (Fig. 3A, C). The pattern of LER_ECE response was similar for both species, with higher rates under eCO₂, a decline during the ECE, and a final cessation of elongation for all stress treatments under less than 20% of RSWC. At that time, all leaves were mostly senescent for all plants of the swards under stress (Roy et al., 2016). As the stress progressed, the leaf water potential declined down to –7 MPa in both species (Table 2). The leaf RWC was significantly higher in D. glomerata (P=0.03), although a CO₂×ECE interaction was significant for both leaf water potential and RWC (Table 1). Leaf nitrogen content was on average 20% higher in D. glomerata than in H. lanatus, and declined during the ECE for both species, but increased at the lowest RSWC, with a significantly lower nitrogen content in H. lanatus at eCO₂ (Supplementary Fig. S1).

Fig. 3.

Leaf elongation rate during the ECE (A, C) and during the recovery period after the ECE (B, D) as a function of the relative soil water content (RSWC) of two perennial grass species, D. glomerata (A, B) and H. lanatus (C, D). Blue and red symbols correspond to 390 ppm and 520 ppm atmospheric CO₂ concentration, respectively. Squares correspond to control treatments (390 Con and 520 Con), for which only mean values of the three periods are shown. Blue triangles and dotted blue lines correspond to ECE treatment at 390 ppm CO₂. Red triangles and solid red lines correspond to ECE treatment at 520 ppm CO_2. Mean ±SD are shown (n=3). For 390 Con and 520 Con, horizontal (RSWC) and vertical (LER) SD are shown.

Table 2.

Pre-dawn leaf water potential and lamina relative water content (RWC) of D. glomerata and H. lanatus under four climatic scenarios

	390 ppm				520 ppm
	Control	ECE			Control	ECE
	3 dates	Before ECE	D1	D2	3 dates	Before ECE	D1	D2
D. glomerata
Leaf water potential (–MPa)	0.9±0.3	1.1±0.3	3.3±0.9	6.7±0.6	0.9±0.2	0.9±0.3	3.1±0.2	7±0
Lamina RWC (%)	90.9±3.4	80.2±2.7	53.5±6.1	38.2±0.3	84.1±8.8	84.3±1.7	62.3±13.4	39.7±6.7
H. lanatus
Leaf water potential (–MPa)	1.1±0.3	1.3±0.02	3.6±0.6	7.00±0	1.2±0.1	1.4±0.4	2±0.1	7±0
Lamina RWC (%)	84.3±7.4	81.9±5.3	65.2±7.3	10.8±5.9	83.8±5.4	79.8±5.0	65.7±16.7	43.4±18.9

Climatic scenarios were two atmospheric CO₂ concentrations (390, 520 ppm) with or without an ECE at three dates (before ECE, 17 June; during ECE, D1; end of ECE, D2; see Fig. 1). For the control treatment, average values of the three dates are shown. Data are presented as mean ±SD (n=3)

During the recovery period, all treatments and interactions significantly affected the LERr (Table 1). Although some H. lanatus plants in the macrocosms recovered, none of the plants of this species sampled for LERr measurements recovered after the ECE (Fig. 3D). The LERr of D. glomerata reached that of control plants at aCO₂ for a RSWC greater than 50% (Fig. 3B). In addition, the recovery of D. glomerata was greatly enhanced under eCO₂, since its LERr reached up to 12 mm day⁻¹, that is, 2-fold more than the control plants (Fig. 3B). In parallel, leaf nitrogen content more than doubled during the recovery period for both species (Supplementary Fig. S1).

Responses at leaf meristem level

The water content and the CMS of meristematic tissues were not affected by CO₂ concentration but were significantly reduced in both species by the ECE treatment (Table 3, Fig. 4). In both species, fructans were the main WSCs in leaf meristematic tissues (Fig. 4, Supplementary Fig. S2). Levels of fructans, sucrose, and glucose, but not fructose, significantly differed between the species (Table 3). Overall, fructan content was ~30% higher in D. glomerata (mean 526 mg g DM⁻¹) than in H. lanatus (mean 357 mg g DM⁻¹), while sucrose content was 3- to 5-fold lower in D. glomerata (mean 16 mg g DM⁻¹) than in H. lanatus (mean 53 mg g DM⁻¹) (Table 3, Fig. 4). Fructan content was higher under eCO₂ (Table 3, Fig. 4E, F). Moreover, fructan content increased between D1 and D2 in ECE-treated and control plants of D. glomerata (P=0.027; t-test) and H. lanatus (P=0.026; t-test) (Fig. 4E, F). Sucrose content was affected by the ECE, with interactions with CO₂ concentration and date (Table 3, Fig. 4G, H). The ECE increased sucrose content at D2, with a bigger effect under aCO₂ than under eCO₂ (Fig. 4G, H).

Table 3.

Effect of species, CO₂ concentration (390, 520 ppm), climatic condition (Clim) (Con, ECE), date (D1, D2), and their interactions (based on ANOVA) on water content, cell membrane stability (CMS), soluble carbohydrate content, total fructans:sucrose ratio, and low DP:sucrose ratio in leaf meristems of D. glomerata and H. lanatus

Variable		Species	CO2	Clim	Species× CO2	Species× Clim	CO2× Clim	Species× CO2xClim	Date	Date× Species	Date× CO2	Date× Clim
	Water content	0.004	NS	<0.001	NS	NS	0.050	NS	<0.001	NS	NS	<0.001
	CMS	NS	NS	<0.001	NS	0.009	NS	NS	0.003	NS	NS	<0.001
Soluble carbohydrate content (mg g DM−1)	Fructans	<0.001	0.042	NS	NS	NS	NS	NS	<0.001	NS	NS	NS
	Sucrose	<0.001	NS	<0.001	NS	NS	0.038	NS	NS	NS	NS	<0.001
	Glucose	0.027	NS	NS	NS	<0.001	NS	NS	0.004	NS	NS	NS
	Fructose	NS	NS	NS	NS	0.004	NS	NS	<0.001	NS	NS	NS
Fructan composition (relative content, %)	Low DP	<0.001	NS	NS	NS	0.004	NS	NS	NS	NS	0.018	NS
	Medium DP	NR	NS	NS	NR	NR	NS	NR	<0.001	NR	NS	NS
	High DP	NR	NS	NS	NR	NR	NS	NR	0.012	NR	NS	NS
Fructans:sucrose ratios	Total fructans:sucrose	<0.001	NS	<0.001	NS	NS	NS	NS	NS	NS	NS	<0.001
	Low DP:sucrose	0.001	NS	0.002	–	–	–	–	NS	–	–	–
	Medium DP:sucrose	NR	NS	<0.001	NR	NR	NS	NR	NS	NR	NS	<0.001
	High DP:sucrose	NR	NS	NS	NR	NR	NS	NR	NS	NR	NS	0.004
Fructans:fructans ratios	Low DP:Medium DP	NR	NS	NS	NR	NR	–	NR	NS	NR	–	–
	Low DP:High DP	NR	NS	NS	NR	NR	NS	NR	0.0037	NR	0.019	NS
	Medium DP:High DP	NR	NS	NS	NR	NR	–	NR	NS	NR	–	–

For medium and high DP composition, medium DP:sucrose, high DP:sucrose and fructans:fructans ratios, the effect of species was not relevant (NR) since groups of medium and high DP were not similar in the two species. For the ratios of low DP:sucrose; low DP:medium DP, and medium DP:high DP, the effects of CO₂, Clim, and date were analysed independently by non-parametric Kruskal–Wallis tests. P values are shown when <0.05. NS, Non-significant effect. –, Conditions for analysis not met.

Fig. 4.

Water content (A, B), cell membrane stability (C, D), fructan content (E, F), and sucrose content (G, H) in leaf meristems of two perennial grass species, D. glomerata (A, C, E, G) and H. lanatus (B, D, F, H) under four climatic scenarios: 390 (blue) or 520 (red) ppm atmospheric CO₂ concentration, combined with the ECE treatment (ECE) or without ECE (Con). Leaf meristematic tissues of the two species were sampled at D1 (21 July) and D2 (4 August). Mean ±SD are shown (n=3).

The analysis of fructan composition (Fig. 5) showed that fructans consisted of a mixture of oligo- and polysaccharides of mainly β(2,6) fructans (levans) in both species. In D. glomerata, fructans up to DP65 were identified, while in H. lanatus the highest detected DP was only 45. The pattern of fructan DP distribution varied between D1 and D2 and differed among species (Figs 6 and 7; Table 3). In D. glomerata, the relative abundance of DP9–24 increased while that of DP40–55 decreased (Fig. 6A, B), so that their ratio increased (Fig. 6C). Conversely, in H. lanatus, the relative abundance of DP9–12 declined while that of DP16–30 increased (Fig. 7A, B), so that their ratio decreased (Fig. 7C). Fructans were divided into three groups corresponding to low, medium, and high DP. For both species, low DP corresponds to the sum of DP3 and DP4. Medium DP corresponds to DP9–24 for D. glomerata and DP9–12 for H. lanatus. High DP corresponds to DP40–55 for D. glomerata and DP18–30 for H. lanatus. Other fructan DPs were not considered since they were not affected by the environmental factors (CO₂ concentration, climatic condition, date) (Figs 6 and 7). The relative abundance of the three DP groups of fructans was not affected by eCO₂ (Table 3, Supplementary Fig. S3).

Fig. 5.

Distribution of the fructans according to their degree of polymerization (DP). HPAEC-PAD chromatogram profiles of water-soluble carbohydrates in leaf meristems of two perennial grass species, D. glomerata (A) and H. lanatus (B). DP for levans [β(2,6) linked fructans] are indicated above the corresponding peak.

Fig. 6.

Relative abundance of fructans based on their degree of polymerization (DP) (A, B) and ratio of medium- to high-DP fructans (C) in leaf meristems of D. glomerata under four climatic scenarios: 390 (A) or 520 (B) ppm atmospheric CO₂ concentration combined with the ECE treatment (ECE) or without ECE (Con). The dashed boxes include the two groups of fructans with medium and high DP that were the most altered by the treatments. Leaf meristematic tissues were sampled at D1 (21 July) and D2 (4 August). Mean ±SD are shown (n=3).

Fig. 7.

Relative abundance of fructans based on their degree of polymerization (DP) (A, B) and ratio of medium- to high-DP fructans (C) in leaf meristems of H. lanatus under four climatic scenarios: 390 (A) or 520 (B) ppm atmospheric CO₂ concentration combined with the ECE treatment (ECE) or without ECE (Con). The dashed boxes include the two groups of fructans with medium and high DP that were the most altered by the treatments. Leaf meristematic tissues were sampled at D1 (21 July) and D2 (4 August). Mean and ±SD are shown (n=3).

For each species, correlations between CMS and carbohydrate data in leaf meristems measured at D1 and D2 were investigated (Table 4). In both species, CMS was not correlated to fructan content or fructan composition (relative content of low- and high-DP fructans). In H. lanatus, CMS was negatively correlated to relative content of medium-DP fructans. In both species, CMS was lowest when the sucrose content was highest (Fig. 8A, B), and CMS was negatively correlated to sucrose content in H. lanatus (Table 4). CMS was negatively correlated to hexose contents in D. glomerata, and positively correlated to hexose contents in H. lanatus (Table 4). In H. lanatus, CMS was correlated with the low-DP fructans to sucrose ratio and with sucrose content (r=0.70 and –0.65, respectively; Table 4, Fig. 8D).

Table 4.

Spearman correlation coefficients of soluble carbohydrate traits against cell membrane stability (CMS) measured in leaf meristems of D. glomerata and H. lanatus at D1 and D2 (n=24 for each species)

Plant traits		Correlation with CMS
		D. glomerata	H. lanatus
Soluble carbohydrate content (mg g DM−1)	Fructans	–0.34 NS	0.27 NS
	Sucrose	–0.33 NS	–0.65***
	Glucose	–0.56**	0.56**
	Fructose	–0.49*	0.46*
Fructan composition (relative content, %)	Low DP	0.01 NS	0.31 NS
	Medium DP	–0.04 NS	–0.42*
	High DP	0.05 NS	0.19 NS
Fructans:sucrose ratios	Total fructans:sucrose	0.11 NS	0.64***
	Low DP:sucrose	0.36 NS	0.70***
	Medium DP:sucrose	0.37 NS	–0.64***
	High DP:sucrose	0.34 NS	0.57**
Fructans:fructans ratios	Low DP:Medium DP	0.01 NS	0.54**
	Low DP:High DP	0.13 NS	0.17 NS
	Medium DP:High DP	–0.03 NS	–0.27 NS

*P<0.05, **P<0.01, *** P<0.001; high significant Spearman correlations (P<0.001) are in bold. NS, Not significant.

Fig. 8.

Relationship between cell membrane stability and sucrose content (A, B) or the low-DP fructans to sucrose ratio (C, D) in D. glomerata (A, C) and H. lanatus (B, D). Leaf meristematic tissues of the two species were sampled at D1 (21 July) and D2 (4 August). Dashed lines indicate a linear relationship (in B) or a logarithmic relationship (in D) for a significant Spearman correlation.

For the analysis of correlations across species between resilience to the ECE and plant traits, the traits related to medium- and high-DP fructans were not considered, since the DP groups differed according to the species (Figs 6 and 7). The best predictors of resilience were the fructans to sucrose ratios, especially the low-DP fructans to sucrose ratio in leaf meristems at D1 and D2 (Supplementary Table S1, Fig. 9).

Fig. 9.

Relationship between the resilience index of above-ground biomass of D. glomerata and H. lanatus and the ratio of low-DP fructans to sucrose level in leaf meristems measured at D1 (21 July). The dashed line indicates a logarithmic relationship for a significant Spearman correlation (n=2).

Discussion

This study explored the response of two coexisting perennial grass species growing in a community of a temperate semi-natural grassland grown under the predicted climatic conditions of the 2050s, subjected or not subjected to an ECE including extreme drought and a heat wave, under ambient or elevated CO₂ concentrations. The resistance during the ECE and recovery after the ECE were high for D. glomerata and lower for H. lanatus. These results confirm that most native grasslands are likely to contain plants with a high diversity of drought resistance (Craine et al., 2013) and also drought survival after severe water deficit (Pérez-Ramos et al., 2013; Zwicke et al., 2015). Contrary to our hypothesis, we found no significant mitigating effect of eCO₂ on the mortality of H. lanatus. However, eCO₂ had a positive effect on both resistance and recovery of D. glomerata. These results are discussed in light of the water, nitrogen, and carbon status response of the plants to the treatments.

Elevated CO₂ increases the drought resistance of H. lanatus and D. glomerata during moderate drought

Both grass species had higher leaf growth and leaf RWC under a combination of eCO₂ and moderate drought during the first stage of ECE. As shown at community level, these positive effects should have slowed leaf senescence and maintained photosynthesis for a longer period during drought than under aCO₂ (Roy et al., 2016), leading to the higher content of fructans in leaf meristems. This confirms the importance of improved water relations under eCO₂, which to some degree sustain photosynthesis in dry periods (Albert et al., 2011b). In addition, as leaf N content was significantly higher in D. glomerata than in H. lanatus, it may be hypothesized that the photosynthetic activity of D. glomerata is higher than that of H. lanatus, especially under eCO₂ (CO₂×species interaction: Table 1; Harmens et al., 2000). Altogether, our results underline and confirm that eCO₂ alleviates the effects of drought stress by conservation of water (Morgan et al., 2001, 2011; Robredo et al., 2007; Holtum and Winter, 2010), higher carbon fixation, and higher fructan accumulation; thus, eCO₂ contributed to increased drought resistance. Previous studies have shown that the level of WSC accumulation in response to eCO₂ largely depended on N availability in the leaves (Dumont et al., 2015). Here, the relatively low WSC accumulation under eCO₂ may indicate that N availability was high enough to sustain leaf growth (Fig. 3), root growth (Roy et al., 2016), and root exudation (Pendall et al., 2004), which limited storage of WSCs in both species.

H. lanatus and D. glomerata show contrasting ECE survival

Although H. lanatus and D. glomerata are both C₃ perennial grasses co-occurring in temperate grassland with similar rooting depth (Pages and Picon-Cochard, 2014), they showed contrasting survival after the ECE. It was previously shown that although species with different growth habits varied in their responses to CO₂ and nitrogen, there was also substantial variation in responses among species within groups (Reich et al., 2001). In the present study, almost all H. lanatus plants died, whereas all D. glomerata survived. This stark contrast between a ‘loser’ and a ‘winner’ species (Dukes, 2007) strongly affected the botanical composition of the grassland macrocosms following the treatments (data not shown). As the grassland macrocosms had a realistic soil depth of 0.6 m, it can be assumed that both species had a rooting depth able to explore the full soil profile and therefore that their potential water uptake was comparable under drought. It has been suggested that long-term responsiveness to rising CO₂ concentrations differs between slow- and fast-growing plants (Ali et al., 2013). Our results also tend to suggest that growth patterns, with a much greater growth potential for D. glomerata than H. lanatus, may be associated with greater ECE survival, irrespective of the treatments. Although above-ground growth of both species was maintained for a similar period in the summer, the water status of leaves and meristems of D. glomerata was higher at some dates, especially under eCO₂. Altogether, the species with a greater acquisition strategy and higher growth rate (D. glomerata) survived better than the species with a more resource-conservative strategy (H. lanatus). Assuming that this pattern for temperate ecotypes of perennial grasses can be generalized, it differs from the pattern found for Mediterranean ecotypes of perennial grasses such as D. glomerata, for which a reduction or cessation of leaf growth (summer dormancy) are the most efficient strategies to conserve soil water resources and therefore to survive severe drought (Volaire, 1995; Volaire and Norton, 2006).

Since the leaves were mostly senescent at the end of the ECE for all species present in the community (Roy et al., 2016), plant survival relied on maintenance of the viability of leaf meristems (Volaire, 1995; Volaire et al., 2014). Our study found major differences between the tested species in WSC accumulation in meristematic tissues of leaves. The stress-tolerant species D. glomerata accumulated a larger amount of fructans than the sensitive species H. lanatus, which in turn accumulated a larger amount of sucrose. These differences in WSC level and composition might be associated with the two species’ differential survival and recovery after the ECE. In particular, the level of sucrose, which accumulated to a much greater extent in the sensitive species, was highly and negatively correlated to the maintenance of cell stability of the meristems at the end of the ECE, confirming previous results on temperate ecotypes of herbaceous species (Zwicke et al., 2015). The higher degree of polymerization of fructans in D. glomerata (up to DP65) could contribute to its greater survival compared with H. lanatus, which accumulated smaller fructans (up to DP45). Similar differences have been observed among the Asteraceae family, in which drought-resistant species (Echinops ritro and Vigueira discolor) accumulate fructans of higher DP (30–100) (Itaya et al., 1997; Vergauwen et al., 2003) than species with lower drought resistance (Cichorium intybus, Helianthus tuberosus; Van den Ende et al., 1996; Marx et al., 1997). It has been observed in vitro that highly polymerized fructans (from bacteria) had a greater protective effect against water stress than lower DP fructans (from Cichorium) (Demel et al., 1998), and that positive synergistic effects on membrane stabilization were obtained with a mixture of DP<7 and DP>7 fructans (from oat or rye) compared with either DP<7 or DP>7 fructans alone (Hincha et al., 2007). In D. glomerata, the high range of DP (from DP3 to DP65) could lead to a better synergistic effect than in H. lanatus, with its smaller range of DP, and this could partly explain the difference in membrane stability between the two species. Fructans can act on membranes either directly, by inserting into the lipid headgroups (Livingston et al., 2009), or indirectly, by reducing lipid peroxidation through ROS scavenging (Peshev et al., 2013). In D. glomerata, the maintenance of CMS observed under ECE might be due to the accumulation of highly polymerized fructans.

According to current knowledge, levans are synthesized by sucrose:fructan 6-fructosyltransferase (6-SFT) and degraded by fructan 6-exohydrolase (6-FEH) (Vijn and Smeekens, 1999). 6-SFT catalyses the transfer of a fructosyl residue from sucrose to sucrose or from sucrose to fructans by forming a β(2,6) fructosyl-fructose linkage. 6-SFT allows both the initiation of levan synthesis and the elongation of the levan chain. In H. lanatus, the relative abundance of the different fructan polymers changed during the ECE, resulting in a more polymerized assemblage. Given that at the same time, total fructan content did not increase, the entry of carbon (in the form of the fructosyl residue from sucrose) into the fructan pool via 6-SFT activity for levan elongation was balanced by an exit of carbon via 6-FEH activity, which catalyses the release of fructose from fructans. This fructan trimming strategy may be associated with the low ECE survival of H. lanatus. In D. glomerata, the total fructan content increased during the ECE and the relative abundance of the different fructan polymers changed, resulting in a more depolymerized assemblage. This might be due to the fact that all newly synthesized fructans were of medium DP or that, concomitantly with fructan synthesis, the pre-existing high-DP fructans were depolymerized. Such a strategy combining fructan trimming and accumulation may have enhanced the ECE survival of D. glomerata. In plant species that accumulate inulin [fructan consisting of linear β(2,1) linked fructosyl units], fructan synthesis requires two enzymes, a sucrose:sucrose 1-fructosyltransferase for initiation and a fructan:fructan 1-fructosyltransferase for elongation (Van den Ende et al., 1996; Vijn and Smeekens, 1999). To date, proteins with fructan:fructan 6-fructosyltransferase (6-FFT) activity have not been described in levan-accumulating plant species. If 6-FFT exists in H. lanatus and D. glomerata, the results obtained would be interpreted differently. Indeed, contrary to 6-SFT, which catalyses both the initiation and elongation of fructans, 6-FFT is responsible only for fructan elongation without concomitant entry of carbon into the fructan pool, the fructosyl donor being fructan instead of sucrose. Consequently, changes in the relative abundance of fructans could be supported by the induction of 6-FFT activity in H. lanatus and by the increase of 6-SFT, 6-FFT, and/or 6-FEH activity in D. glomerata. Further research is needed to unravel levan metabolism and its regulation in levan-accumulating grass species.

In addition, the WSC composition was associated with resilience. Indeed, the ratio of low-DP fructans to sucrose present in the leaf meristems during the ECE was the best predictor of resilience. Altogether, the results showed that a fine-tuning of the relative content and composition of WSCs is crucial for ECE survival. As previously shown in wheat (Joudi et al., 2012; Zhang et al., 2015; Hou et al., 2018) and among carbohydrate-metabolizing enzymes, FEHs are key proteins for ECE survival, since they are involved in fructan size adjustment for cell protection during the ECE and fructan mobilization for carbon feeding of growing cells after the ECE. Comparison of a larger inter- and intra-specific range of populations of grasses should allow further exploration of the mechanistic relationships between drought resilience and fructan metabolism.

Elevated CO₂ enhances strong compensatory growth after stress for the species surviving ECE

During the ECE, the overall water use efficiency at the community level was greater under eCO₂ (Roy et al., 2016), while leaf extension rate in water-stressed plants of both species were also enhanced, as found previously (Casella and Soussana, 1997; Drake et al., 1997). In our study, and as hypothesized, the most striking result was the enhancement of growth recovery of D. glomerata after the ECE under eCO₂. Under aCO₂, no compensatory growth of this species was observed after the ECE. These results contradict previous studies showing compensatory growth after drought in herbaceous species at aCO₂ but not eCO₂ (Newton et al., 1996; Clark et al., 1999). This discrepancy may be due to the different nature and intensity of the stress to which plants were subjected, but also to the duration of acclimation to eCO₂. Compared with the moderate drought applied in previous studies, the severity of the ECE in the present study resulted in full above-ground senescence of all plants. It was only under eCO₂ that resource use was efficiently enhanced after the ECE. The rapid regrowth of D. glomerata during the recovery period might take advantage of the pool of accumulated WSCs in leaf meristematic tissues together with a putative enhancement of 6-FEH expression allowing the release of fructose through fructan breakdown. In addition, the strong recovery of D. glomerata could also be explained by the capacity of roots to take up nitrogen, as shown at the community level in previous studies (Roy et al., 2016; Carlsson et al., 2017) and reflected in our study by the high leaf nitrogen content in this species (Supplementary Fig. S1B). Moreover, this study shows the effects of eCO₂ and extreme stress over an entire growing season, but the long-term impact of the stimulated resilience identified for one species and for the entire plant community (Roy et al., 2016) should be tested on a longer-term time scale due to a possible lag effect of ECE on biomass and species composition (Lee et al., 2011; Zwicke et al., 2013).

Although the ability of species to survive drastic climate change was claimed to be generally greater than hitherto recognized (Hof et al., 2012), our results also highlight a strong inter-specific variability that should be further explored experimentally and taken into account in models. D. glomerata, although present at low abundance, responded in the same way as the whole plant community in the samples, whereas H. lanatus tended to disappear after the extreme climatic stress, hence modifying the overall species abundance, with potential long-term impacts on grassland properties linked to ecosystem services. Our results, showing strong effects of plant carbon, nitrogen, and water status, cast a light on the crucial and timely question of the mechanisms underlying plant resilience under extreme stress (Smith, 2011b; Nimmo et al., 2015).

Supplementary data

Supplementary data are available at JXB online.

Table S1. Correlations between plant traits and resilience.

Fig. S1. Leaf nitrogen content in two grasses during and after an extreme climatic event.

Fig. S2. Monosaccharide contents in grass meristems during an extreme climatic event.

Fig. S3. Relative abundance of fructans in grass meristems during an extreme climatic event.