Contrasting responses of steppe Stipa ssp. to warming and precipitation variability

Abstract Climate change, characterized by warming and precipitation variability, restricted the growth of plants in arid and semiarid areas, and various functional traits are impacted differently. Comparing responses of functional traits to warming and precipitation variability and determining critical water threshold of dominate steppe grasses from Inner Mongolia facilitates the identification and monitoring of water stress effects. A combination of warming (ambient temperature, +1.5°C and +2.0°C) and varying precipitation (−30%, −15%, ambient, +15%, and +30%) manipulation experiments were performed on four Stipa species (S. baicalensis, S. bungeana, S. grandis, and S. breviflora) from Inner Mongolia steppe. The results showed that the functional traits of the four grasses differed in their responses to precipitation, but they shared common sensitive traits (root/shoot ratio, R/S, and specific leaf area; SLA) under ambient temperature condition. Warming increased the response of the four grasses to changing precipitation, and these differences in functional traits resulted in changes to their total biomass, with leaf area, SLA, and R/S making the largest contributions. Critical water thresholds of the four grasses were identified, and warming led to their higher optimum precipitation requirements. The four steppe grasses were able to adapt better to mild drought (summer precipitation decreased by 12%–28%) when warming 1.5°C rather than 2.0°C. These results indicated that if the Paris Agreement to limit global warming to 1.5°C will be accomplished, this will increase the probability for sustained viability of the Stipa steppes in the next 50–100 years.


| INTRODUC TI ON
Global climate projection models suggest that spatial and temporal patterns of temperature and precipitation will become increasingly heterogeneous, with arid areas in the lower latitudes seeing more frequent and severe droughts. Warm and dry climates are increasingly prevalent across much of China (Tan et al., 2017) and have seriously affected the terrestrial ecosystem and food security (Lesk, Rowhani, & Ramankutty, 2016;Reichstein et al., 2013;Sun et al., 2017). Studies have revealed a greater frequency of droughts in the semiarid areas of northern China (Wu, Dijkstra, Koch, Peñuelas, & Hungate, 2011), which will become even more common in the future as a result of climate change (Battisti & Naylor, 2009;Ma, Zhou, Angélil, & Shiogama, 2017). Steppe has an important strategic position for ecological security in China, covering an area of about 4 × 108 ha, or about 41.67% of the territory of China. Steppes in Inner Mongolia account for nearly 67% of the total area of temperate steppe in China (Kang, Han, Zhang, & Sun, 2007) and are highly affected by climate change (Sui, Zhou, & Zhuang, 2013). Decreases in water availability have led to the migration of steppe vegetation and a reduction in steppe area, with additional changes to the structure and productivity of steppe plant communities (Wertin, Reed, & Belnap, 2015;Wilcox, Fischer, Muscha, Petersen, & Knapp, 2015).
Plant functional traits are the internal or external adaptive characters formed by the interaction of plants and environment (Salgado-Negret, Canessa, Valladares, Armesto, & Perez, 2015).
Previous studies have revealed that mild decreases in precipitation do not significantly affect plant growth; however, extreme drought significantly reduces plant height, leaf area, and photosynthetic capacity, negatively impacting the production capacity of the plants (Bret-Harte et al., 2001;Cowling et al., 2015;Munson et al., 2013;Phillips et al., 2009;Shi, Zhou, Jiang, Wang, & Xu, 2016;Xu et al., 2014). For the scientific management of plant resources in our changing climate, a detailed comparison of the responses of functional traits to precipitation variability in different steppe grasses is needed. This will help to guide the application of drought control and resistance methods in a timely manner (Nepstad et al., 2004). Biomass reflects the growth and resource utilization of plants and is an important research focus of steppe ecology (Manea, Sloane, & Leishman, 2016). Plants respond to reduced precipitation mainly by inhibiting gas exchange while concurrently adjusting their physiological characteristics, thereby altering plant functions such as net primary production or biomass (Brodribb, Bowman, Nichols, Delzon, & Burlett, 2010;Li, Zhao, & Liu, 2015; Shafran-Nathan, Svoray, & Perevolotsky, 2013). In addition, differences in the functional traits to precipitation variability mean that the final total biomass of a plant is not necessarily altered, which reflects the adaptive strategies of the plant species to changing environments (Brunner, Herzog, Dawes, Arend, & Sperisen, 2015;Chapin, Bloom, Field, & Waring, 1987). However, quantitatively assessing the contribution of various functional traits to biomass has drawn little attention to date.
In response to precipitation variability, plants are either adaptive or stressed; however, even resilient plants have a limited capability to adapt to reduced precipitation, and the growth and survival of plants are inhibited by prolonged or extreme changes in precipitation that exceed their adaptation and recovery abilities (Barkaoui, Navas, Roumet, Cruz, & Volaire, 2017;Sankaran et al., 2005). The determination of these limits confirms the drought threshold of plant and is essential for the application of drought control and resistance methods. Especially, the response range of the dominate plant species to precipitation variability is particularly representative of the adaptability of the natural ecosystem (Cavin, Mountford, Peterken, & Jump, 2013;Waddington et al., 2015). Identifying critical precipitation thresholds of biomass in dominant steppe species to precipitation variability is therefore important for the evaluation of the effects of climate change on these ecosystems (IPCC, 2014;Palumbi, Barshis, Traylor-Knowles, & Bay, 2014). A recent study revealed that a linear model was not applied to describe temporal precipitationaboveground net primary production (ANPP) relationships, but a nonlinear form of the temporal precipitation-ANPP relationship would better predict responses of ANPP to changing precipitation regimes because the precipitation variability and extreme precipitation were forecast to increase with climate change (Knapp, Ciais, & Smith, 2017). Leymus chinensis and S. grandis reached their peak biomass when the relative soil moisture was 66% and 54.7%, respectively, via a simulated experiment (Xu & Zhou, 2011); thus, the biomass decreased with decreasing precipitation and there existed the response threshold to drought. But warming strengthened the negative effects of drought to plant biomass (Hoover, Knapp, & Smith, 2016;Yang, Wang, Yang, & Guo, 2013). These studies showed that biomass responded differently to precipitation in different species and more focused on the responding extent of plants to precipitation variability. Few studies have reported the critical precipitation threshold of biomass at different warming levels.
Stipa L. dominates the zonal steppe communities of Inner Mongolia, presenting a regular zonal distribution from east to west in the steppes (Shi et al., 2016;. Among them, S. baicalensis dominates in semiarid and semihumid meadow steppes, S. bungeana grows in the semiarid warm temperate typical steppes, S. grandis grows in the semiarid typical steppe, and S. breviflora dominates in the arid and semiarid desert steppes (Qi et al., 2010). Here, these four representative Stipa species as subjects in climate simulation experiments under ambient and warming conditions with precipitation variability. The main objectives of this study were to identify difference in responses of functional traits and drought threshold to precipitation variability at different warming levels. Three hypotheses were tested as follows: (a) Among the functional traits, the growth strategy traits (root/shoot ratio, R/S) may be more sensitive to precipitation than the other traits and warming may improve the responses of those traits; (b) warming may lead to higher optimum water requirements for total biomass and critical water threshold may differ in the four Stipa species; (c) Stipa species may be able to adapt better to drought when temperatures were increased by 1.5°C rather than 2.0°C.
These results will help to identify and monitor the impacts of drought, and provide evidence for dealing with drought in Inner Mongolia steppe.

| Experiment design and environmental conditions
The manipulation experiment was conducted in the Institute of  Figure 1). All the seeds were sterilized with 5% potassium permanganate and subsequently washed with water before sowing. The polyethylene pots (10.9 cm in diameter, 9.5 cm in height, 0.71 L in total volume) were used as the experimental containers, which were wrapped with plastic film to prevent water leakage. Each pot was filled with approximately 0.61 kg of dry chestnut soil (organic carbon content 12.3 g/kg, total nitrogen content 1.45 g/kg, and soil field capacity 24.8%-26.8%), and 10 seeds were planted per pot. All pots were first placed in glasshouse (day/night temperature of 26-28/18-20°C, maximum photosynthetic photon flux density of 1,000 mol m -2 s -1 ) and wellwatered to complete the growth of the seedlings (Lv, Zhou, Wang, & Song, 2016;Song et al., 2016). Until the third leaf emergence (about 3 weeks after sowing), the seedlings were thinned to four plants per pot. Six replicates were used for the five precipitation and the three temperature treatments. Then, the 90 pots (four plants per pot) per Stipa plant were randomly transferred into three climate control chambers (RXZ-500D, The southeast instruments Inc.). Different temperature and water treatments were set based on the average monthly temperature (T0) and water (W0) during the four species' own growth stage in the past 30 years , which were shown in Tables 1 and 2. Three temperature treatments were set: (ambient temperature (T0), +1.5°C (T1), +2.0°C (T2). Five water treatments were set: average monthly water over 30 years (W0), the average decreased by 30% and 15% (W1 and W2), and the average F I G U R E 1 The site of seed collecting in the four Stipa species increased by 15% and 30% (W3 and W4). During the experiment, precipitation was added to the pots by a sprayer every 3 days, as described in our previous similar experiments (Shi et al., 2016;Xu et al., 2014).

| Soil water content measurement
The soil relative water content (SRWC, %), the ratio between the current soil moisture and the field capacity, was measured using the gravimetric method from 0 to 10 cm soil layer. The measurement was determined 90 days after the plants were subjected to a relative long-term soil water treatment. Six replicates were used for SWC determination. The SRWC is calculated as follows: where W c is the weight for empty soil pot and wet soil (g), W p is the empty soil pot weight (g), W d is the weight for empty soil pot and dry soil (g), and FC is the soil field capacity (%).

| Plant measurements
At the end of the experiment, all pots of each treatment were harvested to measure response characteristics (morphological characteristics and biomass). Leaf number and plant height were measured before harvest. Leaf area per plant was measured with a WinFOLIA system to measure the area of the blade part per plant (WinRHIZO, Regent Instruments). Plants were separated into three parts to harvest: leaf, stem, and root, and dried at 80°C to a constant weight, then weighted to get aboveground (leaf and stem) and belowground (root) biomass separately. Growth indices per plant were calculated as follows: specific leaf area (SLA; cm 2 /kg) = leaf area/leaf biomass; leaf area ratio (LAR; cm 2 /kg) = leaf area/plant total biomass; and root-shoot ratio (R/S) = belowground biomass/aboveground biomass.

| Statistical analysis
A method for assessing sensitivity was applied to study the differed responses of functional traits to precipitation variability in the four Stipa species. The response coefficient, which also could be called

T d /T n (°C) T m (°C) T d /T n (°C) T m (°C) T d /T n (°C) T m (°C)
S TA B L E 2 The ambient precipitation during the four Stipa species' own growth stage in the past 30 years  and the water regimes plasticity index, was defined as the absolute value of the relative change in a functional trait per unit of precipitation change compared with ambient precipitation level. A higher response coefficient (plasticity index) indicates a more response sensitivity to precipitation variability of this functional trait (Luers, 2005).
The tipping point of the four Stipa species responding to precipitation variability is defined as the very point when the total biomass starts to deviate significantly from those with sufficient water supply, resulting from a decline in precipitation amount below a critical level (Czajkowski, Ahrends, & Bolte, 2009;Thompson, Gallardo, Valdez, & Fernández, 2007). This method named as one-side lower tolerance limits for normal population was newly adopted to identify the tipping point of total biomass in the four species. The tolerance interval is an estimated interval within which at least a certain proportion β of the population falls at a given level of confidence γ; tolerance limit refers to either of the two endpoints of a tolerance interval (Krishnamoorthy & Mathew, 2009;Shi et al., 2016;Xu et al., 2009;Young, 2014).
The rigorous statistical definitions could be found in Young (2014).
Computational formulas of tolerance intervals and limits for different where x L is one-side lower tolerance limits for total biomass; x is the sample mean; s is the standard deviation of the sample variance; n is the sample size; x is the sample value; 1-β is the least proportion of sample population within the tolerance interval, defined as 95%; 1-γ is the confidence level, defined as 95%; and k is the tolerance coefficient, which could refer to ISO 16269-6 (2005) (Xu et al., 2009) or be calculated directly (Krishnamoorthy & Mathew, 2009).
The relationship between total biomass and precipitation under different temperatures was analyzed using a regression analysis. The tipping points of total biomass in the four species were further quantified by their values of critical water thresholds, which were calculated in terms of the tipping points and the quadratic polynomial regression models between total biomass and precipitation under ambient and warming temperature treatments for the four species (Equation 5).
where X is the observed values of the total biomass under different temperature treatment for the four species; Pre is the precipitation amount (mm); and a, b, and c are the fitting coefficients of the regression model.
All collected data were processed using the statistical software SPSS 17.0 (SPSS) and Origin 9.0 (Origin Lab). Two-way analyses of variance (ANOVAs) were used to examine the effects of warming and precipitation variability. The differences between the means among the precipitation treatments were compared using Duncan's multiple range tests. The effects of different response characteristics on the plant biomass were analyzed using principal component and pathway analyses. All statistical significances were denoted at p < .05 unless otherwise noted.

| Changes in relative soil water content under warming and varying water
Under the same temperature treatment, the changes in SRWC were associated with water regimes in the four Stipa plants (Figure 2).
Under every temperature treatment, increased water treatments (W3 and W4) were in favor of increasing SRWC, but decreased water reduced SRWC. Likewise, SRWC decreased with increased temperature under the same water treatment in the four species ( Figure 2).  (Table 4).

| Sensitivity of functional traits to precipitation variability under warming
(2)

| Relationship between various functional traits and total biomass
Based on a principal component analysis of functional traits in the four species, the variance explained by principal components 1 (production capacity) and 2 (growth strategy) was 44% and 26%, respectively, meaning that these two factors accounted for 70% of the variation arising from the nine variables investigated ( Figure 4a). The production capacity was closely related to BB, AB, LA, and LN. The growth strategy was closely related to the PH,    (Table 5).

| Effects of different precipitation treatments on total biomass
Duncan multiple comparisons showed that, under T0 treatment, the different precipitation treatments did not influence the total biomass of S. breviflora, while the total biomass of S. bungeana, S. grandis, and S. baicalensis was significantly lower under the treatment of W1 and W2 ( Figure 5). Under T1 and T2 treatments, the total biomass of S. breviflora treated with −30% precipitation (W1) was significantly lower than in the other precipitation  treatments; in S. bungeana and S. grandis, the total biomass was significantly lower in decreased precipitation treatments (W1 and W2) than in increased precipitation treatments; the total biomass of S. baicalensis treated with −30% precipitation (W1) was significantly lower than under the −15% and control precipitation (W2 and W0) conditions ( Figure 5).

| Relationship of total biomass and precipitation in the four Stipa species
Under T0 treatment, there was a quadratic curve relationship between the total biomass of S. bungeana and S. grandis and the precipitation rate (p < .05), but no significant relationship was observed for S. breviflora and S. baicalensis (p > .05). Under warming conditions (T1 and T2), the total biomass of the four Stipa species all showed a significant quadratic curve relationship with changing precipitation (p < .05, Table 6 and Figure 6).
Warming increased the optimum water levels and decreased the critical water thresholds in the four Stipa species. As warming, the critical water thresholds of the total biomass were −23% (T1) and −24% (T2) in S. breviflora; −28% (T1) and + 26% (T2) in S. bungeana; −27% (T1) and −36% (T2) in S. baicalensis; and −12% (T1) and −17% (T2) in S. grandis compared with the ambient water (Table 7).  (Table 3 and Figure 3), but differences existed between the different plant species. Under the ambient temperature treatment (T0), all the four grasses showed no or modest responses to changes in precipitation. Growth strategy traits (R/S and SLA, Figure 3) showed strongest sensitivity to changes in precipitation, indicating that

| D ISCUSS I ON
Stipa species may adapt their growth under precipitation variability by altering their allocation of above/belowground growth under ambient temperatures, as previously reported (Chapin et al., 1987;Xu et al., 2014). S. baicalensis, S. grandis, and S. breviflora showed gradually declining responses to precipitation variability under the ambient temperature (Table 4), which reflected the eastto-west precipitation gradients in their respective native areas in the steppes of Inner Mongolia (Qi et al., 2010). Under the warming by 1.5 and 2.0°C, the responses of the four Stipa species to precipitation variability increased to varying degrees. S. bungeana was the most precipitation-sensitive of the four species under ambient and warming (Table 4), which is probably related to its semiarid warm temperate steppe habitat and the resulting higher requirements for precipitation and temperature (Hu, Zhou, Li, Wu, & Wang, 2013). With a 1.5°C increase in temperature, the response of S. breviflora was more sensitive to precipitation variability than S. baicalensis, but the opposite with a 2.0°C increase (Table 4). So the sensitive characteristics of S. breviflora and S. baicalensis could indicate the increased range of temperature. When exposed to warming, leaf area and biomass, which reflect the production capacity, were the most sensitive traits in S. breviflora, S. baicalensis, and S. grandis, while R/S was the most sensitive trait in S. bungeana (Figure 3). This, to some extent, reflected the differences between the four species in their biological characteristics and sensitivity to precipitation, and supports the previous studies that showed that the tissues with the greatest response to changes in precipitation varied between species (Shafran-Nathan also are a reflection of plant growth strategy to changing environment (Brunner et al., 2015;Chapin et al., 1987;Li et al., 2015). The  TA B L E 6 Relationships between total biomass and precipitation under different temperature treatments in the four Stipa species resources, thus ensuring the maximum growth of individuals (Wright et al., 2004). The four Stipa plants were distributed on different scoring axes of the principal components, which indicated that the trade-off adaptation strategies adopted between their functional traits were different when dealing with the combined changes in temperature and precipitation, and this was the reasons for different adaptability to hydrothermal changes and mainly distributed in different grassland types (Donovan et al., 2011).  (Lanari, Silvestroni, Palliotti, Green, & Sabbatini, 2015;Wright et al., 2004). The functional traits that significantly influenced the total biomass of the four grasses were LA, R/S, SLA, and LAR (Table 5), mainly reflecting on the growth allocation strategy. The functional traits had different direct and indirect contributions to the total biomass of the Stipa plants, indicating their varied biological effects (Hu, Liu, Li, Lu, & Gao, 2014). This corresponded to the distribution of the four Stipa species in the different score axes of the principle component analyses, indicating that, under changes in precipitation, their different trade-off adaptive strategies would be achieved through their response characteristics, which influence their total biomass.
This was a reflection of their different adaptabilities to precipitation changes (Donovan et al., 2011;Xu et al., 2014).
Plant biomass is a combined product of photosynthesis and respiration, reflecting the comprehensive response of plants to changing environmental factors (Andresen et al., 2017;Myneni, Keeling, Tucker, Asrar, & Nemani, 1997), and is also the main research focus in steppe studies (Manea et al., 2016). Studies of the critical precipitation threshold of the four grasses would help to determine how steppe plants adapt to changing climates. This study found that the total biomass responses to changing precipitation were parabolic in Stipa species (Figure 6), indicating the existence of a critical point where the total biomass was significantly affected by precipitation variability. That was to say, Stipa plants were therefore adapted to an optimum precipitation threshold, and deviations from this value will inhibit their growth (Badger, Björkman, & Armond, 2006;Xu & Zhou, 2011 (Table 7), indicating that a 1.5°C increase in temperature alongside a mild reduced precipitation would promote biomass accumulation rather than inhibit it. However, as warming 2.0°C, critical precipitation threshold of total biomass of S. grandis and S. baicalensis was decreased by 17% and 38%, respectively, compared with these in warming 1.5°C, while their critical precipitation thresholds were decreased by 6% and 11%. Also, the critical precipitation threshold of S. bungeana was increased by 75% relative to T1 condition and did not change significantly in S. breviflora (Table 7). Thus, Stipa plants can better adapt to a certain degree of precipitation reduction if the temperature increase is limited to 1.5°C instead of 2.0°C. If the Paris Agreement to limit global warming to 1.5°C will be accomplished, this will increase the probability for sustained viability of the Stipa steppes in the next 50-100 years (Guiot & Cramer, 2016;Mitchell et al., 2016;Pecl et al., 2017 In conclusion, manipulation experiments were done under ambient and warming conditions with water variability, the response sensitivity of functional traits in the four Stipa species to water variability was relatively low in ambient temperature and increased in warming condition. R/S, SLA, and LA were the most response sensitive to water among all the functional traits and also made greater contribution to total biomass of the four grasses. The order of response sensitivity to water variability in the four species under ambient temperature treatment was ranked as follows: S. bungeana > S. baicalensis > S. grandis > S. breviflora, while S. bungeana had the greatest response sensitivity, and S. grandis had the lowest response sensitivity under warming treatments. The Stipa species were able to adapt better to mild decreased water when temperatures increased by 1.5°C rather than 2.0°C, indicating that future climate change will be more deleterious to Stipa growth (especially S. bungeana) if temperatures rise by 2.0°C rather than by 1.5°C. These findings contribute to our understanding of grassland species responses to global climate change and may be useful in providing experimental evidence for the Paris agreement.

ACK N OWLED G M ENTS
This study was jointly funded by National Natural Science

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
GZ, QH, and XL designed the experiments. QH and XL conducted the experiments and collected the data; GZ, QH, and XL analyzed the data. XL wrote the manuscript; all authors contributed to manuscript editing and approved of the final version.

E TH I C A L A PPROVA L
This study does not involve human participants or use of vertebrates.

DATA ACCE SS I B I LIT Y
All the data used in this study will be accessible in Dryad data repository https ://doi.org/10.5061/dryad.gm1h74f.