Review Research review Increasing climatic sensitivity of global grassland vegetation biomass and species diversity correlates with water availability Author for correspondence: Daijun Liu1,2,3,4 , Chao Zhang4,5 , Roma Ogaya4,6 , Chao Zhang Email: chao.x.zhang@helsinki.fi Marcos Fernandez-Martınez7 , Thomas A. M. Pugh2,3,8 and nuelas4,6 Josep Pe~ 1 Received: 27 July 2020 Department of Botany and Biodiversity Research, University of Vienna, Rennweg 14, Vienna 1030, Austria; 2School of Geography, Accepted: 3 February 2021 Earth and Environmental Sciences, University of Birmingham, Birmingham, B15 2TT, UK; 3Birmingham Institute of Forest Research, University of Birmingham, Birmingham, B15 2TT, UK; 4CSIC, Global Ecology Unit, CREAF-CSIC-Universitat Autonoma de Barcelona, Bellaterra (Catalonia) 08193, Spain; 5Optics of Photosynthesis Laboratory, Institute for Atmospheric and Earth System Research (INAR)/Forest Sciences, Viikki Plant Science Centre, University of Helsinki, PO Box 27, Helsinki 00014, Finland; 6CREAF, Cerdanyola del Valles (Catalonia) 08193, Spain; 7PLECO (Plants and Ecosystems), Department of Biology, University of Antwerp, Wilrijk 2610, Belgium; 8Department of Physical Geography and Ecosystem Science, Lund University, Lund 22362, Sweden Summary New Phytologist (2021) 230: 1761–1771 Grasslands are key repositories of biodiversity and carbon storage and are heavily impacted by doi: 10.1111/nph.17269 effects of global warming and changes in precipitation regimes. Patterns of grassland dynamics associated with variability in future climate conditions across spatiotemporal scales are yet to be Key words: carbon storage, ecological adequately quantified. Here, we performed a global meta-analysis of year and growing season sensitivity, global warming, precipitation sensitivities of vegetation aboveground biomass (AGB), aboveground net primary productivity alteration, structural changes, temporal (ANPP), and species richness (SR) and diversity (Shannon index, H) to experimental climate dynamics. warming and precipitation shifts. All four variables were sensitive to climate change. Their sensitivities to shifts in precipitation were correlated with local background water availability, such as mean annual precipitation (MAP) and aridity, and AGB and ANPP sensitivities were greater in dry habitats than in nonwater-limited habitats. There was no effect of duration of experiment (short vs long term) on sensitivities. Temporal trends in ANPP and SR sensitivity depended on local water availability; ANPP sensitivity to warming increased over time and SR sensitivity to irrigation decreased over time. Our results provide a global overview of the sensitivities of grassland function and diversity to climate change that will improve the understanding of ecological responses across spatiotemporal scales and inform policies for conservation in dry climates. wetter (Pe~nuelas et al., 2017; Hoegh-Guldberg et al., 2018; Nolan Introduction et al., 2018). Studies have shown that rises in carbon dioxide (CO2) Ongoing global climate change, characterised by warming and emissions have led to increasing levels of biosphere productivity, spatiotemporal shifts in patterns of precipitation, is affecting and warmer temperatures may further increase productivity in cold species diversity and composition, and plant carbon accumulations regions by enhancing photosynthesis at the regional scale across communities and ecosystems (Hooper et al., 2012; Scheffers (Fernandez-Martınez et al., 2019). However, decreases in vegeta- et al., 2016; Thakur et al., 2017; Nolan et al., 2018). These changes tion growth and reductions in terrestrial carbon uptake have also in temperature and precipitation are likely to increase in both been observed and are projected to continue under future climate frequency and intensity in the coming decades, during which change scenarios, alongside increases in water limitation and warmer and drier climates are predicted to prevail across large areas nutrient mineralisation (Reich & Hobbie, 2013; Pe~ nuelas et al., of the globe, with conditions at higher latitudes likely to become 2017; Fernandez-Martınez et al., 2019), large losses in species Ó 2021 The Authors New Phytologist (2021) 230: 1761–1771 1761 New Phytologist Ó 2021 New Phytologist Foundation www.newphytologist.com This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. New 1762 Review Research review Phytologist diversity, and disruptions to community assemblage. These ecosystem respiration) that have used the ln or log ratio response impacts may be especially severe under the high-emission RCP of treatments to that of controls (Wu et al., 2011; DeMalach et al., 8.5 (representative concentration pathway) scenario (Nolan et al., 2017) may hinder the comparison of response variables under 2018; Berdugo et al., 2020; Trisos et al., 2020). contrasting magnitudes of change in simulated climate conditions, However, many of these predicted biodiversity and ecosystem leading to uncertainty of predicted ecosystem dynamics. function responses to future climate change are inconsistent with Assessment of the sensitivity of vegetation responses, such as empirical evidence from natural ecosystems (Vellend et al., 2013; change in output per unit change in input, is a promising approach Estiarte et al., 2016; Messier et al., 2017). For example, plant to quantify ecological impacts of climate change (Huxman et al., growth may be regulated by plant acclimation and plasticity in 2004; Wilcox et al., 2015, 2017), including across space and time physiology, phenology, and evolutionary adaptation, and may (Smith et al., 2017). While attempts have been made to determine resist and recover from environmental disturbance (Jump & sensitivities of aboveground net primary productivity (ANPP) and Pe~nuelas, 2005; Beier et al., 2012; Morecroft et al., 2019). belowground net primary productivity (BNPP) to extreme drought Furthermore, species interactions (e.g. facilitation) may reduce and increases in levels of local mean annual precipitation (MAP) at the magnitude of species diversity loss and biomass declines due to study sites (Wilcox et al., 2015, 2017; Smith et al., 2017), positive impacts of belowground mycorrhizal and rhizobacteria comparison of plant communities across the study sites may be associations (Wright et al., 2017). Climate mitigation and inappropriate, due to divergence in species assemblages and sustainable policies based on model predictions of biomass and successional processes driven by historical land use and other types species diversity declines that do not reflect real-world responses of disturbance (Reinsch et al., 2017; Cramer et al., 2018; Langley of plant communities and ecosystems to climate change may lack et al., 2018; Liu et al., 2020). A novel metric of sensitivity to changes effectiveness. Therefore, the development of approaches that fully in climate, which calculates the proportional change in vegetation detect and reflect the impacts of climate change on species response per unit net change in climate and standardises response diversity and function for natural ecosystems are urgently variables across study sites, has been proposed for global grasslands, required. desert, and forests (Song et al., 2019). It allows the accurate Climate manipulation experiments, in which temperature and/ quantification and comparison of climate change studies of plant or precipitation are controlled, are essential for the elucidation of community dynamics, including biomass accumulation, and plant adaptation and species interaction responses and mecha- species diversity and composition, and analysis of global-scale nisms, and they reduce degrees of uncertainty of vegetation vegetation sensitivity across spatiotemporal gradients (Halbritter dynamics under specific climate scenarios (Jentsch et al., 2007; Wu et al., 2020). et al., 2011; Beier et al., 2012; Langley et al., 2018). For example, Grassland ecosystems are one of the largest terrestrial biomes by empirical experimental data show contrasting spatiotemporal area (occupying c. 40% of the Earth’s surface) (Gibson, 2009) and patterns and directions in plant community responses to climate are crucially important for carbon storage and biodiversity, yet are change (Andresen et al., 2016; Estiarte et al., 2016; Harrison et al., highly vulnerable to climate perturbations (Seddon et al., 2016; 2020), such as increases in biomass accumulation at high latitudes Yuan et al., 2016; Hungate et al., 2017; Wang et al., 2019). and altitudes (Elmendorf et al., 2012a,b, 2015; Metcalfe et al., Application of the novel metric of Song et al. (2019) to data from 2018) and decreases in plant growth under dry and arid conditions empirical studies has indicated that sensitivities of carbon-cycling (Kr€oel-Dulay et al., 2015; Ladron de Guevara et al., 2018; Liu et al., variables in desert, wetland, grassland and forest ecosystems, such as 2020). Although changes in species diversity, richness, and aridity, are strongly associated with study site water availability. composition are expected with increasing temperature and greater However, it is unclear whether similar associations of local water variability in precipitation (Knapp et al., 2002; Smith et al., 2009; availability with ecosystem function and with species diversity Komatsu et al., 2019), the temporal patterns of change are not well sensitivity exist at individual grassland sites. understood. While some long-term studies report a lack of species Temporal variation in directions of effect size (increase, decrease, diversity losses under decadal climate manipulations in arctic no net response) have been reported (Leuzinger et al., 2011; Liu tundra (Hudson & Henry, 2010; Elmendorf et al., 2012a) and et al., 2017; Komatsu et al., 2019) and tend to be related to habitat Mediterranean shrublands (Tielb€orger et al., 2014), others have factors, such water or nutrient availability (Grime et al., 2008; demonstrated reordering shifts in species diversity and composition Tielb€orger et al., 2014; Andresen et al., 2016). However, the ability in grassland (Knapp et al., 2002; Harte et al., 2015; Shi et al., 2018), to identify temporal trends in grassland plant growth and species shrubland (Liu et al., 2017, 2018c, 2020), and forest (Liu et al., diversity responses are likely to depend on the duration of an 2018b) ecosystems. experiment (Grime et al., 2008; Leuzinger et al., 2011; Liu et al., The integration and synthesis of empirical data from a range of 2018a), as reported for the dampening effects on growth while study sites may allow the analysis of ecological responses over larger higher species diversity and composition shifts over time spatial scales (Kr€oel-Dulay et al., 2015; Estiarte et al., 2016; (Leuzinger et al., 2011; Beier et al., 2012; Liu et al., 2020). While Halbritter et al., 2020). However, statistical confidence in the an understanding of temporal trends in sensitivity remains lacking, extrapolation of pattern and trend data from a reduced number of possibly due to insufficient analysis of long-term data, it is likely study sites to regional and global scales is unclear. Meta-analyses of that site water availability may also be a key driver. size effects of warming and/or shifts in precipitation on ecosystem Although several global meta-analyses have been conducted to responses (e.g. biomass accumulation, species diversity, and date, how grassland sensitivity changes across spatiotemporal scales New Phytologist (2021) 230: 1761–1771 © 2021 The Authors www.newphytologist.com New Phytologist © 2021 New Phytologist Foundation New Phytologist Research review Review 1763 remains uncertain. Most of the previous meta-analyses (such as communities and we included studies of climatic experiments Komatsu et al., 2019, Song et al., 2019 and Wu et al., 2011) conducted in ecosystems by a single factor, such as grazing or fire; analysed a mixture of terrestrial ecosystems (shrublands, grasslands multiple experiments of different vegetation communities at the and forests), which may have led to some uncertainties on their same location (latitude and longitude) were recorded as a single results. The impacts of climate change not only on the responses of study site. As a result of these criteria, we acquired data from 138 ANPP and AGB, but also on species richness (SR) and community study sites, comprising 50 warming, 54 drought, and 63 irrigation composition (Shannon index, H), were not included in most (or water addition) single treatments, with five warming 9 drought previous meta-analyses (Wilcox et al., 2017; Wang et al., 2019). A and eight warming 9 irrigation treatment interactions. For warm- recent study has reported the effect on SR over time, but for mixed ing experiments, we considered the impacts of elevated temperature biomes and global ecological drivers (Komatsu et al., 2019). To compared with control plots, irrespective of manipulation method, address these unknowns, we conducted a global synthesis of the such as open top chamber (OTC), cable, infrared and reflector. effects of warming, drought, and irrigation on plant responses using Data were mostly derived from experiments conducted in North a large dataset of recently published grassland experiments (138 America, Europe and China (Fig. 1a; Supporting Information study sites) with durations ranging from 1 to 23 yr. We tested the Table S1). hypotheses that the sensitivities of grassland function and diversity When possible, we acquired vegetation dynamics and climate to climate change are significant, the variability is affected by the data from the published data sets associated with the papers; local climate and magnitude of sensitivity is related to duration of alternatively, we extracted data from figures using WEBPLOTDIG- experimental period. Using the dataset, the specific objectives were ITIZER (v.4.2, 2019; https://automeris.io/WebPlotDigitizer) and to assess: (1) the mean sensitivities of AGB, ANPP, SR and H to averaged control and simulated plot data across the study period for warming, drought, and irrigation; (2) the strength of associations each study site. We mainly focused on net changes in manipulated between sensitivities and local climate conditions across spatial climates on species diversity and function of plant communities, so gradients; and (3) temporal trends in sensitivity under continued responses of variables to climate manipulations for whole years climate manipulation. We therefore aim to advance the under- (WY) and growing seasons (GS) were treated differently. For standing of spatiotemporal sensitivities and vulnerabilities of global warming, drought and irrigation treatments over WYs, we derived grassland structure and function to climate change and to inform data for 72, 84 and 97 plant communities, respectively, while for and improve mitigation strategies. GS, we derived data for 28, 25 and 38 communities, respectively. We used study site mean annual temperature (MAT) and ground surface temperature during the growing season from warming Materials and Methods experiments; if air temperature was not reported, we used soil temperature (5–10 cm depth) of warming and control plots, Data collection because the temperatures of shallow soil and surface air at < 10 cm First, we searched the Web of Science database for studies published height tend to be similar (Reinsch et al., 2017). Local abiotic between January 1980 and April 2019 using the following single factors, comprising MAT, MAP, and growing season temperature and combined keywords for manipulated climate and vegetation and precipitation (Tgs and Pgs, respectively) were obtained from responses: ‘experiment*’, ‘treatment*’, ‘warm*’, ‘increase temper- the Methods and Materials sections of the studies; if these data were ature’, ‘drought*’, ‘rainfall reduction’, ‘decrease precipitation’, lacking, we used control plot means. To avoid negative temper- ‘watering*’, ‘irrigation*’, ‘precipitation*’, ‘rainfall addition’, atures, we transformed MAT and Tgs (+ 15°C; MATm and Tgsm, ‘biomass’, ‘growth’, ‘productivity’, ‘production’, ‘ANPP’, ‘rich- respectively) to allow us to develop logistic regression models for ness’, ‘diversity’, ‘sensitivity’, ‘community’, ‘composition shifts’, the study sites. When MAT and/or MAP were not reported (e.g. ‘herb’, and ‘grass’. We also searched Google Scholar using the Toolik lake, Finse, Haibei, Hongyuan, Beiluhe), they were derived Advanced search function for the keywords ‘experiment*’, from WorldClim (http://worldclim.org/version2, resolution ‘manipulation*’, ‘biomass’, ‘grass’, ‘diversity’, ‘richness’, and c. 1 km2) that contains mean climate data for the period 1970– ‘composition’ in the title of the article. These terms were used to 2000 (Fick & Hijmans, 2017). The aridity index (AI) for all the identify responses of the aboveground biomass (AGB) (the study sites was obtained from Global Aridity and PET Dataset standing biomass), ANPP (the net accumulation rate of above- (https://cgiarcsi.community/data/global-aridity-and-pet-databa ground biomass plus litterfall), SR and diversity (as indicated by the se/), which provides local water availability (c. 1 km resolution) Shannon index H) to changes in climate (warming, and decreases considering both evapotranspiration processes and rainfall deficits and increases in precipitation). We selected climate manipulation (Zomer et al., 2008). data from recent peer-reviewed publications (Wu et al., 2011; Andresen et al., 2016; Wilcox et al., 2017; Song et al., 2019) and Sensitivity calculation and analysis datasets (https://drought-net.colostate.edu/). We selected and used the most recently published papers on the topics and cross-checked We modified the calculation of community sensitivities from experimental sites, using the Advanced search function of Google previous studies (Wu et al., 2011; Smith et al., 2017; Wilcox et al., Scholar, to acquire additional publications that described study site 2017; Song et al., 2019) as the plant communities across the vegetation variables. We mainly focused on AGB or ANPP different study sites may diverge in species assemblages and measured from annual harvests of global grassland plant successional stages. Sensitivity was calculated as the proportion (%) © 2021 The Authors New Phytologist (2021) 230: 1761–1771 New Phytologist © 2021 New Phytologist Foundation www.newphytologist.com New 1764 Review Research review Phytologist (a) Fig. 1 Study sites and analyses of sensitivity to climate treatments. (a) Distribution of grassland ecosystem study sites testing effects of warming (n = 50), drought (n = 54), irrigation (n = 63), warming and drought (n = 5) and warming and irrigation (n = 8). Colours represent the climate treatments: W, (b) (c) warming; D, drought; I, irrigation; WD, warming + drought; and WI, warming + irrigation. Sensitivity of aboveground biomass (AGB), aboveground net primary productivity (ANPP), species richness (SR) and species diversity (H) to climate treatments for whole years (WY) (b) and growing seasons (GS) (c). Error bars represent the standard error. The significance was tested by weighted Student’s t-tests ((*), P < 0.1; *, P < 0.05; **, P < 0.01; and ***, P < 0.001). of magnitude of community metric (AGB, ANPP, SR, and H) to Knapp et al. (2015), we separated climate conditions into dry net change in temperature [(Xt Xc)/Xc)/(Tt Tc) 9 100] and (MAP < 500 mm) and nonwater limited (MAP ≥ 500 mm) to test precipitation [(Xt Xc)/Xc)/(Pt Pc) 9 100], where Xt and Xc are for regulation of sensitivities by site water availability. mean AGB, ANPP, SR and H across all treatment and control years; Tt and Pt are annual mean temperature (or mean temperature Definition of temporal trends of the growing season, Tgs) and annual mean precipitation (or mean precipitation of the growing season, Pgs) for treatment plots Based on duration of global climate manipulation experiments across the study period, respectively; and, Tc and Pc are annual (Beier et al., 2012), we classified experimental temporal scales as mean temperature (or mean temperature of the growing season, short term (1–4 yr) or long term (≥5 yr), and selected long-term Tgs) and annual mean precipitation (or precipitation of growing experiments with ≥ 4 yr measurements to analyse temporal trends season, Pgs) for control plots across the study period, respectively. in ANPP and SR sensitivity; there were insufficient data for Therefore, the sensitivity for warming was expressed as the temporal analysis of H sensitivity. When climate data in control proportion of vegetation response per degree (%/+°C). The and treatments were available for study sites, we used net changes in sensitivity for drought and irrigation treatments were expressed as temperature and precipitation during the study year; alternatively, the proportion of vegetation response per 100 mm change in we used mean magnitudes of climate change across study periods as precipitation (%/100 mm). Sensitivities to warming 9 drought clarified in the published literature. Temporal trends in ANPP and and warming 9 irrigation interactions were calculated as the SR sensitivity (SEN) to climate manipulation treatments were proportion of vegetation response per degree per 100 mm change corrected by differences between single-year sensitivity and the in precipitation or irrigation (%/+°C  100 mm). Positive and mean sensitivity of each community across all years negative sensitivities indicated increasing and decreasing effects of (SEN = SENi SENmean, where SENi and SENmean are sensitivity climate change, respectively. for a specific year and mean sensitivity across all years, respectively), Across sites, we analysed the sensitivity of AGB, ANPP, SR and to allow linear regression analysis of temporal sensitivity among the H to changes in climate, while controlling for local climate communities, using a reduced range of variables. conditions using linear regressions. We did not analyse the sensitivities to warming 9 drought or warming 9 irrigation across Statistical analyses spatial scales due to lack of available data. We tested year and growing season sensitivities of AGB and ANPP to climate change Climate treatment effects on year and growing season AGB, ANPP, treatments separately using local MATm, MAP and AI. Correla- SR, and H were analysed using weighted Student’s t-tests in the R tions between treatments and SR and H were only analysed for WYs WEIGHTS package (Pasek et al., 2020) to test degree of variance in due to lack of growing season data, while those for AGB and ANPP average sensitivity to treatments from 0, with study year as a were tested by year and growing season. Based on definitions by weighting factor to improve precision of estimated sensitivity. We New Phytologist (2021) 230: 1761–1771 © 2021 The Authors www.newphytologist.com New Phytologist © 2021 New Phytologist Foundation New Phytologist Research review Review 1765 then tested for associations between year and growing season WY AGB and ANPP sensitivity to drought was positively sensitivities of AGB, ANPP, SR and H with habitat factors, using correlated with site MAP, respectively (R2 = 0.38, P < 0.01 and linear models (lm function in R), testing for treatment effects on R2 = 0.15, P < 0.01, respectively) (Fig. 2b,e; Table S3), and was more MATm, MAP and AI for WYs and those on Tgs and Pgs for negative in dry sites than nonwater-limited sites (dfHSD = 25.8, growing season; vegetation metrics were response variables and P < 0.05 and dfHSD = 23.5, P < 0.01; respectively). But both climate factors were explanatory variables. We selected the best- sensitivities were negatively correlated with local MAP and AI in fitting models based on the lowest Akaike information criterion irrigated sites, respectively (R2 = 0.16, P < 0.01 and R2 = 0.31, (AIC) and considered models with an AIC within two units of the P < 0.001; Fig. 2c,f). Sensitivity of plant communities (AGB and lowest value. Habitat type differences (dry vs nonwater limited) and ANPP) to irrigation was positive and greater at relatively dry sites temporal differences (short term vs long term) in sensitivities were than nonwater-limited sites (dfHSD = 22.7, P < 0.01 and tested using analysis of variance (ANOVA), with Tukey’s honest dfHSD = 39.2, P < 0.01; respectively; Fig. 3; Table S4). Growing significant difference (HSD) tests; we analysed temporal trends in season ANPP sensitivity to drought and irrigation was correlated with ANPP and SR sensitivity across the duration of study programmes, local Pgs and MAP (Fig. S1b,c; R2 = 0.40, P < 0.021 and R2 = 0.27, with sensitivity as the response variable and year as the explanatory P < 0.05; respectively). In other words, AGB and ANPP sensitivity to variable. We used linear mixed modelling (LME4 package in R) to changes in precipitation was correlated with the site water availability. test overall trends in sensitivities, with study site as a random factor. The trends for individual communities were tested using simple Sensitivity of species diversity to study site climate factors linear modelling. All analyses were performed in R (v.3.5.0; R Core Team, 2018). We did not find significant relationships between SR and H sensitivities and local climate factors under warming (Fig. 4a,d; Table S6). Under drought conditions, WY sensitivities of SR and H Results were positively related to local MAP (R2 = 0.18, P < 0.05 and R2 = 0.57, P < 0.01, respectively; Fig. 4b,e) and under irrigation, Sensitivity of vegetation to climate change they were negatively related to local MAP (R2 = 0.13, P < 0.05 and Vegetation function and species diversity were sensitive to climate R2 = 0.37, P < 0.05; respectively; Fig. 4c,f). WY sensitivity of SR treatments (Fig. 1b,c; Table S2), where AGB and ANPP were most under irrigation was greater in dry sites than in the nonwater- sensitive. Over WYs, results showed positive response under limited sites (dfHSD = 12.04, P < 0.01) (Table S7). In other words, climate warming (mean = 7.7, P < 0.01 and mean = 9.0, P < 0.05, as for AGB and ANPP, the sensitivities of SR and H to drought and respectively) and irrigation (mean = 13.7, P < 0.001 and irrigation treatments were related to the site water availability. mean = 20.5, P < 0.01, respectively) and negative response under drought conditions (mean = 8.0, P < 0.05 and mean = 11.7, Temporal trends in sensitivity P < 0.01, respectively) (Fig. 1b). Sensitivity of SR was negative under climate warming (mean = 4.2, P < 0.05), drought condi- We observed that the overall trend in ANPP sensitivity under tions (mean = 5.3, P < 0.001) and the interaction of warm- climate warming increased over time (R2 = 0.07, P < 0.05; Fig. 5a), ing 9 drought (mean = 9.5, P < 0.05), whereas it was positive but there was no such relationship for SR sensitivity (Fig. 5c). under irrigation (mean = 4.6, P < 0.05). There were no effects of Increasing temporal ANPP sensitivity under climate warming was warming or drought on H sensitivity over WYs, but sensitivity was positively associated with study site water availability at nonwater- significantly negative in the interaction of warming and drought limited sites (MAP = 750 mm, R2 = 0.38, P < 0.01 and (mean = 11.8, P < 0.01). Growing season sensitivities of vegeta- MAP = 914 mm, R2 = 0.31, P < 0.05; Table S8), whereas SR tion growth and community composition were affected by climate sensitivity decreased over time at dry sites (MAP = 66.6 mm). treatments (Fig. 1c), where AGB and ANPP sensitivity were There was no overall temporal trend in WY ANPP sensitivity under positive under irrigation (mean = 18.2, P < 0.05 and mean = 24.6, irrigated conditions (Fig. 5b), but there was a tendency for WY SR P < 0.05, respectively). No influences of climate treatments on SR sensitivity to decrease over time (Slope = 0.9, P < 0.05) (Fig 5d; and H sensitivity were significant at the 5% level. Table S9). There were no local climate factor effects on temporal trends in year ANPP sensitivity, however for SR sensitivity there was a temporal decrease in nonwater-limited habitats, especially Sensitivity of AGB and ANPP to site characteristics those with high levels of rainfall (Fig. 5d; MAP = 2378 mm, Over WYs, neither AGB nor ANPP sensitivity to experimental R2 = 0.59, P < 0.01 and MAP = 700 mm, R2 = 0.49, P < 0.01). climate warming was correlated with local climate conditions (Fig. 2a,d; Table S3). WY AGB sensitivity to warming was Discussion marginally higher at dry sites than at nonwater-limited study sites (dfHSD = 8.2, P < 0.1; Table S4) and overall temporal variation in Importance of quantifying vegetation sensitivity to climate AGB sensitivity increased with duration of experiment change (dfHSD = 9.0, P < 0.05; Table S5). Growing season ANPP sensitivity to climate warming was negatively correlated with Tgsm The application of a sensitivity-based metric is essential for the (Fig. S1a; R2 = 0.53, P < 0.05). development of a global assessment of vegetation function and © 2021 The Authors New Phytologist (2021) 230: 1761–1771 New Phytologist © 2021 New Phytologist Foundation www.newphytologist.com New 1766 Review Research review Phytologist (a) (b) (c) (d) (e) (f) Fig. 2 Correlations between aboveground biomass (AGB) (a–c) and aboveground net primary production (ANPP) (d–f) sensitivity to treatments over whole- year and local climate conditions. AI and MAP represent the aridity index and mean annual precipitation of the study sites. The x-axis shows the best-fitted climate variables based on model selection; R2 is the coefficient of determination for the regression. Circle sizes (small and large) represent short-term (1–4 yr) and long-term (≥5 yr) studies, respectively. Statistical significance across spatial scales was tested using a general linear model and only the significant relationships (P < 0.05) are shown (brown and blue lines). The shading with the lines indicates the standard error. The dashed line (y = 0) separates positive and negative values. Fig. 3 Comparison of aboveground biomass (AGB) and aboveground net primary production (ANPP) sensitivity for whole year (WY) to treatments at dry (mean annual precipitation: MAP < 500 mm) and nonwater-limited (MAP > 500 mm) study sites, tested using analyses of variance and Tukey’s honest significant difference test. Error bars represent the standard error. The asterisk indicates the different significance analysed with Tukey’s honest significant difference (HSD) tests ((*), P < 0.1; *, P < 0.05; **, P < 0.01; and ***, P < 0.001). species diversity responses to changes in climate. To our knowl- associated with elevated water and nutrient availability (Wu et al., edge, this study has compiled one of the largest datasets (209 plant 2011; DeMalach et al., 2017; Pe~ nuelas et al., 2017; Song et al., communities across 138 study sites) for sensitivity analysis of 2019; Wang et al., 2019). However, AGB and ANPP sensitivity grassland biomass and productivity (AGB and ANPP) and species were negatively affected under drought conditions, probably due to diversity (SR and H) to experimental changes in temperature and constrained photosynthesis, higher levels of mortality, and reduced precipitation. We modified the sensitivity index to calculate the biomass storage (Wu et al., 2011; Andresen et al., 2016; Song et al., vegetation response (proportion) to net climate changes, which is 2019; Wang et al., 2019). Our finding of a lack of interaction appropriate to compare and explore the response over spatial- between effects of temperature and precipitation on sensitivities temporal scales. We found that, assessed over WYs, AGB and (Fig. 1b) was not consistent with previous research that has shown ANPP increased (i.e. had a positive sensitivity) under warming and interactive effects of multiple climate factors (Komatsu et al., increased precipitation, possibly due to enhanced biomass storage 2019). We found that SR and H were sensitive to warming and New Phytologist (2021) 230: 1761–1771 © 2021 The Authors www.newphytologist.com New Phytologist © 2021 New Phytologist Foundation New Phytologist Research review Review 1767 (a) (b) (c) (d) (e) (f) Fig. 4 Correlations between species richness (SR) (a–c) and species diversity (H) (d–f) sensitivities for whole year (WY) under climate change treatments and local climate factors. MATm, MAP and AI represent the modified mean annual temperature (MAT + 15), mean annual precipitation and aridity index, respectively. Relationships with MATm and MAP were tested by logistic linear regression analysis; the best-fit regressions are shown. Solid and dashed fitted regression lines indicate the significance P < 0.05 and P < 0.1, respectively, and small and large circles represent short-term (< 5 yr) and long-term (≥ 5 yr) experiments. The shading with the lines indicates the standard error. The dashed black line at y = 0 separates positive and negative values. (a) (b) Fig. 5 Overall temporal trends in year aboveground net primary production (ANPP) (c) (d) and species diversity (SR) sensitivity under experimental climate warming (a, c) and irrigation (b, d) treatment. Blue and brown dots represent nonwater-limited (MAP > 500 mm) and dry (MAP < 500 mm) study sites, respectively. The red line (a, d) indicates the significant trend tested using a linear mixed model. The shading with the lines indicates the standard error. shifts in precipitation. WY SR sensitivity was negative under the AGB and ANPP sensitivity strongly related to water single and combined treatments of warming and drought, possibly availability due to reduced seedling germination and establishment and ultimate local extinctions of highly sensitive species (Sullivan et al., There were no associations between WY AGB and ANPP 2016; LaForgia et al., 2018; Liu et al., 2018c), while WY H sensitivities under climate warming with gradients of MAT or sensitivity was also negative under drought, indicating strong water availability, although there was a greater number of study sites compositional shifts under harsh conditions. We observed high with positive sensitivities (increasing effects of warming) than levels of growing season AGB and ANPP sensitivity to shifts in negative (decreasing effects of warming) (Fig. 2a,d). Indeed, precipitation (Fig. 1c) that reflected the impacts of within-season previous syntheses of large-scale observations and climatic exper- changes in precipitation on biomass accumulation and ecosystem iments have reported that warmer temperatures may lead to net productivity (Bai et al., 2004; Pe~ nuelas et al., 2007, 2017; Wan increases in carbon assimilation and biomass storage, favouring et al., 2009). Overall, our findings, particularly those showing shrub encroachment (Wu et al., 2011; Elmendorf et al., 2012a, species diversity responses to changes in climate, have not been 2015; Myers-Smith et al., 2015). By contrast, WY sensitivities of reported in previous studies. AGB and ANPP to shifts in precipitation shifts were correlated with © 2021 The Authors New Phytologist (2021) 230: 1761–1771 New Phytologist © 2021 New Phytologist Foundation www.newphytologist.com New 1768 Review Research review Phytologist site parameters of MAP and aridity. Although these results are storage of vegetation biomass under elevated temperatures in tundra inconsistent with a previous study that found no relationship (Elmendorf et al., 2012a, 2015) and alpine (Steinbauer et al., 2018) between grassland ANPP sensitivity and MAP under extreme ecosystems. We also found that the increase in ANPP sensitivity to drought conditions (Wilcox et al., 2017), our finding that AGB and warming over time (Fig. 5a) was greater in high MAP sites ANPP sensitivities to irrigation are related to MAP and aridity is (>700 mm), indicating water availability is a key driver of temporal supported by previous studies (Huxman et al., 2004; Wilcox et al., patterns of biomass accumulation (Harte et al., 2015; Kr€ oel-Dulay 2015, 2017). Growing season ANPP sensitivities under warming et al., 2015; Estiarte et al., 2016; Shi et al., 2018). By contrast, we were strongly correlated with study site temperature during the found that the overall trend in SR sensitivity to warming and ANPP growing season (Tgsm) (Fig. S1). This finding has not been sensitivity to irrigation did not change over time (Fig. 5c), possibly reported in previous studies (Wu et al., 2011; Song et al., 2019; due to the plasticity of plant physiology and phenology that drive Wang et al., 2019). Perhaps more importantly, we found that WY demography and species reordering (Leuzinger et al., 2011; Bellard AGB and ANPP sensitivity was greater under drought and irrigated et al., 2012; Liu et al., 2016; Walker et al., 2020) and adaptability of conditions at dry study sites than at nonwater-limited sites (Fig. 3), long-living species as a result of functional trait compensatory illustrating the significant impact of increased levels of precipitation responses (Smith et al., 2015; Sullivan et al., 2016). Our analysis in dry habitats on biomass storage. showed an overall decreasing trend in SR sensitivity to irrigation (Fig. 5d) that was greater in study sites with high levels of water availability; these results may reflect increases in abundance and SR and H sensitivity is strongly related to local climate standing growth of dominant species and losses of sensitive species conditions (Collins et al., 2012; Harpole et al., 2016; Komatsu et al., 2019). Over WYs, sensitivities of SR and H under climate warming were This study shows that temporal trends in species diversity are more not correlated with local climate conditions (Fig. 4a,d). However, sensitive at sites with high levels of water and/or nutrient availability, under drought (Fig. 4b,e) and irrigation (Fig. 4c,d), they depended as supported by previous reports of lack of effects on species diversity on site MAP. These results highlighted the importance of studying and composition of climatic change treatments in arid (Tielb€orger patterns of change in species diversity in water-limited ecosystems et al., 2014) and nutrient-poor (e.g. phosphorus) (Grime et al., (Kr€oel-Dulay et al., 2015; Liu et al., 2017; Cramer et al., 2018). 2008) grassland ecosystems. Although many studies have reported the effects of changes in climate on terrestrial carbon cycling, including ANPP and soil Conclusion and future implications respiration (Jentsch et al., 2007; Reinsch et al., 2017; Song et al., 2019; Wang et al., 2019), few have quantified impacts on vegetation Our global meta-analysis quantified responses of grassland vege- composition, despite overwhelming evidence for climate change- tation biomass and species diversity to manipulated climate mediated biogeographic range shifts (Lenoir et al., 2008; Crimmins conditions using the proportional change in vegetation functioning et al., 2011; Rumpf et al., 2018), species invasion (Van Kleunen (AGB and ANPP) and species diversity (SR and H) per unit net et al., 2015; Seebens et al., 2017), and shifts in dominance (Harte change in manipulated temperature or precipitation. We demon- et al., 2015; Liu et al., 2017; Shi et al., 2018). High levels of global strated that this novel metric is appropriate for the quantitative declines in SR and shifts in composition due to historical climate assessment and comparison of ecosystem-level responses to climate change have been reported, along with predictions of further large change across heterogeneous study sites and showed that vegetation magnitudes of change in response to future climate change scenarios, sensitivity is correlated with local abiotic factors (MAP and aridity), especially under high levels of CO2 emissions (Nolan et al., 2018). while contrasting temporal trends in sensitivity depended on local We suggest that further research should focus on the potential levels of water availability. Vegetation sensitivity to climate change impacts of climate change on species diversity and community varied with local water availability, especially for communities structure, and associated impacts on higher trophic levels, species distributed in dry habitats. Variability in our findings may be interactions and carbon cycling across different ecosystems. explained by the lack of study of species interactions, diversity and functions of soil fungi and bacteria, and nutrient cycling, and that data were derived from studies with contrasting experimental Dependence of temporal trends in AGB and SR sensitivities approaches (active vs passive warming), soil properties (soil types on local water availability and water-holding capacity), sample sizes and vegetation measure- A recent meta-analysis has reported that the effects on plant SR and ment protocols. Future studies should focus on the variability of community composition were greater for decadal processes (>10 yr) sensitivity metrics to quantify impacts of climate, including and when combined with drivers of global change (Komatsu et al., patterns and timing combined with mixed drivers of climate, land 2019). However, and with the exception of AGB sensitivity to use, and species invasion and the need to study the heterogeneity of climate warming, we did not detect differences in vegetation fine scale abiotic and biotic habitat parameters. Therefore, broad- sensitivity to changes in climate between short-term and long-term scale and comprehensive evaluations of ecological sensitivities to studies (Table S5), possibly because the dataset comprised fewer global environmental change would further facilitate identifying long-term experiments than short-term experiments. The overall generality across spatial gradients, allowing the better identification trend in ANPP sensitivity under climate warming increased over of vulnerable habitats and the design and implementation of time, consistent with the expectation of an accelerated increase in effective protection and restoration programmes. New Phytologist (2021) 230: 1761–1771 © 2021 The Authors www.newphytologist.com New Phytologist © 2021 New Phytologist Foundation New Phytologist Research review Review 1769 Cramer W, Guiot J, Fader M, Garrabou J, Gattuso JP, Iglesias A, Lange MA, Lionello Acknowledgements P, Llasat MC, Paz S et al. 2018. Climate change and interconnected risks to sustainable development in the Mediterranean. Nature Climate Change 8: 972–980. We thank all the people who managed and maintained the climate Crimmins SM, Dobrowski SZ, Greenberg JA, Abatzoglou JT, Mynsberge AR. experiments and collected data. Funding for this research was 2011. Changes in climatic water balance drive downhill shifts in plant species’ provided by the Spanish government project PID2019- optimum elevations. Science 331: 324–327. 110521GB-I00, Catalan government project SGR2017-1005, DeMalach N, Zaady E, Kadmon R. 2017. Contrasting effects of water and nutrient European Research Council Synergy grant ERC-2013-SyG- additions on grassland communities: a global meta-analysis. Global Ecology and Biogeography 26: 983–992. 610028 IMBALANCE-P, the European Research Council (ERC) Elmendorf SC, Henry GHR, Hollister RD, Bj€ork RG, Bjorkman AD, Callaghan under the European Union’s Horizon 2020 research and innova- TV, Collier LS, Cooper EJ, Cornelissen JHC, Day TA et al. 2012a. Global tion programme (grant agreement no. 758873, TreeMort) and the assessment of experimental climate warming on tundra vegetation: heterogeneity FWF Austria Science Fund (Lise Meitner Programme M2714- over space and time. Ecology Letters 15: 164–175. B29). MFM is a postdoctoral fellow of the Research Foundation – Elmendorf SC, Henry GHR, Hollister RD, Bj€ork RG, Boulanger-Lapointe N, Cooper EJ, Cornelissen JHC, Day TA, Dorrepaal E, Elumeeva TG et al. 2012b. Flanders (FWO). This paper is number 51 of the Birmingham Plot-scale evidence of tundra vegetation change and links to recent summer Institute of Forest Research. warming. Nature Climate Change 2: 453–457. Elmendorf SC, Henry GHR, Hollister RD, Fosaa AM, Gould WA, Hermanutz L, Hofgaard A, Jonsdottir II, Jorgenson JC, Levesque E et al. 2015. Experiment, Author contributions monitoring, and gradient methods used to infer climate change effects on plant communities yield consistent patterns. Proceedings of the National Academy of DL and CZ developed the ideas, analysed the data and wrote the Sciences, USA 112: 448–452. manuscript; JP and TAMP conceived the ideas and provided in- Estiarte M, Vicca S, Pe~nuelas J, Bahn M, Beier C, Emmett BA, Fay PA, Hanson PJ, depth suggestions for the manuscript; and, RO and MF-M helped Hasibeder R, Kigel J et al. 2016. Few multiyear precipitation-reduction experiments find a shift in the productivity-precipitation relationship. Global analyse the data and revised the manuscript. Change Biology 22: 2570–2581. Fernandez-Martınez M, Sardans J, Chevallier F, Ciais P, Obersteiner M, Vicca S, Canadell JG, Bastos A, Friedlingstein P, Sitch S et al. 2019. Global trends in ORCID carbon sinks and their relationships with CO2 and temperature. Nature Climate Change 9: 73–79. Marcos Fernandez-Martınez https://orcid.org/0000-0002- Fick SE, Hijmans RJ. 2017. WorldClim 2: new 1-km spatial resolution climate 5661-3610 surfaces for global land areas. International Journal of Climatology 37: 4302–4315. Daijun Liu https://orcid.org/0000-0002-0993-0832 Grime JP, Fridley JD, Askew AP, Thompson K, Hodgson J, Bennett C. 2008. Roma Ogaya https://orcid.org/0000-0003-4927-8479 Long-term resistance to simulated climate change in an infertile grassland. Josep Pe~ nuelas https://orcid.org/0000-0002-7215-0150 Proceedings of the National Academy of Sciences, USA 105: 10028–10032. Halbritter AH, De Boeck HJ, Eycott AE, Reinsch S, Robinson DA, Vicca S, Thomas A. M. Pugh https://orcid.org/0000-0002-6242-7371 Berauer B, Christiansen CT, Estiarte M, Gr€ unzweig JM et al. 2020. The Chao Zhang https://orcid.org/0000-0001-7327-9477 handbook for standardized field and laboratory measurements in terrestrial climate change experiments and observational studies (ClimEx). Methods in Ecology and Evolution 11: 22–37. Data availability Harpole WS, Sullivan LL, Lind EM, Firn J, Adler PB, Borer ET, Chase J, Fay PA, Hautier Y, Hillebrand H et al. 2016. Addition of multiple limiting resources All the data and code for making graphs are available to readers reduces grassland diversity. Nature 537: 93–96. upon reasonable request (d.liu@creaf.uab.es and Harrison S, Spasojevic MJ, Li D. 2020. Climate and plant community diversity in chao.x.zhang@helsinki.fi). space and time. Proceedings of the National Academy of Sciences, USA 117: 4464–4470. Harte J, Saleska SR, Levy C. 2015. Convergent ecosystem responses to 23-year ambient and manipulated warming link advancing snowmelt and shrub References encroachment to transient and long-term climate-soil carbon feedback. Global Andresen LC, de Dato G, Dukes JS, Emmett BA, Estiarte M, Jentsch A, Change Biology 21: 2349–2356. Kroel-Dulay G, Luscher A, Niu S, Penuelas J et al. 2016. Shifting impacts Hoegh-Guldberg O, Jacob D, Taylor M, Bindi M, Brown S, Camilloni I, of climate change: Long-term patterns of plant response to elevated CO2, Diedhiou A, Djalante R. 2018. Chapter 3: impacts of 1.5°C global warming on drought, and warming across ecosystems. Advances in Ecological Research 55: natural and human systems. In: Masson-Delmotte V, Zhai P, P€ortner HO, 437–473. Roberts D, Skea J, Shukla PR, Pirani A, Moufouma-Okia W, Pean C, Pidcock R Bai Y, Han X, Wu J, Chen Z, Li L. 2004. Ecosystem stability and compensatory et al. eds. Global warming of 1.5 °C. Special report, Intergovernmental Panel on effects in the Inner Mongolia grassland. Nature 431: 181–184. Climate Change. Geneva, Switzerland: World Meteorological Organization Beier C, Beierkuhnlein C, Wohlgemuth T, Pe~ nuuelas J, Emmett B, Korner C, de Technical Document, 175–311. Boeck H, Christensen JH, Leuzinger S, Janssens IA et al. 2012. Precipitation Hooper DU, Adair EC, Cardinale BJ, Byrnes JEK, Hungate BA, Matulich KL, manipulation experiments-challenges and recommendations for the future. Gonzalez A, Duffy JE, Gamfeldt L, Connor MI. 2012. A global synthesis reveals Ecological Letters 15: 899–911. biodiversity loss as a major driver of ecosystem change. Nature 486: 105–108. Bellard C, Bertelsmeier C, Leadley P, Thuiller W, Courchamp F. 2012. Impacts of Hudson JMG, Henry GHR. 2010. High Arctic plant community resists 15 years of climate change on the future of biodiversity. Ecology Letters 15: 365–377. experimental warming. Journal of Ecology 98: 1035–1041. Berdugo M, Delgado-Baquerizo M, Soliveres S, Hernandez-Clemente R, Zhao Y, Hungate BA, Barbier EB, Ando AW, Marks SP, Reich PB, van Gestel N, Tilman Gaita n JJ, Gross N, Saiz H, Maire V, Lehman A et al. 2020. Global ecosystem D, Knops JMH, Hooper DU, Butterfield BJ et al. 2017. The economic value of thresholds driven by aridity. Science 367: 787–790. grassland species for carbon storage. Science Advances 3: e1601880. Collins SL, Koerner SE, Plaut JA, Okie JG, Brese D, Calabrese LB, Carvajal A, Huxman TE, Smith MD, Huxman TE, Loik ME, Smith SD, Williams DG, Sala Evansen RJ, Nonaka E. 2012. Stability of tallgrass prairie during a 19-year OE, Haddad BM, Fay PA, Schwinning S et al. 2004. Convergence across biomes increase in growing season precipitation. Functional Ecology 26: 1450–1459. to a common rain-use efficiency. Nature 429: 651–654. © 2021 The Authors New Phytologist (2021) 230: 1761–1771 New Phytologist © 2021 New Phytologist Foundation www.newphytologist.com New 1770 Review Research review Phytologist Jentsch A, Kreyling J, Beierkuhnlein C. 2007. A new generation of climate-change sampling biases understanding of climate change impacts across the Arctic. Nature experiments: events, not trends. Frontiers in Ecology and the Environment 5: 365– Ecology and Evolution 2: 1443–1448. 374. Morecroft MD, Duffield S, Harley M, Pearce-Higgins JW, Stevens N, Watts O, Jump AS, Pe~ nuelas J. 2005. Running to stand still: adaptation and the response of Whitaker J. 2019. Measuring the success of climate change adaptation and plants to rapid climate change. Ecology Letters 8: 1010–1020. mitigation in terrestrial ecosystems. Science 366: 6471. Van Kleunen M, Dawson W, Essl F, Pergl J, Winter M, Weber E, Kreft H, Weigelt Myers-Smith IH, Elmendorf SC, Beck PSA, Wilmking M, Hallinger M, Blok D, P, Kartesz J, Nishino M et al. 2015. Global exchange and accumulation of non- Rayback SA, Ravolainen V, Lantz TC, Baittinger C et al. 2015. Climate native plants. Nature 525: 100–103. sensitivity of shrub growth across the tundra biome. Nature Climate Change 5: Knapp AK, Harper CW, Danner BT, Lett MS, Fay PA, Blair JJM, Collins 887–891. SL, Smith MD, Carlisle JD, Harper CW et al. 2002. Rainfall variability, Nolan C, Overpeck J, Allen JRM, Anderson P, Betancourt JL, Binney HA, Simon carbon cycling, and plant species diversity in a mesic grassland. Science 298: B, Bush MB, Chase BM. 2018. Past and future global transformation of 2202–2205. terrestrial ecosystems under climate change. Science 361: 920–923. Knapp AK, Hoover DL, Wilcox KR, Avolio ML, Koerner SE, La Pierre KJ, Loik Pasek J, Pasek MJ, Hmisc D. 2020. Package ‘weights’. [WWW document] URL ME, Luo Y, Sala OE, Smith MD. 2015. Characterizing differences in https://www.rdocumentation.org/packages/weights/versions/1.0.1. precipitation regimes of extreme wet and dry years: implications for climate Pe~ nuelas J, Ciais P, Canadell JG, Janssens IA, Ferna ndez-Martınez M, Carnicer J, change experiments. Global Change Biology 21: 2624–2633. Obersteiner M, Piao S, Vautard R, Sardans J. 2017. Shifting from a fertilization- Komatsu KJ, Avolio ML, Lemoine NP, Isbell F, Grman E, Houseman GR, dominated to a warming-dominated period. Nature Ecology and Evolution 1: Koerner SE, Johnson DS, Wilcox KR, Alatalo JM et al. 2019. Global change 1438–1445. effects on plant communities are magnified by time and the number of global Pe~ nuelas J, Prieto P, Beier C, Cesaraccio C, de Angelis P, de Dato G, Emmett BA, change factors imposed. Proceedings of the National Academy of Sciences, USA 116: Estiarte M, Garadnai J, Gorissen A et al. 2007. Response of plant species richness 17867–17873. and primary productivity in shrublands along a north-south gradient in Europe to Kr€oel-Dulay G, Ransijn J, Schmidt IK, Beier C, De Angelis P, de Dato G, Dukes seven years of experimental warming and drought: reductions in primary JS, Emmett B, Estiarte M, Garadnai J et al. 2015. Increased sensitivity to climate productivity in the heat and drought year of 2003. Global Change Biology 13: change in disturbed ecosystems. Nature Communications 6: 6682. 2563–2581. Ladron de Guevara M, Gozalo B, Raggio J, Lafuente A, Prieto M, Maestre FT. Reich PB, Hobbie SE. 2013. Decade-long soil nitrogen constraint on the CO2 2018. Warming reduces the cover, richness and evenness of lichen-dominated fertilization of plant biomass. Nature Climate Change 3: 278–282. biocrusts but promotes moss growth: insights from an 8 yr experiment. New Reinsch S, Koller E, Sowerby A, De Dato G, Estiarte M, Guidolotti G, Kova cs- Phytologist 220: 811–823. La ng E, Kr€oel-Dulay G, Lellei-Kova cs E, Larsen KS et al. 2017. Shrubland LaForgia ML, Spasojevic MJ, Case EJ, Latimer AM, Harrison SP. 2018. Seed primary production and soil respiration diverge along European climate gradient. banks of native forbs, but not exotic grasses, increase during extreme drought. Scientific Reports 7: 1–7. Ecology 99: 896–903. Rumpf SB, H€ ulber K, Klonner G, Moser D, Sch€ utz M, Wessely J, Willner W, Langley JA, Chapman SK, La Pierre KJ, Avolio M, Bowman WD, Johnson DS, Zimmermann NE, Dullinger S. 2018. Range dynamics of mountain plants Isbell F, Wilcox KR, Foster BL, Hovenden MJ et al. 2018. Ambient changes decrease with elevation. Proceedings of the National Academy of Sciences, USA 115: exceed treatment effects on plant species abundance in global change experiments. 1848–1853. Global Change Biology 24: 5668–5679. Scheffers B, De Meester L, Tom B, Hoffmann A, Pandolfi JM, Corlett RT, Lenoir J, Gegout JC, Marquet PA, de Ruffray P, Brisse H. 2008. A significant Butchart SHM. 2016. The broad footprint of climate change from genes to upward shift in plant species optimum elevation during the 20th century. Science biomes to people. Science 354: 6313. 320: 1768–1771. Seddon AWR, Macias-Fauria M, Long PR, Benz D, Willis KJ. 2016. Sensitivity of Leuzinger S, Luo Y, Beier C, Dieleman W, Vicca S, Ko C. 2011. Do global change global terrestrial ecosystems to climate variability. Nature 531: 229–232. experiments overestimate impacts on terrestrial ecosystems? Trends in Ecology and Seebens H, Essl F, Blasius B. 2017. The intermediate distance hypothesis of Evolution 26: 236–241. biological invasions. Ecology Letters 20: 158–165. Liu D, Estiarte M, Ogaya R, Yang X, Pe~ nuelas J. 2017. Shift in community Shi Z, Lin Y, Wilcox KR, Souza L, Jiang L, Jiang J, Jung CG, Xu X, Yuan M, Guo X structure in an early-successional Mediterranean shrubland driven by long-term et al. 2018. Successional change in species composition alters climate sensitivity of experimental warming and drought and natural extreme droughts. Global Change grassland productivity. Global Change Biology 24: 4993–5003. Biology 23: 4267–4279. Smith MD, Knapp AK, Collins SL. 2009. A framework for assessing ecosystem Liu D, Llusia J, Ogaya R, Estiarte M, Llorens L, Yang X, Pe~ nuelas J. 2016. dynamics in response to chronic resource alterations induced by global change. Physiological adjustments of a Mediterranean shrub to long-term experimental Ecology 79: 3–24. warming and drought treatments. Plant Science 252: 53–61. Smith MD, La Pierre KJ, Collins SL, Knapp AK, Gross KL, Barrett JE, Frey SD, Liu D, Ogaya R, Barbeta A, Yang X, Pe~ nuelas J. 2018b. Long-term experimental Gough L, Miller RJ, Morris JT et al. 2015. Global environmental change and the drought combined with natural extremes accelerate vegetation shift in a nature of aboveground net primary productivity responses: insights from long- Mediterranean holm oak forest. Environmental and Experimental Botany 151: 1– term experiments. Oecologia 177: 935–947. 11. Smith MD, Wilcox KR, Power SA, Tissue DT, Knapp AK. 2017. Assessing Liu D, Pe~ nuelas J, Ogaya R, Estiarte M, Tielb€orger K, Slowik F, Yang X, Bilton community and ecosystem sensitivity to climate change – toward a more MC. 2018c. Species selection under long-term experimental warming and comparative approach. Journal of Vegetation Science 28: 235–237. drought explained by climatic distributions. New Phytologist 217: 1494–1506. Song J, Wan S, Piao S, Knapp AK, Classen AT, Vicca S, Ciais P, Hovenden MJ, Liu D, Zhang C, Ogaya R, Estiarte M, Pe~ nuelas J. 2020. Effects of decadal Leuzinger S, Beier C et al. 2019. A meta-analysis of 1,119 manipulative experimental drought and climate extremes on vegetation growth in experiments on terrestrial carbon-cycling responses to global change. Nature Mediterranean forests and Shrublands. Journal of Vegetation Science 31: 768–779. Ecology and Evolution 3: 1309–1320. Liu H, Mi Z, Lin L, Wang Y, Zhang Z, Zhang F, Wang H, Liu L. 2018a. Shifting Steinbauer MJ, Grytnes JA, Jurasinski G, Kulonen A, Lenoir J, Pauli H, Rixen C, plant species composition in response to climate change stabilizes grassland Winkler M, Bardy-Durchhalter M, Barni E et al. 2018. Accelerated increase in primary production. Proceedings of the National Academy of Sciences, USA 115: plant species richness on mountain summits is linked to warming. Nature 556: 4051–4056. 231–234. Messier J, Myers-Smith IH, McCune JL, Becker-Scarpitta A, Sax DF, Vellend M, Sullivan MJP, Thomsen MA, Suttle KB. 2016. Grassland responses to increased Baeten L, Boucher-Lalonde V. 2017. Plant biodiversity change across scales rainfall depend on the timescale of forcing. Global Change Biology 22: 1655–1665. during the Anthropocene. Annual Review of Plant Biology 68: 563–586. Thakur MP, Tilman D, Purschke O, Ciobanu M, Cowles J, Isbell F, Wragg PD, Metcalfe DB, Hermans TDG, Ahlstrand J, Becker M, Berggren M, Bj€ork RG, Eisenhauer N. 2017. Climate warming promotes species diversity, but with greater Bj€orkman MP, Blok D, Chaudhary N, Chisholm C et al. 2018. Patchy field taxonomic redundancy, in complex environments. Science Advances 3: 1–10. New Phytologist (2021) 230: 1761–1771 © 2021 The Authors www.newphytologist.com New Phytologist © 2021 New Phytologist Foundation New Phytologist Research review Review 1771 Tielb€orger K, Bilton MC, Metz J, Kigel J, Holzapfel C, Lebrija-Trejos E, Konsens Fig. S1Relationships between the sensitivity of aboveground net I, Parag HA, Sternberg M. 2014. Middle-Eastern plant communities tolerate 9 primary production (ANPP) and habitat contexts. years of drought in a multi-site climate manipulation experiment. Nature Communications 5: 1–9. Trisos CH, Merow C, Pigot AL. 2020. The projected timing of abrupt ecological Table S1 Detailed information for the study site. disruption from climate change. Nature 580: 496–501. Vellend M, Baeten L, Myers-Smith IH, Elmendorf SC, Beausejour R, Brown CD, Table S2 Results of t-test for the vegetation sensitivity to De Frenne P, Verheyen K, Wipf S. 2013. Global meta-analysis reveals no net experimental climate change. change in local-scale plant biodiversity over time. Proceedings of the National Academy of Sciences, USA 110: 19456–19459. Walker TWN, Janssens IA, Weedon JT, Sigurdsson BD, Richter A, Pe~ nuelas J, Table S3 Relationships between sensitivity of aboveground Leblans NIW, Bahn M, Bartrons M, De Jonge C et al. 2020. A systemic biomass (AGB) and aboveground net primary production (ANPP) overreaction to years versus decades of warming in a subarctic grassland ecosystem. in treatments for whole-year (WY) and habitat contexts. Nature Ecology and Evolution 4: 101–108. Wan S, Xia J, Liu W, Niu S. 2009. Photosynthetic overcompensation under nocturnal warming enhances grassland carbon sequestration. Ecology 90: 2700–2710. Table S4 Differences between dry and nonwater-limited sites for Wang N, Quesada B, Xia L, Butterbach-Bahl K, Goodale CL, Kiese R. 2019. vegetation variables to climatic experiments. Effects of climate warming on carbon fluxes in grasslands – a global meta-analysis. Global Change Biology 25: 1839–1851. Table S5 Differences between short-term and long-term climatic Wilcox KR, von Fischer JC, Muscha JM, Petersen MK, Knapp AK. 2015. experiments for the whole year. Contrasting above- and belowground sensitivity of three Great Plains grasslands to altered rainfall regimes. Global Change Biology 21: 335–344. Wilcox KR, Shi Z, Gherardi LA, Lemoine NP, Koerner SE, Hoover DL, Bork E, Table S6 Relationships between sensitivity of species richness (SR) Byrne KM, Cahill J, Collins SL et al. 2017. Asymmetric responses of primary and composition (H) to climatic change for whole-year (WY) and productivity to precipitation extremes: a synthesis of grassland precipitation habitat contexts. manipulation experiments. Global Change Biology 23: 4376–4385. Wright AJ, Wardle DA, Callaway R, Gaxiola A. 2017. The overlooked role of facilitation in biodiversity experiments. Trends in Ecology and Evolution 32: 383– Table S7 Differences between dry and nonwater-limited sites for 390. vegetation variables to climatic experiments. Wu Z, Dijkstra P, Koch GW, Pe~ nuelas J, Hungate BA. 2011. Responses of terrestrial ecosystems to temperature and precipitation change: a meta-analysis of Table S8 Overall and individual trend of aboveground net primary experimental manipulation. Global Change Biology 17: 927–942. production (ANPP) and species diversity (SR) sensitivity to climate Yuan ZY, Jiao F, Li YH, Kallenbach RL. 2016. Anthropogenic disturbances are key to maintaining the biodiversity of grasslands. Scientific Reports 6: 1–8. treatments over time. Zomer RJ, Trabucco A, Bossio DA, Verchot LV. 2008. Climate change mitigation: a spatial analysis of global land suitability for clean development mechanism Table S9 Overall and individual trend of SR sensitivity to warming afforestation and reforestation. Agriculture, Ecosystems & Environment 126: 67–80. and irrigation over time. Supporting Information Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the Additional Supporting Information may be found online in the authors. Any queries (other than missing material) should be Supporting Information section at the end of the article. directed to the New Phytologist Central Office. © 2021 The Authors New Phytologist (2021) 230: 1761–1771 New Phytologist © 2021 New Phytologist Foundation www.newphytologist.com