Rapid shifts in grassland communities driven by climate change

Climate change is expected to alter species distributions, with cascading risks to biodiversity and ecosystem functioning. Terrestrial plant communities—especially forests—often show lagged compositional responses relative to the rapid pace of warming, implying future extinctions and community reshuffling. Grasslands, in contrast, may respond more rapidly because they are dominated by short-lived species and are more directly exposed to macroclimatic variation. Yet, evidence for grassland compositional tracking of climate is limited and mixed, with observational studies and manipulative experiments sometimes yielding inconsistent results. This study asks whether grassland communities in the California Floristic Province (CFP) closely track climate change by shifting composition consistently toward species associated with warmer and drier conditions (thermophilization and xerophilization), and whether the pace of these compositional shifts matches the rate of observed climate change.

Prior work has documented slow or lagged community responses to climate change in forests, including understories, across temperate and tropical regions, often framed as climatic debt. In grasslands, fewer studies have quantified the pace of compositional change relative to climate trends. Some long-term observations reported increases in grasses at the expense of forbs under warming and drying, whereas manipulative experiments have sometimes indicated different or context-dependent responses. This inconsistency underscores the need for extensive, long-term records and robust inference methods to evaluate whether and how grassland communities track climate change. Additionally, in temperate systems, thermophilization has sometimes correlated with mesophilization (increasing wetness), but Mediterranean climates like the CFP may exhibit a different coupling due to negative correlations between temperature and precipitation.

Study system and data: The study focuses on coastal grasslands within the California Floristic Province (CFP), a biodiversity hotspot spanning ~300,000 km² with strong geographic and climatic gradients and a Mediterranean climate that has warmed and dried over recent decades. The team compiled 12 long-term observational datasets (8–33 years; 1983–2021) across varied sites (coastal prairies, valley grasslands, serpentine grasslands) and three global change experiments. Species climatic niches: Occurrence records were downloaded from GBIF (filtered for quality and uncertainty <10 km; duplicates removed; synonyms harmonized). Species with ≥100 occurrences were retained, yielding 829,337 records across 349 species (including 104 non-native). Climatic niches were quantified using CHELSA 1981–2010 climatologies; for each species, the niche centroid was defined as the median mean annual temperature and median annual precipitation across occurrences. Robustness was assessed against alternative datasets, variables, summaries, and thinning approaches. Temperature and precipitation niche centroids were significantly negatively correlated across species (r = −0.671, P ≤ 0.001). Community indices: For each community (plot-year), the Community Temperature Index (CTI) and Community Precipitation Index (CPI) were calculated as community-weighted means of species’ temperature and precipitation niche centroids, respectively, with species’ relative abundances as weights (point-intercept hits or percent cover), following CTI_sti = Σ A_stij T_j / Σ A_stij and CPI_sti = Σ A_stij P_j / Σ A_stij. Observational analyses: Site-specific linear regressions summarized CTI and CPI trends over time. Overall trends across the 12 sites were estimated using linear mixed-effects models with site-level random intercepts and slopes to account for spatial and temporal correlation: CTI_sti = β0 + β0s + (β1 + β1s)t + ε_sti (analogous model for CPI). Climate trends (mean annual temperature, annual precipitation) for 1980–2019 at each site were derived from CHELSA and modeled with mixed-effects to estimate overall warming and drying rates across sites. Experimental analyses: The Jasper Ridge Global Change Experiment (JRGCE) warming treatment (heaters; 16 growing seasons, 1999–2014) progressed through three phases: +80 W m−2 (~+1 °C; 1999–2002), +100 W m−2 (~+1.5 °C; 2003–2009), and +250 W m−2 (~+2 °C; 2010–2014). Community composition was measured via point-intercept. Using the factorial design, 72 ambient and 64 warming subplots were analyzed. Treatment effects on CTI and CPI were estimated with linear mixed-effects models including plot random intercepts and year effects; annual ambient vs warming contrasts were also examined. Additional precipitation manipulations included: (1) JRGCE watering (+50% ambient rainfall plus late-season additions); (2) McLaughlin Water Experiment (2015–2021): watering to bring totals to the 30-year weekly average (serpentine and non-serpentine sites) and drought via shelters simulating ~70% exclusion on serpentine soils; and (3) Santa Cruz International Drought Experiment (2015–2021) at three sites using shelters intercepting ~60% precipitation. Effects were analyzed with similar mixed-effects models. Species-level responses: For observations, species that increased or decreased over time, established (appeared after first 5 years), or extirpated (absent in last 5 years) were identified via tests on abundance trends. For experiments, species abundance differences between warming and ambient were tested. Climatic niche centroids among increasing, decreasing, and unchanged species were compared using Wilcoxon tests. Community evenness and rank abundance curves were assessed to evaluate whether dominant or rare species drove shifts. Significance was evaluated with two-sided tests at P ≤ 0.05 without multiple-comparison adjustment, given independent datasets/experiments. Taxonomy and standardization: Names were harmonized using automated tools and expert curation (Jepson eFlora), including consolidation of synonyms and assignment of dummy species for some genus-level identifications (Avena, Festuca, Hypochaeris) by averaging congeners’ niches. Abundance standardized as relative intercepts or cover.

- Species’ climatic niches spanned wide gradients, with temperature niche medians from 8.15 °C to 17.3 °C and precipitation niche medians from 266 to 1,410 mm; temperature and precipitation niches were negatively correlated (r = −0.671, P ≤ 0.001). - Across 12 long-term observational sites (176 total plot-years), communities thermophilized at 0.0216 ± 0.00592 °C yr−1 (P = 0.0053) and xerophilized at −3.04 ± 0.742 mm yr−1 (P = 0.0019), comparable in magnitude to concurrent climate trends at the sites (warming 0.0177 ± 0.00260 °C yr−1; drying −4.22 ± 1.48 mm yr−1, 1980–2019). - Eight of 12 sites showed significant increases in CTI and decreases in CPI over time; four sites showed no significant trends, potentially reflecting influences of soils (e.g., serpentine), microclimate, topography, propagule pressure, disturbance, and native vs non-native dominance. - Experimental warming (JRGCE): Significant effects emerged under the strongest warming (phase 3, +250 W m−2, ~+2 °C): CTI increased by 0.148 ± 0.0249 °C and CPI decreased by −17.3 ± 2.76 mm relative to ambient. No significant differences were detected in phases 1–2. Shifts could occur within a year under strong warming. - Precipitation manipulations: Drought treatments led to no change or warmer/drier compositions (higher CTI, lower CPI), whereas watering generally led to cooler/wetter compositions (lower CTI, higher CPI), consistent with warming-driven thermophilization and xerophilization. - Coupled shifts: Across observations and experiments, thermophilization and xerophilization followed a common trajectory in CTI–CPI space; a 0.1 °C thermophilization corresponded to a −12.3 mm median xerophilization in observations and −9.28 mm in the experiment. The magnitude of shifts scaled with the strength/duration of climate change. - Species-level drivers: Species that increased in abundance over time or under warming were generally associated with warmer and drier niches (Wilcoxon P ≤ 0.05). No consistent pattern indicated that only dominant or only rare species drove community shifts. - The approach detected nuanced compositional changes not evident when grouping species by native/non-native, annual/perennial, or grass/forb guilds.

Grassland communities in the coastal CFP shifted composition rapidly and consistently toward species associated with warmer and drier conditions, at rates comparable to observed climatic warming and drying. This contrasts with lagged responses widely reported for forests and forest understories. Several mechanisms may underpin the rapid response of grasslands: dominance by annuals and short-lived perennials that allow fast population turnover; limited buffering by microclimates relative to forests, leading to greater exposure to macroclimate; and the prevalence of non-native species adapted to warmer, drier conditions, especially on non-serpentine soils. Region-wide xerophilization is a notable feature of this Mediterranean climate system and reflects the negative coupling between species’ temperature and precipitation niches and concurrent warming and drying. The strong coupling and the narrow available niche space suggest that as climates move beyond current species’ niche bounds, communities may face constraints in tracking multidimensional climate change, potentially limiting future compositional adjustment. The consistency between long-term observations and manipulative experiments strengthens causal inference that warming and drying are driving these compositional shifts.

This study integrates long-term observations across 12 grassland sites and multiple climate manipulation experiments to demonstrate that coastal CFP grassland communities rapidly and consistently track climate change via thermophilization and xerophilization, at rates similar to background warming and drying. The shifts are evident in community-weighted climatic indices and are driven by species associated with warmer and drier niches, with magnitudes scaling with the strength of climate forcing. These findings imply imminent and potentially large biodiversity and interaction changes in grassland ecosystems and highlight constraints on communities’ capacity to track multidimensional climatic change due to the structure of available niche space. Future research should incorporate intraspecific variation and acclimation/adaptation, expand coverage to interior grassland types (e.g., valley grasslands, oak–grass savannas), and further disentangle the separate and interactive effects of warming and drying.

- Climatic niche estimates did not incorporate intraspecific variation or adaptive/acclimatory responses, potentially limiting generality. - Data were concentrated in coastal grasslands; interior grassland types (e.g., valley grasslands, oak–grass savannas) were underrepresented. - Experimental warming also induced drying, limiting the ability to fully isolate independent effects of temperature and precipitation changes.