Terrestrial ecosystems are vital for global food and water security, and play a critical role in carbon sequestration through CO2 uptake. Their functionality is profoundly impacted by soil moisture availability. Climate change is expected to alter soil moisture patterns, but the exact consequences for vegetation remain uncertain. This study aims to investigate how climate change will affect soil moisture limitation on vegetation growth and ecosystem function. Understanding this relationship is crucial because the provision of food and water, CO2 uptake, and evaporative cooling all hinge on adequate moisture supply. Climate change influences moisture supply, and in combination with rising atmospheric CO2, affects ecosystem function. The water and carbon cycles are interconnected through vegetation, which assimilates CO2 during photosynthesis while simultaneously transpiring water through stomata. Transpiration (T) cools the surface, reducing surface heating. Changes in soil moisture influence evaporative cooling and therefore surface warming through water-vegetation-climate feedback. However, regional changes in water availability don't uniformly impact ecosystem function; responses depend on whether a region is energy-limited or water-limited. Rising atmospheric CO2 is anticipated to influence physiological processes, leading to more favorable photosynthesis conditions and plant growth. This, however, has contrasting impacts on plant transpiration and thus the energy and water cycles. A key factor in predicting the future of the terrestrial carbon sink is the degree to which ecosystems become water-limited. There's currently a lack of agreement on trends in individual water-related variables like soil moisture and terrestrial evaporation (evapotranspiration). This discrepancy extends to traditional drought indices whether using observations, reanalyses, climate model simulations, or future climate projections. Analysis is complicated by various processes operating at different temporal scales and in different directions. For example, while widespread vegetation greening doesn't support increased water limitation, it's primarily driven by CO2 fertilization which can temporarily overshadow changes in water availability. Large-scale variability patterns like El Niño/Southern Oscillation influence inter-annual variability in ecosystem water limitation. Reconciling how ecosystem water limitation impacts vegetation (through drought stress, tree mortality, and changes in surface albedo and roughness) is essential. Existing uncertainties are partly due to differing approaches; some studies analyse water supply through soil moisture, while others focus on water demand using precipitation alongside (potential) evaporation. Increasing trends in energy availability, however, are clear, consistent, and coincide with rising temperatures. This affects ecosystems in multiple ways. RuBisCO enzyme activity (crucial for photosynthesis) is temperature-sensitive. Temperature also influences vapor pressure deficit; higher temperatures increase atmospheric evaporative demand, increasing ecosystem water limitation and causing plants to close stomata to prevent excessive water loss. This underscores the need to consider energy and water variables together to characterize ecosystem water limitation.

The existing literature presents conflicting views on the impact of climate change on terrestrial water resources and ecosystem function. Some studies suggest an expansion of global drylands and increased water limitation on vegetation growth, while others show that CO2 fertilization may temporarily offset the negative effects of water scarcity. This lack of consensus highlights the complexities involved in modeling and predicting the response of terrestrial ecosystems to a changing climate. Traditional drought indices, based primarily on meteorological data, have also proven to be insufficient to capture the full complexity of ecosystem responses to hydrological and meteorological variables. The studies reviewed used varying approaches to assess water limitation, with some focusing on soil moisture supply and others on water demand. This discrepancy contributes to inconsistencies in results, necessitating a more integrated approach to analyze energy and water limitation simultaneously.

This study uses an ecosystem limitation index (ELI) to investigate the interplay of energy and water limitations on ecosystem function. The ELI is calculated as the difference between the correlation of terrestrial evaporation anomalies with soil moisture anomalies (cor(SM', ET')) and the correlation of terrestrial evaporation anomalies with incoming shortwave radiation anomalies (cor(SW, ET')). Positive ELI values indicate water limitation, and negative values indicate energy limitation. The ELI moves beyond traditional drought indices by evaluating the functional ecosystem response to hydrometeorological conditions using monthly anomalies to analyze deviations from the seasonal cycle. Terrestrial evaporation represents the total ecosystem response, encompassing bare soil evaporation, canopy interception, and plant transpiration. Soil moisture reflects water available for evaporation, and incoming shortwave radiation serves as a proxy for energy availability, directly used by plants for photosynthesis. The study uses historical and 'worst-case' (SSP 5-8.5) simulations from the Coupled Model Intercomparison Project Phase 6 (CMIP6) from 11 models for the period 1980-2100. The ELI is calculated using monthly anomalies of soil moisture (SM), terrestrial evaporation (ET), and incoming shortwave radiation (SW). Air temperature is used as an alternative energy variable to ensure robustness of the results. Spatial analyses are performed on the warm land area (grid cells with air temperature > 10°C for at least four months in at least four models). Regions of interest are defined to facilitate comparison. The timing of regime shifts from energy to water limitation is determined by identifying the first decade after which ELI becomes positive. Trends in water-limited months-of-year are analyzed to investigate seasonal changes in water limitation. To attribute ELI trends to land-atmosphere variables, multivariate linear regression is employed, considering various hydrological, meteorological, and ecological predictors. The importance of individual predictors is evaluated using the Akaike Information Criterion (AIC) and variance explained. The robustness of the analyses is checked by considering different thresholds for model performance and using air temperature as an alternative energy proxy. The study also accounts for uncertainty related to different CMIP6 model representations of processes such as CO2 fertilization, water use efficiency, and dynamic vegetation.

The study reveals a consistent increase in the ELI from 1980 to 2100, signifying a widespread shift towards water limitation. This increase reflects a weakening correlation between terrestrial evaporation and incoming shortwave radiation and a strengthening correlation between terrestrial evaporation and soil moisture. Across approximately 73% of the warm land area, positive ELI trends (indicating increasing water limitation) are more widespread than negative trends. ELI increases tend to be strongest in regions with substantial tree coverage. The transition to water limitation is more pronounced in regions where increasing incoming shortwave radiation coincides with decreasing soil moisture. While there's an increase in global terrestrial evaporation until approximately 2030, there is substantial uncertainty from 2030 onward with some models suggesting increases and others decreases. Leaf area index (LAI) increases steadily, reflecting CO2 fertilization, contributing to increased plant transpiration. The fraction of transpiration relative to total terrestrial evaporation also increases over time. Spatial analysis reveals that positive ELI trends are widespread and stronger than negative trends, especially in regions with substantial tree coverage. This leads to an expansion of water-limited area, particularly in the Northern Hemisphere. Analysis of water-limited months-of-year shows that in many regions, the duration of water-limited periods increases substantially, sometimes extending into seasons that were previously energy-limited. Attribution analysis shows that incoming shortwave radiation is the most significant predictor of ELI trends across most regions, although other variables such as soil moisture, terrestrial evaporation, LAI, and aridity index are also important, particularly in specific regions and for explaining global patterns. The importance of the combined effect of changes in energy and water availability alongside ecosystem feedbacks is emphasized by the fact that the normalized ELI trend is even more prominent than its individual components.

The findings demonstrate a global increase in ecosystem water limitation driven largely by global warming. The strongest regional trends are attributed to a combination of reduced energy limitation and intensified water limitation. Incoming shortwave radiation is the most important predictor, but other variables are needed for a complete explanation of global patterns. This study resolves the debate regarding the relative importance of energy and water limitations in affecting terrestrial evaporation and ecosystem productivity by simultaneously considering both. While soil moisture globally plays a crucial role in determining whether ecosystems are water- or energy-limited, the trends in incoming shortwave radiation are more consistent and dominant in shaping ecosystem function trends. This widespread increase in ecosystem water limitation has far-reaching consequences for food and water security, land degradation, CO2 sequestration, biodiversity, and the frequency and intensity of extreme events.

This research provides robust evidence of a global shift toward ecosystem water limitation, primarily attributed to climate change. The findings highlight the critical need for considering both energy and water limitations to fully understand and predict future changes in terrestrial ecosystem function. Further research is needed to refine model representations of relevant processes, incorporate nutrient limitations, and investigate the specific impacts of this shift on various ecosystem services.

The study's accuracy is inherently limited by uncertainties in CMIP6 models. Differences in model representations of CO2 fertilization, water use efficiency, dynamic vegetation, and evaporative regime changes introduce uncertainties, primarily affecting the magnitude but not the sign of ELI trends. Nutrient limitation, a potentially significant factor in the future, is not considered due to data limitations. Despite these limitations, the multi-model mean terrestrial evaporation closely resembles state-of-the-art datasets, suggesting that the findings are robust.