Soil moisture dominates dryness stress on ecosystem production globally

Dryness stress significantly impacts vegetation growth, posing substantial threats to agricultural production and driving widespread tree mortality. The capacity of land ecosystems to act as a future carbon sink is also heavily reliant on soil moisture's influence on ecosystem carbon fluxes. Understanding the relative contributions of low soil moisture (SM) and high atmospheric water demand (vapor pressure deficit, VPD) to dryness stress is crucial for managing drought risks and improving predictions of future land carbon uptake. While low SM directly affects plant water availability, high VPD can induce stomatal closure, limiting photosynthesis. The scientific literature and models show disagreement on the relative importance of these factors. Some studies highlight SM's direct impact on plant water supply as the primary driver of dryness stress, while others emphasize VPD's role in constraining photosynthesis through stomatal regulation. This discrepancy stems from the strong coupling between SM and VPD through land-atmosphere interactions, making it challenging to disentangle their individual effects. This study uses combined satellite observations of solar-induced chlorophyll fluorescence (SIF) and climate datasets to decouple the SM and VPD correlations and to determine their respective contributions to limiting ecosystem production globally.

Existing studies present conflicting views on the relative roles of soil moisture (SM) and vapor pressure deficit (VPD) in limiting ecosystem production under dryness stress. Some research emphasizes the direct impact of SM on plant water availability, showcasing its success in capturing the consequences of dryness on vegetation productivity. These studies often use low precipitation or SM availability as indicators of vegetation dryness stress, which are well documented to successfully capture the consequences of dryness on vegetation productivity, and result in feedbacks of plants' activity to climate. Other studies highlight the importance of VPD, suggesting it might exert stronger effects than SM on ecosystem water and carbon fluxes by inducing stomatal closure and constraining photosynthesis. This leads to diverse representations of dryness stress in terrestrial ecosystem models (TEMs), with some using SM only, others VPD only, and some incorporating both. The lack of consensus underscores the need for a global-scale analysis that decouples the influence of SM and VPD to accurately assess their relative contributions to dryness stress on ecosystem production.

This study utilizes multiple independent satellite observations of solar-induced chlorophyll fluorescence (SIF) and climate datasets to decouple the strong correlations between SM and VPD and to disentangle their respective effects on ecosystem production. The primary SIF data used is a spatially continuous OCO-2 SIF dataset (CSIF) generated using a machine-learning algorithm, trained on observations from the Orbiting Carbon Observatory-2 (OCO-2). Additional SIF data from GOME-2 and SCIAMACHY missions are also used for comparison. Soil moisture data are obtained from ERA-Interim, MERRA-2 reanalysis products, and the ESA CCI satellite product. Daily precipitation, near-surface temperature, and photosynthetically active radiation (PAR) are acquired from ERA-Interim, MERRA-2, and NASA's CERES. VPD is calculated using temperature and humidity data. To address the strong SM-VPD coupling, the data are binned into percentiles of either SM or VPD, effectively decoupling the two variables at daily or finer scales. The analysis assumes that if SM dominates, low SM will limit production regardless of VPD variations; conversely, if VPD dominates, high VPD will limit production regardless of SM. The effects of SM and VPD on SIF are quantified using two approaches: (1) calculating the difference in SIF between highest and lowest VPD or SM bins within each SM or VPD bin, respectively, and (2) using linear regression to estimate the changes in SIF caused by VPD or SM. Analysis focuses on growing seasons and days where SM and VPD are likely to be the dominant drivers, filtering out days when other meteorological variables are more influential. The sensitivity of SIF to SM is also calculated to compare across vegetation types and climate gradients. The study further analyzes the dependence of SM stress on climate and vegetation gradients by examining variations in SM limitation effects across aridity and tree cover gradients, as well as exploring variations in SIF sensitivity to SM within different plant functional types.

The analysis reveals that soil moisture (SM) is the dominant driver of dryness stress on ecosystem production across most vegetated land areas (71.3% with valid data), significantly exceeding the impact of vapor pressure deficit (VPD). The strong negative correlation between yearly SM and VPD highlights the confounding effect of their coupling. By binning daily data according to SM or VPD percentiles, the study decouples these variables and demonstrates that the apparent VPD limitation on ecosystem production is largely a byproduct of the SM-VPD coupling. In many regions, the effect of VPD on solar-induced chlorophyll fluorescence (SIF), a proxy for gross primary production (GPP), is negligible when SM is controlled for. SM limitation effects are particularly pronounced in semi-arid ecosystems, which are also major contributors to interannual variability in global terrestrial CO2 flux. Areas with lower tree cover also show a larger response to SM stress. The sensitivity of ecosystem production to changes in SM varies considerably even within the same plant functional type (PFT), highlighting the influence of plant-specific traits and processes not fully captured by PFT classifications. A change from the wettest to the driest SM under constant VPD reduces SIF by up to 14.9% on average, while a change in VPD from lowest to highest quantiles under constant SM has a much smaller effect on SIF (-3.8%) on average. The spatial patterns of SM and VPD effects on SIF are robust across different data sets and analysis methods.

These findings challenge previous studies that overestimated the role of VPD in limiting ecosystem production by not adequately accounting for the strong SM-VPD coupling. The results highlight the crucial need for models to accurately disentangle the respective limitations of SM and VPD to reliably predict dryness stress on ecosystems and associated drought risks. The strong dominance of SM in most vegetated regions underscores the importance of considering SM dynamics in climate change impact assessments. The observed variation in SM sensitivity within the same PFTs suggests the need to incorporate more plant-specific traits and hydraulic processes into ecosystem models. This research provides valuable insights for improving models and predictions of terrestrial carbon fluxes and their contributions to climate change.

This study demonstrates that soil moisture, rather than vapor pressure deficit, is the dominant driver of drought-induced limitations on vegetation productivity at the ecosystem scale across most vegetated areas. The strong SM-VPD coupling has confounded previous assessments. The findings underscore the need to revise models to properly account for the independent roles of soil moisture and vapor pressure deficit, improving drought risk management and predictions of terrestrial carbon fluxes. Future research should focus on incorporating observed variations in SM sensitivity across PFTs and incorporating more plant hydraulic processes into the next generation of terrestrial ecosystem models.

The study primarily relies on satellite-based observations of soil moisture and SIF, which may have limitations in spatial and temporal resolution and accuracy compared to in-situ measurements. The analysis focuses on relatively shallow soil water; deep soil moisture or other water storage mechanisms may also be relevant and could potentially influence the results. The use of SIF as a proxy for GPP introduces uncertainties related to the complex relationship between fluorescence and photosynthesis. Further, the analysis assumes a linear relationship between variables in some instances; non-linear relationships may affect the interpretation of results.