Climate extremes are on the rise. Impacts of extreme climate and weather events on ecosystem services and ultimately human well-being can be partially attenuated by the organismic, structural, and functional diversity of the affected land surface. However, the ongoing transformation of terrestrial ecosystems through intensified exploitation and management may put this buffering capacity at risk. Here, we summarise the evidence that reductions in biodiversity can destabilise the functioning of ecosystems facing climate extremes. We then explore if impaired ecosystem functioning could, in turn, exacerbate climate extremes. We argue that only a comprehensive approach, incorporating both ecological and hydrometeorological perspectives, enables to understand and predict the entire feedback system between altered biodiversity and climate extremes. This ambition, however, requires a reformulation of current research priorities to emphasise the bidirectional effects that link ecology and atmospheric processes.

The land sink of anthropogenic carbon emissions, a crucial component of mitigating climate change, is primarily attributed to the CO₂ fertilization effect on global gross primary productivity (GPP). However, direct observational evidence of this effect remains scarce, hampered by challenges in disentangling the CO₂ fertilization effect from other long-term drivers, particularly climatic changes. Here, we introduce a novel statistical approach to separate the CO₂ fertilization effect on GPP and daily maximum net ecosystem production (NEPmax) using eddy covariance records across 38 extratropical forest sites. We find the median stimulation rate of GPP and NEPmax to be 16.4 ± 4% and 17.2 ± 4% per 100 ppm increase in atmospheric CO₂ across these sites, respectively. To validate the robustness of our findings, we test our statistical method using factorial simulations of an ensemble of process-based land surface models. We acknowledge that additional factors, including nitrogen deposition and land management, may impact plant productivity, potentially confounding the attribution to the CO₂ fertilization effect. Assuming these site-specific effects offset to some extent across sites as random factors, the estimated median value still reflects the strength of the CO₂ fertilization effect. However, disentanglement of these long-term effects, often inseparable by timescale, requires further causal research. Our study provides direct evidence that the photosynthetic stimulation is maintained under long-term CO₂ fertilization across multiple eddy covariance sites. Such observation-based quantification is key to constraining the long-standing uncertainties in the land carbon cycle under rising CO₂ concentrations.

Fire is a crucial factor in terrestrial ecosystems playing a role in disturbance for vegetation dynamics. Process-based fire models quantify fire disturbance effects in stand-alone dynamic global vegetation models (DGVMs) and their advances have incorporated both descriptions of natural processes and anthropogenic drivers. Nevertheless, these models show limited skill in modeling fire events at the global scale, due to stochastic characteristics of fire occurrence and behavior as well as the limits in empirical parameterizations in process-based models. As an alternative, machine learning has shown the capability of providing robust diagnostics of fire regimes. Here, we develop a deep-learning-based fire model (DL-fire) to estimate daily burnt area fraction at the global scale and couple it within JSBACH4, the land surface model used in the ICON ESM. The stand-alone DL-fire model forced with meteorological, terrestrial and socio-economic variables is able to simulate global total burnt area, showing 0.8 of monthly correlation (rm) with GFED4 during the evaluation period (2011-15). The performance remains similar with the hybrid modeling approach JSB4-DL-fire (rm=0.79) outperforming the currently used uncalibrated standard fire model in JSBACH4 (rm=-0.07). We further quantify the importance of each predictor by applying layer-wise relevance propagation (LRP). Overall, land properties, such as fuel amount and water content in soil layers, stand out as the major factors determining burnt fraction in DL-fire, paralleled by meteorological conditions over tropical and high latitude regions. Our study demonstrates the potential of hybrid modeling in advancing fire prediction in ESMs by integrating deep learning approaches in physics-based dynamical models.

The process of evapotranspiration transfers water vapour from vegetation and soil surfaces to the atmosphere, the so-called latent heat flux (𝑄 LE), and thus crucially modulates Earth’s energy, water, and carbon cycles. Vegetation controls 𝑄 LE through regulating the leaf stomata (i.e., surface resistance 𝑟 s) and through altering surface roughness (aerodynamic resistance 𝑟 a). Estimating 𝑟 s and 𝑟 a across different vegetation types proves to be a key challenge in predicting 𝑄 LE. Here, we propose a hybrid modeling approach (i.e., combining mechanistic modeling and machine learning) for 𝑄 LE where neural networks independently learn the resistances from observations as intermediate variables. In our hybrid modeling setup, we make use of the Penman-Monteith equation based on the Big Leaf theory in conjunction with multi-year flux measurements across different forest and grassland sites from the FLUXNET database. We follow two conceptually different strategies to constrain the hybrid model to control for equifinality arising when estimating the two resistances simultaneously. One strategy is to impose an a priori constraint on 𝑟 a based on our mechanistic understanding (theory-driven strategy), while the other strategy makes use of more observational data and adds a constraint in predicting 𝑟 a through multi-task learning of the latent as well as the sensible heat flux (𝑄 H ; data-driven strategy). Our results show that all hybrid models exhibit a fairly high predictive skill for the target variables with 𝑅 2 = 0.82-0.89 for grasslands and 𝑅 2 = 0.70-0.80 for forests sites at the mean diurnal scale. The predictions of 𝑟 s and 𝑟 a show physical consistency across the two regularized hybrid models, but are physically implausible in the under-constrained hybrid model. The hybrid models are robust in reproducing consistent results for energy fluxes and resistances across different scales (diurnal, seasonal, interannual), reflecting their ability to learn the physical dependence of the target variables on the meteorological inputs. As a next step, we propose to test these heavily observation-informed parameterizations derived through hybrid modeling as a substitute for overly simple ad hoc formulations in Earth system models.

Elevated atmospheric CO2 (eCO2) influences the carbon assimilation rate and stomatal conductance of plants, and thereby can affect the global cycles of carbon and water. However, the extent to which these physiological effects of eCO2 influence the land-atmosphere exchange of carbon and water is uncertain. In this study, we aim at developing a method to detect the emergence of the physiological CO2 effects on various variables related to carbon and water fluxes. We use a comprehensive process-based land surface model QUINCY (QUantifying Interactions between terrestrial Nutrient CYcles and the climate system) to simulate the leaf-level effects of increasing atmospheric CO2 concentrations and their century-long propagation through the terrestrial carbon and water cycles across different climate regimes and biomes. We then develop a statistical method based on the signal-to-noise ratio to detect the emergence of the eCO2 effects. The signal in gross primary production (GPP) emerges at relatively low eCO2 (Δ[CO2] ~ 20 ppm) where the leaf area index (LAI) is relatively high. Compared to GPP, the eCO2 effect causing reduced 28 transpiration water flux (normalized to leaf area) emerges only at relatively high CO2 increase (Δ[CO2] >> 40 ppm), due to the high sensitivity to climate variability and thus lower signal-to-noise ratio. In general, the response to eCO2 is detectable earlier for variables of the carbon cycle than the water cycle, when plant productivity is not limited by climatic constraints, and stronger in forest-dominated rather than in grass- dominated ecosystems. Our results provide a step towards when and where we expect to detect physiological CO2 effects in in-situ flux measurements, how to detect them and encourage future efforts to improve the understanding and quantification of these effects in observations of terrestrial carbon and water dynamics.