🌿 ecology

Reconciling intrinsic and ecosystem water-use efficiency reveals physiological mechanisms underlying carbon-water coupling across biological scales

By integrating tree-ring data, triple stable isotopes, eddy covariance observations, and process-based modeling in a mature beech forest, this study reveals that long-term increases in intrinsic water-use efficiency are driven by enhanced photosynthesis rather than stomatal closure, while highlighting that discrepancies between leaf- and ecosystem-scale efficiency metrics arise primarily from the model's failure to account for diverse vegetation and hydrological processes.

Original authors: Puchi, P. F., Dalmonech, D., Castagneri, D., Vangi, E., Lehmann, M. H., Saurer, M., Klesse, S., Hortnagl, L., Collalti, A.

Published 2026-10-08

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Puchi, P. F., Dalmonech, D., Castagneri, D., Vangi, E., Lehmann, M. H., Saurer, M., Klesse, S., Hortnagl, L., Collalti, A.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Forests are the lungs of the planet, breathing in carbon dioxide and releasing oxygen, but they are also thirsty drinkers, pulling vast amounts of water from the soil and releasing it into the air. The balance between how much carbon a tree captures and how much water it loses is a critical measure of its health, known as water-use efficiency. Scientists have long used two different ways to measure this balance. One method looks at the leaves themselves, tracking how a single tree regulates its tiny pores to trade carbon for water. The other method looks at the entire forest as a whole system, measuring the total carbon absorbed and the total water evaporating from the ground, soil, and all the plants combined. For a long time, researchers have struggled to make these two perspectives agree, often finding that the story told by a single tree does not match the story told by the entire ecosystem. This disconnect matters because as the climate changes and the air becomes drier and hotter, understanding whether forests are becoming more efficient or struggling to survive is essential for predicting the future of our planet's carbon storage.

A team of researchers set out to solve this puzzle by studying a mature beech forest in the mountains of Switzerland over a twenty-year period. They combined three different tools to get a complete picture: they analyzed the chemical fingerprints locked inside the annual rings of the trees, they watched the real-time exchange of gases and water vapor from a tall tower overlooking the forest, and they ran a sophisticated computer simulation of the forest's biology. The goal was to see if the efficiency of the individual trees matched the efficiency of the whole forest, and to understand why the numbers might differ. The researchers found that when they looked strictly at the beech trees, the computer model and the tree rings told the same story: over the last two decades, these trees have become significantly more efficient at using water. This increase, roughly twenty percent, was not caused by the trees closing their pores to save water, as some might expect in a warming world. Instead, the trees were simply getting better at photosynthesis, capturing more carbon without needing to restrict their water loss.

However, when the team compared these tree-level findings to the measurements taken from the tower, which represent the entire ecosystem, the story changed dramatically. The computer model, which only simulated the mature beech trees, predicted a high level of efficiency that did not match the reality measured by the tower. The actual forest was far less efficient than the model suggested. The researchers discovered the reason for this gap: the model was missing a huge part of the forest. The computer simulation only accounted for the old, tall beech trees, but the real forest floor was bustling with other life. It included other types of deciduous trees, evergreen conifers, younger saplings, and a thick carpet of wild garlic that grows densely in the spring and early summer. All of these plants were drinking water and releasing it into the air, but because the model ignored them, it underestimated the total amount of water the forest was losing. Since the model thought the forest was losing less water than it actually was, it calculated a falsely high efficiency.

The study also revealed that the way we measure time changes what we see. When the researchers looked at the data day by day, the model and the real-world measurements barely agreed, likely because the weather changes too quickly for a simple model to catch every gust of wind or shift in sunlight. But when they looked at the data month by month, the model began to align much better with reality, capturing the general seasonal rhythms of the forest. This suggests that while computers are good at understanding the broad strokes of how a forest works over a season, they struggle with the minute-by-minute chaos of daily weather. Furthermore, the chemical analysis of the tree rings showed that the water the trees drank came from different sources depending on the year. In wet years, the trees drank fresh rain, but in dry years, they tapped into older water stored deep in the soil. This ability to switch water sources helped the trees maintain their health even when the surface soil was dry, a nuance that the simple model could not fully replicate.

Ultimately, the research clarifies that the efficiency of a single tree is not the same as the efficiency of the forest it lives in. The forest is a complex community where different plants, different ages, and the soil itself all play a role in how water moves. The increase in efficiency seen in the beech trees is a real physiological success driven by rising carbon dioxide levels, but the forest as a whole is losing more water than the trees alone would suggest because of the hidden contributions of the understory and other species. This means that to predict how forests will respond to climate change, scientists cannot just look at the tallest trees or rely on models that only simulate one species. They must account for the entire living community and the different ways water moves through the soil and the air. By reconciling these different scales, the study provides a clearer, more honest framework for understanding how our forests will fare in a warmer, drier future.

Technical Summary: Reconciling Intrinsic and Ecosystem Water-Use Efficiency

Problem Statement
Understanding how carbon–water coupling scales from individual leaves to entire ecosystems remains a critical challenge in ecohydrology. Water-use efficiency (WUE) is a key metric for plant physiological functioning, yet it is not scale-invariant. At the leaf level, intrinsic water-use efficiency (iWUE) is defined as the ratio of net photosynthetic assimilation ( $A$ ) to stomatal conductance ( $g_s$ ), often reconstructed from tree-ring carbon isotopes ( $\delta^{13}C$ ). In contrast, ecosystem-scale WUE ( $WUE_{eco}$ ) is typically derived from eddy covariance (EC) measurements as the ratio of gross primary productivity (GPP) to total evapotranspiration (ET). These two metrics often exhibit weak or inconsistent relationships in long-term studies. This discrepancy arises because $WUE_{eco}$ integrates non-transpirational water fluxes (soil evaporation, canopy interception, understory transpiration) and structural complexities that are absent in leaf-level $iWUE$ reconstructions. Furthermore, distinguishing whether long-term $iWUE$ trends are driven by stomatal closure or enhanced photosynthetic capacity requires disentangling physiological signals from hydrological controls, a task complicated by the lack of unified frameworks integrating multiple data sources.

Methodology
The study integrates three complementary approaches over a 20-year period (2004–2023) in a mature, mixed-species deciduous forest dominated by European beech (Fagus sylvatica) at the Lägeren site (CH-Lae) in Switzerland:

Tree-Ring Analysis: Stable isotopes ( $\delta^{13}C$ , $\delta^{18}O$ , $\delta^{2}H$ ) were measured in cellulose extracted from 15 F. sylvatica trees. $\delta^{13}C$ was used to calculate observed intrinsic water-use efficiency ( $iWUE_{obs}$ ), while the triple-isotope approach was employed to distinguish between stomatal and photosynthetic drivers of $iWUE$ variability.
Eddy Covariance (EC): Continuous half-hourly flux measurements provided ecosystem-level GPP and ET data, from which observed ecosystem water-use efficiency ( $WUE_{eco}^{obs}$ ) was calculated using a ratio-of-sums approach to avoid biases associated with averaging daily ratios.
Process-Based Modeling: The 3D-CMCC-FEM model was used to simulate carbon and water cycles. The model was configured specifically for mature F. sylvatica to generate modeled intrinsic water-use efficiency ( $iWUE_{mod}$ ) and modeled ecosystem water-use efficiency ( $WUE_{eco}^{mod}$ ).
Statistical Integration: The study compared these datasets across daily, monthly, and annual temporal scales. It utilized a dual-isotope framework (Scheidegger et al., 2000) and triple-isotope correlations to attribute $iWUE$ trends to physiological mechanisms (stomatal regulation vs. photosynthetic capacity) and hydrological controls (source water dynamics).

Key Results

Leaf-Scale Consistency: There was a strong positive correlation ( $r = 0.61, p < 0.05$ ) between isotope-derived $iWUE_{obs}$ and model-simulated $iWUE_{mod}$ . Both datasets indicated a long-term increase of approximately 20% over the study period. However, absolute values differed significantly, with $iWUE_{mod$ averaging higher (172.7 $\mu$ mol CO $_2$ mol $^{-1}$ H $_2$ O) than $iWUE_{obs}$ (98.9 $\mu$ mol CO $_2$ mol $^{-1}$ H $_2$ O), attributed to the model capturing peak summer conditions while tree rings integrate annual wood formation.
Ecosystem-Scale Divergence: A marked divergence was observed between observed and modeled ecosystem WUE. $WUE_{eco}^{obs}$ showed a significant positive trend (increasing ~55%), whereas $WUE_{eco}^{mod}$ remained stable. The model systematically underestimated ET by approximately 50%, leading to inflated $WUE_{eco}^{mod}$ values. This underestimation was attributed to the model's single-species configuration, which excluded co-occurring species (e.g., ash, spruce), younger trees, and understory vegetation (notably Allium ursinum), all of which contribute significantly to total ecosystem water loss.
Physiological Drivers: Triple-isotope analyses revealed that the long-term increase in $iWUE$ was driven primarily by enhanced photosynthetic assimilation ( $A$ ) rather than sustained stomatal closure. This conclusion was supported by the lack of a positive trend in $\delta^{18}O$ (which would indicate evaporative enrichment from stomatal closure) and the finding that $\delta^{18}O$ and $\delta^{2}H$ variability were mainly controlled by source-water dynamics rather than leaf-level physiological regulation.
Temporal Scaling: Agreement between observed and modeled $WUE_{eco}$ improved with temporal aggregation. Correlations were weak at the daily scale but strengthened at the monthly scale, suggesting that high-frequency meteorological noise and measurement uncertainties obscure short-term coupling, while seasonal patterns are more reliably captured by the model.

Significance and Claims
The paper claims to provide the first mechanistic framework for reconciling leaf-level and ecosystem-scale WUE by integrating tree-ring isotopes, eddy covariance, and process-based modeling. The authors assert that:

Scale Dependency is Fundamental: $iWUE$ and $WUE_{eco}$ are not interchangeable metrics; they reflect distinct processes, spatial domains, and temporal integrations. Direct extrapolation of tree-ring $iWUE$ trends to ecosystem-scale carbon–water exchange is invalid without accounting for hydrological partitioning and community composition.
Mechanistic Attribution: The study demonstrates that long-term increases in $iWUE$ in this temperate forest are driven by CO $_2$ -fertilization-enhanced photosynthesis rather than stomatal drought avoidance, a conclusion made possible by the decoupling of carbon and oxygen isotope signals.
Model Limitations: The divergence in ecosystem WUE highlights the critical need for process-based models to incorporate multi-species canopies, understory vegetation, and explicit hydrological partitioning to accurately simulate total ecosystem water fluxes.
Methodological Integration: The study validates the use of process-based models as a "mechanistic bridge" to interpret observational data, provided that model limitations regarding species representation and flux partitioning are explicitly acknowledged.

The authors conclude that while tree-ring isotopes and EC observations capture different aspects of the carbon–water cycle, their integration offers a robust framework for improving predictions of forest responses to rising atmospheric CO $_2$ and increasing atmospheric water demand.

More like this