Search

A steep decline in archiving could make large tree-ring datasets irrelevant. But increased spatiotemporal coverage, the addition of novel parameters at sub-annual resolution, and integration with other in situ and remote Earth observations will elevate tree-ring data as an essential component of global-change research.

Terrestrial ecosystem processes, and the associated vegetation carbon dynamics, respond differently to hydrometeorological variability across timescales, and so does our scientific understanding of the underlying mechanisms. Long-term variability of the terrestrial carbon cycle is not yet well constrained and the resulting climate–biosphere feedbacks are highly uncertain. Here we present a comprehensive overview of hydrometeorological and ecosystem variability from hourly to decadal timescales integrating multiple in situ and remote-sensing datasets characterizing extra-tropical forest sites. We find that ecosystem variability at all sites is confined within a hydrometeorological envelope across sites and timescales. Furthermore, ecosystem variability demonstrates long-term persistence, highlighting ecological memory and slow ecosystem recovery rates after disturbances. However, simulation results with state-of-the-art process-based models do not reflect this long-term persistent behaviour in ecosystem functioning. Accordingly, we develop a cross-time-scale stochastic framework that captures hydrometeorological and ecosystem variability. Our analysis offers a perspective for terrestrial ecosystem modelling and paves the way for new model–data integration opportunities in Earth system sciences. The atmosphere and biosphere are intrinsically coupled systems. Here, the authors integrate multiple datasets from hourly to decadal timescales and show that a hydrometerological envelope constrains ecosystem variability through time.

The total uptake of carbon dioxide by ecosystems via photosynthesis (gross primary productivity, GPP) is the largest flux in the global carbon cycle. A key ecosystem functional property determining GPP is the photosynthetic capacity at light saturation (GPPsat), and its interannual variability (IAV) is propagated to the net land–atmosphere exchange of CO2. Given the importance of understanding the IAV in CO2 fluxes for improving the predictability of the global carbon cycle, we have tested a range of alternative hypotheses to identify potential drivers of the magnitude of IAV in GPPsat in forest ecosystems. Our results show that while the IAV in GPPsat within sites is closely related to air temperature and soil water availability fluctuations, the magnitude of IAV in GPPsat is related to stand age and biodiversity (R2 = 0.55, P < 0.0001). We find that the IAV of GPPsat is greatly reduced in older and more diverse forests, and is higher in younger forests with few dominant species. Older and more diverse forests seem to dampen the effect of climate variability on the carbon cycle irrespective of forest type. Preserving old forests and their diversity would therefore be beneficial in reducing the effect of climate variability on Earth's forest ecosystems. Interannual variability (IAV) of the net carbon dioxide exchange over land is globally the main determinant of the variability of atmospheric CO2 growth rate1,2. So understanding the factors controlling the IAV in CO2 fluxes is essential to improve the predictability of the global carbon cycle3. Ecosystem biotic properties — such as soil and canopy nutrient status, rates of change in physiological properties of the vegetation, or the sensitivity of these properties to environmental factors — influence ecosystem CO2 exchange. Recent studies have shown that the IAV of the carbon budget can be better explained by variation in biotic properties of ecosystems such as photosynthetic capacity (GPPsat) than directly by environmental and climatic drivers4,​5,​6. GPPsat is defined as the value of gross primary productivity (GPP) at saturating light under non-stressed conditions, minimizing the influence of anomalous hydrometeorological conditions (for example, droughts and heatwaves), which potentially affect photosynthesis. A robustly retrieved characterization of GPPsat can be regarded as an ecosystem functional property reflecting the physiological response of the ecosystem to the environment. Given that IAV of GPPsat must propagate to observed GPP, this quantity is thought to be a key variable in understanding IAV of carbon fluxes7. In fact, recent studies demonstrated that GPPsat correlates more strongly than any climatic variable with annual GPP8, but also correlates with net ecosystem CO2 exchange5. The magnitude of IAV in GPPsat has been shown to exhibit considerable variation across ecosystems9, yet no obvious explanation for this pattern has been reported in the literature. However, the consequences are important: a low IAV in GPPsat would suggest that ecosystem functioning is not very sensitive to climatic variability, and that it preserves its functionality under the influence of that variability — and, likewise, high IAV is a consequence of high sensitivity. The capability of an ecosystem to preserve its functioning and structure over time (after external disturbances or climate extremes), is often defined as ecosystem stability and is linked to ecosystem resilience10. Using this terminology, low values of IAV in GPPsat can be understood as a characterization of high ecosystem functional stability. The relation of ecosystem functionality, structure and stability has been a matter of debate for many decades in the field of ecology. In particular, the diversity of vascular plants has been investigated as a stabilizing factor with respect to variations in productivity, for example by buffering the ecosystem’s sensitivity to climate extremes11. However, it is also well known that plant diversity is co-limited by soil properties12, ecosystem management, and climate conditions. Another variable to consider is stand age (the mean age of the forest stand or the number of years after a major stand replacement after disturbance), which may affect ecosystem stability through adaptation, particularly of trees to their environment — hence increasing ecosystem resilience to climate variability13. Moreover, structural parameters such as canopy cover, rooting depth, canopy height or leaf area index (LAI), which also depend on tree species diversity and stand age14, have an important effect on the ecosystem response to variation in environmental drivers since they define the capacity of trees to access resources such as water and light15. For instance, a regional study in the Amazon basin has shown that GPP, derived from the remotely sensed enhanced vegetation index, is less sensitive to environmental influences in regions with high canopy cover16. Despite this growing body of ecological knowledge, it remains largely uncertain which factors stabilize ecosystem functional properties at the global scale. In particular, we do not understand the causes of variability of specific ecosystem functional properties, such as photosynthetic capacity across ecosystem types, which ultimately controls ecosystem productivity. Here we hypothesize that stand age and species diversity play an important role in stabilizing ecosystem photosynthetic capacity. We test this hypothesis while also considering other factors related to climate, water availability, forest structure and soil properties that might have direct or indirect effects on ecosystem photosynthetic capacity. In this study, we used measurements of ecosystem-level fluxes, and climate variables (temperature, precipitation, and water availability), species richness, stand age, forest structure (canopy cover, height, and LAI), and soil properties (nutrient availability17) derived from satellite data, in situ observations and the literature (see the Methods). We used half-hourly ecosystem-level GPP fluxes estimated by the means of the eddy-covariance technique at 50 FLUXNET sites18 with at least 4 years of measured fluxes, and with different vegetation types across different climatic regions. We included data from evergreen forest (EF) as well as deciduous broadleaved and mixed forest (DBMF) located in temperate, boreal, mediterranean, tropical and dry climate regions (Supplementary Fig. 1 and Table 1). All 50 sites have information on stand age (referred to simply as ‘age’ in Figs 1–3) and species richness in addition to the CO2 flux data. Species richness (‘sp. no.’ in Figs 1–3) is the number of dominant plant species (for example tree or herb) that account for a cumulative abundance of 90 percent at a given site. We collected additional information on (i) canopy cover, (ii) canopy height, (iii) LAI, (iv) temperature and precipitation, and (v) soil water availability index (WAI) for a subset of 44 sites; and (vi) an index of nutrient availability for 36 sites compiled from the literature17 (see Methods).