Landscape heterogeneity is widely regarded as a key driver of biodiversity, yet it remains unclear whether increasing heterogeneity is always beneficial or whether excessive heterogeneity instead leads to fragmentation. We hypothesized that mixed heterogeneity among land-cover types and within-habitat heterogeneity have distinct effects on biodiversity, potentially explaining contrasting heterogeneity–biodiversity relationships reported across studies. To test this hypothesis, we analysed breeding bird richness from 1,010 1-km² landscapes across Germany together with high-resolution spatial datasets. Mixed heterogeneity was quantified by land-cover diversity and edge density, whereas within-habitat heterogeneity was quantified by tree diversity and edge density in forests and crop diversity and edge density in croplands. We used boosted generalized additive models to assess the effects of different heterogeneity components while accounting for habitat amount and other environmental predictors, complemented by interaction analyses among key predictors. Land-cover diversity consistently increased bird species richness, whereas land-cover edge density showed predominantly hump-shaped relationships, with richness peaking at intermediate levels. The apparent positive effects of land-cover edge density in compositionally simple landscapes largely reflected its strong correlation with land-cover diversity; after accounting for this coupling, its independent effects were negligible or negative. In contrast, within-habitat configurational heterogeneity, represented by tree and crop edge density, was more important for habitat-specialist richness than within-habitat compositional heterogeneity. These positive effects generally weakened or saturated rather than becoming negative with increasing habitat amount. Our results demonstrate that the biodiversity effects of configurational heterogeneity fundamentally differ between mixed and within-habitat heterogeneity. Mixed configurational heterogeneity may ultimately act as fragmentation, whereas greater within-habitat configurational heterogeneity may enhance habitat diversity for specialist birds. Distinguishing mixed from within-habitat heterogeneity and separating compositional from configurational components provides a mechanistic basis for understanding contrasting biodiversity responses to landscape heterogeneity.

Tropical rainforests harbor exceptional biodiversity and function, but are increasingly threatened by agricultural expansion. Whether landscape heterogeneity mitigates these impacts, as in temperate systems, remains unclear. Here, we quantified how local land use and landscape heterogeneity shape multidiversity and ecosystem multifunctionality, using 34 biodiversity metrics and 21 functions across 128 plots in Sumatra. Relative to rainforests, plantations reduced multidiversity and multifunctionality by ~25%, with stronger aboveground declines, lower plant and animal but higher microbial diversity. Contrary to temperate systems, landscape heterogeneity did not buffer local land-use effects but exacerbated declines in multidiversity and multifunctionality, and benefits of surrounding-rainforest cover were confined to rainforest fragments rather than plantations. Our results highlight the irreplaceability of continuous tropical rainforests, and the limited transferability of temperate-based landscape conservation strategies.

Accurate estimation of the fraction of absorbed photosynthetically active radiation (FAPAR) is crucial for understanding plant productivity and ecosystem dynamics. A number of indirect measurement techniques are used for estimating FAPAR with hand-held instruments, but researchers have identified discrepancies among different techniques when using them to validate satellite land products. Many researchers have also utilised photosynthetically active radiation (PAR) sensors to obtain quantitative measurements of PAR, but these lack robust measurement frameworks and protocols. Only very limited research has started on automated wireless PAR network systems to measure at finer temporal scales as well as to reduce human error and logistical costs. This study evaluates the performance of two flux (2f) and four flux (4f) FAPAR measurement systems and digital hemispherical photography (DHP) across multiple vegetation types (e.g., vineyard, broadleaf deciduous forest, savanna woodland) and different temporal scales (instantaneous and daily integrated). Results reveal strong agreement (R2 > 0.99, RMSE ≤ 0.04) between 2f- and 4f-FAPAR for all three study sites, with minimal overestimation (bias ≤ 0.04) by the 2f systems, suggesting that it can substitute, over similar environments, the more complex and costly 4f setup without substantially compromising accuracy. Daily integrated FAPAR exhibited greater stability and lower uncertainty compared to instantaneous FAPAR, underscoring its importance for long-term ecosystem monitoring. However, instantaneous FAPAR remains essential for satellite product validation due to its alignment with satellite overpass times. Additionally, 2f-FAPAR showed a good relationship with DHP-derived FAPAR. The findings highlight the potential of the 2f wireless PAR network as an automated, cost-effective, and reliable tool for canopy light absorption studies, offering substantial advantages for both ground-based ecosystem monitoring and remote sensing applications.

Forest ecosystems are increasingly stressed through heatwaves, drought periods, and other factors such as ozone pollution or insect infestations. These stressors have a profound impact on the emissions of biogenic volatile organic compounds (BVOC) from trees, which in turn influence aerosol formation and atmospheric oxidation cycles and thus feedback on the atmospheric cleansing capacity and climate change itself. While previous studies have investigated the impacts of specific stressors on BVOC emissions, analyses of combined stress effects are rare, even though the stressors seldomly occur in isolation. This study investigates the impact of heat and (nighttime) ozone stress, both individually and in combination, on BVOC emissions from two ecologically significant temperate tree species: European beech (Fagus sylvatica L.) and English oak (Quercus robur L.). In a climate-controlled chamber, both tree species were subjected to heat stress (38 +/- 3.3 degrees C) and ozone stress (similar to 120 ppb), separately and in combination. BVOC emission rates were measured using proton transfer reaction time-of-flight mass spectrometry, and the results were compared across pre-stress, heat, ozone, and combined heat-ozone conditions.Heat stress elicited the strongest emission increases of isoprene, monoterpene, and green leaf volatiles in both species, while ozone suppressed the emissions of most BVOCs. Combined stress led to non-additive responses different from those in single-stress scenarios. Both machine learning and positive matrix factorization analyses were performed to identify key VOC fingerprint markers that may be applied to identify stress-impacted emissions from field data, and both methods showed good agreement. The OH reactivity of the emissions, which serves as a measure for their atmospheric chemistry and ozone formation impacts, was consistently highest under heat stress for both species. However, nighttime ozone stress led to reduced OH reactivity of emissions (by 10 %-18 %).Our results underscore that the study of realistic combinations of stressors is crucial to understand future BVOC emissions and indicate that BVOC emissions could alter atmospheric chemistry and feedback with air quality and climate as heatwaves and pollutant-induced stress become more frequent due to climate change.

Agroforestry systems are considered suitable nature-based solutions to mitigate climate change. Long-term measurements of CO2 flux densities, evapotranspiration and sensible heat flux densities are, however, largely still missing. Here we present a unique eddy covariance and meteorological dataset from a total of ten stations paired over agroforestry and open cropland or grassland agricultural sites located in Northern Germany. The data were harmonized to create a consistent dataset which includes gap-filled time series of meteorological and lower-cost eddy covariance measurements with identical instrumentation, accounting for a total of seventy eight site-years of data. The objective of this dataset is to provide observational data on the differences of meteorological conditions, carbon, water and energy balances of adjacent agroforestry and open cropland or grassland sites in five distinct regions of Germany. This extensive, continuous dataset can be used to study ecosystem properties and the potential benefits of agroforestry. It can also be used to parametrize models on crop and biomass productivity, or to evaluate the response of such agroecosystems to climate change scenarios, among other applications. Anticipated key users of this dataset are researchers in the fields of micrometeorology, eddy covariance, agronomy, and ecosystem modeling. This dataset can be accessed through 10.25625/A2Z8T8 .

Disturbances such as extreme drought stress are becoming more frequent globally and pose a critical threat to boreal and temperate forests. Forest resistance to disturbances is influenced by multiple factors, including soil, climate, and structural complexity. Since a substantial portion of ecosystem functioning variability is related to maximum ecosystem productivity and water-use strategies, as part of WP2 of the CLIMB-FOREST EU project we calculated ecosystem functional properties (EFPs) that reflect these processes. Specifically, we used eddy covariance flux data to quantify photosynthetic capacity (NEPsat), underlying water-use efficiency (uWUE), and evaporative fraction (EFrac) for 71 forest sites across boreal and temperate regions of Europe and North America. To describe functional stability, we analyzed both mean EFPs and their inter-annual variability for each site. To examine which scales of structural complexity are associated with EFP stability, we used satellite-based indices describing vegetation structure and heterogeneity, including Rao’s Q of the Enhanced Vegetation Index (EVIRao), normalized near-infrared reflectance of vegetation (NIRvN), near-infrared entropy (NIRent), and maximum leaf area index (LAI). We applied generalized additive models (GAMs) combined with bootstrap-based variable importance analysis to evaluate associations between EFPs and structural complexity.We found that associations between EFPs and structural complexity metrics varied among ecosystem properties, with predictors more frequently meeting bootstrap-based importance criteria for mean EFPs than for their inter-annual variability. Maximum LAI and NIRvN were consistently retained as important predictors for mean NEPsat and mean EFrac, whereas no structural complexity metrics met the importance criteria for uWUE or for most variability metrics. Smooth-term estimates indicated directional partial associations, with higher LAI and NIRvN corresponding to higher modelled values of NEPsat and EFrac, while EVIRao and NIRent showed weaker or inconsistent partial trends. Overall, the results suggest that quantity of leaves and their spatial arrangement might be more important for EFPs than horizontal heterogeneity. Forests with denser and more organized canopies tended to function at higher levels of productivity and evaporation, without showing stronger inter-annual variability.

Land use intensity (LUI) significantly influences the biophysical and biogeochemical properties of the global landscape. The impact of LUI is exceptionally strong in Southeast Asia (SEA), where forests are increasingly being replaced by intensively managed plantations. Despite these transformations, comprehensive studies on how different LUI regulate carbon, energy, and water fluxes in this region remain scarce. In this study, we examine data from 16 eddy covariance (EC) flux tower sites in SEA, representing a total of 112 years of measurements. We aim to assess trade-offs in carbon fluxes, light use efficiency (LUE), and water use efficiency (WUE) across a gradient of LUI: low (primary forests), medium (secondary forests), and high (plantations). Our findings reveal that mature high LUI sites on mineral soil act as carbon sinks; however, their high evapotranspiration rates often exceed site-specific precipitation, making them susceptible to water stress. For example, mean annual carbon uptake at high LUI sites (ranged from -1.19 to -0.74 kg C m-2 year-1) outperformed low LUI sites (-0.85 to -0.02 kg C m-2 year-1). The strong carbon uptake on high LUI had an exception when the ecosystem was in the establishment phase and managed on peat soil (0.98 kg C m-2 year-1). However, WUE was greater at low LUI (mean annual: 2.63 to 6.50 g C kg-1 H2O-1) compared to high LUI sites (2.08 to 3.53 g C kg-1 H2O-1), illustrating a trade-off between carbon uptake and water use. Additionally, while high LUI sites required less radiation to achieve maximum gross primary productivity, their mean daily LUE was not higher compared to low LUI sites. These findings underscore the importance of carefully balancing carbon sequestration goals with water resource considerations and drought resilience when promoting plantation systems. Conversely, forest conservation offers advantages for water security and ecosystem resilience to face climate change.

Natural bogs have high water level (WL) that prevents peat decomposition and promotes bog-specific vegetation. Especially in rewetted bogs, increasing tree encroachment is observed in combination with low WL. Simultaneously, a general shift in vegetation and microform distribution is visible. The effects of this development on evapotranspiration (ET) are not clear.Two sites were established at a former peat extraction site, one with high WL and a Sphagnum-dominated vegetation (open site), one with varying and lower WL and a dense birch population (tree site). Energy and CO2 exchange were measured with eddy covariance towers. For one growing season, ET of the different microforms (hummocks and hollows) and birches was measured with closed chambers to model their contribution to growing season ET.ET was lower at the tree site in all months and consequently in all three investigation years (620 and 387 mm at the open and tree site, respectively, averaged over 3 years). ET of birches was 104 mm, which was 28% of growing season site ET, while microform ET was strongly reduced at the tree site. ET at the open site was high due to high Sphagnum cover and WL. The open site generally had a higher proportion of latent heat, leading to a lower Bowen ratio, while the water use efficiency was higher at the tree site. Consequently, the birches are not the primary cause of the lower WL, but rather the additional water loss due to site conditions and incomplete rewetting. Therefore, despite the higher ET, the open site is in a better state of conservation.

Assessing ecosystem functioning is crucial for managing and conserving ecosystems and their services. Numerous ways to evaluate ecosystem functioning have been developed, using species traits, such as Plant Functional Types (PFTs), flux measurements with the Eddy Covariance (EC) technique, and remote sensing techniques. We propose that the spatial heterogeneity in ecosystem functioning at a regional scale can be assessed and monitored using satellite-derived Ecosystem Functional Types (EFTs): groups of ecosystems or patches of the land surface that share similar dynamics of matter and energy exchanges. We hypothesize that, as observed for PFTs, different EFTs should have distinct patterns and magnitudes of Net Ecosystem Exchange (NEE) of carbon dioxide measured using the EC technique. We derived EFTs from 2001-2014 time-series of satellite images of the Enhanced Vegetation Index (EVI) and compared them with NEE measurements (derived from in situ field observations using the EC technique) across 50 European sites. Our results show that distinct EFTs classes display significantly different dynamics and magnitudes of NEE and that EFTs perform marginally better than PFTs in explaining NEE regional patterns. Land-cover maps based on PFTs are difficult to update on an annual basis and are not sensitive to changes in ecosystem performance (e.g., droughts or pests) that do involve short-term changes in PFT composition. In contrast, satellite-derived EFTs are sensitive to short-term changes in ecosystem performance. Satellite-derived EFTs are an ecosystem functional classification built from satellite observations that allow the identification of homogeneous land patches based on ecosystem functions, e.g., ecosystem net productivity measured on the ground as NEE. Satellite-derived EFTs can be recalculated annually, providing a straightforward way to assess and monitor interannual changes in ecosystem functioning and functional diversity.

Croplands are among the land systems most vulnerable to shifts in precipitation regimes and prolonged droughts, particularly in temperate climates. Characterizing root water uptake patterns is therefore essential to understand how crops maintain function and sustain transpiration under drought stress. We investigated water uptake patterns of winter cereals (wheat, barley) across two contrasting growing seasons (2024, 2025) at a temperate cropland in Central Germany. Within the footprint of an eddy covariance (EC) tower, we sampled plant leaves, precipitation, soil water, and groundwater for stable water isotope analysis at natural abundance and estimated xylem water isotope composition using the Craig–Gordon model. Soil moisture was monitored at nine depths (5–100 cm) at three profiles. Using multi-layer soil moisture time series and a dual-isotope mixing model, we identified the contribution of different water sources to transpiration, and quantified water uptake depth patterns.In 2024, more frequent rainfall maintained wetter soil layers and higher evapotranspiration (ET), whereas 2025 was marked by longer dry spells and consistently lower ET, although average water-use efficiency remained similar with greater variability in 2025. Both crop types exhibited flexibility in water uptake depth in response to dry spells, including access to deeper soil layers (> 60 cm). However, the median uptake depth ranged from 20–60 cm for wheat in 2024, whereas barley met transpiration demand primarily from 10–40 cm despite drier conditions in 2025. Our findings suggest that water uptake strategies are tightly linked to plant traits, access to deeper water sources, and dry spell characteristics, which together shape root plasticity and modulate the impact of drought on transpiration and ecosystem productivity.

Abstract. Ecosystem-atmosphere interactions remain poorly quantified in temperate agroforestry systems despite their potential as nature-based solutions against climate change. In this study, we present CO2 and H2O balances measured with eddy covariance over four pairs of agroforestry with open cropland or grassland systems in northern Germany from 2019 to 2024. Annual sums of net ecosystem CO2 uptake were generally 1.2- to 4-fold larger at agroforestry sites (average of -113.9 g C m-2 yr-1) compared to open cropland or grassland (average of 20.4 g C m-2 yr-1), while differences in evapotranspiration between systems were comparatively small (averages of 473.7 mm yr-1 and 423.5 mm yr-1, respectively). Water-use efficiency was thus generally higher at agroforestry, driven primarily by enhanced photosynthetic activity. Site-years with winter crops showed stronger carbon uptake, mostly due to reduced soil emissions in winter. Overall, the inclusion of trees resulted in an extended growing season and reduced sensitivity of these agroecosystems to high air temperature and low soil moisture. Furthermore, although net biome production was positive for most of the site-years, indicating a net carbon release, its magnitude was two or three times higher at open croplands and grassland than agroforestry, indicating stronger carbon losses from the reference systems that always showed a positive balance. These results reveal the potential of agroforestry systems as nature-based solution for climate change mitigation in agriculture, without compromising water availability.

Abstract. Expansion of oil palm plantations is a dominant driver of land-use change in equatorial Asia and has contributed substantially to deforestation in Indonesia. Most dynamic global vegetation models (DGVMs), however, still represent oil palm plantations as generic cropland, thereby contributing to uncertainty in carbon emissions from land-use change (ELUC). We develop a new oil palm plant functional type (PFT) using the JULES land surface model, called JULES-Palm. This PFT incorporates oil palm-specific physiological, structural and plantation-management traits, including rotation-based harvest/replanting cycles, and is parameterised from literature-derived values before calibration using eddy covariance (EC) and allometric observations, including leaf area index (LAI), above-ground biomass and canopy height. This calibration improves simulated photosynthesis and the vegetation structure. We then ran JULES-Palm to quantify the contribution of oil palm plantations to Indonesian ELUC, using land-use drivers from the Mapbiomas Indonesia 2.0 land cover maps (2000–2022). We investigated three scenarios: (i) oil palm represented as a generic crop, (ii) oil palm omitted and treated as natural vegetation and (iii) explicit oil palm PFT. The mean and interannual variability of Indonesian ELUC (without peat emissions) is 0.058 ± 0.012 PgC yr-1 with the oil palm PFT and 0.052 ± 0.010 PgC yr-1 when oil palm is omitted; representing oil palm as a generic cropland gives 0.066 ± 0.011 PgC yr-1. Cumulatively, the expansion of oil palm plantations by 10.5 Mha since 2000 contributed approximately 0.14 PgC, accounting for 10.3 % of Indonesia’s total ELUC including legacy fluxes, but approximately 89 % of land-use change attributable to conversions occurring during 2000–2022. Representing oil palm as a dedicated PFT reduces the estimated oil-palm-related ELUC relative to the generic-cropland representation by capturing plantation-specific biomass accumulation and carbon turnover. Thus, using a specific PFT for oil palm is preferable to providing a more accurate, process-based representation of plantation carbon dynamics, but treating oil palm as cropland still captures a first-order estimation of ELUC when a dedicated PFT is unavailable.

Forests are aerodynamically rough surfaces above which turbulent exchange is enhanced, generating smaller vertical gradients in windspeed and air temperature than those predicted by the Monin-Obukhov Similarity Theory (MOST). Roughness sublayer (RSL) corrections, based on canopy structure, have been proposed to account for this enhancement in turbulent exchange, but evaluation of these RSL corrections above a range of forest types is still lacking. In this study, we mobilised multiyear datasets of canopy structure, turbulent fluxes and microclimate gradients from the ICOS European network to evaluate two widely-used RSL corrections in a range of climates, forest structures and atmospheric conditions. As expected, observed windspeed gradients above these forest sites were smaller than MOST predictions. The two RSL corrections improved the windspeed gradient predictions with similar accuracy, irrespective of atmospheric stability conditions. The need of RSL corrections for predicting air temperature gradients above forests was more contrasted than for windspeed, and depended on the site and the atmospheric stability conditions. Overall, the RSL corrections at canopy height remained relatively small, around 0.5°C for air temperature and 0.5 m s-1 for windspeed on average, and were most pronounced under stable atmospheric conditions. Based on the evaluation of the simplifying assumptions behind each RSL correction, we built recommendations to implement those corrections in models. We also discussed the potential impact of wind sensor positions to estimate aerodynamic parameters needed to apply MOST and RSL corrections and provide recommendations for improvement of the ICOS network. Our results highlight the difficulty to estimate displacement height and relate it to canopy structure, and question the idea that the drag coefficient does not change with leaf area or clumping.

Forests are increasingly exposed to extreme events and disturbances like droughts, storms, fires, pathogens, and others. At the same time, forests are expected to act as important carbon sinks with the corresponding climate change mitigation capacity. What are the links between forest structure and ecosystem functional properties and the resilience against disturbances and extreme events? What are the options for forest management in this context?Using data from flux towers and field experiments from 90 sites in 16 countries, mostly in Europe, and remote sensing observations, we investigate the role of forest structure as buffer of climate extremes; link light-use efficiency to stand characteristics and management; elucidate the role of climate effects of short-lived climate forcers and their feedback due to a warming climate, stress and disturbances, and evaluate the impact of extreme drought, fire disturbances and forest management on soil organic carbon (SOC) and nitrogen dynamics.Combining GAMs with bootstrap-based variable importance analysis, we could show that there are associations between the means of selected Ecosystem Functional Properties of boreal and temperate forests, like photosynthetic capacity (NEPsat) or underlying water-use efficiency (uWUE), and structural complexity metrics, like Leaf Area Index or Near-Infrared Reflectance of Vegetation. With increasing drought stress, higher canopies, LAI and species number stabilizes the forest response both for NEPsat and uWUE.Work on entangling the climate effects of short-lived climate forcers (SLCFs) is progressing with measurements of terpene concentrations and emissions and aerosol particle dynamics process modelling. Model evaluation of the climate effect from afforestation in the Nordic countries with coniferous trees on previous grassland shows that the climate cooling effect of increased terpene emissions and aerosol formation outweighs the warming effect due to the filtering of aerosol particles by trees.Field experiments on Spanish sites indicate that drought (induced through precipitation exclusion) significantly reduces the litter decomposition rate, and that thinning increases SOC content; however, differences in SOC between management regimes are often masked by high spatial variability.The work presented has emerged within the Work Package “Data assessment of processes and their impacts on biodiversity and climate effects on forests” of the CLIMB-FOREST H2020 EU project.

The O2:CO2 exchange ratio (ER) between terrestrial ecosystems and the atmosphere is a key parameter for partitioning global ocean and land carbon fluxes. The long-term terrestrial ER is considered to be close to 1.10 mol of O2 consumed per mole of CO2 produced. Due to the technical challenges in measuring directly the ER of entire terrestrial ecosystems (EReco), little is known about variations in ER at hourly and seasonal scales, as well as how different ecosystem and flux components (e.g., vegetation and soil, assimilation and respiration) contribute to EReco. In this modeling study, we explored the variability in and drivers of EReco and evaluated the hypothetical uncertainty in determining ecosystem O2 fluxes based on current instrument precision used in micrometeorological methods such as the flux-gradient approach. We updated the one-dimensional, multilayer atmosphere–biosphere gas exchange model “CANVEG” by 1) implementing ER for various ecosystem components in the model; 2) implementing the control of triose phosphate utilization (TPU) and Medlyn’s stomatal conductance equation to CO2 assimilation; and 3) linking photosynthetic O2 emission to nitrogen (N) assimilation sources. The model study was conducted at the Leinefelde FLUXNET site, a temperate beech forest in Germany, where eddy covariance, profile, and gas exchange chamber measurements were available. We found that when assuming fixed ER for CO2 assimilation and respiration, the hourly EReco showed strong variations over diel and seasonal cycles and within the vertical canopy profile, indicating the potential to partition eddy-covariance derived CO2 fluxes with corresponding O2 flux measurements. The O2 and CO2 mole fraction ratio of canopy air (ERconc) showed different values and mechanisms from EReco. The model showed more robust performances in future CO2, temperature and air humidity conditions when taking into account TPU limitation and Medlyn’s stomatal conductance algorithm in CO2 assimilation processes. The predicted net carbon sink under elevated atmospheric CO2 mole fraction increased less with TPU limitation than without. The most significant impacts on photosynthetic O2 emission and hence the ER of CO2 assimilation resulted from variation in nitrogen assimilation sources. The ER of net assimilation measured with branch-level gas exchange chambers showed little variation from 1.0 mol mol-1, indicating ammonia as the main N assimilation source. The model indicated that the O2 emission would increase by up to 23% if nitrate was used as N assimilation source. Our study successfully coupled oxygen with carbon fluxes within a multilayer atmosphere–biosphere gas exchange model. The modeling study yielded that the application of the flux-gradient measurement approach is feasible to derive ecosystem O2 fluxes. To achieve better model behavior, it is necessary to incorporate TPU limitation in the assimilation model and to properly consider N assimilation during photosynthesis.

Understanding drought stress responses and adaptation mechanisms in forest ecosystems towards climate extremes is crucial. This knowledge aids in assessing adaptive capacities and developing supportive management measures. Here, a comprehensive long-term data set obtained in the Hainich National Park, an old-growth mixed-beech forest in Central Germany (DE-Hai), gives the opportunity to investigate stress effects and transformation processes caused by the 2018 and 2019 summer droughts on tree and stand scale. The forest displays a near-natural, diverse system with a range of tree age classes and species (main species: Fagus sylvatica, Fraxinus excelsior, Acer pseudoplatanus). In this study, we combined long-term observations of stand-level CO2 exchange obtained with the eddy covariance method with annual growth records and structural indices of 80 trees obtained from dendrometer bands and terrestrial laser scans. Further, drone and satellite imagery provided estimates of tree mortality through canopy gap dynamics. Based on this multi-scale data set, we strive to better understand the link between forest CO2 uptake and tree response dynamics under the influence of a severe drought. During the drought events in 2018 and 2019, we observed that the forest remained a net CO2 sink, but the CO2 uptake strength was considerably diminished (up to -30%) in comparison to the reference period of the previous 17 years. Moreover, the reduction in CO2 uptake extended beyond the duration of the droughts, which implies significant changes in the mechanisms and dynamics of the forest. Further, an increase in the canopy gap fraction by more than 50% in 2021 indicated a significant increase in tree mortality. Surviving trees were affected differently by the droughts depending on species-specific stress response strategies and a tree’s role as competitor or suppressed individual. In particular, the growth of older and larger trees, mostly Fraxinus excelsior, was impaired during and after the drought period. However, approximately half of the observed trees, mostly suppressed, vital Fagus sylvatica, showed a positive growth trend during and after the drought period. The structural diversity of the old-growth mixed forest could buffer the drought-induced outage in the CO2 uptake strength, though the increased growth of a large cohort of surviving, suppressed Fagus sylvatica could not compensate for the diminished CO2 uptake by a few dying dominant trees. The natural succession dynamics in the forest ecosystem seem to be accelerated due to drought events. A continuous and consistent long-term monitoring of forest ecosystems is needed to further investigate the initiated transformation processes, the stand and tree resilience, and additionally the impact of legacy effects.

Eddy covariance (EC) studies typically involve the use of one or maximum two measuring towers, which leads to a low level of spatial replication, compromising the statistical representativity of EC measurements, especially above highly heterogeneous ecosystems, such as agroforestry systems. Lower-cost eddy covariance setups (LC-EC) represent a potential solution to this problem, since their affordability allows for the installation of multiple EC towers to study heterogeneity at the landscape scale. In the last years, several LC-EC setups have been successfully validated against conventional EC setups (CON-EC), with the main difference being the use of slower gas analyzers. These introduce a higher uncertainty due to the enhanced high-frequency spectral attenuation in the turbulent energy spectrum. In this study, we analyzed turbulent fluxes of CO2 and H2O and turbulence characteristics measured by three flux towers equipped with LC-EC setups above one agroforestry system located in Wendhausen, Germany. The agroforestry system was a Short Rotation Alley Cropping (SRAC) system, consisting of alternating rows of trees and crops. The three flux towers were installed at different North-South aligned tree stripes. Additionally, we compared the results of the three LC-EC setups above the SRAC with another LC-EC setup installed at an adjacent monocropping (MC) field. The objectives of the study were: (i) to evaluate the spatial variability of EC fluxes from the three flux towers above the SRAC system; (ii) to compare the variability of fluxes within the SRAC to the variability of fluxes between SRAC and MC; (iii) to quantify whether the use of several LC-EC setups counteracts the higher uncertainty associated to LC-EC, due to the increased statistical robustness of the measurement network compared to the hypothetical use of just one EC station. The highest spatial variability across the SRAC was measured for CO2 fluxes, followed by latent heat (LE) flux, with coefficients of variation, calculated following Oren et al. (2006) (https://doi.org/10.1111/j.1365-2486.2006.01131.x), of 2.3 and 1.4 (dimensionless), respectively. The spatial variability in CO2 and LE fluxes within the SRAC was similar to the variability between MC and SRAC, and was attributed to the different land cover types around the towers. On the other hand, the spatial variability in sensible heat flux (H), momentum flux and turbulence characteristics (such as friction velocity and variance of vertical wind speed), within the SRAC, was smaller than the variability between SRAC and MC, likely explained by the development of an internal boundary layer (IBL) above the SRAC. Our results show that the heterogeneity of the SRAC, despite not affecting significantly the turbulence characteristics across the site, leads to a large spatial variation in CO2 and LE fluxes. Therefore, a distributed network of several EC systems is necessary to properly quantify patterns and drivers of CO2 and latent heat fluxes above such heterogeneous land-use systems.

Forests are important natural carbon sinks and can help mitigate climate change. The drought and heat waves of recent years have severely affected forests in Germany, resulting in reduced net CO2 uptake. How forest management, age and species composition moderate the negative impacts of weather extremes on net CO2 uptake or its recovery is still unknown.For around 25 years, gross primary production, ecosystem respiration and net ecosystem exchange as well as evapotranspiration have been studied in a unmanaged, uneven-aged and mixed beech stand in the Hainich National Park (DE-Hai) and a managed, even-aged and pure beech stand near Leinefelde (DE-Lnf) in Thuringia, Germany.Both forest stands were and are a substantial CO2 sink (DE-Hai: 512±89 gC m-2 a-1; DE-Lnf: 590±190 gC m-2 a-1), whereby the annual CO2 uptake of the managed stand varied significantly more than that of the unmanaged stand. The CO2 sink function of both stands persisted even in the extremely dry and hot year 2018, though the annual CO2 uptake was reduced by 27% (DE-Hai) and 64% (DE-Lnf) compared to the long-term average from 2002-2017. A reduction in CO2 uptake was also evident in the following year, which can mainly be attributed to persistently low soil water availability. In addition, a loss of tree vitality was observed, which affected the CO2 balance in the following years. In contrast to the unmanaged stand, however, the managed stand already reached higher uptake rates again in 2020. The differences between the stands can mainly be explained by differences in tree age and stand structure. With a mean age of about 130 years, the managed stand consists almost exclusively of vigorous beech trees (optimal phase), whereas the unmanaged stand comprises all age classes and developmental stages, in particular a high proportion of very old trees (> 180 years), which were particularly badly damaged by the drought.Long-term flux measurement covering 25 years revealed divergent responses of the two differently managed and structured forest stands to drought. In a next step, more sites covering a range of management strategies, species and ages need to be included.