From rain to radiance - Drivers of tropical carbon flux interannual variability shift with local climate

doi:10.21203/rs.3.rs-5040347/v1

Download PDF

Article

From rain to radiance - Drivers of tropical carbon flux interannual variability shift with local climate

https://doi.org/10.21203/rs.3.rs-5040347/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Tropical land carbon fluxes drive the inter-annual variability (IAV) of the atmospheric CO2 growth rate. However, what mechanistic processes and climate variables drive the IAV of tropical land carbon exchange remains debated. Here, we use a Bayesian framework that ingests a suite of satellite-based carbon and water observables to quantify the relative contributions of different climate variables to the tropical land carbon exchange IAV. We find the dominant climate drivers for carbon exchange depend on climate states. Radiation dominates the IAV of gross primary production (GPP) in wetter tropical forests, especially during La Niña and wet seasons, while VPD dominates GPPIAV in drier forests, with more pronounced influence during El Niño and dry seasons. Precipitation predominantly drives heterotrophic respiration IAV. Net ecosystem exchange IAV is more closely linked to ecosystem respiration than GPP. Our findings yield novel insights into the complex interplay among tropical climate gradients and IAV in climate states and carbon fluxes.

Earth and environmental sciences/Climate sciences/Climate change/Climate and Earth system modelling

Earth and environmental sciences/Climate sciences/Biogeochemistry/Carbon cycle

Tropics

interannual-variation

carbon cycle

rainforest

semi-arid

grassland

El Niño

La Niña

The dominant climate driver for GPP interannual variability (IAV) shifts with mean climate states
The dominant climate driver for NEE_IAV more closely mirrors ER than GPP
The dominant climate driver for carbon fluxes IAV is climate-state-dependent across both space and time

Mean climate states determine the dominant climate drivers of interannual variability of tropical carbon exchange

The sensitivity of net carbon fluxes in the tropical terrestrial biosphere to climate interannual variability (IAV) is closely linked to how the tropical land carbon pools will respond to future climate change (Cox et al., 2013). Thus, understanding how climate drivers of IAV in tropical terrestrial biosphere carbon flux has been a critical but contentious area of research, primarily due to the region’s heterogeneous tropical climate states, rich biodiversity, and the limited availability of observations across the tropics. Earlier studies found a substantial correlation between the mean tropical land temperature and tropical NEE (Anderegg et al., 2015; Cox et al., 2013; Kindermann et al., 1996; W. Wang et al., 2013, X. Wang et al., 2014). In recent years, an increasing number of studies have shown a tight coupling between water availability and tropical net carbon flux variability (Humphrey et al., 2018; Liu et al., 2023). The conclusions of these studies are limited to the pan-tropical scale since the atmospheric CO₂ growth rate was used as a proxy for the IAV of tropical net land carbon fluxes. Using both machine learning algorithms and process-based global land models, Jung et al. (2017) showed that the choice of spatial scales matters: when integrated globally, temperature is the main driver of NEE_IAV, while at latitudinal scale, water availability is the dominant driver of the IAV of gross primary productivity (GPP), ecosystem respiration (ER), and NEE.

A great number of scattered studies that have quantified how climate drivers control carbon flux interannual variability within specific biomes or subregions in the tropics. For instance, Saleska et al. (2007) pioneered a debate about the role of radiation in controlling photosynthetic variability in the wet Amazon (Huete et al., 2006; Nemani et al., 2003; Wang et al., 2023; Braghiere et al., 2020). Poulter et al., (2014) and Ahlström et al. (2015) argued that variations in water availability critically shape the net ecosystem exchange (NEE) IAV in semi-arid and sub-tropical ecosystems. These studies suggest that the dominant climate drivers for carbon fluxes vary spatially. More recently, Liu et al. (2024) found that the sensitivity of carbon fluxes, including gross primary product (GPP) and net biosphere exchange (NBE), to vapor pressure deficit and total water storage increased during the severe drought in tropical South America in 2015-2016, indicating that the sensitivity of carbon fluxes to climate states may vary temporally.

However, comprehensive quantitative analyses of the relationships between spatiotemporal changes in the dominant climate drivers and the mean climate state have not been systematically explored. Understanding how these drivers evolve with mean climate across space and time is essential for predicting future changes in carbon flux IAV. This insight is especially critical given that climate projections suggest uneven changes in tropical climates: most of tropical South America is expected to experience drying trends, while parts of tropical Africa may see increased rainfall. Additionally, rising temperatures across the tropics are likely to drive higher atmospheric water demand, as reflected by greater vapor pressure deficits (VPD) (IPCC AR6).

To fill this knowledge gap, we systematically investigate the dominant climate drivers of interannual variability in tropical carbon fluxes using a Bayesian modeling framework CARDAMOM (Bloom et al., 2020). Our analysis focuses on (1) how the climate drivers for tropical carbon flux IAV, including net ecosystem exchange (NEE) and its component fluxes, vary across space, time, and vegetation types during dry/wet seasons and El Niño/La Niña events at local and regional scales; (2) the relationships between spatiotemporal variations in dominant climate drivers and climate states; and (3) the contributions of major ecosystems to tropical carbon flux IAV, and their links with climate sensitivity and dominant climate drivers. By integrating a suite of satellite observations to constrain variables such as NEE, GPP, and total water storage in a process-based land model, CARDAMOM effectively addresses the challenge posed by the spatial sparsity of in-situ data for model validation and parameterization. Additionally, the modeling framework allows us to mechanistically quantify the contributions of multiple climate factors to tropical carbon flux IAV and to understand how the dominant climate factors vary across space and time, thereby enhancing the granularity of climate controls on tropical carbon flux IAV. We examine NEE and its component fluxes (GPP, ER) to understand the compensating/aggregating effect of climate drivers – such as temperature, participation, VPD, radiation, and their interactions. Our findings enhance the understanding of climate controls on carbon cycle variability, offering critical insights for improving predictions of tropical carbon fluxes in future climate scenarios.

Dominant climate factors for GPP_IAV, ER_IAV, and NEE_IAV

The seasonal and IAV of observational data at each grid cell is captured by the Data Assimilation Linked Ecosystem Carbon model (DALEC) at both the local and continental scales (Figure S1-S2). Detailed results on data-model integration are in the supplementary (Text S1). To evaluate the relative importance of precipitation (P), VPD (V), shortwave radiation (R), temperature (T), and their interactions (I), on C flux variability, we replace one climatic driver at a time with their climatological means in the perturbed model runs and calculate their carbon fluxes (dC) differences from the control model runs (scenarios are listed in Table S1). The carbon response index (K) is calculated as the time series mean of the absolute value of dC. The K index essentially quantifies the importance of each climate driver on carbon flux variabilities.

We identified the three most important climate factors out of the five investigated (precipitation, VPD, radiation, temperature, interactions) - precipitation, VPD, and radiation - and show their relative importance for each 4° (latitude) x 5° (longitude) model grid in ternary maps that graphically depicts the ratios of the three climate factors as positions in an equilateral triangle (Figure 1A, B, C). The spatial map showing the vegetation types of each 4°x5° degree pixel is in Figure S3. The dominant climate factors for NEE and its component fluxes are heterogeneous. Radiation dominates GPP in the moist rainforests in the Amazon and part of Congo, while VPD drives the variability of GPP in the seasonally moist rainforests on both continents. In southeast tropical South America, as well as the north- and south-east of tropical Africa, where the lowest Mean Annual Precipitations (MAP) occurs, GPP_IAV is driven by precipitation. A similar but weaker pattern is found for ecosystem respiratory fluxes (ER), mainly due to the strong dominance of precipitation over heterotrophic respiration (Rh, Figure S4). The dominant climate factor for NEE_IAV is the net outcome of ER_IAV - GPP_IAV (Figure 1C). We find that the IAV of NEE, GPP, and ER in tropical savannas is predominantly influenced by precipitation (Figure 1).

To understand how the dominant climate factors of C flux IAV shifts in the tropic regions, we calculated correlations between relative contributions of climate factors to C flux IAV and their environmental and biological mean states (mean monthly minimum temperature, mean monthly maximum temperature, total precipitation, VPD, radiation, above- and belowground biomass (ABGB), soil organic C (SOC)) (Figure S5). We find that dominant climate factors shifted with mean precipitation (r=-0.98) and the mean radiation gradient (r=0.87) (Figure S5-AB) such that the IAV in wetter and drier environments was driven by different factors. Other mean climate states, as well as ABGB and SOC, are to a lesser degree correlated with the relative importance (Figure S5-CDEFG). We use mean annual precipitation in relevant figures and discussions hereafter, to highlight the determinant role of mean climate state in determining the relative contributions of different climate factors to C fluxes IAV.

The dominant climate factors for different PFTs on each continent also differed and with a consistent shift of dominant climate factors across functional types over all the tropical continents, which coincides with their mean annual precipitation gradients (Figure 2). GPP IAV is primarily controlled by the variability of precipitation over all the grassland and cropland pixels, African shrubland, and savanna pixels, where MAP is less than 1400mm on average. Over American shrubland and savanna pixels and the seasonally moist rainforests in America and Africa where MAP is higher, VPD is the dominant climate factor for GPP IAV. In the wettest part of the tropical world, where both air and soil are relatively wet all year long, radiation becomes the dominant factor for the IAV of photosynthesis. In summary, for GPP_IAV, the dominant climate factor shifts from radiation (MAP>2800mm), to VPD (1400<MAP<=2800mm), and to precipitation (MAP<=1400m), whereas the contribution of precipitation IAV to NBE_IAVand ER_IAV becomes larger, as the mean annual precipitation decreases (Figure 2A, B, C).

Dependence of dominant climate factors on dry and wet season

Previous section show that the spatial distributions of dominant climate factors depend on the spatial distributions of annual mean precipitation (Figure 1-2). Precipitation also has large seasonal and interannual variations. In this and the next section, we evaluate whether the dominant climate factors also change with season and climate interannual variations. The question we want to address is whether the carbon flux anomalies caused by climate anomalies depend not only on where the climate anomalies occur but also on when they occur. In this section, we compared the dominant climate factors in the dry/wet seasons during 2010-2021. We consider consecutive months with precipitation greater than the mean value (climatology) as the wet season, and consecutive months with precipitation below the mean value as the dry season, as was adopted by Green et al (2020).

The general heterogeneous pattern does not change across the annual average, dry seasons, or wet seasons (Figure 1 and Figure 3). However, precipitation becomes less important, and radiation and VPD become more important for GPP_IAV, NBE_IAV, and ER_IAV during the wet seasons over semi-arid ecosystems, corresponding to the higher precipitation states. Radiation becomes the dominant climate factor for GPP_IAV even over the seasonally moist forest in the Amazon, and for NBE_IAV and ER_IAV over the moist forest in the Amazon during the wet season (Figure 4).

During the dry season, VPD or precipitation instead of radiation makes the largest contribution to changing GPP_IAV, NBE_IAV, and ER_IAV in the moist and seasonally moist pixels, and the relative contributions of precipitation become much larger to GPP_IAV, NEE_IAV,and ER_IAVin the semi-arid pixels (Figures 2-4). The dominant role of radiation on GPP_IAV is less obvious even in the moist Amazon Forest (AMEF-M, mean MAP>2800mm, Figure 1,3) during the dry seasons. The radiation only contributes 30% (Figure 4) to GPP_IAV during the dry season, less than the 50% contribution from the annual average (Figure 2), and the 70% contribution during the wet season. The relative contributions of radiation to NEE_IAValso reduces from 40% during the wet season to 20% during the dry season, while the relative contributions of VPD to NEE_IAVincreases from 13% to 23% in the moist Amazon Forest. The similar changes also occur over the seasonally moist forest in both tropical South America and tropical Africa. Interestingly, across all three tropical continents, the relative contributions of precipitation to carbon flux IAV over the grassland become much larger during the dry season than during the wet season, which is especially obvious for ecosystem respiration and NEE. Specifically, the relative contributions of precipitation to ER_IAV and NEE_IAVover tropical American grassland are 75% and 70% respectively, while they are only about 33% and 50% respectively during the wet season. Similar magnitude changes are also observed for tropical African and Asian grassland. For shrubland and savanna ecosystems, the changes of relative contributions of different climate factors between dry season and wet season are not as obvious as over the forest and grassland (Figure 4). This contrast of the dominant climate factors between dry seasons and wet seasons was also found for El Niño vs La Niña (next section).

Change of dominant climate factors during El Niño and La Niña climate events

El Niño and La Niña have the largest impact on the tropical climate interannual variations, with drier climate anomalies during El Niño and wetter climate anomalies during La Niña. The last section shows that the dominant climate factors for carbon flux interannual variations vary among dry seasons, wet seasons, and annual averages, implying the dependence of dominant climate factors on the mean seasonal climate states. In this section, we investigate whether the dominant climate factors for the response of the carbon cycle would depend on whether it is dry climate anomalies – El Niño, wet climate anomalies – La Niña, or a mean climate state. As in previous sections, we looked at the relative importance of precipitation (P), VPD (V), radiation (R), temperature (T), and their interactions (I), on carbon flux variability during the strongest El Niño (drier, May 2015-April 2016) and La Niña (wetter, July 2010-December 2011) months within our data-constrained period (2010-2021).

The general heterogeneous pattern does not change across the 2010-2021 annual average (Figure 1), 2015-2016 El Niño (Figures S6), and 2010-2011 La Niña months (Figure S6). However, we find similar contrasting changes between El Niño and La Niña as between wet and dry seasons. During El Niño, the dominant role of radiation on GPP_IAV is less obvious even in the wettest part of the Amazon Forest (mean MAP>2800mm, Figure 3). It changes from a median value of 50% in the annual average to a median value of 37% during the 2015-2016 El Niño (Figure S7). VPD becomes an even more important driver during El Niño for the seasonally moist rainforest in the Amazon and Congo as well as the wetter semi-arid ecosystems and grassland (mean MAP 1400-2800mm) (Figure S6). For example, the VPD contributions change from a median value of 25% to GPP_IAVover American semi-arid ecosystems during the annual average to a median value of 40% during 2015-2016 El Niño (Figure S7). The relative importance of precipitation in NEE_IAV during 2015-2016 El Niño does not change significantly from the annual average, but the temperature becomes a more important factor, attributing to the increasing contributions of temperature in ER anomaly during that period (Figure S7 and Figure 2).

During the 2010-2011 La Niña, when vegetation is less stressed due to the larger-than-usual amount of precipitation across the tropics, VPD takes over the dominance of precipitation on GPP_IAVin the driest pixels (mean MAP<1400mm), especially over African semi-arid ecosystems and grassland (right panel in Figure S7 and Figure 2). From July 2010 to December 2011, the MAPs over African semi-arid ecosystems and grassland, and Asian/Australian grassland were 14mm, 59mm, and 342mm higher than the climatology precipitation during the same months (Figure S7). Precipitation also becomes a less important driver for GPP_IAV in the same regions during La Niña climate anomalies. It changed from a median value of 35% to 10%. Even though the shifting of relative contributions of climate drivers from El Nino to La Nina is broadly consistent with the changes from dry season to wet season, the contrast of relative contribution is much smaller. The changes in relative contributions of climate drivers to the carbon flux anomalies between dry and wet seasons and between El Niño vs. La Niña imply that the dependence of dominant climate factors for carbon flux interannual variability on mean climate also applies in time.

Climate anomalies vs. sensitivity of carbon flux to climate anomalies

The contribution of any climate driver to the carbon flux interannual variability is the product of the magnitude of climate anomaly and the sensitivity of carbon flux anomalies to that climate anomaly. In this section, we disentangle whether the dominance of climate drivers in both space and time can be attributed to either the larger magnitude of that climate anomaly or to the magnitude of sensitivity of carbon flux anomalies to that climate anomaly.

We find that radiation anomaly is the weakest in both the Amazon and Congo forests compared to elsewhere, whereas the highest GPP sensitivities to radiation are also found in the same regions. This is direct evidence that the reason for the dominant radiation impact on Amazon GPP_IAV is due to its high sensitivity to radiation, not the high radiation anomaly (Figure 5). We also find the IAV of GPP, ER, and NEE in the driest tropical pixels (semi-arid and grassland/cropland) are dominated by precipitation due to the extremely high sensitivity not due to the magnitude of precipitation anomalies, since the precipitation anomalies in those regions are the weakest. Both the wet seasons and the 2010-2011 La Niña event have relatively lower GPP sensitivity to precipitation in the same regions. This suggests that GPP sensitivity to precipitation depends on the mean precipitation state (Figure S8, 9) the contrast of relative contributions different climate factors to carbon flux IAV arises from the contrast in the sensitivity to these climate factors. High-elevation regions (southwest tropical South America and southeast tropical Africa) show higher GPP sensitivity to temperatures (Figure S10). Last but not least, we find negative GPP sensitivity to VPD across all tropical ecosystems (Figure S10).

NEE sensitivity to precipitation and the relationship with the dominant role of semi-arid ecosystems in tropical NEE IAV

Tropical semi-arid ecosystems together contribute 51% to the tropical NEE IAV, 2-3 times larger than tropical forest (19%) and tropical grassland and cropland (16%), despite just a ~25% contribution to land area (Figure 6), mainly due to their high sensitivity to precipitation (Figure 5). In the next two sections, we compare the synchrony and climate drivers of GPP and ER_IAV in semi-arid and evergreen forest ecosystems to explain the large differences found in NBE_IAV.

1) Semi-arid ecosystems

The large NEE IAV in the American and African semi-arid regions have distinctly differently causes. In the African semi-arid region, although only encompassing approximately 17% of the pan-tropical area, NEE accounts for 41% of the tropical IAV (Figure 6). The large NEE year-to-year anomaly in the tropical African semi-arid ecosystems is due to the asynchrony of GPP and ER IAV. Although both GPP and ER IAV are dominated by precipitation, ER (80%), however, has much higher sensitivity than GPP (40%) (Figure 5b). The monthly anomaly of ER has much higher variability than GPP, indicating ER has a sudden drop when drought starts and an instant, dramatic rise when soil rewets (Figure S11).

Unlike the African semi-arid region where both GPP and ER IAV are driven by precipitation, in the American semi-arid region, the large NEE IAV is due to different climate drivers that determine the IAV of GPP (VPD) and ER (temperature). We show asynchronous VPD and temperature IAV in Figure S12.

2) Evergreen forest ecosystems

Contrary to the asynchronous GPP and ER_IAV in the semi-arid regions, the pulse responses of ER in evergreen forest ecosystems are rarely found (Figure S11). Based on the in situ and satellite observations-informed GPP and ER, GPP_IAV and ER_IAVdisplay a notably high degree of synchronization within the African evergreen forest (AF) (Figure 6). The synchronization in the American evergreen forest (AM) is weaker but still positive. The synchronization of GPP and ER helps elucidate the lower proportion of NEE IAV attributed to both American (17%) and African continents (2%) evergreen forest ecosystems, despite the substantial contributions to GPP IAV (33% and 9%) and ER IAV (32% and 9%) from evergreen forest ecosystems in both continents.

Mean climate states determine the dominant climate drivers for tropical carbon exchange IAV across both space and time

A number of recent studies show conflicting conclusions on temperature/precipitation being the dominant driver of tropical carbon cycle IAV, using machine learning algorithms, atmospheric inversions, and process-based global land models (Ahlström et al., 2015; Bousquet et al., 2000; Gurney et al., 2008; Peylin et al., 2013, 2005; Poulter et al., 2014; Tian et al., 1998; Wang et al., 2016; Zeng et al., 2005). By constraining a land model with a Bayesian approach using satellite observations, including NEE anomaly, GPP, SIF, biomass, mean LAI, carbon monoxide, and total water storage anomaly, we demonstrate that at the grid-cell level, the dominant climate factor for tropical carbon IAV is heterogeneous, not limited to temperature and precipitation, and determined by the local mean climate state, we show that radiation dominates GPP_IAVandNEE_IAV in the moist tropical forests (northwest Amazon, western Congo forest, Figure 1A). Our finding is consistent with some previous studies showing that the wetter part of the Amazon rainforest is photosynthetically light limited, due to frequent cloud covering that reduces light availability (Huete et al 2006; Nemani et al 2003; Saleska et al 2007). Although Jung et al (2017) found water availability predominantly drives GPP IAV in the tropics based on FLUXCOM and TREnDY models, the possibility of an underestimated role of radiation in tropical carbon cycle IAV due to the lower radiation interannual variability in the CRUNCEP data was discussed in the same study (Jung et al 2017).

We find contrasting results in the seasonally moist rainforests (MAP<2800mm), where VPD, instead of radiation, dominated GPP_IAVandNEE_IAV (Figure 1-2). The average precipitation of seasonally moist rainforest is 1740mm in the Congo rainforest and 2342mm in the Amazon, much lower than the moist Amazon rainforest (3024mm), during our investigated period (2010-2021, ERA5 data, Figure 2). Congo Forest and the seasonally moist Amazon Forest have much lower water availability in the soil and are more vulnerable to hydraulic damage when VPD is high (Blackman et al., 2019; Brodribb and Cochard, 2009; Johnson et al., 2022; Li et al., 2020; McDowell, 2011), which explains why these regions have more negative GPP sensitivity to VPD (Figure 4). We also find that the seasonally moist rainforest has higher VPD_IAV compared to the moist rainforest. Increasing VPD causes plants to close their stomata to minimize water loss, at a cost of hydraulic stress and less substrate for dark reaction. High VPD under warm temperatures can even lead to hydraulic damage, which can reduce the leaf biomass and the efficiency of photosynthesis (Gentine et al., 2019; Grossiord et al., 2020; Massmann et al., 2019; Schönbeck et al., 2022).

In tropical semiarid and grassland/cropland regions, where the annual precipitation is below a threshold of 1400mm, GPP_IAVis driven by precipitation year-to-year fluctuation. This finding is consistent with Jung et al (2017) where they find water availability is the most important factor controlling the local IAV of GPP in tropical semiarid and grassland/cropland regions, based on FLUXCOM and TRENDYv3. Satellite-based solar-induced chlorophyll fluorescence (SIF) also indicates that semiarid region GPP fluctuations are mostly controlled by precipitation (Zhang et al., 2016). It is widely acknowledged that vegetation productivity is primarily controlled by soil moisture limitation in low-rainfall areas. (Dubey and Ghosh, 2023; Krueger et al., 2021). We further show that for semiarid and grassland/cropland in southern tropical Africa and east South America, plant photosynthesis is very sensitive to soil water availability (Figure 4), which contributes to a precipitation-driven GPP_IAV.

The main focus of our study is to discuss the dominant climate factors for net ecosystem uptake of carbon (NEE), instead of the net biosphere C exchange (NBE). We looked at the relative importance of fire on carbon fluxes IAV. We find that fire is less important than any other climate factor, including the interaction effects (Figure S13), in most regions, except semi-arid regions in North Africa.

Dominate climate factor for NEE_IAV mirrors ER_IAV, instead of GPP_IAV

Dominant climate factor for NEE_IAV is the net outcome of GPP_IAV and ER_IAV. Although the distribution of climate factors for NEE_IAV and ER_IAV follows a similar pattern as GPP_IAV, where it shows some impact from radiation for both NEE and ER in the moist rainforests and VPD for the seasonally moist rainforests (Figure 1), the dominant climate factor for NEE_IAV more closely mirrors ER than GPP. Aggregating pixel level dominant drivers to latitudinal mean / biogeographic regions, both NEE_IAV and ER_IAV are dominated by precipitation (Figure 2), which is consistent with a study that found water availability drives NEE and ER IAV, using data from FLUXCOM and TRENDYv3 (Jung 2017). This result is also consistent with a few studies that find water availability becomes a more dominant driver for global CO₂ growth rate in the recent decade (Liu et al., 2023; He et al., 2023). Separate controllers compensate the climate driver for ecosystem respiration (ER) for its aboveground (autotrophic respiration, Ra) and belowground (Ra from root and heterotrophic respiration from soil microbes, Rh) processes (Figure S4). The dominant climate factor for Ra is in line with GPP, while the dominant climate factor for Rh is precipitation.

Essentially, our results suggest that in tropical regions, the dominant climate factor influencing NEE_IAV is its Rh component. This finding appears controversial compared to the current understanding of carbon flux IAV, where globally averaged data show a stronger GPP/NEE correlation than ER/NEE, as indicated by land carbon cycle models and FLUXCOM (Ahlström et al., 2015; Jung et al., 2011). However, several points provide context. First, in these studies, the wettest parts of the Amazon exhibit a low GPP/NEE correlation or a higher ER/NEE correlation according to land models (Ahlström et al., 2015; Piao et al., 2019). Additionally, the estimated contribution of GPP_IAV varies significantly among land model members, ranging from 56% to over 90% (Ahlström et al., 2015). Both land models and FLUXCOM show that the GPP/NEE correlation is weak in the dry parts of tropical regions, including eastern South America, North America, and South Africa. Despite over 72% of FLUXNET sites indicating a larger GPP contribution than ER to the IAV of net land carbon flux (Baldocchi, Chu, & Reichstein, 2018; Jensen, Herbst, & Friborg, 2017; Marcolla et al., 2017; Wu et al., 2012), none of these sites are from the Amazon or Congo forests, where both land models and FLUXCOM members show low agreement on the sign of net carbon flux anomalies, further increasing the uncertainty of GPP/NEE and ER/NEE correlation estimations.

NEE sensitivity to precipitation and its relationship with the dominant role of semi-arid ecosystems in tropical NEE IAV

While the evergreen forests are the largest contributor to GPP and ER IAV, the IAV of tropical land carbon fluxes is driven by semiarid ecosystems (Figure 6). We find predominant roles of semi-arid regions, contributing half of the tropical NEE_IAV, whereas tropical evergreen forests together contribute 19%. Similar to what we find, previous studies show that although the tropical forest is the largest C sink of all the biogeographic regions in the globe, the interannual variability in global terrestrial CO2 uptake is dominated by semi-arid regions (Poulter et al 2014; Ahlstrom et al 2015; Jung et al 2017; Humphrey et al 2021). To explain such phenomenon, we look into the monthly observations of GPP and ER in the semi-arid region, based on satellite and in situ data, and find a fast and abrupt drop of ER in response to drought, and pulse response when soil rewet (Figure S11). This asynchrony between GPP and ER likely results in a larger variation of NEE in the semi-arid regions. The respiration pulses shortly after the onset of rainfall are described extensively in local studies of water-limited ecosystems, and their large-scale relevance was recently reported by Metz et al (2023) using space-borne air CO2 measurements over the Australian semi-arid regions.

Future directions

Humphrey et al. highlight the need to use land-atmosphere coupled models to account for feedback between soil and atmospheric dryness when estimating the response of the carbon cycle to climatic change globally (2021). Our study aims to understand climate controls on carbon flux IAV with a model informed with a wide range of space-borne observations in a Bayesian model-data fusion framework. This methodology requires a relatively simple model to reduce both the optimization dimensions as well as the computation cost. Therefore, the model we use here is an offline terrestrial ecosystem model (DALEC). The next-generation earth system models aim to learn from observations and high-resolution simulations using GPUs are currently in developing, e.g. the Climate Modeling Alliance (CliMA) model (Wang et al. 2023; Schneider et al. 2017), which eventually will enable revisiting the climate controls on carbon flux IAV while taking into account the soil-atmosphere feedback. The tradeoff between model complexity and data availability in producing realistic model predictions is discussed by Famiglietti et al. (2021).

Our study reveals the complex and heterogeneous nature of climate drivers influencing carbon flux interannual variability (IAV) in tropical ecosystems. It highlights the importance of considering local climate conditions, seasonal variations, and climate events in understanding these dynamics. Furthermore, our study emphasizes the critical role of radiation, VPD, and precipitation in shaping carbon fluxes IAV and calls for a comprehensive understanding of the interplay between climate and ecosystem processes in tropical regions. The contributions of climate drivers to carbon flux interannual variability facilitate our understanding of the relative contributions of different PFTs to the carbon flux interannual variability. These findings contribute to our understanding of the climate factors driving carbon cycling in tropical ecosystems, which is essential for accurately predicting future climate change impacts and guiding conservation efforts in these vital regions.

We use a process-based model (DALEC) embedded in a data-model integration system (CARbon Data Model fraMework, CARDAMOM) and a suite of satellite constraints to investigate the effect of IAV on temperature, precipitation, vapor pressure deficit (VPD), and radiation on gross primary productivity (GPP), ecosystem respiration (ER), and net ecosystem exchange (NEE) at local and regional scales. We describe the model structure, forcings, and observational constraints in section 4.1; we describe the Bayesian Model-data fusion framework in section 4.2; climate perturbation methods are in section 4.3; we then use the data-constrained DALEC model to determine the dominant climatic drivers of the tropical carbon IAV in section 4.4.

5.1 CARDAMOM model-data fusion system and DALEC-JCR model structure

CARDAMOM is a model-data fusion (MDF) system that constrains model parameters and initial states in a process-oriented model - the Data Assimilation Linked Ecosystem Carbon model (DALEC, Figure S14) - and a Bayesian model-data integration algorithm using adaptive MHMCMC (Haario et al., 2001). In contrast to more complex process-based terrestrial biosphere models, the relatively parsimonious DALEC C cycle model structures allow for joint optimization of highly uncertain process parameters (such as initial carbon and water states, photosynthetic parameters, C allocation, turnover rates, and their dependencies on soil moisture) based on the available observations of carbon and water variables (Bloom & Williams, 2015; Famiglietti et al., 2021; Quetin et al., 2020; Williams et al., 2005; Yang et al., 2021; Norton et al. 2022). While typically incorporation of complexity in models is beneficial for process representation, Famiglietti et al. (2021) show that model complexity advantages are fundamentally limited when insufficient data is available to support parameter inference. The DALEC model-embedded CARDAMOM MDF system has been applied to a range of spatial scales with a suite of eddy covariance and satellite datasets to (i) represent C cycles, (ii) optimize critical parameters and states, and (iii) infer C cycle-associated climate sensitivities (Bloom & Williams, 2015; Famiglietti et al., 2021; Quetin et al., 2020; Williams et al., 2005; Yang et al., 2021; Norton et al. 2022; Ma et al., 2023).

Here we briefly introduce how CO2 fluxes are simulated and scaled by environmental drivers in the DALEC model, which is most relevant to this study. Complete descriptions of model structure can be found in Bloom et al. (2020) and Levine et al. (2024 in prep). GPP is calculated based on the biochemistry of C3 and C4 photosynthesis to determine potential leaf-level photosynthesis, which is calculated in terms of two potentially limiting rates: Rubisco limited rate and light-limited rate. Moisture stress factor is related to the mean soil moisture concentration in the root zone and affects the light-limited rate. The stomatal conductance is calculated using the Medlyn stomatal conductance model (Medlyn et al., 2011). DALEC represents water stress constraints on GPP and stomatal conductance based on water potential along the soil-plant continuum. Autotrophic respiration is calculated as a proportional flux to GPP. Transpiration is computed using the Penman–Monteith equation. Heterotrophic CO2 is respired both anaerobically and aerobically, both as a function of C turnover rate, soil temperature, soil moisture, and the volumetric fraction of aerobic/anaerobic soil.

Following Bloom & Williams (2015), we provide 4x5 degree resolution, monthly meteorological forcing dataset for DALEC (namely monthly temperature, total precipitation, snowfall, global incident shortwave radiation, vapor pressure deficit, CO₂, and burned area) obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis v5 (ERA5) (Berrisford et al., 2011) to drive DALEC model at each grid-cell. Satellite observations we use to constrain the model are listed in Table 1.

Table 1 Observation datasets used in data-model fusion

Observations (abbreviation)	Observation data sources	Number of observational constraints
Leaf area index (LAI)	MODIS LAI retrievals¹ (Myneni et al. 2015)	1
Soil organic carbon (SOC)	HWSD for initial SOC (Hiederer and Köchy, 2012)	1
Above- and Below- ground biomass (ABGB)	Annual 2001-2020 ABGB (Xu et al., 2021)	20
Solar-induced fluorescence (SIF)	Monthly part 2018-2021 TROPOMI retrievals of SIF² (Doughty et al., 2019)
Fire C emissions (BB)	Mean 2001-2015 4° x 5° inverse estimates of fire C emissions (Worden et al., 2017; Bowman et al., 2017)	1
Net biospheric exchange (NBE)	Monthly 2010-2018 4° x 5° inverse estimates of terrestrial NBE CMS-FLUX NBE 2020³ (Liu et al., 2021)	108
Total water storage anomaly (TWS)	Monthly 2002-2017, 2018-2021 GRACE/GRACE-FO retrieval of terrestrial water storage anomaly (Wiese 2018)	237
Gross primary production (GPP)	Monthly 2001-July2020 FluxSat GPP v2 Based on MODIS and FLUXNET (Joiner and Yoshida, 2021)	235

¹only mean 2001-2021 LAI is assimilated into CARDAMOM, to minimize the influence of seasonal LAI retrieval biases (Bi et al., 2015). ²time-resolved SIF is assimilated as a relative constraint on the temporal variability of GPP (Section 2.4 in Bloom et al lagged effect). ³annual means and monthly variabilities of observed NBE are used to constrain model (Section 2.4 in Bloom et al lagged effect).

5.2 Bayesian Model-data fusion framework

The initial states and parameters of DALEC-JCR are constrained in a Bayesian optimization framework (CARbon Data Model fraMework, CARDAMOM). The CARDAMOM framework uses Bayes' theorem to optimize the posterior probability of initial states and time-invariant process parameters (y), given observations O, p(y|O), as follows:

p(y|O) ∝ p(y)p(O|y) (1)

where p(y) is the prior probability distribution of y, and p(O|y) is proportional to the likelihood of y given O, L(y|O). At each grid-cell, the observation vector O consists of measurements including the NEE, and CH₄. Assuming errors are uncorrelated, the overall likelihood of y given O can be expressed as:

We derive the corresponding likelihood function L_* (i.e., ) as follows:

where o_i and m_i(y) correspond to the ith observation and the corresponding modeled quantity derived from control vector y, respectively; σ_i accounts for the combined errors from the DALEC model structure, forcing drivers, and observations. Following (Bloom et al., 2020; Ma et al. 2023), we retrieve the distribution of p(y|O) at each grid-cell by running four adaptive Metropolis-Hastings Markov Chain Monte Carlo (MHMCMC) chains for 10⁸ iterations.

5.3 Climate perturbations

Perturbation method

To characterize the relative importance of climate drivers variability to tropical caron IAV, we designed 5 climate perturbation scenarios, listed in Table S1. We denote the modeled C fluxes, C, as a function of time-dependent meteorological drivers (M) and time-invariant model parameters and initial states (y):

C = DALEC(M,y) (4)

where C represents carbon fluxes or carbon pools. The DALEC operator denotes the DALEC model simulation driven by forcing M and parameters y. Model simulation scenarios: Original model run (M_ori) use ERA5 reanalysis data. Perturbed model runs: replace climate drivers with 2010-2021 climatology (M_pert), including four single-factor perturbations (P, T, R, and V are short for precipitation, temperature, radiation, and VPD), and one full-factor perturbation (PTRV). Although the model is run from 2001 to 2021 for validation, we use 2010-2021 for our climate perturbation analysis due to a stronger constraint from observations as NBE from CMS Flux starts in 2010.

We take two steps to evaluate the relative importance of climate drivers' variability to tropical caron IAV:

STEP 1: Calculate the mean monthly climatology of Precipitation, Temperature, Radiation, and VPD. We then replace the original drivers with climatology mean in 5 scenarios.

STEP 2: We then calculate the difference of carbon fluxes/pool between original and perturbed climates:

dC_M = C_ori– C_pert (5)

where M could be any single perturbation scenario: P, T, R, V, or the full climate factors perturbation scenario: PTRV. By definition, we can derive the interactive effect of climate factors on carbon fluxes (dCi) using Eqn 6:

dCptrv = dCi + dCt+ dCp+ dCr + dCv (6)

We use the carbon response index K to represent the anomaly of carbon responses to climate variables:

K_M= time series mean(|dC_M|) (7)

To evaluate the relative importance of the interaction effect as well as each single climate factor to the change of carbon fluxes/states, we use eqn (8):

K_dC_M/(K_dCt+ K_dCp+ K_dCr+ K_dCv+ K_dCi) (8)

Note that because we took the absolute value of dC at each time step, (K_dCt+ K_dCp+ K_dCr+ K_dCv+ K_dCi) /K_dCptrv does not equal one, K_dCmis different from dC_m.

Carbon flux sensitivities to climate change

To answer if it is the magnitude of climate IAV, or the flux sensitivity to the climate anomaly, that led to heterogeneous dominant climate factors for carbon fluxes in different tropical ecosystems. We use climate anomaly standardized by the mean value to represent the magnitude of climate IAV (met_ano, Figure 4A). We then calculate carbon fluxes sensitivity to climate change (Figure 4B) by dividing carbon response dC_M with time series mean of normalized climate anomaly. Normalization was done by dividing the climate anomaly with the average magnitude of climate anomalies. Purpose of our normalization is to make flux sensitivities to climate drivers inter-comparable, by bringing the magnitudes each climate driver on the same scale.

Regional contribution to the interannual variability of the tropical land carbon cycle

We further partitioned IAV in global NBE among land cover classes according to the contribution of individual regions to global NBE IAV, using an index . We use the same algorithm as in Ahlström et al (2015). x_jt is the flux anomaly (departure from a long-term trend) for region j at time t; X_t is the global flux anomaly, where X_t = ∑_jxjt. By this definition is the average relative anomaly X_jt /X_tfor region j, weighted with the absolute global anomaly |X_t|.

To find out if the disparities of evergreen forest and semi-arid ecosystems contribution to NBE IAV is due to differential synchrony between GPP- and ER- anomaly. We calculated the correlations of C fluxes anomalies using Pearson’s correlation.

Data and code availability:

Data and code for generating figures are available at Dryad.

Acknowledgment

Part of the research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration, (80NM0018D0004).

Author contributions:

Conceptualization: J.L;A.A.B;S.M;D.S;

Methodology: J.L;A.A.B;S.M;

Investigation: J.L;A.A.B;S.M;L.P;R.D;A.N;

Visualization: J.L;A.A.B;S.M;R.B;E.B;

Supervision: J.L;A.A.B;S.M;

Writing—original draft: J.L;A.A.B;S.M;

Writing—review & editing: J.L;A.A.B;S.M;R.B;E.B;J.W;R.D;A.N;

Competing interests:

There is no competing interests.

Ahlström, A., Raupach, M.R., Schurgers, G., Smith, B., Arneth, A., Jung, M., Reichstein, M., Canadell, J.G., Friedlingstein, P., Jain, A.K., Kato, E., Poulter, B., Sitch, S., Stocker, B.D., Viovy, N., Wang, Y.P., Wiltshire, A., Zaehle, S., Zeng, N., 2015. The dominant role of semi-arid ecosystems in the trend and variability of the land CO2 sink. Science 348, 895–899. https://doi.org/10.1126/science.aaa1668
Anderegg, W.R.L., Ballantyne, A.P., Smith, W.K., Majkut, J., Rabin, S., Beaulieu, C., Birdsey, R., Dunne, J.P., Houghton, R.A., Myneni, R.B., Pan, Y., Sarmiento, J.L., Serota, N., Shevliakova, E., Tans, P., Pacala, S.W., 2015. Tropical nighttime warming as a dominant driver of variability in the terrestrial carbon sink. Proceedings of the National Academy of Sciences 112, 15591–15596. https://doi.org/10.1073/pnas.1521479112
Berrisford, P., Dee, D.P., Poli, P., Brugge, R., Fielding, M., Fuentes, M., Kållberg, P.W., Kobayashi, S., Uppala, S., Simmons, A., 2011. The ERA-Interim archive Version 2.0, ERA Report Series, ERA Report. ECMWF, Shinfield Park, Reading.
Blackman, C.J., Creek, D., Maier, C., Aspinwall, M.J., Drake, J.E., Pfautsch, S., O’Grady, A., Delzon, S., Medlyn, B.E., Tissue, D.T., Choat, B., 2019. Drought response strategies and hydraulic traits contribute to mechanistic understanding of plant dry-down to hydraulic failure. Tree Physiology 39, 910–924. https://doi.org/10.1093/treephys/tpz016
Bloom, A.A., Bowman, K., Liu, J., Konings, A., Worden, J., Parazoo, N., Meyer, V., Reager, J., Worden, H., Jiang, Z., Quetin, G., Smallman, T.L., Exbrayat, J.-F., Yin, Y., Saatchi, S., Williams, M., Schimel, D., 2020. Lagged effects dominate the inter-annual variability of the 2010&#8211;2015 tropical carbon balance. Biogeosciences Discussions 1–49. https://doi.org/10.5194/bg-2019-459
Bloom, A.A., Williams, M., 2015. Constraining ecosystem carbon dynamics in a data-limited world: integrating ecological "common sense" in a model–data fusion framework. Biogeosciences 12, 1299–1315. https://doi.org/10.5194/bg-12-1299-2015
Braghiere, R. K., Yamasoe, M. A., Évora do Rosário, N. M., Ribeiro da Rocha, H., de Souza Nogueira, J., & de Araújo, A. C. (2020). Characterization of the radiative impact of aerosols on CO2 and energy fluxes in the Amazon deforestation arch using artificial neural networks. Atmospheric Chemistry and Physics, 20(6), 3439–3458. https://doi.org/10.5194/acp-20-3439-2020
Brodribb, T.J., Cochard, H., 2009. Hydraulic Failure Defines the Recovery and Point of Death in Water-Stressed Conifers. Plant Physiology 149, 575–584. https://doi.org/10.1104/pp.108.129783
Barkhordarian, A., Bowman, K.W., Cressie, N., Jewell, J., Liu, J., 2021. Emergent constraints on tropical atmospheric aridity—carbon feedbacks and the future of carbon sequestration. Environ. Res. Lett. 16, 114008. https://doi.org/10.1088/1748-9326/ac2ce8
Bousquet, P., Peylin, P., Ciais, P., Le Quéré, C., Friedlingstein, P., Tans, P.P., 2000. Regional Changes in Carbon Dioxide Fluxes of Land and Oceans since 1980. Science 290, 1342–1346.
Poulter, D. Frank, P. Ciais, R. B. Myneni, N. Andela, J. Bi, G. Broquet, J. G. Canadell, F. Chevallier, Y. Y. Liu, S. W. Running, S. Sitch, G. R. van der Werf, Contribution of semi-arid ecosystems to interannual variability of the global carbon cycle. Nature509, 600–603 (2014).
Cox, P.M., Pearson, D., Booth, B.B., Friedlingstein, P., Huntingford, C., Jones, C.D., Luke, C.M., 2013. Sensitivity of tropical carbon to climate change constrained by carbon dioxide variability. Nature 494, 341–344. https://doi.org/10.1038/nature11882
Doughty, R., Köhler, P., Frankenberg, C., Magney, T.S., Xiao, X., Qin, Y., Wu, X., Moore, B., 2019. TROPOMI reveals dry-season increase of solar-induced chlorophyll fluorescence in the Amazon forest. Proceedings of the National Academy of Sciences 116, 22393–22398. https://doi.org/10.1073/pnas.1908157116
Dubey, N., Ghosh, S., 2023. The relative role of soil moisture and vapor pressure deficit in affecting the Indian vegetation productivity. Environ. Res. Lett. 18, 064012. https://doi.org/10.1088/1748-9326/acd2ef
Famiglietti, C. A., Smallman, T. L., Levine, P. A., Flack-Prain, S., Quetin, G. R., Meyer, V., Parazoo, N. C., Stettz, S. G., Yang, Y., Bonal, D., Bloom, A. A., Williams, M., and Konings, A. G.: Optimal model complexity for terrestrial carbon cycle prediction, Biogeosciences, 18, 2727–2754, https://doi.org/10.5194/bg-18-2727-2021, 2021.
Gentine, P., Green, J.K., Guérin, M., Humphrey, V., Seneviratne, S.I., Zhang, Y., Zhou, S., 2019. Coupling between the terrestrial carbon and water cycles—a review. Environ. Res. Lett. 14, 083003. https://doi.org/10.1088/1748-9326/ab22d6
Grossiord, C., Buckley, T.N., Cernusak, L.A., Novick, K.A., Poulter, B., Siegwolf, R.T.W., Sperry, J.S., McDowell, N.G., 2020. Plant responses to rising vapor pressure deficit. New Phytologist 226, 1550–1566. https://doi.org/10.1111/nph.16485
Gurney, K.R., Baker, D., Rayner, P., Denning, S., 2008. Interannual variations in continental-scale net carbon exchange and sensitivity to observing networks estimated from atmospheric CO2 inversions for the period 1980 to 2005. Global Biogeochemical Cycles 22. https://doi.org/10.1029/2007GB003082
He, B., Xie, X., & Guo, L. (2023). A shift from temperature to water as the primary driver for interannual variability of the tropical carbon cycle. Geophysical Research Letters, 50, e2023GL102812. https://doi.org/10.1029/2023GL102812
Hiederer, R., Köchy, M., 2012. Global Soil Organic Carbon Estimates and the Harmonized World Soil Database [WWW Document]. JRC Publications Repository. https://doi.org/10.2788/13267
Humphrey, V., Berg, A., Ciais, P. et al. Soil moisture–atmosphere feedback dominates land carbon uptake variability. Nature 592, 65–69 (2021). https://doi.org/10.1038/s41586-021-03325-5
Humphrey, V. et al. Sensitivity of atmospheric CO2 growth rate to observed changes in terrestrial water storage. Nature 560, 628-631 (2018).
Ishibashi, M., Terashima, I., 1995. Effects of continuous leaf wetness on photosynthesis: adverse aspects of rainfall. Plant, Cell & Environment 18, 431–438. https://doi.org/10.1111/j.1365-3040.1995.tb00377.x
Johnson, D.M., Katul, G., Domec, J., 2022. Catastrophic hydraulic failure and tipping points in plants. Plant Cell & Environment 45, 2231–2266. https://doi.org/10.1111/pce.14327
Joiner, J., Yoshida, Y., 2021. Global MODIS and FLUXNET-derived Daily Gross Primary Production, V2. ORNL DAAC. https://doi.org/10.3334/ORNLDAAC/1835
Jung, M., Reichstein, M., Schwalm, C.R., Huntingford, C., Sitch, S., Ahlström, A., Arneth, A., Camps-Valls, G., Ciais, P., Friedlingstein, P., Gans, F., Ichii, K., Jain, A.K., Kato, E., Papale, D., Poulter, B., Raduly, B., Rödenbeck, C., Tramontana, G., Viovy, N., Wang, Y.-P., Weber, U., Zaehle, S., Zeng, N., 2017. Compensatory water effects link yearly global land CO2 sink changes to temperature. Nature 541, 516–520. https://doi.org/10.1038/nature20780
Jung, M., Koirala, S., Weber, U. et al. The FLUXCOM ensemble of global land-atmosphere energy fluxes. Sci Data 6, 74 (2019). https://doi.org/10.1038/s41597-019-0076-8
Kindermann, J., Würth, G., Kohlmaier, G.H., Badeck, F.-W., 1996. Interannual variation of carbon exchange fluxes in terrestrial ecosystems. Global Biogeochemical Cycles 10, 737–755. https://doi.org/10.1029/96GB02349
Krueger, E.S., Ochsner, T.E., Levi, M.R., Basara, J.B., Snitker, G.J., Wyatt, B.M., 2021. Grassland productivity estimates informed by soil moisture measurements: Statistical and mechanistic approaches. Agronomy Journal 113, 3498–3517. https://doi.org/10.1002/agj2.20709
Li, M.-Y., Fang, L.-D., Duan, C.-Y., Cao, Y., Yin, H., Ning, Q.-R., Hao, G.-Y., 2020. Greater risk of hydraulic failure due to increased drought threatens pine plantations in Horqin Sandy Land of northern China. Forest Ecology and Management 461, 117980. https://doi.org/10.1016/j.foreco.2020.117980
Letts, M.G., Mulligan, M., 2005. The impact of light quality and leaf wetness on photosynthesis in north-west Andean tropical montane cloud forest. Journal of Tropical Ecology 21, 549–557. https://doi.org/10.1017/S0266467405002488
Liu, J., Baskaran, L., Bowman, K., Schimel, D., Bloom, A.A., Parazoo, N.C., Oda, T., Carroll, D., Menemenlis, D., Joiner, J., Commane, R., Daube, B., Gatti, L.V., McKain, K., Miller, J., Stephens, B.B., Sweeney, C., Wofsy, S., 2021. Carbon Monitoring System Flux Net Biosphere Exchange 2020 (CMS-Flux NBE 2020). Earth System Science Data 13, 299–330. https://doi.org/10.5194/essd-13-299-2021
Liu, J., Bowman, K.W., Schimel, D.S., Parazoo, N.C., Jiang, Z., Lee, M., Bloom, A.A., Wunch, D., Frankenberg, C., Sun, Y. and O’Dell, C.W., 2017. Contrasting carbon cycle responses of the tropical continents to the 2015–2016 El Niño. Science, 358(6360), p.eaam5690.
Liu, L., Ciais, P., Wu, M. et al. Increasingly negative tropical water–interannual CO2 growth rate coupling. Nature 618, 755–760 (2023). https://doi.org/10.1038/s41586-023-06056-x
Liu, J., Bowman, K., Palmer, P. I., Joiner, J., Levine, P., Bloom, A. A., et al. (2024). Enhanced carbon flux response to atmospheric aridity and water storage deficit during the 2015–2016 El Niño compromised carbon balance recovery in tropical South America. AGU Advances, 5, e2024AV001187. https://doi.org/10.1029/2024AV001187
Ma, S., Bloom, A. A., Watts, J. D., Quetin, G. R., Donatella, Z., Euskirchen, E. S., et al. (2023). Resolving the carbon-climate feedback potential of wetland CO2 and CH4 fluxes in Alaska. Global Biogeochemical Cycles, 37, e2022GB007524. https://doi.org/10.1029/2022GB007524
Massmann, A., Gentine, P., Lin, C., 2019. When Does Vapor Pressure Deficit Drive or Reduce Evapotranspiration? Journal of Advances in Modeling Earth Systems 11, 3305–3320. https://doi.org/10.1029/2019MS001790
McDowell, N.G., 2011. Mechanisms Linking Drought, Hydraulics, Carbon Metabolism, and Vegetation Mortality. Plant Physiology 155, 1051–1059. https://doi.org/10.1104/pp.110.170704
Myneni, R., Knyazikhin, Y., Park, T. (2015). MOD15A2H MODIS Leaf Area Index/FPAR 8-Day L4 Global 500m SIN Grid V006. NASA EOSDIS Land Processes DAAC. http://doi.org/10.5067/MODIS/MOD15A2H.006 (Terra)
O'Sullivan, M., Smith, W. K., Sitch, S., Friedlingstein, P., Arora, V. K., Haverd, V., et al. (2020). Climate-driven variability and trends in plant productivity over recent decades based on three global products. Global Biogeochemical Cycles, 34, e2020GB006613. https://doi.org/10.1029/2020GB006613
Peylin, P., Bousquet, P., Le Quéré, C., Sitch, S., Friedlingstein, P., McKinley, G., Gruber, N., Rayner, P., Ciais, P., 2005. Multiple constraints on regional CO2 flux variations over land and oceans. Global Biogeochemical Cycles 19. https://doi.org/10.1029/2003GB002214
Peylin, P., Law, R.M., Gurney, K.R., Chevallier, F., Jacobson, A.R., Maki, T., Niwa, Y., Patra, P.K., Peters, W., Rayner, P.J., Rödenbeck, C., van der Laan-Luijkx, I.T., Zhang, X., 2013. Global atmospheric carbon budget: results from an ensemble of atmospheric CO2 inversions. Biogeosciences 10, 6699–6720. https://doi.org/10.5194/bg-10-6699-2013
Piao S, Wang X, Wang K, et al. Interannual variation of terrestrial carbon cycle: Issues and perspectives. Glob Change Biol. 2020; 26: 300–318. https://doi.org/10.1111/gcb.14884
Poulter, B., Frank, D., Ciais, P., Myneni, R.B., Andela, N., Bi, J., Broquet, G., Canadell, J.G., Chevallier, F., Liu, Y.Y., Running, S.W., Sitch, S., Van Der Werf, G.R., 2014. Contribution of semi-arid ecosystems to interannual variability of the global carbon cycle. Nature 509, 600–603. https://doi.org/10.1038/nature13376
Quetin, G.R., Bloom, A.A., Bowman, K.W., Konings, A.G., 2020. Carbon Flux Variability From a Relatively Simple Ecosystem Model With Assimilated Data Is Consistent With Terrestrial Biosphere Model Estimates. Journal of Advances in Modeling Earth Systems 12. https://doi.org/10.1029/2019MS001889
Schimel, D., Pavlick, R., Fisher, J. B., Asner, G. P., Saatchi, S., Townsend, P., Miller, C., Frankenberg, C., Hibbard, K., & Cox, P. (2015). Observing terrestrial ecosystems and the carbon cycle from space. Global Change Biology, 21(5), 1762–1776. https://doi.org/10.1111/GCB.12822
Schneider, T., Lan, S., Stuart, A., & Teixeira, J. (2017). Earth system modeling 2.0: A blueprint for models that learn from observations and targeted high-resolution simulations. Geophysical Research Letters, 44, 12,396–12,417. https://doi.org/10.1002/2017GL076101
Schönbeck, L.C., Schuler, P., Lehmann, M.M., Mas, E., Mekarni, L., Pivovaroff, A.L., Turberg, P., Grossiord, C., 2022. Increasing temperature and vapour pressure deficit lead to hydraulic damages in the absence of soil drought. Plant Cell Environ 45, 3275–3289. https://doi.org/10.1111/pce.14425
Simard, M., Pinto, N., Fisher, J.B., Baccini, A., 2011. Mapping forest canopy height globally with spaceborne lidar. Journal of Geophysical Research: Biogeosciences 116. https://doi.org/10.1029/2011JG001708
Tian, H., Melillo, J.M., Kicklighter, D.W., McGuire, A.D., Helfrich, J.V.K., Moore, B., Vörösmarty, C.J., 1998. Effect of interannual climate variability on carbon storage in Amazonian ecosystems. Nature 396, 664–667. https://doi.org/10.1038/25328
Wang, Y., Liu, J., Wennberg, P. O., He, L., Bonal, D., Köhler, P., Frankenberg, C., Sitch, S., & Friedlingstein, P. (2023). Elucidating climatic drivers of photosynthesis by tropical forests. Global Change Biology, 29, 4811–4825.
Wang, J., Zeng, N., Wang, M., 2016. Interannual variability of the atmospheric CO2 growth rate: roles of precipitation and temperature. Biogeosciences 13, 2339–2352. https://doi.org/10.5194/bg-13-2339-2016
Wang, W., Ciais, P., Nemani, R.R., Canadell, J.G., Piao, S., Sitch, S., White, M.A., Hashimoto, H., Milesi, C., Myneni, R.B., 2013. Variations in atmospheric CO2 growth rates coupled with tropical temperature. Proceedings of the National Academy of Sciences 110, 13061–13066. https://doi.org/10.1073/pnas.1219683110
Wang, X., Piao, S., Ciais, P., Friedlingstein, P., Myneni, R.B., Cox, P., Heimann, M., Miller, J., Peng, S., Wang, T., Yang, H., Chen, A., 2014. A two-fold increase of carbon cycle sensitivity to tropical temperature variations. Nature 506, 212–215. https://doi.org/10.1038/nature12915
Wang, Y., Braghiere, R. K., Longo, M., Norton, A. J., Köhler, P., Doughty, R., et al. (2023). Modeling global vegetation gross primary productivity, transpiration and hyperspectral canopy radiative transfer simultaneously using a next generation land surface model—CliMA Land. Journal of Advances in Modeling Earth Systems, 15, e2021MS002964. https://doi.org/10.1029/2021MS002964
Williams, M., Schwarz, P.A., Law, B.E., Irvine, J., Kurpius, M.R., 2005. An improved analysis of forest carbon dynamics using data assimilation. Global Change Biology 11, 89–105. https://doi.org/10.1111/j.1365-2486.2004.00891.x
Humphrey, A. Berg, P. Ciais, P. Gentine, M. Jung, M. Reichstein, S. I. Seneviratne, C. Frankenberg, Soil moisture-atmosphere feedback dominates land carbon uptake variability. Nature592, 65–69 (2021).
Xu, L., Saatchi, S.S., Yang, Y., Yu, Y., Pongratz, J., Bloom, A.A., Bowman, K., Worden, J., Liu, J., Yin, Y., Domke, G., McRoberts, R.E., Woodall, C., Nabuurs, G.-J., de-Miguel, S., Keller, M., Harris, N., Maxwell, S., Schimel, D., 2021. Changes in global terrestrial live biomass over the 21st century. Science Advances 7, eabe9829. https://doi.org/10.1126/sciadv.abe9829
Yang, Y., Bloom, A.A., Ma, S., Levine, P., Norton, A., Parazoo, N., Reager, J., Worden, J., Quetin, G., Smallman, T.L., Williams, M., Xu, L., Saatchi, S., 2021. CARDAMOM-FluxVal Version 1.0: a FLUXNET-based Validation System for CARDAMOM Carbon and Water Flux Estimates. Geoscientific Model Development Discussions 1–25. https://doi.org/10.5194/gmd-2021-190
Zeng, N., Mariotti, A., Wetzel, P., 2005. Terrestrial mechanisms of interannual CO2 variability. Global Biogeochemical Cycles 19. https://doi.org/10.1029/2004GB002273
Zhang, Y., Xiao, X., Guanter, L., Zhou, S., Ciais, P., Joiner, J., Sitch, S., Wu, X., Nabel, J., Dong, J., Kato, E., Jain, A.K., Wiltshire, A., Stocker, B.D., 2016. Precipitation and carbon-water coupling jointly control the interannual variability of global land gross primary production. Sci Rep 6, 39748. https://doi.org/10.1038/srep39748

There is NO Competing Interest.

suppMaetal2024Sep5NC.docx
supplemantary figures and tables

Download PDF

Version 1

posted

You are reading this latest preprint version

From rain to radiance - Drivers of tropical carbon flux interannual variability shift with local climate

Status:

Version 1

Abstract

Figures

Key Points

Teaser

1. Introduction

2. Results

3 Discussion

4. Conclusion

5. Methods and Data Description

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1