S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E AT M O S P H E R I C S C I E N C E Copyright © 2023 The Authors, some Large diurnal compensatory effects mitigate the rights reserved; exclusive licensee response of Amazonian forests to atmospheric warming American Association for the Advancement and drying of Science. No claim to original U.S. Government Works. Distributed Zhaoying Zhang1,2,3, Alessandro Cescatti4, Ying-Ping Wang5, Pierre Gentine6, Jingfeng Xiao7, under a Creative Luis Guanter8, Alfredo R. Huete9, Jin Wu10,11, Jing M. Chen12, Weimin Ju1,3, Josep Peñuelas13,14, Commons Attribution Yongguang Zhang1,3,15* NonCommercial License 4.0 (CC BY-NC). Photosynthesis and evapotranspiration in Amazonian forests are major contributors to the global carbon and water cycles. However, their diurnal patterns and responses to atmospheric warming and drying at regional scale remain unclear, hindering the understanding of global carbon and water cycles. Here, we used proxies of photosynthesis and evapotranspiration from the International Space Station to reveal a strong depression of dry season afternoon photosynthesis (by 6.7 ± 2.4%) and evapotranspiration (by 6.1 ± 3.1%). Photosynthesis positively responds to vapor pressure deficit (VPD) in the morning, but negatively in the afternoon. Further- more, we projected that the regionally depressed afternoon photosynthesis will be compensated by their in- creases in the morning in future dry seasons. These results shed new light on the complex interplay of climate with carbon and water fluxes in Amazonian forests and provide evidence on the emerging environmental con- straints of primary productivity that may improve the robustness of future projections. INTRODUCTION Given these potential climate impacts on Amazon rainforests, it Amazonian rainforests play an important role in the climate system is becoming critical to understand and predict the direction and by exchanging large amounts of energy, water, and carbon dioxide magnitude of photosynthesis and ET changes in response to atmo- (CO2) with the atmosphere and acting as a major carbon stock. spheric warming and drying in Amazonian forests (7, 8). Dynamic Climate change is expected to make the Amazon forest substantially global vegetation models were used to predict forest response to warmer and potentially drier over the 21st century (1) with potential long-term climate change, whereas there were substantial uncer- warring impacts on primary productivity and the regional carbon tainties in the prediction due to their poor responses to short- budget. Photosynthesis and evapotranspiration (ET) may decline term variations in environmental drivers (9). For example, under future atmospheric warming and drying due to the stomatal because land-surface models do not accurately account for the closure, hydraulic failure, and Rubisco enzymatic capacity impair- daily hysteretic response of stomatal conductance to compound at- ment (2, 3). Large-scale decreases in ET might further feed back to mospheric and soil dryness, most of them underestimate and over- the regional climate by reducing cloud formation and subsequent estimate latent heat flux (LE; equivalent to ET) in the morning and rainfall events (4), causing a higher risk of drought and tree mortal- afternoon, respectively (10). These mechanisms also highlight diffi- ity (5). Moreover, the negative impact on primary productivity of culties in predicting the responses of photosynthesis and transpira- Amazon forests could substantially weaken their carbon uptake (6). tion to warmer and drier conditions. In addition to model simulations, numerous studies based on ground and satellite measurements have quantified the underlying 1 Jiangsu Center for Collaborative Innovation in Geographical Information Resource mechanisms (biotic or abiotic) of the Amazon seasonality (11–13). Development and Application, International Institute for Earth System Sciences, Nanjing University, Nanjing, Jiangsu 210023, China. 2Yuxiu Postdoctoral Institute, The photosynthetic seasonality in Amazonian forests is strongly Nanjing University, Nanjing, Jiangsu 210023, China. 3Jiangsu Provincial Key Labo- related to phenology (that is, leaf development and demography) ratory of Geographic Information Science and Technology, Key Laboratory for Land (11), rather than directly to the physiological response to environ- Satellite Remote Sensing Applications of Ministry of Natural Resources, School of mental drivers that explain only 3% of the photosynthesis variability Geography and Ocean Science, Nanjing University, Nanjing, Jiangsu 210023, China. 4European Commission, Joint Research Centre (JRC), Ispra, Italy. 5CSIRO, at the monthly scale (14). As a result, the response of Amazonian Oceans and Atmosphere, Private Bag 1, Aspendale, Victoria 3195, Australia. forests to atmospheric warming and drying remains controversial, 6 Department of Earth and Environmental Engineering, Columbia University, because conventional assessments of these environmental effects on New York, NY, USA. 7Earth Systems Research Center, Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, NH, USA. photosynthesis and ET, such as the changes in the phenology of 8 Research Institute of Water and Environmental Engineering (IIAMA), Department leaves and their age demography (young or mature) at seasonal of Applied Physics, Polytechnic University of Valencia, Valencia, Spain. 9School of scale, often masked out their responses to warming or drying at Life Sciences, University of Technology Sydney, Sydney, New South Wales, Austra- the diurnal time scale. lia. 10School of Biological Sciences, The University of Hong Kong, Pokfulam, Hong Kong. 11State Key Laboratory of Agrobiotechnology, The Chinese University of Nevertheless, environmental drivers explain most of the diurnal Hong Kong, Shatin, Hong Kong, China. 12Department of Geography and Planning, variations in photosynthesis when leaf phenology barely changes University of Toronto, Toronto, Ontario, Canada. 13CSIC, Global ecology Unit within a day (14). Air temperature (Tair) and atmospheric dryness CREAF-CSIC-UAB, Bellaterra 08193, Catalonia, Spain. 14CREAF, Cerdanyola del Vallès 08193, Catalonia, Spain. 15International Joint Carbon Neutrality Laboratory, [or vapor pressure deficit (VPD)] are higher in the afternoon than Nanjing University, Nanjing, Jiangsu 210023 China. in the morning (15, 16), while the foliar water potential decreases *Corresponding author. Email: yongguang_zhang@nju.edu.cn Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 1 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E from morning to afternoon due to the loss of water through tran- perturbations in environmental drivers [Tair, VPD, and soil mois- spiration and then recovers overnight (17–19). These processes can ture (SM)] and their differences between morning and afternoon. induce different responses of photosynthesis and ET to Tair and We further generated temporally and spatially continuous hourly VPD between morning and afternoon. The difference in the total canopy SIF emission (denoted as SIF from here onward) and fluxes of carbon and water between morning and afternoon pro- ET during 2015–2021 using random forest (RF) models trained with vides key information on the physiological responses of plants to the response variables of OCO-3 SIF and ECOSTRESS ET and with environmental drivers and potential constraints on primary pro- the drivers of Tair, VPD, SM, radiation, and fraction of photosyn- ductivity (15). For example, an afternoon depression of photosyn- thetic active radiation (FPAR) (see Materials and Methods). Using thesis is a typical indicator of stomatal stress due to high VPD, the RF models, we also projected changes in future photosynthesis whereas a symmetric pattern between morning and afternoon sug- and ET from 2015 to 2100 under a scenario of climate warming and gests that photosynthesis is predominantly light limited (15). increasing VPD based on the sixth phase of the Coupled Model In- Diurnal variations in photosynthesis and ET have been mea- tercomparison Project (CMIP6). The limited biases related to the sured at a few eddy covariance (EC) flux sites across the Amazon spatial extrapolations, temporal extrapolations, and future extrapo- forest, where afternoon reductions in photosynthesis and ET have lations does not affect our results and conclusions using RF models been observed (16). However, these flux sites are sparsely distribu- (see Material and Methods). ted across the region and may not fully represent the complex geo- graphical patterns in vegetation and climate of the Amazon forest (20), resulting in a clear gap in scaling up from site observations RESULTS to regional estimates of photosynthesis and ET. At present, large- Diurnal patterns of photosynthesis and ET in scale observations of diurnal variations in photosynthesis or ET Amazon forests of Amazon forests are lacking, limiting our understanding of the First, we explored the diurnal patterns of clear-sky SIF, ET, and physiological regulation of carbon-water fluxes and their feedbacks WUESIF across the Amazon forest (fig. S1) based on the RF model- to the climate. ing from 2015 to 2021 (Fig. 1). Both carbon and water fluxes typi- The International Space Station (ISS) hosts a series of instru- cally peaked around noon (11:00 to 13:00) (Fig. 2, A and D), and ments that observe the Earth at different times throughout the their diurnal variations were driven mainly by solar radiation and day (21). Among these instruments, the Ecosystem Spaceborne air temperature. SIF slightly increased from 7.69 mW · m−2 · nm−1 Thermal Radiometer Experiment (ECOSTRESS) has been retriev- in the morning to 7.70 mW · m−2 · nm−1 in the afternoon in the wet ing ET measurements since 2018 (22), and the Orbiting Carbon Ob- season but decreased from 8.12 to 7.76 mW · m−2 · nm−1 in the dry servatory-3 (OCO-3) has been measuring solar-induced season (Fig. 2A), which was consistent with CMIP6 earth system chlorophyll fluorescence (SIF) since 2019 (23). SIF is electromag- model (ESM) simulations: slightly increasing in the morning and netic radiation emitted by leaf chlorophyll during the light reaction decreasing in the afternoon (Fig. 2C). Site-based gross primary pro- of photosynthesis (24) and can be retrieved from high spectral res- duction (GPP) also showed a larger afternoon decrease in the dry olution atmospheric and meteorological space missions (25, 26). season than in the wet season (Fig. 2B), which can be attributed Over the past decade, SIF has been shown to be a promising indi- to the much higher VPD and Tair in the dry season afternoon cator of terrestrial photosynthesis (25, 27, 28) and for studying the (fig. S2). Such a higher water stress condition of the dry season af- seasonality of Amazonian forests (29, 30) and their responses to ternoon is further observed to accompany with a reduction in ET by drought (31, 32) and aridity (33). However, because traditional RF models (Fig. 2D) and a weak ET increase by site level flux obser- sun-synchronous satellites pass over a given location on Earth at vations (Fig. 2E), both of which are in large contrast to the substan- the same time each day, they cannot provide observations for differ- tial afternoon increase in wet season ET. CMIP6 ESMs simulated a ent times of day over the Amazon. similar afternoon increase in ET between the wet and dry seasons Hence, the combination of ECOSTRESS ET and OCO-3 SIF pro- (Fig. 2F), showing that ESMs could not fully capture the effects of vides a unique opportunity for studying the large-scale fluxes of water stress on ET. In addition, the morning WUESIF was always carbon and water at the diurnal time scale. The diurnal photosyn- higher than the afternoon WUESIF, consistent with WUEGPP from thesis and ET at regional scales are informative as how land-atmo- both site-based flux observations and CMIP6 ESM simulations sphere interactions are regulated by changes in physiological (Fig. 2, G to I). Because we simulated clear-sky SIF and ET using responses to varying environmental conditions (34). In addition, the symmetrical radiation around local solar noon, the differences the high Tair and VPD in the afternoon can mimic future conditions in SIF and ET between morning and afternoon were not due to light compared with current status. Thus, an analysis of the diurnal pat- effects but rather to increased Tair and VPD and the related stomatal terns of photosynthesis and ET from morning to afternoon can regulation on photosynthesis and transpiration. offer a unique view of their future trends and responses to climate Next, we investigated the diurnal and seasonal patterns of carbon change. Therefore, we aim to investigate the large-scale diurnal pat- and water fluxes across the Amazon forest and the hysteretic re- terns of photosynthesis and ET in the Amazon forest and their re- sponses of WUE to VPD, which have not yet been observed at sponses to atmospheric warming and drying by combining the regional scale before. The amplitudes of the diurnal variations ECOSTRESS ET with OCO-3 SIF. in both SIF and ET were greater than their respective seasonal var- Building on these ata streams, we investigated the large-scale iations (Fig. 3, A and B and fig. S3), and the relationships between diurnal patterns of photosynthesis and ET in the Amazon forest SIF and ET differed significantly (P < 0.001, using one-way analysis and SIF-based water use efficiency (WUESIF = SIF/ET) (34), of covariance) between the morning and afternoon in both the wet which is a crucial indicator of the water use strategy in photosynthe- and dry seasons (Fig. 3, C and D). The results of the linear regres- sis (35). We also analyzed the sensitivities of SIF, ET, and WUESIF to sion modeling showed that SIF was lower in the late afternoon than Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 2 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E Fig. 1. Diurnal patterns of SIF (mW · m−2 · nm−1), ET (W · m−2), and WUESIF (mW · m−2 · nm−1/W · m−2) in Amazonian forests throughout 2015–2020. W · m−2 Diurnal (A to E) SIF, (F to J) ET, and (K to O) WUESIF (=SIF/ET) at 2-hour intervals: 7:00 to 9:00, 9:00 to 11:00, 11:00 to 13:00, 13:00 to 15:00, and 15:00 to 17:00. The mean and SD are shown in the bottom right of each panel. The spatial resolution is 1° × 1°. in the early morning for the same level of ET, supporting the idea (fig. S2). Marginal diurnal changes in SIF were observed during that WUESIF declined during the day as VPD increased (Fig. 3, E the wet season (Fig. 5A), indicating that photosynthesis was and F). Moreover, the regional diurnal WUESIF exhibited an mostly light limited when Tair and VPD were low (fig. S2, A and oblique U-shaped, hysteretic relationship with VPD (Fig. 3, G and F) and SM was high (fig. S2K). However, during the dry season, H). The hysteresis of WUESIF-VPD suggests that the relationship SIF decreased from morning to afternoon (Fig. 5B), especially in between photosynthesis and ET as quantified by WUESIF changes the southeastern part of the Amazon where rainfall was low and during the day, most likely because of a decline of water content Tair and VPD were high. ET increased and decreased diurnally in the plant system and an increased sensitivity of stomatal conduc- over much of the Amazon in the wet and dry seasons, respectively tance to VPD in the afternoon. (Fig. 5, C and D), with a shift from a demand-limited regime to a To assess how the hysteresis of WUESIF-VPD varied spatially, we supply-limited regime. Therefore, these results revealed varying selected three subregions in eastern, northwestern, and southwest- effects of Tair and VPD on photosynthesis and ET between the ern Amazon (three boxes in fig. S4) that differ in canopy structure, wet and dry seasons. phenology, and rainfall and found that the hysteretic responses of WUESIF to VPD varied spatially across the three regions (Fig. 4), Sensitivities of photosynthesis and ET to indicating that the limited site observation cannot adequately rep- environmental drivers resent all the forests in the Amazon region. The strongest hysteretic The afternoon depressions of photosynthesis and ET in the dry response of WUESIF to VPD was observed in drier regions, such as season (Fig. 5, B and D) provide unique insights on how land-atmo- the eastern Amazon in the dry season (Fig. 4E). The oblique U- sphere interactions will respond to future climatic warming and at- shaped relationships between WUESIF and VPD across different mospheric drying, but it is unclear how individual environmental subregions were consistent with surface flux observations at three drivers (such as Tair, VPD, and SM) affect the fluxes of carbon Amazonian sites (BR-Sa1, BR-Sa3, and GF-Guy) (figs. S5 to S7). and water at the diurnal scale. Particularly, the effects of Tair, These results indicated that our regional hourly SIF and ET cap- VPD and incoming solar radiation were commonly mixed due to tured well the diurnal, spatial, and seasonal variations in photosyn- their high correlations (fig. S10). To distinguish between the inter- thesis and ET and their responses to environmental drivers. dependent effects of environmental drivers on the fluxes of carbon To quantify the Amazonian diurnal variations in the ecosystem and water, we analyzed the sensitivities of SIF, ET, and WUESIF fluxes of carbon and water, we compared SIF, ET, and WUESIF using RF models (see Materials and Methods). between morning (7:00 to 12:00) and afternoon (12:00 to 17:00). The results showed positive sensitivity of SIF to VPD (in other Clear spatial variations were observed for SIF, ET, and WUESIF in words, SIF would increase with VPD) in the morning more than both the wet and dry seasons (fig. S8). For example, WUESIF during 98.7% and 70.9% of the region in the wet and dry seasons, respec- the dry season was generally lowest in the east with relatively tively (Fig. 6, A, B, and D), after factoring out the effects of other younger forest (fig. S9), where VPD was high and SM was low drivers. The negative sensitivity of photosynthesis to VPD in the Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 3 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E Fig. 2. Diurnal patterns of energy, carbon, and water fluxes in Amazonian forests for wet and dry seasons. The RF models generated (A) SIF (mW · m−2 · nm−1), (D) ET (W · m−2), and (G) WUESIF (mW · m−2 · nm−1/W · m−2) averaged over 2015–2021. The shaded area represented the one SD across the Amazon forest. Site-based eddy flux-derived (B) GPP (μmol · m−2 · s−1), (E) ET (W · m−2), and (H) WUEGPP (μmol · m−2 · s−1/W · m−2). The shaded area represented the one SD across three sites (BR-Sa1, BR- Sa3, and GF-Guy). CMIP6 ESM simulated historical (C) GPP (μmol · m−2 · s−1), (F) ET (W · m−2), and (I) WUEGPP (μmol · m−2 · s−1/W · m−2) averaged over 2000–2014. The shaded area represented the one SD across ESMs. More details about the sites and CMIP6 ESM can be found in tables S1 and S2, respectively. The averages of morning (a.m.) and afternoon (p.m.) were shown in blue and orange for the wet and dry seasons, respectively. dry season morning mainly occurred in the nonforest regions (such afternoon led to a decrease in SIF, which is a sign of what could as grass and savanna). The sensitivity of SIF to VPD also showed a happen in Amazonian forests under the projected atmospheric clear northwest-southeast gradient from being positive to being warming and drying in the future. negative in the dry season morning, indicating that the super-wet Next, we quantified the large-scale afternoon depressions of SIF, northwest Amazon was not VPD stressed. However, the afternoon ET, and WUESIF following increases in Tair and VPD. For this SIF exhibited negative sensitivity to VPD more than 74.5 and 67.4% purpose, we applied the afternoon RF model to the morning con- of the region in the wet and dry seasons, respectively (Fig. 6, A, C, ditions during 2015–2021 (see Materials and Methods). The results and E). Moreover, asymmetric responses of SIF to Tair between of this process were considered in the “modelPM + conditionAM” morning and afternoon were observed in the dry season (Fig. 7, simulations and were compared with the “modelAM + conditionAM” A, D, and E). Different from SIF, ET showed consistently negative simulations. SIF and ET decreased by 4.4 ± 1.5 and 2.1 ± 2.8%, re- sensitivities to VPD (Fig. 6, F to J) in the morning or afternoon in spectively, from the modelAM + conditionAM simulations to the both dry and wet seasons but positive sensitivities to Tair (Fig. 7, F to modelPM + conditionAM simulations in the wet season (Fig. 8, A J). These results stressed the importance of stomatal control on the and C); these decreases were stronger in the dry season (Fig. 8, B ET of the Amazon that is able to reverse the sensitivity of ET to VPD and D). The stronger reduction in SIF than ET led to decreases in (i.e., plants are transpiring less when the atmospheric demand is WUESIF by 2.2 ± 3.6 and 0.6 ± 3.5% in the wet and dry seasons, re- higher). Consequently, WUESIF was positively sensitive to VPD spectively (Fig. 8, E and F). Therefore, the strong afternoon reduc- (Fig. 6, K to O) but negatively sensitive to Tair (Fig. 7, K to O); there- tions in SIF and ET point to the stronger asymmetric responses of fore, forests varied their water use strategies under different temper- photosynthesis and ET to atmospheric warming and drying in the ature and atmospheric dryness stresses. Given the small variations dry season than in the wet season in the Amazon forest. in SM during the course of the day, SIF and ET exhibited weaker sensitivities to SM (fig. S11) than to VPD and Tair. The combined negative sensitivities of SIF to Tair and VPD in the dry season Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 4 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E Fig. 3. Seasonal and diurnal patterns (shown as fingerprints) of SIF (mW · m−2 · nm−1), ET (W · m−2), WUESIF (mW · m−2 · nm−1/W · m−2), and VPD (kPa) through- out the Amazon forest. (A) SIF, (B) ET, (E) WUESIF, and (F) VPD. The red vertical lines represent the dry season from June to October. The SIF and ET were derived on the basis of the RF models during 2015 and 2021, and VPD was from the ERA5 reanalysis data during 2015 and 2021. The linear relationships between SIF and ET in the (C) wet and (D) dry seasons, with crosses (morning) and squares (afternoon) representing average hourly values at a monthly temporal resolution in (A) and (B). The relationships between hourly WUESIF and VPD averaged over the (G) wet and (H) dry seasons. The black arrows represent the directions of variations in diurnal WUESIF and VPD. Fig. 4. Relationships between WUESIF and VPD for the different regions. (A to C) Northwestern, eastern, and southwestern Amazon Basin (red, blue, and black boxes, respectively, in fig. S4). The dots represent the seasonal averages, with the colors indicating the time of day. (D to F) similar to (A) to (C) but for the dry season. Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 5 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E ET was nearly constant. As a result, daily ET increased (Fig. 9, M and N). The widespread reductions in SIF (89%) and ET (77%) in the dry season afternoon were predicted by the end of this century under SSP585 (figs. S12 and S13). On the basis of our data stream of diurnal SIF and ET, our results revealed stronger sensitivity to atmo- spheric dryness leading to overall a neutral response of Amazonian photosynthesis to atmospheric warming and drying in the future dry season. DISCUSSION Differences in diurnal patterns of photosynthesis and ET between wet and dry seasons This study uses a combination of satellite-based data from two in- struments (ECOSTRESS and OCO-3) aboard the ISS to study the large-scale diurnal patterns of photosynthesis and ET over the Amazon forest and their responses to environmental drivers. We reveal different large-scale diurnal patterns of photosynthesis and ET between wet and dry seasons in the Amazon forest (Fig. 3), which have not been observed at regional scale. The photosynthesis and ET across the Amazon forest did not decrease from morning to afternoon in the unstressed wet season but decreased in the dry season (Fig. 2, A and D), which could be attributed to the decreases in mesophyll (37) and stomatal conductance (15) in response to high VPD (fig. S2J) and decreased leaf water potential during the day. The stronger stresses on photosynthesis and ET in the dry season than in the wet season were also supported by site-based flux observations: Photosynthesis decreased more, and ET in- Fig. 5. Difference in SIF, ET, and WUESIF between morning (7:00 to 12:00) and creased less from morning to afternoon in the dry season than in afternoon (12:00 to 17:00) in Amazonian forests. (A and B) Spatial distributions the wet season (Fig. 2, B and E). of percentage changes in SIF from morning to afternoon in the wet and dry The different diurnal patterns of ET between wet and dry seasons seasons, respectively. The mean and SD are shown in the bottom right of each could be partly attributed to the strong interplay between SM and panel. (C to F) are similar to (A) and (B) but for ET and WUESIF, respectively. VPD on leaf water potential and therefore on stomatal behavior (15). Because the decrease in SM from morning to afternoon was very small, SM had a weak direct effect on ET at the diurnal scale. Projections of photosynthesis and ET under atmospheric However, the difference in SM between wet and dry season regulates warming and drying predawn leaf water potential (38). In the wet season (SM is high), the The RF models were used to predict future changes in Amazonian high predawn leaf water potential resulted in low stomatal sensitiv- morning and afternoon photosynthesis and ET from 2015 to 2100 ity to VPD. In the dry season with low SM, leaf water potential could using the CMIP6 model simulations as inputs for the shared socio- reach a critical threshold, inducing partial stomatal closure at high economic pathways (SSPs) SSP245, SSP370, and SSP585 (36). We VPDs and hence the decrease in ET. A recent study also pointed out did not consider the CO2 fertilization effect and changes in forest that plants could either positively or negatively regulate water use composition or structure. The morning SIF in the wet season in- depending on the water potential of the leaves (39). In other creased by 4, 6, and 8% from 2015 to 2100 under SSP245, SSP370, words, ET increased when water stress was limited and ET was and SSP585, respectively (Fig. 9A), but the increases in the after- demand limited (e.g., in the wet season) following an increase in noon SIF were weaker (Fig. 9B). Accordingly, the daily SIF in the evaporative demand. In contrast, ET decreased due to stomatal wet season increased (Fig. 9C), which was consistent with the closure in response to substantial water limitation, leading to a de- CMIP6 model simulations of GPP, although there was a difference crease in ET (e.g., in the dry season). in magnitude between our results and CMIP6 simulations (Fig. 9D). The changes from 2015 to 2100 in ET in the wet season were much Asymmetric responses of photosynthesis to VPD and Tair smaller than those in SIF or GPP (Fig. 9, E to H). In the dry season, These results raise a question of whether the response of photosyn- an increase in the morning SIF (Fig. 9I) was offset by the decrease in thesis to VPD would be different between morning and afternoon in the afternoon SIF (Fig. 9J). Consequently, only marginal changes in the humid topics; for example, photosynthesis could be positively the daily SIF in the dry season were projected (Fig. 9K). However, and negatively sensitive to atmospheric dryness in the morning CMIP6 models predicted increasing trends of primary productivity and afternoon, respectively (Fig. 6). The positive sensitivity of pho- under future climate warming in the Amazon (Fig. 9L), which could tosynthesis to VPD in wet environment could not be attributed to be dominantly driven by the CO2 fertilization effects. The morning the correlation between VPD and radiation in the morning because ET in the dry season increased by 3, 4, and 5% from 2015 to 2100 radiation remained unchanged using the sensitivity analysis (see under SSP245, SSP370, and SSP585, respectively, but the afternoon Materials and Methods). Rainfall interception by leaves and Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 6 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E Fig. 6. RF-based sensitivity analysis of SIF, ET, and WUESIF to VPD. (A) Box plots of the sensitivity of SIF to an increase in VPD by one SD for the wet season morning (WM), wet season afternoon (WA), dry season morning (DM), and dry season afternoon (DA, the period when foliar and soil water potentials are lowest and stress is highest). (B to E) The spatial distribution of the sensitivity of SIF to VPD. (F to J) and (K to O) are similar to (A) to (E) but display the sensitivities of ET and WUESIF to VPD, respectively. These sensitivities were derived as partial dependences, assuming that other drivers did not vary with VPD. The red line in each box plot indicates the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. The stippling on the maps represents regions where at least 95% of the RF models agree on the sign of the sensitivity. The numbers in the bottom right corner of each map panel are the area percentages of positive (+, red) and negative (−, blue) sensitivities. morning dew can limit photosynthesis by reducing light availability atmospheric warming and drying and their compound effects and slowing carbon diffusion (40). Drier air can facilitate the rate of with soil drying, offering a window onto future physiological evaporation from wet foliar surfaces, thus reducing morning fog changes in the Amazon. Although these results indicate that SIF and increasing light absorption earlier in the day (41) and hence from existing sun-synchronous satellites cannot fully reveal the photosynthesis and SIF. The positive sensitivity of morning SIF to physiological responses to varying environmental drivers at the VPD in this work is consistent with a recent study reporting that SIF diurnal scale, it is still feasible to use SIF based on sun-synchronous derived from Global Ozone Monitoring Experiment–2 (~9:30 over- satellites to study the seasonality of photosynthesis and phenology pass time) was positively sensitive to VPD in the Amazon rainforest (29, 30, 43) because the diurnal bias does not affect the overall sea- (33). However, we further found that the positive effect of VPD on sonality of photosynthesis. photosynthesis mostly disappeared in the afternoon, by then the dew would already have evaporated. In addition, we also found Different water use strategies under temperature and the asymmetric responses of SIF to Tair in the dry season but not atmospheric dryness stresses in the wet season. This could depend on whether Tair reached the On the basis of the sensitivity analysis, this study identified the dif- optimum temperature for photosynthesis (Topt, estimated at around ferent large-scale water-use strategies of Amazonian forests under 30°C for tropical forests) (42), indicating that Tair exceeded Topt in temperature and atmospheric dryness stresses: WUESIF increased the dry season afternoon and hence photosynthesis reduced. with VPD (being conservative; Fig. 6, K to O) and decreased with The asymmetric responses of photosynthesis to VPD and Tair Tair (Fig. 7, K to O). Therefore, ET decreased to reduce water losses between morning and afternoon in the humid tropics were not ob- to reduce the risk of hydraulic failure under high VPD (44), leading served at regional scale based on SIF from sun-synchronous satel- to increasing WUESIF. However, under high temperature and ample lites that passed over at a fixed time, thus highlighting the water supply, humid tropical forests reduced their WUESIF (less importance of OCO-3 SIF and ECOSTRESS ET observations for conservative) by increasing ET to lower foliar temperature (for studying the physiological responses of Amazonian forests to cooling) for maintaining high photosynthetic rates (45). The Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 7 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E Fig. 7. RF-based sensitivity analysis of SIF, ET, and WUESIF to . Tair (A) Box plots of the sensitivity of SIF to an increase in Tair by one SD for the wet season morning (WM), wet season afternoon (WA), dry season morning (DM) and dry season afternoon (DA, the period when foliar and soil water potentials are lowest and stress is highest). (B to E) Spatial distribution of the sensitivity of SIF to Tair. (F to J) and (K to O) are similar to (A) to (E) but display the sensitivities of ET and WUESIF to Tair, respectively. These sensitivities were derived as partial dependences, assuming that other drivers did not vary with Tair. The red line in each box plot indicates the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. The stippling on the maps represents regions where at least 95% of the RF models agree on the sign of the sensitivity. The numbers in the bottom right corner of each map panel are the area percentages of positive (+, red) and negative (−, blue) sensitivities. decreases in WUESIF from morning to afternoon indicate a domi- supply (48). However, this positive effect will not occur in the dry nant role of the cooling mechanism in regulating WUESIF for season afternoon due to increasing temperature (beyond Topt) and humid tropical forests. The decrease in WUESIF is opposite to atmospheric dryness, which will reduce the photosynthetic capacity what would be predicted by the widely used model of stomatal con- and stomatal conductance (3, 44). Therefore, we predict large-scale ductance based on optimality principles (46), i.e., minimizing water afternoon depressions of photosynthesis in the dry season. These losses while maximizing photosynthesis. Such an inconsistency diurnal variations do not contradict the dry season increase of pho- with optimal stomatal models was also observed from a warming tosynthesis and ET (12) because the seasonality of Amazon ecosys- experiment (47). Our results suggest that stomatal models based tems is largely controlled by phenology (i.e., leaf quality and on optimality principles may underestimate the role of ET as a demography) (14), although the global seasonal and interannual key cooling mechanism in the humid tropical forests in response variations in photosynthesis are also related to radiation (49) and to temperature stress. water availability (50). Our results highlight the importance of in- corporating more accurate representation of physiological respons- Neutral changes of Amazon photosynthesis and ET in the es driving carbon and water fluxes in “next-generation” Earth dry season system models. We projected that both diurnal (morning and afternoon) and sea- Diurnal CO2 data have not been available at the regional scale, sonal (wet and dry) variations of photosynthesis and ET in the and therefore, atmospheric CO2 concentrations were not included Amazon forest will change under future atmospheric warming as a driver in our RF model, and the potential impact of increasing and drying conditions (Fig. 9). Assuming a similar species compo- atmospheric CO2 concentration on our simulations was therefore sition and no major CO2 fertilization effect, photosynthesis will not considered. Because we mainly investigated the diurnal patterns likely increase in the unstressed wet season and in the mornings of photosynthesis and ET and their responses to environmental of the dry season in the humid tropics, supporting the positive drivers, ignoring CO2 effects has little effect on the diurnal patterns. effect of warming on photosynthesis under adequate soil water Therefore, the projections still present valuable information on the Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 8 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E their drivers over the entire Amazon forest. We identified reduc- tions in regional photosynthesis and ET from morning to afternoon in the dry season and quantified this effect at an unprecedented spatial scale through an observation-driven analysis. Our large- scale diurnal patterns of photosynthesis and ET provided a strong observational basis for modeling the exchanges of carbon and water fluxes between the land and atmosphere in the Amazon forest. This study would also improve our understanding of the direction and magnitude of ecosystem fluxes in response to climate change in other tropical forests. We envisage that these findings will pave the way for the use of high-frequency observations made by satel- lite-based sensors in studies of plant functioning and ecosystem processes to better project future changes in the terrestrial carbon and water cycles. MATERIALS AND METHODS OCO-3 SIF SIF is tightly linked to photosynthesis and is a good proxy of GPP at the ecosystem scale (25, 27, 28). The varying overpass time of the ISS allowed us to derive the diurnal pattern of SIF over the Amazon forest from OCO-3 SIF data (v10) (23, 51), which has been used in a recent perspective article (34). Figure S1B shows the footprints of OCO-3 from January to February 2020 as an example, which present the variations in observation time (local solar time) for each path. We used OCO-3 SIF located in the Amazon forest from August 2019 to December 2021. OCO-3 provides SIF data at both 757 and 771 nm, but we used only the 757-nm SIF data due to Fig. 8. Simulated percentage changes in SIF, ET, and WUESIF from the their stronger signal (52). We also retained only the observations at modelAM+ conditionAM simulation to the modelPM+ conditionAM simulation nadir (viewing zenith angle near zero) and removed low-quality due to the increase in and VPD and decline in leaf water potential. Tair The data, which were characterized as having a quality flag of >1. modelAM + conditionAM and modelPM + conditionAM simulations were obtained We investigated the diurnal patterns of SIF over the Amazon by driving the morning and afternoon RF models using the morning inputs (in- forest using observations from OCO-3 on the ISS. However, the cluding radiation, FPAR, VPD, Tair, and SM), respectively. (A and B) Percentage sun-target-viewing geometry effects on satellite SIF observations change in SIF from the modelAM + conditionAM to modelPM + conditionAM simu- lations in the wet and dry seasons, respectively. (C to F) are similar to (A) and (B) but are particularly relevant for OCO-3, given the wide range of solar for ET and WUESIF, respectively. The mean and SD are shown in the bottom right of zenith angles (SZA) observed during the diurnal cycle. Therefore, each panel. we corrected the sun-target-viewing geometry effects in SIF data ac- quired by the OCO-3 sensor on the ISS (SIFobs) by deriving the total canopy SIF emission (SIFtotal) based on a reflectance-based method, large-scale diurnal patterns of photosynthesis and ET. For example, which has been widely used in previous studies (52–54) the future increases in photosynthesis in the wet season are higher in the morning than in the afternoon. Even if considering the CO2 SIFtotal ¼ SIFobs =fesc � π ð1Þ fertilization effects, our projection still holds because of the strong and negative influences from the high afternoon Tair and VPD. As for the dry season, it is unclear whether the CO2 fertilization effect fesc ¼ NIRv =i0 � K ð2Þ could mitigate afternoon decreases of photosynthesis. This issue is beyond the scope of this study and deserves future study. CMIP6 ESMs include additional processes (such as CO2 fertilization NIRv ¼ ðNIR RÞ=ðNIR þ RÞ � NIR ð3Þ effects and atmospheric feedbacks), and they predict increasing GPP in both the wet and dry seasons. Given the notable atmospher- i0 ¼ 1 expð G � CI � LAI=cosSZAÞ ð4Þ ic drying in the future dry season, the simulated similar increase in the dry season photosynthesis (Fig. 9L) as the wet season photosyn- where fesc is the probability of the escape of fluorescence from the thesis (Fig. 9D) by the CMIP6 ESMs suggests that ESMs did not canopy in the direction of observation, NIRV is the near-infrared fully accurately represent the negative VPD effects on photosynthe- reflectance of vegetation, R and NIR are the reflectance at red and sis, and likely overestimate photosynthesis and potentially also the near-infrared bands, respectively, i0 is canopy interception, ω is terrestrial sink strength of the Amazon forest under future climatic foliar leaf albedo at the band of 757 nm, G is the value of a G func- conditions. tion, CI is clumping index, LAI is leaf area index, and cosSZA is the In summary, we used a combination of satellite observations to cosine of SZA. R and NIR were simulated with the RossThick-LiS- assess the large-scale diurnal patterns of photosynthesis and ET and parseR (RTLSR) Bidirectional Reflectance Distribution Function Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 9 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E Fig. 9. Projections of relative changes in SIF and GPP from 2015 to 2100 across the Amazon forest. Projected changes in SIF in the wet season morning (A), wet season afternoon (B), wet season daily average (C), and CMIP6 daily average GPP in the wet season (D). The gray areas are the overlapped regions of one SD across different CMIP6 models for three SSP scenarios. To reduce the interannual variation, these data were smoothed using a 5-year running average. (E to H) similar to (A) to (D) but for ET. (I to P) similar to (A) to (H) but for the dry season. model using the sun-target-viewing geometry of OCO-3 SIF. The the world (22). In addition to ET, ECOSTRESS level 3 products parameters provided by the MCD19A3 BRDF/albedo product also provide the fraction of transpiration (T) in ET (T/ET), but were used to drive the RTLSR model. For Amazonian forests, K the retrieval accuracy of T/ET has not been extensively assessed was set at 1.2, G was set at 0.5, and CI was set at 0.66. LAI data due to the lack of ground validation data. In addition, the T/ET pro- were obtained from Copernicus LAI products (see below). SIFtotal vided by ECOSTRESS (fig. S14) was lower than previously reported is referred to as SIF for simplicity elsewhere in the manuscript. values as it is less constrained by observations (56). Therefore, we did not use T. ECOSTRESS ET and T The ECOSTRESS, a thermal radiometer built by the Jet Propulsion Calculation of SIF-based WUE Laboratory of the National Aeronautics and Space Administration, The coupling of carbon and water was studied by developing a is also aboard the ISS. ECOSTRESS measures thermal infrared ra- metric, SIF-based water-use efficiency (WUESIF = SIFtotal/ET) diation in five bands, which are used to generate higher level prod- (34), which is a proxy for GPP-based WUE (WUEGPP = GPP/ ucts, such as surface temperature and ET. ECOSTRESS and OCO-3 ET). ECOSTRESS’ level 4 WUE was not used because it was calcu- follow similar acquisition schedules at different times of day, which lated from ET and eight-day MODIS GPP. WUEGPP is an indicator enabled us to explore the diurnal cycle of ET (fig. S1C). We used the of the adjustment of photosynthesis to the loss of water, and has ECOSTRESS ET product computed with the ET_PT–JPL algorithm been studied many times at the seasonal and yearly scales globally (55). We used ECOSTRESS ET located in the Amazon forest from (35). The magnitude of WUE often corresponds to the strategy of July 2018 to December 2021. Observations under cloudy conditions water use by plants, with higher WUEs indicating a more conserva- were excluded from the analysis based on the quality control flag. A tive strategy and lower WUEs indicating a less conservative strategy. preliminary validation study demonstrated that ECOSTRESS ET The derived patterns of the relationship between VPD and WUESIF performed well against site measurements at 82 EC sites around (Fig. 3, G and H) were consistent with those between VPD and Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 10 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E WUEGPP based on the surface EC flux measurements (figs. S5 to generate temporally and spatially continuous SIF and ET fields. S7), and uncorrected SIFobs-based WUE (WUESIFobs = SIFobs/ET) OCO-3 SIF and ECOSTRESS ET are released at different spatial res- had reverse relationships with VPD (fig. S15), demonstrating the olutions. To maintain consistency, we averaged instantaneous necessity of correcting the sun-target-viewing effects of SIFobs. OCO-3 SIF and ECOSTRESS ET into 0.1° × 0.1° grids and then used the gridded SIF and ET as response variables in the RF Copernicus LAI model. FPAR derived from Copernicus LAI and clear-sky DSSR, We used LAI data from the Copernicus Global Land Service LAI Tair, VPD, and SM from the ERA5 reanalyzed data were used as pre- (version GEOV2). The Copernicus LAI dataset was derived using dictor variables. Note that SM was 0- to 100-cm volumetric SM, a near real-time algorithm based on an artificial neural network. which was weighted average from three levels of ERA5 SM (0 to Copernicus LAI was released at 10-day intervals and was linearly 7, 7 to 28, and 28 to 100 cm) according to their respective depth. interpolated into a daily interval. Copernicus LAI was used to esti- Because of the hysteretic and asymmetric responses of photosyn- mate i0 in Eq. 4 above and then used to estimate diurnal FPAR (57) thesis and ET to environmental drivers, particularly between morning and afternoon (15), we trained the RF model individually FPAR ¼ ð1 0:05Þ ð1 0:06Þ � ð1 i0 Þ ð5Þ for the wet season morning, wet season afternoon, dry season morning, and dry season afternoon. Precipitation was used to ERA5 reanalysis define the wet and dry seasons at the grid cell level in each month We acquired air temperature (Tair), dew-point temperature (Tdew), (fig. S19): Months with precipitation rates higher than the annual downward surface solar radiation (DSSR), and clear-sky DSSR from mean were defined as the wet season, and months with precipitation the fifth ECMWF ReAnalysis (ERA5) hourly data, which cover the rates lower than the annual mean were defined as the dry season global surface and have a spatial resolution of 0.25°. Tair and Tdew (33). For the whole Amazon forest, the dry season was defined were used to calculate the VPD using the equation described in a from June to October when >50% grid cells were in the dry previous study (58). To match satellite SIF and ET observations, season. The 100 RF models were used to predict SIF and ET in 1- linear interpolation was used to interpolate these meteorological hour intervals each month from 2015 to 2021 at a spatial resolution data based on the latitude, longitude, and overpass time of the sat- of 1° × 1°. The RF-based SIF and ET values were averaged to repre- ellite observations. The DSSR and clear-sky DSSR, as proxies of sent the climatological values over 2015–2021 without the interan- PAR, were used to calculate the absorbed PAR (APAR). Because nual variations. The climatological SIF and ET values in the grid both the satellite SIF and ET data were filtered for cloudy condi- cells where the three ground flux sites (BR-Sa1, BR-Sa3, and GF- tions, they were better correlated with clear-sky APAR than with Guy) were located were compared with the tower GPP and ET all-sky APAR (fig. S16), and hence clear-sky DSSR was used. data averaged from multiple years. The RF-based SIF and ET were well correlated with the GPP and ET data from the towers, with the Eddy flux data correlation (r) of 0.86 and 0.92, respectively (fig. S20, A and B). Al- To compare with results from satellite observations, we used EC though the magnitude of RF-based ET was higher, such a bias had measurements from three flux sites (BR-Sa1, BR-Sa3, and GF- little effect on the diurnal trends or interannual trends expressed as Guy) located in Amazonian forests (red dots in fig. S1A and table percentages (%). The GPP-based WUE (WUEGPP = GPP/LE) was S1) and acquired the data from the FLUXNET2015 TIER ONE also moderately correlated with the SIF-based WUE (WUESIF = dataset (59). We picked up half-hourly or hourly GPP, LE, short- SIF/ET) (R 2 = 0.44; fig. S20C). These results supported the use of wave radiation (Rg), and VPD and removed low-quality data RF-based SIF and ET for the diurnal and regional analysis of using quality control flag. The clearness index, which is the ratio carbon and water fluxes. The interpolation and spatial and temporal of shortwave radiation to potential radiation at the top of the atmo- extrapolation abilities of the RF models were evaluated in text S1. sphere, was calculated to remove data collected under cloudy skies Only a small absolute relative bias (|rBias| < 0.2%) was observed (clearness index < 0.3). GPP-based WUE (WUEGPP) was calculated for interpolation ability of RF models (fig. S21). In addition, the as the ratio of GPP to LE. The local time was converted to local solar |rBias| was less than 0.8% for most cases in terms of spatial (fig. S22) time based on the time zone and longitude to enable comparisons and temporal extrapolation (fig. S23). We used the changes in SIF, with data acquired by the OCO-3 and ECOSTRESS instruments. All ET, and WUESIF from morning to afternoon as the diurnal varia- data for the available period were averaged into hourly data from tions and from wet to dry season as the seasonal variations (Fig. 2). 7:00 to 17:00 and per month to represent the climatology. Sensitivity analysis RF model The RF model was also used to calculate the sensitivities of SIF, ET, The available OCO-3 SIF and ECOSTRESS ET data were sparse in and WUESIF to perturbations in Tair, VPD, and SM, separately. We time and space and had substantial gaps (figs. S17 and S18) due to use the calculation of the sensitivity of ET to VPD as an example the high cloud cover over the study region and the ISS orbit. Dis- below. On the basis of the trained RF model described above, the crete sampling of SIF and ET might produce biased diurnal patterns FPAR and ERA-5 climatic data were used to produce ET and spatial distributions after simple aggregation if data are not (denoted as ETO). Then, we produced ET (denoted ETVPD) again properly processed. However, the thousands of satellite-based SIF by artificially adding one standard deviation to VPD, representing and ET data points we acquired were sufficient to train a machine a disturbance, while keeping the other variables unchanged (33). learning model, such as the RF model (60). RF models can represent Last, the sensitivity of ET to the disturbance in VPD was calculated the nonlinear interactions among various variables and are relative- using Eq. 6. The process was repeated using 100 RF models, and the ly insensitive to outliers. Accordingly, we used RF models to means were analyzed to increase the robustness of the results (Figs. 6 and 7 and fig. S11). We also calculated the sensitivity of SIF and ET Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 11 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E to Tair by adding one SD and to SM by decreasing one SD. results indicate that the limited biases related to the future extrapo- � � lations would not affect our results and conclusions using ETVPD ETO RF models. Sens:ðETÞ to VPD ¼ mean � 100% ð6Þ ETO Comparison of modelAM + conditionAM and modelPM + Supplementary Materials This PDF file includes: conditionAM simulations Tables S1 to S3 The foliar water potential decreases from morning to afternoon due Figs. S1 to S27 to the loss of water and then recovers during nighttime (17–19). To Text S1 evaluate the afternoon reductions in photosynthesis and ET in re- sponse to the drop in foliar water potential and increases in Tair and VPD, we performed the modelAM + conditionAM and modelPM + REFERENCES AND NOTES conditionAM simulations, in which the morning and afternoon 1. W. Yuan, Y. Zheng, S. Piao, P. Ciais, D. Lombardozzi, Y. Wang, S. Yang, Increased atmo- spheric vapor pressure deficit reduces global vegetation growth. Sci. Adv. 5, RF models were driven using the same morning inputs (including eaax1396 (2019). radiation, FPAR, Tair, VPD, and SM), respectively. By using the 2. N. McDowell, C. D. Allen, K. Anderson-Teixeira, P. Brando, R. Brienen, J. Chambers, same morning inputs for both the morning and afternoon RF B. Christoffersen, S. Davies, C. Doughty, A. Duque, F. Espirito-Santo, R. Fisher, C. G. Fontes, models, the effects of varying environmental drivers between D. Galbraith, D. Goodsman, C. Grossiord, H. Hartmann, J. Holm, D. J. Johnson, A. R. Kassim, morning and afternoon can be factored out. As an example, Eq. 7 M. Keller, C. Koven, L. Kueppers, T. Kumagai, Y. Malhi, S. M. McMahon, M. Mencuccini, P. Meir, P. Moorcroft, H. C. Muller-Landau, O. L. Phillips, T. Powell, C. A. Sierra, J. Sperry, was used to calculate the percentage change in ET from the J. Warren, C. Xu, X. Xu, Drivers and mechanisms of tree mortality in moist tropical forests. modelAM + conditionAM simulations to the modelPM + New Phytol. 219, 851–869 (2018). conditionAM simulations 3. B. E. Medlyn, E. Dreyer, D. Ellsworth, M. Forstreuter, P. C. Harley, M. U. F. Kirschbaum, D. Loustau, Temperature response of parameters of a biochemically based model of ETPM ETAM photosynthesis. II. A review of experimental data. Plant Cell Environ. 25, 1167–1179 (2002). Change in ET ð%Þ ¼ � 100% ð7Þ ETAM 4. A. Staal, O. A. Tuinenburg, J. H. C. Bosmans, M. Holmgren, E. H. van Nes, M. Scheffer, S. C. Dekker, Forest-rainfall cascades buffer against drought across the Amazon. Nat. Clim. where ETAM and ETPM are the predicted ET using the modelAM + Chang. 8, 539–543 (2018). conditionAM and modelPM + conditionAM simulations, respectively. 5. D. C. Nepstad, I. M. Tohver, D. Ray, P. Moutinho, G. Cardinot, Mortality of large trees and lianas following experimental drought in an Amazon forest. Ecology 88, 2259–2269 (2007). Projection of photosynthesis and ET 6. V. A. Maia, A. B. M. Santos, N. de Aguiar-Campos, C. R. de Souza, M. C. F. de Oliveira, The RF-based SIF and ET models were also used to predict diurnal P. A. Coelho, R. M. dos Santos, The carbon sink of tropical seasonal forests in southeastern Brazil can be under threat. Sci. Adv. 6, eabd4548 (2020). SIF and ET from 2015 to 2100 under different combined SSPs and 7. R. R. Nemani, C. D. Keeling, H. Hashimoto, W. M. Jolly, S. C. Piper, C. J. Tucker, S. W. Running, representative concentration pathways, including SSP245, SSP370, Climate-driven increases in global terrestrial net primary production from 1982 to 1999. and SSP585, from CMIP6 (36). The climatic data (Tair, SM, specific Science 300, 1560–1563 (2003). humidity, surface pressure, and clear-sky radiation) with a 3-hour 8. P. Gentine, J. K. Green, M. Guérin, V. Humphrey, S. I. Seneviratne, Y. Zhang, S. Zhou, Cou- interval from CMIP6 models were resampled to hourly values pling between the terrestrial carbon and water cycles—A review. Environ. Res. Lett. 14, 083003 (2019). (CMIP6 models can be found in table S2). Tair, specific humidity, 9. N. Restrepo-Coupe, N. M. Levine, B. O. Christoffersen, L. P. Albert, J. Wu, M. H. Costa, and surface pressure were used to calculate VPD. Changing species S. R. Saleska, Do dynamic global vegetation models capture the seasonality of carbon composition could have a large impact on SIF and ET but is beyond fluxes in the Amazon basin? A data-model intercomparison. Glob. Chang. Biol. 23, the scope of this study and thus is not considered here. FPAR was 191–208 (2017). kept constant at the values observed in 2015. Then, the hourly cli- 10. A. M. Matheny, G. Bohrer, P. C. Stoy, I. T. Baker, A. T. Black, A. R. Desai, H. Verbeeck, Char- matic data and FPAR were used in the trained RF model to simulate acterizing the diurnal patterns of errors in the prediction of evapotranspiration by several land-surface models: An NACP analysis. Eur. J. Vasc. Endovasc. Surg. 119, hourly SIF and ET. We analyzed relative changes in morning (7:00 1458–1473 (2014). to 12:00) and afternoon (12:00 to 17:00) SIF and ET from 2015 to 11. J. Wu, L. P. Albert, A. P. Lopes, N. Restrepo-Coupe, M. Hayek, K. T. Wiedemann, N. Prohaska, 2100 across the Amazon forest. The morning and afternoon SIF and Leaf development and demography explain photosynthetic seasonality in Amazon ever- ET were averaged as daily averages, which were also compared with green forests. Science 351, 972–976 (2016). daily averaged GPP and ET predicted by CMIP6 models (see table 12. N. Restrepo-Coupe, H. R. da Rocha, L. R. Hutyra, A. C. da Araujo, L. S. Borma, B. Christoffersen, S. R. Saleska, What drives the seasonality of photosynthesis across the S2) These comparisons will improve our understanding of the dif- Amazon basin? A cross-site analysis of eddy flux tower measurements from the Brasil flux ferences in photosynthesis and ET between satellite observation- network. Agric. For. Meteorol. 182-183, 128–144 (2013). based and model-based simulations. In addition, we used different 13. C. von Randow, A. O. Manzi, B. Kruijt, P. J. de Oliveira, F. B. Zanchi, R. L. Silva, P. Kabat, datasets of training (mimicking the current conditions) and valida- Comparative measurements and seasonal variations in energy and carbon exchange over tion (mimicking the extrapolation to future higher VPD conditions forest and pasture in South West Amazonia. Theor. Appl. Climatol. 78, 5–26 (2004). than those used for calibration) to assess the uncertainty of RF 14. J. Wu, K. Guan, M. Hayek, N. Restrepo-Coupe, K. T. Wiedemann, X. Xu, S. R. Saleska, Parti- tioning controls on Amazon forest photosynthesis between environmental and biotic models when making predictions of future scenarios (text S1 and factors at hourly to interannual timescales. Glob. Chang. Biol. 23, 1240–1257 (2017). fig. S24). Under the future conditions of SSP245 and SSP370, the 15. C. Lin, P. Gentine, C. Frankenberg, S. Zhou, D. Kennedy, X. Li, Evaluation and mechanism |rBias| was less than 2% in all cases (figs. S25 and S26). Under the exploration of the diurnal hysteresis of ecosystem fluxes. Agric. For. Meteorol. 278, future conditions of SSP585, the highest |rBias| (4.03%; fig. S27E) 107642 (2019). was obtained for predicting ET in the wet season morning, followed 16. M. L. Goulden, S. D. Miller, H. R. da Rocha, M. C. Menton, H. C. de Freitas, A. M. e Silva Figueira, C. A. D. de Sousa, Diel and seasonal patterns of tropical forest CO2 exchange. Ecol. by the dry season afternoon with the |rBias| of 3.17% (fig. S27H). In Appl. 14, 42–54 (2004). other cases, the |rBias| was less than 2% (fig. S27). Overall, these Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 12 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E 17. J.-E. Lee, R. S. Oliveira, T. E. Dawson, I. Fung, Root functioning modifies seasonal climate. 42. M. Slot, K. Winter, In situ temperature response of photosynthesis of 42 tree and liana Proc. Natl. Acad. Sci. U.S.A. 102, 17576–17581 (2005). species in the canopy of two Panamanian lowland tropical forests with contrasting rainfall 18. R. S. Oliveira, T. E. Dawson, S. S. O. Burgess, D. C. Nepstad, Hydraulic redistribution in three regimes. New Phytol. 214, 1103–1117 (2017). Amazonian trees. Oecologia 145, 354–363 (2005). 43. J. Joiner, Y. Yoshida, A. P. Vasilkov, K. Schaefer, M. Jung, L. Guanter, L. Belelli Marchesini, The 19. R. B. Neumann, Z. G. Cardon, The magnitude of hydraulic redistribution by plant roots: A seasonal cycle of satellite chlorophyll fluorescence observations and its relationship to review and synthesis of empirical and modeling studies. New Phytol. 194, 337–352 (2012). vegetation phenology and ecosystem atmosphere carbon exchange. Remote Sens. Environ. 20. A. R. Huete, K. Didan, Y. E. Shimabukuro, P. Ratana, S. R. Saleska, L. R. Hutyra, R. Myneni, 152, 375–391 (2014). Amazon rainforests green-up with sunlight in dry season. Geophys. Res. Lett. 33, 44. C. Grossiord, T. N. Buckley, L. A. Cernusak, K. A. Novick, B. Poulter, R. T. W. Siegwolf, L06405 (2006). N. G. McDowell, Plant responses to rising vapor pressure deficit. New Phytol. 226, 21. E. N. Stavros, D. Schimel, R. Pavlick, S. Serbin, A. Swann, L. Duncanson, P. Wennberg, ISS 1550–1566 (2020). observations offer insights into plant function. Nat. Ecol. Evol. 1, 0194 (2017). 45. G. B. Bonan, Forests and climate change: Forcings, feedbacks, and the climate benefits of 22. J. B. Fisher, B. Lee, A. J. Purdy, G. H. Halverson, M. B. Dohlen, K. Cawse-Nicholson, S. Hook, forests. Science 320, 1444–1449 (2008). ECOSTRESS: NASA’s next generation mission to measure evapotranspiration from the in- 46. B. E. Medlyn, R. A. Duursma, D. Eamus, D. S. Ellsworth, I. C. Prentice, C. V. M. Barton, ternational space station. Water Resources Res. 56, e2019WR026058 (2020). L. Wingate, Reconciling the optimal and empirical approaches to modelling stomatal 23. T. E. Taylor, A. Eldering, A. Merrelli, M. Kiel, P. Somkuti, C. Cheng, S. Yu, OCO-3 early mission conductance. Glob. Chang. Biol. 17, 2134–2144 (2011). operations and initial (vEarly) XCO2 and SIF retrievals. Remote Sens. Environ. 251, 47. J. E. Drake, M. G. Tjoelker, A. Vårhammar, B. E. Medlyn, P. B. Reich, A. Leigh, C. V. M. Barton, 112032 (2020). Trees tolerate an extreme heatwave via sustained transpirational cooling and increased leaf 24. M. Meroni, M. Rossini, L. Guanter, L. Alonso, U. Rascher, R. Colombo, J. Moreno, Remote thermal tolerance. Glob. Chang. Biol. 24, 2390–2402 (2018). sensing of solar-induced chlorophyll fluorescence: Review of methods and applications. 48. Q. Quan, D. Tian, Y. Luo, F. Zhang, T. W. Crowther, K. Zhu, S. Niu, Water scaling of ecosystem Remote Sens. Environ. 113, 2037–2051 (2009). carbon cycle feedback to climate warming. Sci. Adv. 5, eaav1131 (2019). 25. C. Frankenberg, J. B. Fisher, J. Worden, G. Badgley, S. S. Saatchi, J. E. Lee, A. Kuze, New 49. G. R. Quetin, A. L. S. Swann, Empirically derived sensitivity of vegetation to climate across global observations of the terrestrial carbon cycle from GOSAT: Patterns of plant fluores- global gradients of temperature and precipitation. J. Climate 30, 5835–5849 (2017). cence with gross primary productivity. Geophys. Res. Lett. 38, 351–365 (2011). 50. K. A. Novick, D. L. Ficklin, P. C. Stoy, C. A. Williams, G. Bohrer, A. C. Oishi, R. P. Phillips, The 26. J. Joiner, Y. Yoshida, A. P. Vasilkov, Y. Yoshida, L. A. Corp, E. M. Middleton, First observations increasing importance of atmospheric demand for ecosystem water and carbon fluxes. Nat. of global and seasonal terrestrial chlorophyll fluorescence from space. Biogeosciences 8, Clim. Chang. 6, 1023–1027 (2016). 637–651 (2011). 51. R. Doughty, T. P. Kurosu, N. Parazoo, P. Köhler, Y. Wang, Y. Sun, C. Frankenberg, Global 27. Y. Sun, C. Frankenberg, J. D. Wood, D. S. Schimel, M. Jung, L. Guanter, T. J. Griffis, L. Gu, GOSAT, OCO-2, and OCO-3 solar-induced chlorophyll fluorescence datasets. Earth Syst. Sci. T. S. Magney, P. Köhler, B. Evans, K. Yuen, OCO-2 advances photosynthesis observation Data 14, 1513–1529 (2022). from space via solar-induced chlorophyll fluorescence. Science 358, eaam5747 (2017). 52. Z. Y. Zhang, Y. G. Zhang, A. Porcar-Castell, J. Joiner, L. Guanter, X. Yang, Y. Goulas, Reduction 28. X. Li, J. Xiao, B. He, M. A. Arain, J. Beringer, A. R. Desai, A. Varlagin, Solar-induced chlorophyll of structural impacts and distinction of photosynthetic pathways in a global estimation of fluorescence is strongly correlated with terrestrial photosynthesis for a wide variety of GPP from space-borne solar-induced chlorophyll fluorescence. Remote Sens. Environ. 240, biomes: First global analysis based on OCO-2 and flux tower observations. Glob. Chang. 111722 (2020). Biol. 24, 3990–4008 (2018). 53. Z. Zhang, Y. Zhang, Solar angle matters: Diurnal pattern of solar-induced chlorophyll 29. R. Doughty, P. Köhler, C. Frankenberg, T. S. Magney, X. Xiao, Y. Qin, B. Moore III, TROPOMI fluorescence from OCO-3 and TROPOMI. Remote Sens. Environ. 285, 113380 (2023). reveals dry-season increase of solar-induced chlorophyll fluorescence in the Amazon 54. X. Liu, L. Liu, J. Hu, J. Guo, S. Du, Improving the potential of red SIF for estimating GPP by forest. Proc. Natl. Acad. Sci. U.S.A. 116, 22393–22398 (2019). downscaling from the canopy level to the photosystem level. Agric. For. Meteorol. 281, 30. N. C. Parazoo, K. Bowman, C. Frankenberg, J. E. Lee, J. B. Fisher, J. Worden, I. T. Baker, In- 107846 (2020). terpreting seasonal changes in the carbon balance of southern Amazonia using mea- 55. J. B. Fisher, K. P. Tu, D. D. Baldocchi, Global estimates of the land–atmosphere water flux surements of XCO2 and chlorophyll fluorescence from GOSAT. Geophys. Res. Lett. 40, based on monthly AVHRR and ISLSCP-II data, validated at 16 FLUXNET sites. Remote Sens. 2829–2833 (2013). Environ. 112, 901–919 (2008). 31. J. E. Lee, C. Frankenberg, C. van der Tol, J. A. Berry, L. Guanter, C. K. Boyce, S. Saatchi, Forest 56. S. P. Good, D. Noone, G. Bowen, Hydrologic connectivity constrains partitioning of global productivity and water stress in Amazonia: Observations from GOSAT chlorophyll fluor- terrestrial water fluxes. Science 349, 175–177 (2015). escence. Proc. R. Soc. Lond. B Biol. Sci. 280, 20130171 (2013). 57. J. M. Chen, J. Liu, S. G. Leblanc, R. Lacaze, J.-L. Roujean, Multi-angular optical remote 32. A. O. Castro, J. Chen, C. S. Zang, A. Shekhar, J. C. Jimenez, S. Bhattacharjee, A. Rammig, sensing for assessing vegetation structure and carbon absorption. Remote Sens. Environ. OCO-2 solar-induced chlorophyll fluorescence variability across ecoregions of the Amazon 84, 516–525 (2003). basin and the extreme drought effects of El Niño (2015–2016). Remote Sens 12, 58. A. Barkhordarian, S. S. Saatchi, A. Behrangi, P. C. Loikith, C. R. Mechoso, A recent systematic 1202 (2020). increase in vapor pressure deficit over tropical south America. Sci. Rep. 9, 15331 (2019). 33. J. K. Green, J. Berry, P. Ciais, Y. Zhang, P. Gentine, Amazon rainforest photosynthesis in- 59. G. Pastorello, C. Trotta, E. Canfora, H. Chu, D. Christianson, Y.-W. Cheah, C. Poindexter, creases in response to atmospheric dryness. Sci. Adv. 6, eabb7232 (2020). J. Chen, A. Elbashandy, M. Humphrey, P. Isaac, D. Polidori, M. Reichstein, A. Ribeca, C. van 34. J. Xiao, J. B. Fisher, H. Hashimoto, K. Ichii, N. C. Parazoo, Emerging satellite observations for Ingen, N. Vuichard, L. Zhang, B. Amiro, C. Ammann, M. A. Arain, J. Ardö, T. Arkebauer, diurnal cycling of ecosystem processes. Nat. Plants 7, 877–887 (2021). S. K. Arndt, N. Arriga, M. Aubinet, M. Aurela, D. Baldocchi, A. Barr, E. Beamesderfer, 35. M. Huang, S. Piao, Z. Zeng, S. Peng, P. Ciais, L. Cheng, Y. Wang, Seasonal responses of L. B. Marchesini, O. Bergeron, J. Beringer, C. Bernhofer, D. Berveiller, D. Billesbach, T. A. Black, terrestrial ecosystem water-use efficiency to climate change. Glob. Chang. Biol. 22, P. D. Blanken, G. Bohrer, J. Boike, P. V. Bolstad, D. Bonal, J.-M. Bonnefond, D. R. Bowling, 2165–2177 (2016). R. Bracho, J. Brodeur, C. Brümmer, N. Buchmann, B. Burban, S. P. Burns, P. Buysse, P. Cale, 36. B. C. O’Neill, C. Tebaldi, D. P. van Vuuren, V. Eyring, P. Friedlingstein, G. Hurtt, M. Cavagna, P. Cellier, S. Chen, I. Chini, T. R. Christensen, J. Cleverly, A. Collalti, C. Consalvo, B. M. Sanderson, The scenario model intercomparison project (ScenarioMIP) for CMIP6. B. D. Cook, D. Cook, C. Coursolle, E. Cremonese, P. S. Curtis, E. D’Andrea, H. da Rocha, X. Dai, Geosci. Model Dev. 9, 3461–3482 (2016). K. J. Davis, B. De Cinti, A. de Grandcourt, A. De Ligne, R. C. De Oliveira, N. Delpierre, 37. C. P. Bickford, D. T. Hanson, N. G. McDowell, Influence of diurnal variation in mesophyll A. R. Desai, C. M. Di Bella, P. di Tommasi, H. Dolman, F. Domingo, G. Dong, S. Dore, P. Duce, conductance on modelled 13C discrimination: Results from a field study. J. Exp. Bot. 61, E. Dufrêne, A. Dunn, J. Dušek, D. Eamus, U. Eichelmann, H. A. M. El Khidir, W. Eugster, 3223–3233 (2010). C. M. Ewenz, B. Ewers, D. Famulari, S. Fares, I. Feigenwinter, A. Feitz, R. Fensholt, G. Filippa, M. Fischer, J. Frank, M. Galvagno, M. Gharun, D. Gianelle, B. Gielen, B. Gioli, A. Gitelson, 38. L. D. Prior, D. Eamus, G. A. Duff, Seasonal trends in carbon assimilation, stomatal con- I. Goded, M. Goeckede, A. H. Goldstein, C. M. Gough, M. L. Goulden, A. Graf, A. Griebel, ductance, pre-dawn leaf water potential and growth in Terminalia ferdinandiana, a de- C. Gruening, T. Grünwald, A. Hammerle, S. Han, X. Han, B. U. Hansen, C. Hanson, J. Hatakka, ciduous tree of northern Australian savannas. Aust. J. Bot. 45, 53–69 (1997). Y. He, M. Hehn, B. Heinesch, N. Hinko-Najera, L. Hörtnagl, L. Hutley, A. Ibrom, H. Ikawa, 39. A. Massmann, P. Gentine, C. Lin, When does vapor pressure deficit drive or reduce M. Jackowicz-Korczynski, D. Janouš, W. Jans, R. Jassal, S. Jiang, T. Kato, M. Khomik, J. Klatt, evapotranspiration? J. Adv. Model. Earth Syst 11, 3305–3320 (2019). A. Knohl, S. Knox, H. Kobayashi, G. Koerber, O. Kolle, Y. Kosugi, A. Kotani, A. Kowalski, 40. M. Ishibashi, I. Terashima, Effects of continuous leaf wetness on photosynthesis: Adverse B. Kruijt, J. Kurbatova, W. L. Kutsch, H. Kwon, S. Launiainen, T. Laurila, B. Law, R. Leuning, aspects of rainfall. Plant Cell Environ. 18, 431–438 (1995). Y. Li, M. Liddell, J.-M. Limousin, M. Lion, A. J. Liska, A. Lohila, A. López-Ballesteros, E. López- 41. U. Anber, P. Gentine, S. Wang, A. H. Sobel, Fog and rain in the Amazon. Proc. Natl. Acad. Sci. Blanco, B. Loubet, D. Loustau, A. Lucas-Moffat, J. Lüers, S. Ma, C. Macfarlane, V. Magliulo, U.S.A. 112, 11473–11477 (2015). R. Maier, I. Mammarella, G. Manca, B. Marcolla, H. A. Margolis, S. Marras, W. Massman, Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 13 of 14 S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E M. Mastepanov, R. Matamala, J. H. Matthes, F. Mazzenga, H. M. Caughey, I. M. Hugh, Acknowledgments A. M. S. McMillan, L. Merbold, W. Meyer, T. Meyers, S. D. Miller, S. Minerbi, U. Moderow, Funding: This research was supported by the National Natural Science Foundation of China R. K. Monson, L. Montagnani, C. E. Moore, E. Moors, V. Moreaux, C. Moureaux, J. W. Munger, (42125105, 42071388, and 42101320). J.W. was in part funded by National Natural Science T. Nakai, J. Neirynck, Z. Nesic, G. Nicolini, A. Noormets, M. Northwood, M. Nosetto, Foundation of China (31922090) and in part supported by the Innovation and Technology Fund Y. Nouvellon, K. Novick, W. Oechel, J. E. Olesen, J.-M. Ourcival, S. A. Papuga, F.-J. Parmentier, (funding support to State Key Laboratories in Hong Kong of Agrobiotechnology) of the Hong E. Paul-Limoges, M. Pavelka, M. Peichl, E. Pendall, R. P. Phillips, K. Pilegaard, N. Pirk, G. Posse, Kong Special Administrative Region, China. P.G. acknowledges funding from USMILE European T. Powell, H. Prasse, S. M. Prober, S. Rambal, Ü. Rannik, N. Raz-Yaseef, C. Rebmann, D. Reed, Research Council grant. S. R. Saleska and N. Restrepo-Coupe were appreciated for providing flux V. R. de Dios, N. Restrepo-Coupe, B. R. Reverter, M. Roland, S. Sabbatini, T. Sachs, data. Author contributions: Z.Z. and Y.Z. conceived the study. Z.Z., Y.Z., A.C., Y.-P.W., P.G., and S. R. Saleska, E. P. Sánchez-Cañete, Z. M. Sanchez-Mejia, H. P. Schmid, M. Schmidt, J.X. contributed to interpreting the results. All authors edited the manuscript. Competing K. Schneider, F. Schrader, I. Schroder, R. L. Scott, P. Sedlák, P. Serrano-Ortíz, C. Shao, P. Shi, interests: The authors declare that they have no competing interests. Data and materials I. Shironya, L. Siebicke, L. Šigut, R. Silberstein, C. Sirca, D. Spano, R. Steinbrecher, availability: All data needed to evaluate the conclusions in the paper are present in the paper R. M. Stevens, C. Sturtevant, A. Suyker, T. Tagesson, S. Takanashi, Y. Tang, N. Tapper, J. Thom, and/or the Supplementary Materials. Details about CMIP6 models used in this study are M. Tomassucci, J.-P. Tuovinen, S. Urbanski, R. Valentini, M. van der Molen, E. van Gorsel, provided in table S2. The data used in this study have been properly cited and are publicly K. van Huissteden, A. Varlagin, J. Verfaillie, T. Vesala, C. Vincke, D. Vitale, N. Vygodskaya, available (see table S3). J. P. Walker, E. Walter-Shea, H. Wang, R. Weber, S. Westermann, C. Wille, S. Wofsy, G. Wohlfahrt, S. Wolf, W. Woodgate, Y. Li, R. Zampedri, J. Zhang, G. Zhou, D. Zona, Submitted 15 April 2022 D. Agarwal, S. Biraud, M. Torn, D. Papale, The FLUXNET2015 dataset and the ONEFlux Accepted 25 April 2023 processing pipeline for eddy covariance data. Sci. Data 7, 225 (2020). Published 26 May 2023 60. L. Breiman, Random Forests. Mach. Learn. 45, 5–32 (2001). 10.1126/sciadv.abq4974 Zhang et al., Sci. Adv. 9, eabq4974 (2023) 26 May 2023 14 of 14