Detecting vulnerability of humid tropical forests to multiple stressors

Abstract

•An index to track vulnerability of global rainforests to climate and land use•Four decades of satellite data show widespread vulnerability across the tropics•Response of rainforests to heat and drying varies across the continents•Early warning from the index can identify regions for conservation and restoration Rainforests are being lost at an alarming rate due to deforestation and degradation. As these forests lose their intactness and diversity, their resilience to climate change declines and they become more vulnerable to droughts and wildfires. Here, we built a spatially explicit tropical forest vulnerability index (TFVI) based on observations of forest cover, carbon, and water fluxes to identify areas where rainforests are losing resilience to disturbance and are changing toward an irreversible state, a “tipping point.” Our findings show how and where tipping points may occur, either as a gradual downhill decline of ecosystem services or an abrupt change. We present TFVI as an index to monitor tropical forests and provide early-warning signals for regions that are in need of policies that simultaneously promote conservation and restoration to increase resilience and climate mitigation. Humid tropical forests play a dominant role in the functioning of Earth but are under increasing threat from changes in land use and climate. How forest vulnerability varies across space and time and what level of stress forests can tolerate before facing a tipping point are poorly understood. Here, we develop a tropical forest vulnerability index (TFVI) to detect and evaluate the vulnerability of global tropical forests to threats across space and time. We show that climate change together with land-use change have slowed the recovery rate of forest carbon cycling. Temporal autocorrelation, as an indicator of this slow recovery, increases substantially for above-ground biomass, gross primary production, and evapotranspiration when climate stress reaches a critical level. Forests in the Americas exhibit extensive vulnerability to these stressors, while in Africa, forests show relative resilience to climate, and in Asia reveal more vulnerability to land use and fragmentation. TFVI can systematically track the response of tropical forests to multiple stressors and provide early-warning signals for regions undergoing critical transitions. Humid tropical forests play a dominant role in the functioning of Earth but are under increasing threat from changes in land use and climate. How forest vulnerability varies across space and time and what level of stress forests can tolerate before facing a tipping point are poorly understood. Here, we develop a tropical forest vulnerability index (TFVI) to detect and evaluate the vulnerability of global tropical forests to threats across space and time. We show that climate change together with land-use change have slowed the recovery rate of forest carbon cycling. Temporal autocorrelation, as an indicator of this slow recovery, increases substantially for above-ground biomass, gross primary production, and evapotranspiration when climate stress reaches a critical level. Forests in the Americas exhibit extensive vulnerability to these stressors, while in Africa, forests show relative resilience to climate, and in Asia reveal more vulnerability to land use and fragmentation. TFVI can systematically track the response of tropical forests to multiple stressors and provide early-warning signals for regions undergoing critical transitions. Humid tropical forests (HTFs) (Figure S1) are hyper-diverse and play a dominant role in the functioning of Earth and regulating its climate by accounting for more than half of its life forms, one-third of its metabolic activity, and storing more than half of its vegetation carbon.1Malhi Y. Gardner T.A. Goldsmith G.R. Silman M.R. Zelazowski P. Tropical forests in the Anthropocene.Annu. Rev. Environ. Resour. 2014; 39: 125-159Crossref Scopus (260) Google Scholar,2Trumbore S. Brando P. Hartmann H. Forest health and global change.Science (80-. 2015; 349: 814-818Crossref PubMed Scopus (493) Google Scholar These forests benefit from relatively warm temperature and high rainfall in equatorial regions that support the storage and processing of larger amounts of carbon via plant productivity and ecosystem respiration than any other biome.3Lewis S.L. Edwards D.P. Galbraith D. Increasing human dominance of tropical forests.Science. 2015; 349: 827-832Crossref PubMed Scopus (403) Google Scholar During the dry season, tropical trees are the primary source of water released to the atmosphere through evapotranspiration (ET), creating rainfall locally and thousand miles away to maintain their moist environment.4Wright J.S. Fu R. Worden J.R. Chakraborty S. Clinton N.E. Risi C. Sun Y. Yin L. Rainforest-initiated wet season onset over the southern Amazon.Proc. Natl. Acad. Sci. U. S. A. 2017; 114: 8481-8486Crossref PubMed Scopus (121) Google Scholar These processes have been maintained regularly for millennia with intact forest structure and diversity and stable ecological functions.2Trumbore S. Brando P. Hartmann H. Forest health and global change.Science (80-. 2015; 349: 814-818Crossref PubMed Scopus (493) Google Scholar However, HTFs are under increasing threats from human-induced and environmental stressors.5Zhou L. Tian Y. Myneni R.B. Ciais P. Saatchi S. Liu Y.Y. Piao S. Chen H. Vermote E.F. Song C. et al.Widespread decline of Congo rainforest greenness in the past decade.Nature. 2014; 508: 86-90Crossref Scopus (260) Google Scholar, 6Song X.P. Hansen M.C. Stehman S.V. Potapov P.V. Tyukavina A. Vermote E.F. Townshend J.R. Global land change from 1982 to 2016.Nature. 2018; 560: 639-643Crossref PubMed Scopus (723) Google Scholar, 7Saatchi S. Asefi-Najafabady S. Malhi Y. Aragão L.E.O.C. Anderson L.O. Myneni R.B. Nemani R. Persistent effects of a severe drought on Amazonian forest canopy.Proc. Natl. Acad. Sci. U. S. A. 2013; 110: 565-570Crossref PubMed Scopus (266) Google Scholar Between 15% and 20% of HTFs have been cleared since the early 1990s and at least an additional 10% have been degraded.8Vancutsem C. Achard F. Pekel J.-F. Vieilledent G. Carboni S. Simonetti D. Gallego J. Aragão L.E.O.C. Nasi R. Long-term (1990–2019) monitoring of forest cover changes in the humid tropics.Sci. Adv. 2021; 7: eabe1603Crossref PubMed Scopus (42) Google Scholar Widespread decline of carbon storage and productivity has been reported regionally due to elevated temperature and droughts.9Yang Y. Saatchi S.S. Xu L. Yu Y. Choi S. Phillips N. Kennedy R. Keller M. Knyazikhin Y. Myneni R.B. Post-drought decline of the Amazon carbon sink.Nat. Comm. 2018; 9: 3172Crossref PubMed Scopus (78) Google Scholar, 10Aleixo I. Norris D. Hemerik L. Barbosa A. Prata E. Costa F. Poorter L. Amazonian rainforest tree mortality driven by climate and functional traits.Nat. Clim. Chang. 2019; 9: 384-388Crossref Scopus (90) Google Scholar, 11Harris N.L. Gibbs D.A. Baccini A. Birdsey R.A. de Bruin S. Farina M. Fatoyinbo L. Hansen M.C. Herold M. Houghton R.A. et al.Global maps of twenty-first century forest carbon fluxes.Nat. Clim. Chang. 2021; 11: 234-240Crossref Scopus (121) Google Scholar Water and energy exchanges are suppressed significantly from forest degradation or drying atmosphere.12Longo M. Saatchi S. Keller M. Bowman K. Ferraz A. Moorcroft P.R. Morton D.C. Bonal D. Brando P. Burban B. et al.Impacts of degradation on water, energy, and carbon cycling of the Amazon tropical forests.J. Geophys. Res. Biogeosciences. 2020; 125https://doi.org/10.1029/2020JG005677Crossref PubMed Scopus (16) Google Scholar Remaining forests are changing into increasingly smaller fragments with substantial biodiversity loss.13Palma A. De Hoskins A. Gonzalez R.E. Börger L. Newbold T. Sanchez-Ortiz K. Ferrier S. Purvis A. Changes in the Biodiversity Intactness Index in tropical and subtropical forest biomes, 2001–2012.bioRxiv. 2019; : 311688Google Scholar With the projected rate of climate change and escalating land-use activities, these forests will lose their current capacity for global carbon sink and may even become a source of carbon to the atmosphere via changing photosynthesis and respiration rates, losing biodiversity, increasing tree mortality from droughts, and widespread forest collapse via fire.14Lewis S. Tropical forests and the changing earth system.Phil Trans. R. Soc. B. 2006; 361: 195-210Crossref PubMed Scopus (234) Google Scholar Over recent decades, HTFs have grown increasingly vulnerable to these pressures with high probability of undergoing regime shifts.15Nobre C.A. Sampaio G. Borma L.S. Castilla-Rubio J.C. Silva J.S. Cardoso M. Land-use and climate change risks in the amazon and the need of a novel sustainable development paradigm.Proc. Natl. Acad. Sci. U. S. A. 2016; 113: 10759-10768Crossref PubMed Scopus (337) Google Scholar Climate models predict either a tipping point of tropical forests in the form of large-scale tree mortality16Malhi Y. Aragão L.E.O.C. Galbraith D. Huntingford C. Fisher R. Zelazowski P. Sitch S. McSweeney C. Meir P. Exploring the likelihood and mechanism of a climate-change-induced dieback of the Amazon rainforest.Proc. Natl. Acad. Sci. U. S. A. 2009; 106: 20610-20615Crossref PubMed Scopus (585) Google Scholar or a more gradual continuous transition to a drier and fire-dominated savanna-like ecosystem,17Levine N.M. Zhang K. Longo M. Baccini A. Phillips O.L. Lewis S.L. Alvarez-Dávila E. De Andrade A.C.S. Brienen R.J.W. Erwin T.L. et al.Ecosystem heterogeneity determines the ecological resilience of the Amazon to climate change.Proc. Natl. Acad. Sci. U. S. A. 2016; 113: 793-797Crossref PubMed Scopus (120) Google Scholar both with large uncertainties in future trajectories.18Huntingford C. Zelazowski P. Galbraith D. Mercado L.M. Sitch S. Fisher R. Lomas M. Walker A.P. Jones C.D. Booth B.B.B. et al.Simulated resilience of tropical rainforests to CO2-induced climate change.Nat. Geosci. 2013; 6: 268-273Crossref Scopus (277) Google Scholar There is ample evidence from local studies and ecosystem modeling that the HTF vulnerability is eminent,18Huntingford C. Zelazowski P. Galbraith D. Mercado L.M. Sitch S. Fisher R. Lomas M. Walker A.P. Jones C.D. Booth B.B.B. et al.Simulated resilience of tropical rainforests to CO2-induced climate change.Nat. Geosci. 2013; 6: 268-273Crossref Scopus (277) Google Scholar, 19Cox P.M. Pearson D. Booth B.B. Friedlingstein P. Huntingford C. Jones C.D. Luke C.M. Sensitivity of tropical carbon to climate change constrained by carbon dioxide variability.Nature. 2013; 494: 341-344Crossref PubMed Scopus (451) Google Scholar, 20Phillips O.L. Aragão L.E.O.C. Lewis S.L. Fisher J.B. Lloyd J. López-González G. Malhi Y. Monteagudo A. Peacock J. Quesada C.A. et al.Drought sensitivity of the amazon rainforest.Science. 2009; 323: 1344-1347Crossref PubMed Scopus (1178) Google Scholar but how forest vulnerability varies across space and time and what level of stress forests can tolerate before facing a tipping point are poorly understood.21Cole L.E.S. Bhagwat S.A. Willis K.J. Recovery and resilience of tropical forests after disturbance.Nat. Commun. 2014; 5: 1-7Crossref Scopus (149) Google Scholar This raises the question of whether one can detect the vulnerability of tropical forests to human- and climate-induced stressors spatially and identify areas of low resilience that may drive the ecosystem to an alternative state. Here, we address this challenge in two ways. First, we used time series of climate data, satellite records, and models to track stressors and ecosystem responses over four decades (1982–2018). We define a stressor as a condition, event, or a trend related to climate variability and change or forest disturbance that can exacerbate hazards on ecosystems. This provides the first pan-tropical test (Figure S1) of the degree to which the putative predictors of HTF changes (e.g., vapor pressure deficit, water balance, and forest cover) vary spatially and remain consistent over time in magnitude and trends. All spatial data were gridded at 0.1°×0.1° (120 km2) resolution (experimental procedures). For climate stress, the predictors include trends in temperature (T), vapor pressure deficit (VPD), and water balance (WB) and for human-induced stress, we used forest tree cover (TC) changes from deforestation, degradation, and fire. For ecosystem responses, we focused on the carbon cycle and included the above-ground live biomass (AGB) (2000–2018),22Xu L. Saatchi S. Yang Y. et al.Changes in Global Terrestrial Live Biomass Over the 21st Century.Sci. Adv. 2021; 7eabe9829https://doi.org/10.1126/sciadv.abe9829Crossref PubMed Scopus (24) Google Scholar gross primary production (GPP) (1982–2018), ET (1982–2018), and the vegetation skin temperature (land surface temperature [LST]) (2000–2018) as variables linking the water and energy balance processes with the forest carbon cycling.23Jin M. Dickinson R.E. Land surface skin temperature climatology: benefitting from the strengths of satellite observations.Environ. Res. Lett. 2010; 5: 044004Crossref Scopus (225) Google Scholar Second, we used AR1 autoregressive models to quantify the temporal autocorrelation and the sensitivity of the ecosystem carbon cycle responses to multiple stressors. We included climate variables (WB, VPD) and TC changes in AR1 model with 1-month-lagged GPP, ET, and LST or 1-year-lagged AGB response variables to identify regions exhibiting amplified responses to land-use and climate variability and trends (experimental procedures). We define vulnerability as the degree to which a system is susceptible to, or unable to cope with, adverse effects of disturbance, such as the stress from climate change, including climate variability and extremes.24Turner B.L. Kasperson R.E. Matsone P.A. McCarthy J.J. Corell R.W. Christensene L. Eckley N. Kasperson J.X. Luers A. Martello M.L. et al.A framework for vulnerability analysis in sustainability science.Proc. Natl. Acad. Sci. U. S. A. 2003; 100: 8074-8079Crossref PubMed Scopus (2598) Google Scholar Vulnerability can be considered as an estimate of the inability of the ecosystem to tolerate stressors.25Folke C. Carpenter S. Walker B. Scheffer M. Elmqvist T. Gunderson L. Holling C.S. Regime shifts, resilience, and biodiversity in ecosystem management.Annu. Rev. Ecol. Evol. Syst. 2004; 35: 557-581Crossref Scopus (2388) Google Scholar This occurs when the ecosystem starts modulating its responses to stressors over time and space and starts losing its resilience or the ability to recover from disturbance.26Williams L.R.R. Kapustka L.A. Ecosystem vulnerability: a complex interface with technical components.Environ. Toxicol. Chem. 2000; 19: 1055-1058Google Scholar To quantify resilience, we evaluated the temporal autocorrelation between each response and stress variables on the monthly time scales after removing the seasonal cycles and trends. Higher temporal autocorrelation points to the slow recovery of the ecosystem if it is exposed to severe stress, which can be considered an important early-warning indicator of critical transitions.27Dakos V. Van Nes E.H. D’Odorico P. Scheffer M. Robustness of variance and autocorrelation as indicators of critical slowing down.Ecology. 2012; 93: 264-271Crossref PubMed Scopus (184) Google Scholar,28Scheffer M. Bascompte J. Brock W.A. Brovkin V. Carpenter S.R. Dakos V. Held H. Van Nes E.H. Rietkerk M. Sugihara G. Early-warning signals for critical transitions.Nature. 2009; 461: 53-59Crossref PubMed Scopus (2439) Google Scholar We then used additive regression AR1 models to simultaneously assess the relationship of temporal autocorrelation to multiple stressors.29Verbesselt J. Umlauf N. Hirota M. Holmgren M. Van Nes E.H. Herold M. Zeileis A. Scheffer M. Remotely sensed resilience of tropical forests.Nat. Clim. Chang. 2016; 6: 1028-1031Crossref Scopus (97) Google Scholar,30Seddon N. Chausson A. Berry P. Girardin C.A.J. Smith A. Turner B. Understanding the value and limits of nature-based solutions to climate change and other global challenges.Philos. Trans. R. Soc. B Biol. Sci. 2020; 375https://doi.org/10.1098/rstb.2019.0120Crossref Scopus (246) Google Scholar The coefficients of the 1-month-lagged response variables can be used to identify regions sensitive to stress and/or with the memory effects and to develop a spatially explicit tropical forest vulnerability index (TFVI) that shows changes of the response variables to long-term trends of multiple stressors. The magnitude and spatial variations of TFVI can identify areas across the tropics that exhibit high vulnerability and risks of a critical transition. We quantify land-cover and land-use change (LCLUC) stress across the tropics using the spatial variations of TC change and burned area from fire from 1982 to 2018 (Figures 1A and 1B ). The percent of TC change at the grid cells includes loss of forest from deforestation and degradation and gain from secondary forest recovery, and afforestation relative to the benchmark forest cover of 1982 (experimental procedures). Changes of TC show that, in the 1980s, more than 340 Mha (106 hectares) of tropical forests experienced a net loss (gain-loss) of tree cover, which became more widespread in the 1990s (∼390 Mha) and 2000s (∼420 Mha), but dropped significantly in 2010s to about 300 Mha (Figure S2). TC loss exhibited uneven patterns globally and across time, with large-scale deforestation and agricultural expansion in the Americas, to small-scale shifting cultivation in Central Africa, and a combination of agroforestry and commodity-driven agriculture in Asia.31Curtis P.G. Slay C.M. Harris N.L. Tyukavina A. Hansen M.C. Classifying drivers of global forest loss.Science. 2018; 361: 1108-1111Crossref PubMed Scopus (700) Google Scholar The net loss of TC was consistently higher in the Americas, with an average rate of 2.5 Mha year−1 in the 1980s, 1.2 Mha year−1 in the 1990s, 2.0 Mha year−1 in the 2000s, and 1.6 Mha year−1 in the 2010s. In Africa, the highest rate of TC loss was about 0.6 Mha year−1 from 1982 to 1999 but it dropped significantly to about 0.15 Mha year−1 from 2000 to 2018. More recently (2000–2018), the gross TC loss was about 7.3 Mha year−1, which is comparable with 7.65 Mha year−1 of average rate of forest clearing extracted from the Landsat-based (30-m resolution) forest cover change products (Figure S3).32Hansen M.C. Potapov P.V. Moore R. Hancher M. Turubanova S.A. Tyukavina A. Thau D. Stehman S.V. Goetz S.J. Loveland T.R. et al.High-resolution global maps of 21st-century forest cover change.Science (80-. ). 2013; 342: 850-853Crossref PubMed Scopus (5806) Google Scholar Wildfire impacts on forest cover remained confined to areas where deforestation and human activities are concentrated or across forest-savanna boundaries (Figure 1B). We found the annual rate of burned area (BA) in forest pixels across all HTFs is about 21 Mha year−1 averaged from 1982 to 2018. This estimate may be larger than real area of fire burns due to the coarse resolution of satellite data (500 m) and partial burns within each pixel. Time series of percent of BA show that fire disturbances in the African tropical forests are twice more prevalent and have larger interannual variability than fires in other continents (Figure S4). Our results suggest that forest fire and other land-use change activities impact approximately 10–15 Mha year−1 (not accounting the overlapped areas) (experimental procedures), causing the reduced intact forests area (>50% TC) from 1,300 Mha in 1982 to less than 1,000 Mha at the end of 2018. Increasing climate variability and change further exacerbates the land-use stress on tropical forests. Trends of air temperature (Figure 1C) show that regions in the southeast and northwest of the Amazon, Central and West Africa, and tropical Asia have experienced an increase of more than 0.4°C per decade. While tropical forests may show strong resilience to increased temperature,33Smith M.N. Taylor T.C. van Haren J. Rosolem R. Restrepo-Coupe N. Adams J. Wu J. de Oliveira R.C. da Silva R. de Araujo A.C. et al.Empirical evidence for resilience of tropical forest photosynthesis in a warmer world.Nat. Plants. 2020; 6: 1225-1230Crossref PubMed Scopus (24) Google Scholar the warming trend is also accompanied by atmospheric drying represented by VPD. We find an extensive increase of VPD (>0.01 hPa year−1) in South America and Central Africa (Figure 1D), including severe anomalies across regions in the southeastern Amazon and Congo Basins during drought years (Figure S5). The VPD trend shows a potential turning point from the early 2000s with approximately 1.5 times larger increase than the first two decades (1982–1999), suggesting a substantially more stress from atmospheric drying across the tropical forests in the last two decades.34Barkhordarian A. Saatchi S.S. Behrangi A. Loikith P.C. Mechoso C.R. A recent systematic increase in vapor pressure deficit over Tropical South America.Sci. Rep. 2019; 9: 1-12Crossref PubMed Scopus (57) Google Scholar Impacts of droughts and water stress on tropical forests are detected by the spatial variations of the WB as the difference between monthly water supply from precipitation and the water loss from potential ET (experimental procedures). Trends of WB from 1982 to 2018 (Figure 1E) show heterogeneous and divergent patterns with negative trends developing across the Congo Basin and areas in the south and southwestern Amazon. In contrast, we found increasing water availability in Asia and the northwestern Amazon. However, trends alone may not capture the water stress across the tropics. Stress from WB increases from seasonal anomalies of rainfall and during episodic droughts.7Saatchi S. Asefi-Najafabady S. Malhi Y. Aragão L.E.O.C. Anderson L.O. Myneni R.B. Nemani R. Persistent effects of a severe drought on Amazonian forest canopy.Proc. Natl. Acad. Sci. U. S. A. 2013; 110: 565-570Crossref PubMed Scopus (266) Google Scholar,35Marengo J.A. Williams E.R. Alves L.M. Soares W.R. Rodriguez D.A. Extreme seasonal climate variations in the Amazon Basin: droughts and floods.in: Interactions between Biosphere, Atmosphere and Human Land Use in the Amazon Basin. Springer, 2016: 55-76Crossref Google Scholar,36Asefi-Najafabady S. Saatchi S. Response of African humid tropical forests to recent rainfall anomalies.Philos. Trans. R. Soc. Lond. B. Biol. Sci. 2013; 368: 20120306Crossref PubMed Scopus (65) Google Scholar The negative trend of rainfall during the driest quarter shows that the western Amazon and its headwaters are emerging as the largest region of seasonal rainfall decline in recent decades (Figure S6). Rainfall trends are more heterogeneous across African and Asian forests, showing patterns of increasing rainfall (Figure S6). Among ecological responses, we focus on processes associated with the forest carbon cycling by including changes in live AGB from 2000 to 2018, the GPP from 1982 to 2018, ET from 1982 to 2018, and day-time LST from 2000 to 2018, which links forest carbon fluxes with energy cycles and is closely related to forest canopy temperature. Trend analysis shows the spatial patterns of these responses across the tropics (experimental procedures) (Figure 2). We found significant biomass loss in the southern and eastern Amazon and along the Andean foothills, Central America, West Africa, and Insular Asia (Figure 2A), where deforestation and degradation have been persistent and widespread (Figure S3). Trends of live biomass across tropics also show extensive areas of forest biomass gain, most importantly in the central and northern Congo Basin and mainland Asia and southern China. Increasing biomass in the Congo Basin may be attributed to the decline of the LCLUC stress6Song X.P. Hansen M.C. Stehman S.V. Potapov P.V. Tyukavina A. Vermote E.F. Townshend J.R. Global land change from 1982 to 2016.Nature. 2018; 560: 639-643Crossref PubMed Scopus (723) Google Scholar and stable productivity due to climate conditions and atmospheric CO2 fertilization.37Hubau W. Lewis S.L. Phillips O.L. Affum-Baffoe K. Beeckman H. Cuní-Sanchez A. Daniels A.K. Ewango C.E.N. Fauset S. Mukinzi J.M. et al.Asynchronous carbon sink saturation in African and Amazonian tropical forests.Nature. 2020; 579: 80-87Crossref PubMed Scopus (209) Google Scholar Increases in biomass in southern China are associated with recent intensive forestry, which significantly expanded across the region over the last 20 years.38Tong X. Brandt M. Yue Y. Ciais P. Rudbeck Jepsen M. Penuelas J. Wigneron J.-P. Xiao X. Song X.-P. Horion S. et al.Forest management in southern China generates short term extensive carbon sequestration.Nat. Commun. 2020; 11: 129Crossref PubMed Scopus (114) Google Scholar Forest productivity shows widespread and strong negative trends in the Amazon Basin over the entire time series (Figure 2B) influenced by frequent droughts, increasing temperature, and VPD since the early 2000s (see supplemental information) (Figure S5). In contrast, forests of the Congo Basin show stable and even increasing carbon productivity in northern and eastern regions (Figure 2B). Forest productivity shows positive trends in large areas of mainland Asia, southern China, and Central America, mostly due to widespread forest recovery from past disturbance, afforestation in recent years, and climate conditions favorable to GPP increase, such as increasing radiation and rainfall.38Tong X. Brandt M. Yue Y. Ciais P. Rudbeck Jepsen M. Penuelas J. Wigneron J.-P. Xiao X. Song X.-P. Horion S. et al.Forest management in southern China generates short term extensive carbon sequestration.Nat. Commun. 2020; 11: 129Crossref PubMed Scopus (114) Google Scholar,39Zhang Y. Song C. Band L.E. Sun G. No proportional increase of terrestrial gross carbon sequestration from the greening Earth.J. Geophys. Res. Biogeosciences. 2019; 124: 2540-2553Crossref Scopus (29) Google Scholar ET shows a declining trend (Figure 2C), mostly in areas impacted by the LCLUC across the tropics. Southern Amazon, with large-scale forest conversions since the early 1980s, areas in Central America, West Africa and Madagascar, and Insular Asia, with significant loss of intact forests appear as hotspots of long-term loss of ET. In contrast, large areas across the tropics show increasing trends of ET due to increasing rainfall from the intensification of the hydrological cycle40Gloor M. Brienen R.J.W. Galbraith D. Feldpausch T.R. Schöngart J. Guyot J.L. Espinoza J.C. Lloyd J. Phillips O.L. Intensification of the Amazon hydrological cycle over the last two decades.Geophys. Res. Lett. 2013; 40: 1729-1733Crossref Scopus (229) Google Scholar (Figure S5) or forest cover gain (e.g., mainland Asia and southern China) from afforestation and plantation systems. We used the current state of HTF biodiversity intactness (BI) varying between 0 and 1 and representing the present state of biological and structural integrity of tropical forests (Figure 2E) (experimental procedures). This index is used in our analysis to examine the link between forest resilience and its BI. We found high values of intactness (>0.7) ubiquitous across the central Amazon, Congo Basin, and Asia. However, there are also large areas with lower scores of (<0.4) in the southern and western Amazon, along the foothills of the Andes, and the coastlines where most forest clearing and degradation are concentrated. Wetlands of the Congo Basin exhibit fragmentation from dense hydrographic networks and extensive small-scale agriculture and settlements along rivers (Figure 1A), but remain relatively intact in biodiversity (>0.7). A clear contrast between TC change and BI exists in the mainland and Insular Asia where the landscape is highly fragmented due to forest cover change but the biodiversity remains relatively high (>0.7). To facilitate a direct comparison of the ecosystem stressors and responses across space and time, we normalized the distribution of each variable to their long-term mean across HTFs using quantile transform and preserved both negative and positive trends (experimental procedures) (Figures 3A and 3B ). We represented the cumulative distributions of normalized stress and response variables for each tropical region separately in cart wheel plots (Figures 3C and 3D). Climate and forest disturbance stressors emerge with similar scores but show larger variability across the tropics (Figure C), while ecosystem responses show relatively higher impacts on functions than states and remain less variable across the tropical regions (Figure 3B). HTFs in the Americas experience relatively similar cumulative climate and LCLUC stress with the exception of fire, whereas forests in Africa and Asia are exposed to relatively higher and lower climate and LCLUC stress, respectively. Overall, we find climate stress in each tropical region rivals LCLUC stress to near 0.4–0.6 of the HTF extent in Americas, about 0.4 in Asia and Oceania, and 0.6 in Africa. Their spatial patterns, however, suggest that the responses of ecosystem functions represented by RF (GPP, ET, LST) are mostly (RF > 0.6) influenced by the patterns of climate stress (Figures S7A and S7C), whereas the ecosystem states (AGB, BI) are mostly (RS > 0.6) impacted by the patterns of forest cover change from LCLUC (Figures S7B and S7D). The temporal autocorrelation of ecosystem responses (AR1 model) shows diverse variations along climate and LCLUC gradients across the three continental regions (Figure 4). The increase in temporal autocorrelation means that the state of the ecosystem on subsequent moments in time become more correlated indicating slower recovery rates (slow down) of the system.29Verbesselt J. Umlauf N. Hirota M. Holmgren M. Van Nes E.H. Herold M. Zeileis A. Scheffer M. Remotely sensed resilience of tropical forests.Nat. Clim. Chang. 2016; 6: 1028-1031Crossref Scopus (97) Google Scholar Tempor