Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Article
  • Published:

Amazon forest biogeography predicts resilience and vulnerability to drought


Amazonia contains the most extensive tropical forests on Earth, but Amazon carbon sinks of atmospheric CO2 are declining, as deforestation and climate-change-associated droughts1,2,3,4 threaten to push these forests past a tipping point towards collapse5,6,7,8. Forests exhibit complex drought responses, indicating both resilience (photosynthetic greening) and vulnerability (browning and tree mortality), that are difficult to explain by climate variation alone9,10,11,12,13,14,15,16,17. Here we combine remotely sensed photosynthetic indices with ground-measured tree demography to identify mechanisms underlying drought resilience/vulnerability in different intact forest ecotopes18,19 (defined by water-table depth, soil fertility and texture, and vegetation characteristics). In higher-fertility southern Amazonia, drought response was structured by water-table depth, with resilient greening in shallow-water-table forests (where greater water availability heightened response to excess sunlight), contrasting with vulnerability (browning and excess tree mortality) over deeper water tables. Notably, the resilience of shallow-water-table forest weakened as drought lengthened. By contrast, lower-fertility northern Amazonia, with slower-growing but hardier trees (or, alternatively, tall forests, with deep-rooted water access), supported more-drought-resilient forests independent of water-table depth. This functional biogeography of drought response provides a framework for conservation decisions and improved predictions of heterogeneous forest responses to future climate changes, warning that Amazonia’s most productive forests are also at greatest risk, and that longer/more frequent droughts are undermining multiple ecohydrological strategies and capacities for Amazon forest resilience.

This is a preview of subscription content, access via your institution

Access options

Buy this article

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Amazon forest remotely sensed responses to droughts.
Fig. 2: The Amazon forest response to the 2005 drought is structured by water-table depth.
Fig. 3: Southern Amazon forest responses to multiple droughts.
Fig. 4: Basin-wide Amazon forest responses to the 2015 drought, structured by ecotopes and predicted by whole-basin GAM analysis.
Fig. 5: A biogeography of Amazon forest drought resilience and vulnerability.

Similar content being viewed by others

Data availability

All remote sensing data and products (vegetation/photosynthetic indices (,, climate variables (,,, land cover (,, tree characteristics (canopy height, and soil texture ( are publicly available online. The ground-based demographic validation data are publicly available in refs. 2,26. The ground-based hydraulic trait validation data are publicly available in ref. 50. The HAND data are from ref. 25, which derived them from the digital elevation model from the Shuttle Radar Topography Mission. The soil fertility data are available in ref. 43.

Code availability

Code for reproducing the modelling analysis and figures is posted at Code Ocean (


  1. Pan, Y. et al. A large and persistent carbon sink in the world’s forests. Science 333, 988–993 (2011).

  2. Brienen, R. J. W. et al. Long-term decline of the Amazon carbon sink. Nature 519, 344–348 (2015).

    ADS  CAS  PubMed  Google Scholar 

  3. Wigneron, J.-P. et al. Tropical forests did not recover from the strong 2015–2016 El Niño event. Sci. Adv. 6, eaay4603 (2020).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  4. Gatti, L. V. et al. Amazonia as a carbon source linked to deforestation and climate change. Nature 595, 388–393 (2021).

    ADS  CAS  PubMed  Google Scholar 

  5. Boulton, C. A., Lenton, T. M. & Boers, N. Pronounced loss of Amazon rainforest resilience since the early 2000s. Nat. Clim. Chang. 12, 271–278 (2022).

    ADS  Google Scholar 

  6. Flores, B. M. et al. Critical transitions in the Amazon forest system. Nature 626, 555–564 (2024).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  7. Oyama, M. D. & Nobre, C. A. A new climate‐vegetation equilibrium state for Tropical South America. Geophys. Res. Lett. 30, 2199 (2003).

  8. Science Panel for the Amazon. Amazon Assessment Report 2021 (UN SDSN, 2021).

  9. Saleska, S. R., Didan, K., Huete, A. R. & da Rocha, H. R. Amazon forests green-up during 2005 drought. Science 318, 612 (2007).

    ADS  CAS  PubMed  Google Scholar 

  10. Brando, P. M. et al. Seasonal and interannual variability of climate and vegetation indices across the Amazon. Proc. Natl Acad. Sci. USA 107, 14685–14690 (2010).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  11. Xu, L. et al. Widespread decline in greenness of Amazonian vegetation due to the 2010 drought. Geophys. Res. Lett. (2011).

  12. Yang, J. et al. Amazon drought and forest response: largely reduced forest photosynthesis but slightly increased canopy greenness during the extreme drought of 2015/2016. Glob. Change Biol. 24, 1919–1934 (2018).

    ADS  Google Scholar 

  13. Anderson, L. O. et al. Vulnerability of Amazonian forests to repeated droughts. Philos. Trans. R. Soc. Lond. B 373, 20170411 (2018).

  14. Anderegg, W. R. L., Trugman, A. T., Badgley, G., Konings, A. G. & Shaw, J. Divergent forest sensitivity to repeated extreme droughts. Nat. Clim. Change 10, 1091–1095 (2020).

  15. Feldpausch, T. R. et al. Amazon forest response to repeated droughts. Glob. Biogeochem. Cycles 30, 964–982 (2016).

  16. Phillips, O. L. et al. Drought sensitivity of the Amazon rainforest. Science 323, 1344–1347 (2009).

    ADS  CAS  PubMed  Google Scholar 

  17. Esquivel-Muelbert, A. et al. Tree mode of death and mortality risk factors across Amazon forests. Nat. Commun. 11, 5515 (2020).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  18. Tansley, A. G. The use and abuse of vegetational concepts and terms. Ecology 16, 284–307 (1935).

    Google Scholar 

  19. Whittaker, R. H., Levin, S. A. & Root, R. B. Niche, habitat, and ecotope. Am. Nat. 107, 321–338 (1973).

    Google Scholar 

  20. Jiménez-Muñoz, J. C. et al. Record-breaking warming and extreme drought in the Amazon rainforest during the course of El Niño 2015–2016. Sci. Rep. 6, 33130 (2016).

    ADS  PubMed  PubMed Central  Google Scholar 

  21. Marengo, J. A. & Espinoza, J. C. Extreme seasonal droughts and floods in Amazonia: causes, trends and impacts. Int. J. Climatol. 36, 1033–1050 (2016).

    Google Scholar 

  22. Longo, M. et al. Ecosystem heterogeneity and diversity mitigate Amazon forest resilience to frequent extreme droughts. N. Phytol. 219, 914–931 (2018).

    Google Scholar 

  23. Violle, C., Reich, P. B., Pacala, S. W., Enquist, B. J. & Kattge, J. The emergence and promise of functional biogeography. Proc. Natl Acad. Sci. USA 111, 13690–13696 (2014).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  24. Costa, F. R. C., Schietti, J., Stark, S. C. & Smith, M. N. The other side of tropical forest drought: do shallow water table regions of Amazonia act as large-scale hydrological refugia from drought?. N. Phytol. 237, 714–733 (2022).

    Google Scholar 

  25. Nobre, A. D. et al. Height above the nearest drainage—a hydrologically relevant new terrain model. J. Hydrol. 404, 13–29 (2011).

  26. Sousa, T. R. et al. Palms and trees resist extreme drought in Amazon forests with shallow water tables. J. Ecol. 108, 2070–2082 (2020).

  27. Esteban, E. J. L., Castilho, C. V., Melgaço, K. L. & Costa, F. R. C. The other side of droughts: wet extremes and topography as buffers of drought negative effects in an Amazonian forest. N. Phytol. 229, 1995–2006 (2020).

  28. Oliveira, R. S. et al. Linking plant hydraulics and the fast-slow continuum to understand resilience to drought in tropical ecosystems. N. Phytol. 230, 904–923 (2021).

    Google Scholar 

  29. Garcia, M. N., Domingues, T. F., Oliveira, R. S. & Costa, F. R. C. The biogeography of embolism resistance across resource gradients in the Amazon. Glob. Ecol. Biogeogr. 32, 2199–2211 (2023).

    Google Scholar 

  30. Chitra-Tarak, R. et al. Hydraulically-vulnerable trees survive on deep-water access during droughts in a tropical forest. N. Phytol. 231, 1798–1813 (2021).

    Google Scholar 

  31. Brum, M. et al. Hydrological niche segregation defines forest structure and drought tolerance strategies in a seasonal Amazon forest. J. Ecol. 107, 318–333 (2019).

    Google Scholar 

  32. Tumber-Dávila, S. J., Schenk, H. J., Du, E. & Jackson, R. B. Plant sizes and shapes above and belowground and their interactions with climate. N. Phytol. 235, 1032–1056 (2022).

    Google Scholar 

  33. Giardina, F. et al. Tall Amazonian forests are less sensitive to precipitation variability. Nat. Geosci. 11, 405–409 (2018).

  34. ter Steege, H. et al. Continental-scale patterns of canopy tree composition and function across Amazonia. Nature 443, 444–447 (2006).

    ADS  PubMed  Google Scholar 

  35. McDowell, N. et al. Mechanisms of plant survival and mortality during drought: why do some plants survive while others succumb to drought? N. Phytol. 178, 719–739 (2008).

    Google Scholar 

  36. Anderegg, W. R. L. et al. Meta-analysis reveals that hydraulic traits explain cross-species patterns of drought-induced tree mortality across the globe. Proc. Natl Acad. Sci. USA 113, 5024–5029 (2016).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  37. Lyapustin, A. I. et al. Multi-angle implementation of atmospheric correction for MODIS (MAIAC): 3. Atmospheric correction. Remote Sens. Environ. 127, 385–393 (2012).

  38. Li, X. & Xiao, J. A Global, 0.05-degree product of solar-induced chlorophyll fluorescence derived from OCO-2, MODIS, and reanalysis data. Remote Sens. 11, 517 (2019).

  39. Aragão, L. E. O. C. et al. Spatial patterns and fire response of recent Amazonian droughts. Geophys. Res. Lett. (2007).

  40. Hastie, T. J. & Tibshirani, R. J. Generalized Additive Models (CRC, 1990).

  41. Pearl, J. Causal inference in statistics: an overview. Statist. Surv. 3, 96–146 (2009).

    MathSciNet  Google Scholar 

  42. Aragão, L. E. O. C. et al. 21st Century drought-related fires counteract the decline of Amazon deforestation carbon emissions. Nat. Commun. 9, 536 (2018).

    ADS  PubMed  PubMed Central  Google Scholar 

  43. Zuquim, G. et al. Making the most of scarce data: mapping soil gradients in data‐poor areas using species occurrence records. Methods Ecol. Evol. 10, 788–801 (2019).

    Google Scholar 

  44. Hengl, T. et al. SoilGrids250m: global gridded soil information based on machine learning. PLoS ONE 12, e0169748 (2017).

    PubMed  PubMed Central  Google Scholar 

  45. Simard, M., Pinto, N., Fisher, J. B. & Baccini, A. Mapping forest canopy height globally with spaceborne lidar. J. Geophys. Res. (2011).

  46. Christina, M. et al. Almost symmetrical vertical growth rates above and below ground in one of the world’s most productive forests. Ecosphere 2, 1–10 (2011).

    Google Scholar 

  47. da Costa, A. C. L. et al. Effect of 7 yr of experimental drought on vegetation dynamics and biomass storage of an eastern Amazonian rainforest. N. Phytol. 187, 579–591 (2010).

    Google Scholar 

  48. Nepstad, D. C., Tohver, I. M., Ray, D., Moutinho, P. & Cardinot, G. Mortality of large trees and lianas following experimental drought in an Amazon forest. Ecology 88, 2259–2269 (2007).

    PubMed  Google Scholar 

  49. Phillips, O. L. et al. Drought–mortality relationships for tropical forests. N. Phytol. 187, 631–646 (2010).

    Google Scholar 

  50. Tavares, J. V. et al. Basin-wide variation in tree hydraulic safety margins predicts the carbon balance of Amazon forests. Nature 617, 111–117 (2023).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  51. Duffy, P. B., Brando, P., Asner, G. P. & Field, C. B. Projections of future meteorological drought and wet periods in the Amazon. Proc. Natl Acad. Sci. USA 112, 13172–13177 (2015).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  52. Wunderling, N. et al. Recurrent droughts increase risk of cascading tipping events by outpacing adaptive capacities in the Amazon rainforest. Proc. Natl Acad. Sci. USA 119, e2120777119 (2022).

    CAS  PubMed  PubMed Central  Google Scholar 

  53. Makarieva, A. M. et al. The role of ecosystem transpiration in creating alternate moisture regimes by influencing atmospheric moisture convergence. Glob. Chang. Biol. 29, 2536–2556 (2023).

    CAS  PubMed  Google Scholar 

  54. Costa, M. H. et al. in Amazon Assessment Report 2021 (eds Nobre, C. et al.) Ch. 7 (UN SDSN, 2021).

  55. Betts, M. G. et al. When are hypotheses useful in ecology and evolution? Ecol. Evol. 11, 5762–5776 (2021).

    PubMed  PubMed Central  Google Scholar 

  56. Glass, D. J. & Hall, N. A brief history of the hypothesis. Cell 134, 378–381 (2008).

    CAS  PubMed  Google Scholar 

  57. Huete, A. et al. Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sens. Environ. 83, 195–213 (2002).

    ADS  Google Scholar 

  58. Santos, V. A. H. F. D. et al. Causes of reduced leaf-level photosynthesis during strong El Niño drought in a Central Amazon forest. Glob. Change Biol. 24, 4266–4279 (2018).

    ADS  Google Scholar 

  59. Wu, J. et al. Partitioning controls on Amazon forest photosynthesis between environmental and biotic factors at hourly to interannual timescales. Glob. Change Biol. 23, 1240–1257 (2017).

    ADS  Google Scholar 

  60. Wu, J. et al. Leaf development and demography explain photosynthetic seasonality in Amazon evergreen forests. Science 351, 972–976 (2016).

    ADS  CAS  PubMed  Google Scholar 

  61. Mohammed, G. H. et al. Remote sensing of solar-induced chlorophyll fluorescence (SIF) in vegetation: 50 years of progress. Remote Sens. Environ. 231, 111177 (2019).

    PubMed  PubMed Central  Google Scholar 

  62. Sun, Y. et al. OCO-2 advances photosynthesis observation from space via solar-induced chlorophyll fluorescence. Science 358, eaam5747 (2017).

    PubMed  Google Scholar 

  63. Shekhar, A., Buchmann, N. & Gharun, M. How well do recently reconstructed solar-induced fluorescence datasets model gross primary productivity? Remote Sens. Environ. 283, 113282 (2022).

    Google Scholar 

  64. Huffman, G. J., Adler, R. F., Bolvin, D. T. & Nelkin, E. J. in Satellite Rainfall Applications for Surface Hydrology (eds Gebremichael, M. & Hossain, F.) 3–22 (Springer, 2010).

  65. Gelaro, R. et al. The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Clim. 30, 5419–5454 (2017).

    ADS  Google Scholar 

  66. Aumann, H. H. Atmospheric infrared sounder on the Earth observing system. Opt. Eng. 33, 776 (1994).

  67. Kahn, B. H. et al. The Atmospheric Infrared Sounder version 6 cloud products. Atmos. Chem. Phys. 14, 399–426 (2014).

  68. Susskind, J., Blaisdell, J. M. & Iredell, L. Improved methodology for surface and atmospheric soundings, error estimates, and quality control procedures: the atmospheric infrared sounder science team version-6 retrieval algorithm. J. Appl. Remote Sens. 8, 084994 (2014).

  69. Sun, J. et al. Global evaluation of terrestrial near-surface air temperature and specific humidity retrievals from the Atmospheric Infrared Sounder (AIRS). Remote Sens. Environ. 252, 112146 (2021).

    Google Scholar 

  70. Bastian, O. et al. In Development and Perspectives of Landscape Ecology (eds Bastian, O. & Steinhardt, U.) 49–112 (Springer, 2002).

  71. Rennó, C. D. et al. HAND, a new terrain descriptor using SRTM-DEM: mapping terra-firme rainforest environments in Amazonia. Remote Sens. Environ. 112, 3469–3481 (2008).

    ADS  Google Scholar 

  72. Brunner, I., Herzog, C., Dawes, M. A., Arend, M. & Sperisen, C. How tree roots respond to drought. Front. Plant Sci. 6, 547 (2015).

    PubMed  PubMed Central  Google Scholar 

  73. Oliveira, R. S. et al. Embolism resistance drives the distribution of Amazonian rainforest tree species along hydro‐topographic gradients. N. Phytol. 221, 1457–1465 (2019).

  74. Levis, C. et al. Persistent effects of pre-Columbian plant domestication on Amazonian forest composition. Science 355, 925–931 (2017).

    ADS  CAS  PubMed  Google Scholar 

  75. Fan, Y., Li, H. & Miguez-Macho, G. Global patterns of groundwater table depth. Science 339, 940–943 (2013).

    ADS  CAS  PubMed  Google Scholar 

  76. Cunha, H. F. V. et al. Direct evidence for phosphorus limitation on Amazon forest productivity. Nature 608, 558–562 (2022).

    ADS  CAS  PubMed  Google Scholar 

  77. Chadwick, O. A., Derry, L. A., Vitousek, P. M., Huebert, B. J. & Hedin, L. O. Changing sources of nutrients during four million years of ecosystem development. Nature 397, 491–497 (1999).

    ADS  CAS  Google Scholar 

  78. Darela-Filho, J. P. et al. Reference maps of soil phosphorus for the pan-Amazon region. Earth Syst. Sci. Data 16, 715–729 (2024).

    ADS  Google Scholar 

  79. Liu, H.-Y., Sun, W.-N., Su, W.-A. & Tang, Z.-C. Co-regulation of water channels and potassium channels in rice. Physiol. Plant. 128, 58–69 (2006).

  80. Fan, Y., Miguez-Macho, G., Jobbágy, E. G., Jackson, R. B. & Otero-Casal, C. Hydrologic regulation of plant rooting depth. Proc. Natl Acad. Sci. USA 114, 10572–10577 (2017).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  81. Hasanuzzaman, M., Araújo, S. & Gill, S. S. The Plant Family Fabaceae: Biology and Physiological Responses to Environmental Stresses (Springer, 2021).

  82. Friedl, M. A. et al. MODIS Collection 5 global land cover: algorithm refinements and characterization of new datasets. Remote Sens. Environ. 114, 168–182 (2010).

    ADS  Google Scholar 

  83. Hess, L. L. et al. Wetlands of the lowland Amazon basin: extent, vegetative cover, and dual-season inundated area as mapped with JERS-1 synthetic aperture radar. Wetlands 35, 745–756 (2015).

    Google Scholar 

  84. Gómez, J., Schobbenhaus, C. & Montes, N. E. Geological Map of South America 2019. Scale 1: 5 000 000. Commission for the Geological Map of the World (CGMW), Colombian Geological Survey and Geological Survey of Brazil (2019);

  85. Vancutsem, C. et al. Long-term (1990–2019) monitoring of forest cover changes in the humid tropics. Sci. Adv. 7, eabe1603 (2021).

    ADS  PubMed  PubMed Central  Google Scholar 

  86., et al. Taking the pulse of Earth’s tropical forests using networks of highly distributed plots. Biol. Conserv. 260, 108849 (2021).

    Google Scholar 

  87. Shi, H. et al. Assessing the ability of MODIS EVI to estimate terrestrial ecosystem gross primary production of multiple land cover types. Ecol. Indic. 72, 153–164 (2017).

    Google Scholar 

  88. Sims, D. A. et al. On the use of MODIS EVI to assess gross primary productivity of North American ecosystems. J. Geophys. Res. 111, G04015 (2006).

  89. Rahman, A. F., Sims, D. A., Cordova, V. D. & El-Masri, B. Z. Potential of MODIS EVI and surface temperature for directly estimating per-pixel ecosystem C fluxes. Geophys. Res. Lett. 32, L19404 (2005).

  90. Sims, D. A. et al. A new model of gross primary productivity for North American ecosystems based solely on the enhanced vegetation index and land surface temperature from MODIS. Remote Sens. Environ. 112, 1633–1646 (2008).

    ADS  Google Scholar 

  91. Huete, A. R. et al. Amazon rainforests green-up with sunlight in dry season. Geophys. Res. Lett. 33, L06405 (2006).

  92. Huete, A. R. et al. Multiple site tower flux and remote sensing comparisons of tropical forest dynamics in Monsoon Asia. Agric. For. Meteorol. 148, 748–760 (2008).

    ADS  Google Scholar 

  93. Huete, A. R. Vegetation indices, remote sensing and forest monitoring. Geogr. compass 6, 513–532 (2012).

    Google Scholar 

  94. Samanta, A. et al. Amazon forests did not green-up during the 2005 drought. Geophys. Res. Lett. 37, L05401 (2010).

  95. Morton, D. C. et al. Amazon forests maintain consistent canopy structure and greenness during the dry season. Nature 506, 221–224 (2014).

    ADS  CAS  PubMed  Google Scholar 

  96. Saleska, S. R. et al. Dry-season greening of Amazon forests. Nature 531, E4–E5 (2016).

    CAS  PubMed  Google Scholar 

  97. Maeda, E. E., Heiskanen, J., Aragão, L. E. O. C. & Rinne, J. Can MODIS EVI monitor ecosystem productivity in the Amazon rainforest? Geophys. Res. Lett. 41, 7176–7183 (2014).

    ADS  Google Scholar 

  98. Magney, T. S., Barnes, M. L. & Yang, X. On the covariation of chlorophyll fluorescence and photosynthesis across scales. Geophys. Res. Lett. 47, e2020GL091098 (2020).

  99. Chambers, J. Q. et al. Respiration from a tropical forest ecosystem: partitioning of sources and low carbon use efficiency. Ecol. Appl. 14, 72–88 (2004).

    Google Scholar 

  100. Smith, M. N. et al. Seasonal and drought-related changes in leaf area profiles depend on height and light environment in an Amazon forest. N. Phytol. 222, 1284–1297 (2019).

    Google Scholar 

  101. Janssen, T. et al. Drought effects on leaf fall, leaf flushing and stem growth in the Amazon forest: reconciling remote sensing data and field observations. Biogeosciences 18, 4445–4472 (2021).

  102. Quesada, C. A. et al. Basin-wide variations in Amazon forest structure and function are mediated by both soils and climate. Biogeosciences 9, 2203–2246 (2012).

    ADS  Google Scholar 

  103. Lloyd, J. et al. Edaphic, structural and physiological contrasts across Amazon Basin forest–savanna ecotones suggest a role for potassium as a key modulator of tropical woody vegetation structure and function. Biogeosciences 12, 6529–6571 (2015).

  104. Wold, S., Esbensen, K. & Geladi, P. Principal component analysis. Chemometr. Intellig. Lab. Syst. 2, 37–52 (1987).

    CAS  Google Scholar 

  105. Doan, H. T. X. & Foody, G. M. Reducing the impacts of intra-class spectral variability on soft classification and its implications for super-resolution mapping. In Proc. IEEE International Geoscience and Remote Sensing Symposium 2585–2588 (IEEE, 2007).

  106. Malhi, A. & Gao, R. X. PCA-based feature selection scheme for machine defect classification. IEEE Trans. Instrum. Meas. 53, 1517–1525 (2004).

    ADS  Google Scholar 

  107. Canty, M. J. Image Analysis, Classification and Change Detection in Remote Sensing: With Algorithms for ENVI/IDL and Python, Third Edition (CRC, 2014).

  108. Grafton, R. Q. et al. Realizing resilience for decision-making. Nat. Sustain. 2, 907–913 (2019).

    Google Scholar 

  109. Nikinmaa, L. et al. Reviewing the use of resilience concepts in forest sciences. Curr. For. Rep. 6, 61–80 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  110. Gunderson, L. H. Ecological resilience—in theory and application. Annu. Rev. Ecol. Syst. 31, 425–439 (2000).

    Google Scholar 

  111. Samanta, A. et al. Amazon forests did not green-up during the 2005 drought. Geophys. Res. Lett. (2010).

  112. Haining, R. P. & Haining, R. Spatial Data Analysis: Theory and Practice (Cambridge Univ. Press, 2003).

  113. Burnham, K. P., Anderson, D. R. & Huyvaert, K. P. AIC model selection and multimodel inference in behavioral ecology: some background, observations, and comparisons. Behav. Ecol. Sociobiol. 65, 23–35 (2011).

    Google Scholar 

  114. Clark, J. S. Models for Ecological Data (Princeton Univ. Press, 2020).

  115. Pearl, J. Causality (Cambridge Univ. Press, 2009).

  116. Ankan, A., Wortel, I. M. N. & Textor, J. Testing graphical causal models using the R package ‘dagitty’. Curr. Protoc. 1, e45 (2021).

    PubMed  Google Scholar 

  117. Arif, S. & MacNeil, M. A. Applying the structural causal model framework for observational causal inference in ecology. Ecol. Monogr. 93, e1554 (2023).

  118. Arif, S. & MacNeil, M. A. Predictive models aren’t for causal inference. Ecol. Lett. 25, 1741–1745 (2022).

    PubMed  Google Scholar 

  119. Wood, S. N. Generalized Additive Models: An Introduction with R, Second Edition (CRC, 2017).

  120. Quaresma, M., Carpenter, J. & Rachet, B. Flexible Bayesian excess hazard models using low-rank thin plate splines. Stat. Methods Med. Res. 29, 1700–1714 (2020).

    MathSciNet  PubMed  Google Scholar 

  121. Schempp, W. & Zeller, K. eds. Constructive Theory of Functions of Several Variables: Proceedings of a Conference Held at Oberwolfach, April 25 - May 1, 1976 (Springer, 1977).

  122. Laurance, W. F. et al. Relationship between soils and Amazon forest biomass: a landscape-scale study. For. Ecol. Manage. 118, 127–138 (1999).

    Google Scholar 

  123. Levine, N. M. et al. Ecosystem heterogeneity determines the ecological resilience of the Amazon to climate change. Proc. Natl Acad. Sci. USA 113, 793–797 (2016).

    ADS  CAS  PubMed  Google Scholar 

  124. Addicott, E. T., Fenichel, E. P., Bradford, M. A., Pinsky, M. L. & Wood, S. A. Toward an improved understanding of causation in the ecological sciences. Front. Ecol. Environ. 20, 474–480 (2022).

    Google Scholar 

  125. Textor, J., van der Zander, B., Gilthorpe, M. S., Liskiewicz, M. & Ellison, G. T. Robust causal inference using directed acyclic graphs: the R package ‘dagitty’. Int. J. Epidemiol. 45, 1887–1894 (2016).

    PubMed  Google Scholar 

  126. de Castilho, C. V. et al. Variation in aboveground tree live biomass in a central Amazonian Forest: effects of soil and topography. For. Ecol. Manage. 234, 85–96 (2006).

    Google Scholar 

  127. Velleman, P. F. & Welsch, R. E. Efficient computing of regression diagnostics. Am. Stat. 35, 234–242 (1981).

    Google Scholar 

  128. Webb, C. O., Cannon, C. H. & Davies, S. J. in Tropical Forest Community Ecology (eds Carson, W. P. & Schnitzer, S. A.) 79–97 (Wiley–Blackwell, 2008).

  129. Humboldt’s legacy. Nat. Ecol. Evol. 3, 1265–1266 (2019).

  130. Reichstein, M., Bahn, M., Mahecha, M. D., Kattge, J. & Baldocchi, D. D. Linking plant and ecosystem functional biogeography. Proc. Natl Acad. Sci. USA 111, 13697–13702 (2014).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  131. Nelson, B. W., Gonçalves, N. B., Chen, S. & Saleska, S. Persistent effect of 2015 El Niño drought on NIR reflectance of central Amazon upland forests? In Anais do XX Simpósio Brasileiro de Sensoriamento Remoto 1636–1638 (INPE, 2023).

  132. Oliveira, R. S., Dawson, T. E., Burgess, S. S. O. & Nepstad, D. C. Hydraulic redistribution in three Amazonian trees. Oecologia 145, 354–363 (2005).

    ADS  PubMed  Google Scholar 

  133. Nunes, M. H. et al. Forest fragmentation impacts the seasonality of Amazonian evergreen canopies. Nat. Commun. 13, 917 (2022).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  134. Van der Meer, T., Te Grotenhuis, M. & Pelzer, B. Influential cases in multilevel modeling: a methodological comment. Am. Sociol. Rev. 75, 173–178 (2010).

    Google Scholar 

  135. Boers, N., Marwan, N., Barbosa, H. M. J. & Kurths, J. A deforestation-induced tipping point for the South American monsoon system. Sci. Rep. 7, 41489 (2017).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

Download references


We thank J. Schietti for early discussions of the idea that remote sensing might be used to investigate the effect of water-table depth on forest drought response; T. R. Sousa for sharing and discussing plot-based forest demographic data (from along the BR-319 road)26; G. Zuquim for sharing an early version of mapped basin-wide soil fertility data43; H. ter Steege for sharing mapped basin-wide tree characteristics data34; T. R. Sousa and J. Schietti for comments on an earlier version of the manuscript; L. Alves for advice on forest demography plots; R. Palacios for recommending the use of GAM models; M. N. Garcia for discussions about soil fertility; N. Boers for advice on the South American monsoon system; T. C. Taylor and V. Ivanov for discussions; J. Cronin and S. McMahon for detailed advice and comments; and S.C.’s doctoral dissertation committee members W. K. Smith, J. Hu and B. Enquist for constructive criticism and advice on the direction of this work. This work was supported by US National Aeronautics and Space Administration, fellowship 80NSSC19K1376 (S.C.); US National Science Foundation, DEB grant 1950080 (S.C.S. and M.N.S.); US National Science Foundation, DEB grant 2015832 (S.R.S.); US National Science Foundation, DEB grant 1754803 (S.R.S., N.R.-C.), US National Science Foundation, DEB grant 1754357 (S.C.S.); Brazil National Council for Scientific and Technological Development (CNPq) scholarships 371626/2022-6, 372734/2021-9,381711/2020-0 (D.d.J.A.); and US Department of Energy’s Next Generation Ecosystem Experiments-Tropics (R.C.-T.).

Author information

Authors and Affiliations



S.C. and S.R.S. designed the analysis, based on early conception by A.D.N. and S.R.S., and on funded proposals to investigate ‘the other side of tropical forest drought’ led by S.C.S., M.N.S. and S.R.S. (from NSF) and by S.C. and S.R.S. (from NASA). A.D.N., L.A.C. and D.d.J.A. updated their HAND data product and interpreted it for this analysis. B.W.N. and N.R.-C. contributed remote-sensing expertise and analysis. R.C.-T. contributed statistical modelling expertise and analysis. H.K. contributed code, especially for the variogram analysis. S.C. organized the datasets (with assistance from N.R.-C.), conducted the analysis and wrote the initial draft. S.C., S.R.S. and S.C.S. revised the draft. All of the authors contributed to writing the final version.

Corresponding authors

Correspondence to Shuli Chen or Scott R. Saleska.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature thanks Christopher Baraloto, David Bauman, Jovan Tadic and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Drought maps 2005, 2010 and 2015/16 droughts and GOSIF-based forest responses droughts.

(a)-(c): Maximum cumulative water deficit (MCWD) standardized anomalies. (relative to the long term mean MCWD across years, blue=positive, orange=negative) during drought for (a) 2005, (b) 2010, and (c) 2015 droughts. MCWD is calculated (see Methods, ‘Climate variables’) as the maximum water deficit reached for each hydrologic year (from May of the nominal year to the following April). The “drought region” is defined as pixels whose MCWD anomaly is more than one SD below the mean (light orange to red). (d)-(f): GOSIF-based forest response to droughts. GOSIF anomalies during drought, relative to the long term mean GOSIF (green=positive, orange=negative) in drought regions for the (d) 2005, (e) 2010 and (f) 2015 droughts, respectively. (g) EVI (left axis) and GOSIF (right axis) anomalies in the 2005 drought elliptical region (as depicted in Figs. 1a, 2a, and here in Extended Data Fig. 1d) show consistent patterns versus HAND (bin averages ±95% CI, with N = 6,547 5-km pixels for both EVI and GOSIF); (h) GOSIF anomalies (bin averages points ±95% CI and solid regression line) vs. water-table depths (indexed by HAND) support hypothesis 1 (with negative slopes, consistent with EVI in Fig. 3a) for the 2005 (green, slope = −0.016 ± 0.006 SE m−1), 2010 (purple, slope = −0.012 ± 0.003 SE m−1), and 2015 (blue, slope = −0.010 ± 0.003 SE m−1) droughts, paired with HAND distributions in each drought region (bottom graphs, right axis, with N = 34,980, 30,004, 43,475 5-km pixels for 2005, 2010, and 2015 droughts, respectively).

Extended Data Fig. 2 Ecotope factors of the Amazon basin.

(a) Height Above Nearest Drainage (HAND), a proxy for water-table depth25; (b) Soil fertility, as exchangeable base cation concentrations43; (c) Average forest heights as acquired by lidar45; (d) Soil sand content44; (e) Proportion of trees belonging to the Fabaceae family34; (f) MCWD variability (see the ‘Climate anomalies for drought definition and mapping’ section of methods), in terms of the standard deviation of the long-term MCWD timeseries. High variance in climate and low soil fertility in Guiana shield might contribute to the greatest proportion of trees belonging to the family Fabaceae with the very high wood density; (g) Averaged minimum monthly precipitation (low=green, high=orange). The north-west everwet Amazon is distinguished by lacking a dry season (precipitation exceeds evapotranspiration). (h) Community-weighted wood density34. Panels a-d are used as ecotope predictors in the GAM analysis of Supplementary Table 1. (Data sources: see the ‘Climate variables’ and ‘Climate anomalies for drought definition and mapping’ sections of methods).

Extended Data Fig. 3 Spatial distributions of climate dynamics in the 2005, 2010 and 2015 droughts.

ai, Spatial distributions of climate dynamics in the 2005 (left column), 2010 (middle column) and 2015 (right column) droughts for: (a)-(i): Drought dynamics showing drought onset date (row 1, a-c), drought end date (row 2, d-f), and drought duration (row 3, g-i, end date minus start date). Pixel-by-pixel drought responses (EVI in Figs. 14; or GOSIF in Extended Data Figs. 1 & 5) are taken as the standardized anomalies that occur during the pixel-specific drought period defined here. (j)-(r): climatic anomalies of: photosynthetic active radiation (PAR) (row 4, j-l), vapor pressure deficit (VPD) (row 5, m-o), and precipitation (row 6, p-r). precipitation (Data source: see the ‘Climate variables’ section of methods).

Extended Data Fig. 4 Regions in the Amazon basin.

that emerge from a principal components analysis (PCA) followed by classification: (a) PCA of the Amazon basin 0.4° x 0.4° pixel data (coloured according to a supervised classification into three classes identified by variance minimization), projected onto their first two principal components, which are composed mainly of three dimensions, one defined by wood density and proportions of the family Fabaceae (first principal component, horizontal axis), one defined by minimum monthly precipitation and MCWD variability (second principal component, vertical axis), and a third defined mainly by soil fertility; the classes are significantly separated in PCA space (psuedo-F ratio =950, df=2, 3805, p ~ 0, permanova test); (b) The Amazon pixels coloured according to their class (corresponding to the colours in a), showing that the classification of (a) maps pixels into distinct, mostly contiguous spatial regions.) (c) Standardized values, for each region, of each group of characteristics (ordered by water availability, soil fertility, and tree traits/characteristics), illustrate distinct regional niches: the everwet Amazon is highest in minimum precipitation and lowest (highest negative) in MCWD variability; the Southern Amazon is moderately high in mean fertility, and the Guiana Shield has the tallest mean forest height and greatest wood density. (d) scree plot of the eigenvalues (principal components) of the PCA shown in (a), plotted in rank order.

Extended Data Fig. 5 Amazon forest drought responses in different regions using the EVI and GOSIF remote sensing indices.

Amazon forest EVI (top row) and GOSIF (bottom row) responses to multiple droughts in the Guiana shield (left column) and the ever-wet northwest (right column). These generally do not support the “other side of drought” hypothesis 1, because they show generally consistently positive slopes with water-table depth (HAND), in contrast to negative slope responses in the Southern Amazon (Fig. 3a). Plots show observations (bin average points ±95% CI, and solid regression lines) and unified multi-drought GAM predictions (±95% CI shaded region, for models in Supplementary Table 1b, c), with climate fixed to region-wide median drought conditions for each drought.) Observations for EVI (a-b): N = 83 and 666 0.4° pixels for 2005 and 2015 droughts respectively, in the Guiana shield (a), and N = 147, 368, and 648 for 2005, 2010 and 2015 droughts respectively in the ever-wet Amazon (b). Observations for GOSIF (c-d): N = 1876, and 25,460 5-km pixels for 2005 and 2015 droughts, respectively, in Guiana shield (c), and N = 1,914, 8,261, and 19,918 for 2005, 2010 and 2015 droughts, respectively, in the ever-wet Amazon (d). Purple points (2010) are not shown in panels a,c, because the 2010 drought did not significantly affect the Guiana shield.

Extended Data Fig. 6 Implementing Structured Causal Modeling (SCM) of Amazon forest drought response using Directed acyclic graphs (DAGs).

ad, Development of a Directed acyclic graph (DAG) representing the structure of factors influencing tropical forest responses to drought. (a) Initially hypothesized DAG characterizing the causal relationships among climatic, environmental, and forest variables (measured variables depicted as blue nodes, unmeasured rooting depth is depicted in grey) leading to forest drought response (other colour node), with arrows representing the hypothesized causal links. (b) DAG-data consistency tests for initial DAG, with the largest 20 approximated non-linear correlation coefficients (estimated via root mean square error of approximation, RMSEA) between unlinked variables in (a). (Note: unlinked variables in a DAG are hypothesized to have zero correlation or zero conditional correlation; thus, the second row of panel b tests “DR_ | | _DSL | DL” -- whether DR is independent of DSL conditioned on DL, by estimating the non-linear correlation between DR and the residuals of DSL regressed on DL.) Correlations greater than an acceptability threshold (dashed vertical lines at ±0.30) fail the test of conditional independence, addressed by adding to the DAG either a direct causal link (indicated by a green symbol), or links to a common cause (pink symbol) (such added arrows are included in panel c). (c) Final DAG after correcting for conditional independency inconsistencies of the initial DAG in A, in light of ecological considerations. Also illustrates use of the backdoor criterion to determine the causal effect of ‘drought length (DL)’ (the exposed predictor node and associated forward causal paths, in green) on forest drought response (corresponding to the model in Extended Data Fig. 10c), while blocking the confounding variable dry season length, DSL (hypothesized to itself affect DL) and its associated causal backdoor paths (which are considered non-causal paths with respect to the exposed variable DL) (in pink). (d) DAG-Data consistency tests for final DAG (panel c), showing the largest 20 RMSEA values. (e)-(j): GAM regression model predictions (±95% CI shaded region) of causal effects of different variables derived from DAG, employing backdoor criterion, for the Southern Amazon, average across all three droughts: (e) of HAND (no backdoor to be blocked) (f) of PAR (adjusting for back door paths through drought length, dry season length) (g) of Drought length (adjusting for back door path through dry season length) on EVI responses (adjusted EVI prediction); the whole Amazon basin during the 2015 drought: (h) of forest height, categorized by shallow (blue, HAND = 0-10 m) and deep (red, HAND = 20–40 m) water tables (adjusting for back door paths through soil fertility, soil texture and dry season length), (i) of soil fertility (adjusting for back door path through dry season length) (j) of soil texture (no backdoor path to be blocked).

Extended Data Fig. 7 The sensitivities of forest drought response to soil texture and drought timing.

(a) The sensitivity of forest response to soil texture (sand content) and water- table depth (HAND) in basin-wide GAM analysis. GAM-predicted adjusted EVI anomaly (left axis) versus soil sand content (%), with water table-depth in colour (shallow=blue to deep=red), paired with distributions of mean forest height in each soil texture bin (bottom graph, right axis, with N = 3,318, and 1,142 0.4° pixels for shallow and deep water tables, respectively). ‘Adjusted’ GAM predictions are made by setting non-displayed predictors (climate variables, tree-height, soil fertility) to their median values during the drought. (b)-(d): The sensitivity of forest responses to dry versus wet season drought periods, across the three-droughts: (b) distribution of the proportion of drought that was in the dry season (0 = all in the wet season to 1= all in the dry season) for drought-affected pixels in each of the three droughts; (c) GAM-predicted EVI anomaly versus PAR, for different proportions of dry season drought (blue=all wet to red=all dry, corresponding to coloured tick marks in the vertical axis of b). (d) Adjusted EVI anomaly from GAM prediction versus drought length, for different proportions of dry-season drought (blue to red, as in panel c).

Extended Data Fig. 8 Scale-dependence of Southern Amazon forest responses to drought, showing that detected response patterns are largely invariant across different scales of analysis.

(a) At 0.4 degree (40-km) scale (across the Southern Amazon. all three droughts): Climate-adjusted EVI responses (standardized anomalies from MODIS) vs. water-table depths (indexed by HAND) for observations (solid points ±95% CI and solid regression line) and for unified multi-drought GAM predictions (model of Supplementary Table 1a, shaded bands and dashed regression line slopes) for the 2005 (green, slope = −0.019 ± 0.001 SE m−1), 2010 (purple, slope = −0.020 ± 0.002 SE m−1), and 2015 (blue, slope = −0.028 ± 0.002 SE m−1) droughts (with N = 1,384, 1,673, and 1,837 0.4° pixels for 2005, 2010, and 2015 droughts, respectively); (b) At 1-km scale (across the Southern Amazon, all three droughts), as in (a): climate-adjusted EVI responses vs. HAND for observations (solid points and regression line) and corresponding GAM (with the same Supplementary Table 1a model now fit at 1 km scale, revealing autocorrelation in observations causing too-narrow confidence bands, and slight model underpredictions of the extremes of the 2005 greenup and the 2010 browdown, but maintaining the similar negative dependence on HAND across all droughts); (c) At 30 to 180 m scales (for a forest region around Manaus, 2015-2016 drought only): Delta EVI, the fraction change in EVI due to the drought = (after-drought EVI (July 2016) - pre-drought EVI (August 2015))/pre-drought EVI (Landsat OLI8, at 30 m resolution) vs. water-table depths (indexed by HAND) for Landsat observations (solid points ±95% CI and solid regression line) at native (30 m) and aggregated to 90 and 180-m scales (with N = 105,359, 11,901, and 2,999 pixels for 30-m, 90-m, and 180-m scales, respectively). Also shown in the bottom of each panel is the distribution of water-table depth (HAND proxy) at each scale. Aggregations to larger (coarser) scales induce an apparent regression towards the mean in the water-table depth distributions (as more extreme water-table depths at finer scales become diluted by averaging to large scales), while similar dilution of extremes in EVI response (not shown) preserves the overall relation between EVI responses and watertable depth (especially evident in the Landsat analysis where the slopes through data aggregated at different scales do not detectably differ).

Extended Data Fig. 9 Remote sensing validation with forest inventory plot demography.

(a) Remotely sensed map of MAIAC EVI (1-km resolution), overlaid with aboveground NPP (ANPP) rates from 321 ground-monitored forest plots (red circles, % standing biomass y−1) as aggregated to 1 degree grid plots (RAINFOR plots in Brienen et al. 2), with both EVI and ANPP taken during the 2000–2011 interval. ANPP rate is calculated as Aboveground Biomass (AGB) gain (Mg/(ha·yr)) (total annual AGB productivity of surviving trees plus recruitment, plus inferred growth of trees that died between censusing intervals) divided by initial AGB (Mg/ha) (standing above ground biomass at the start of the census interval). (b) ANPP rates as predicted by EVI (points from (a) plus solid regression line with statistics; Dashed line and associated statistics in grey represent linear regression without the high leverage point, shown in red, defined by Cook’s distances > 4/n, where n=number of points134). EVI is the mean extracted from intervals matching the average census interval of the corresponding plots in Brienen et al. 2 (c)–(e) MAIAC EVI anomalies (1-km pixels) versus ground-monitored tree demography in shallow water table forests during the 2015-2016 drought26 for: (c) mortality, (d) recruitment, and (e) mortality:recruitment ratios in 1-ha plots. (f)–(h): GOSIF anomalies (5-km pixels) versus ground-monitored (f) mortality, (g) recruitment, and (h) mortality:recruitment ratios; Solid lines and statistics (R2 and p-values) represent standard linear regression fits to all data. Red points, if they exist, are high leverage, i.e. with Cook’s distances > 4/n, where n=number of points134, and dotted lines and associated statistics in grey represent standard linear regressions without such points, showing that remote detection of ground-derived demographic trends is robust. R2 values reported here are consistent with the expectation that they should be less than for remote detection of tropical forest GPP (R2 = 0.5-0.7), because GPP contributes only partially to the NPP driver of demography (as discussed in the 'Validation by forest plot metrics of demography and of physiological drought tolerance' section of Methods). Considering multple comparisons (six regressions), the probability, under the null hypothesis, of seeing five or more significant regresssions out of six is p = 0.000002 (Binomial test).

Extended Data Fig. 10 Modeled forest response to the 2015 drought and implications of the derived map of Amazon forest biogeography.

ac, Forest response to the 2015 drought in drought-affected pixels. (a) Observed EVI anomalies (resampled at 0.4 degrees to match model resolution which accounts for spatial autocorrelation (see Supplementary Fig. 1). (b) GAM-predicted EVI anomalies (model of Supplementary Table 1d). (c) Residual EVI anomalies (panel a observations minus panel b predictions). The GAM well-predicts the pattern of response (Panel b), but under-estimates the extremes of the responses (as evident from residuals in panel c continuing to show greening/browning patterns beyond the predictions). (d) Map of Amazon forest biogeography of resilience/vulnerability, overlaid with mean winds (arrows, at height 650 hPa) and location of the arc of deforestation. The most productive as well as the most vulnerable forests (in red) are also those most experiencing deforestation (in the “arc of deforestation”) which is causing local climatic warming/drying4, further stressing these vulnerable forests. These “arc of deforestation”/vulnerable forests are often upwind forests135 (especially when the Intertropical convergence zone, ITCZ, swings to the south) that are critical for hydrological recycling in the Amazon.

Supplementary information

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Chen, S., Stark, S.C., Nobre, A.D. et al. Amazon forest biogeography predicts resilience and vulnerability to drought. Nature 631, 111–117 (2024).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing