Skip to main content

Thank you for visiting nature.com. 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.

Global patterns of drought recovery

Abstract

Drought, a recurring phenomenon with major impacts on both human and natural systems1,2,3, is the most widespread climatic extreme that negatively affects the land carbon sink2,4. Although twentieth-century trends in drought regimes are ambiguous5,6,7, across many regions more frequent and severe droughts are expected in the twenty-first century3,7,8,9. Recovery time—how long an ecosystem requires to revert to its pre-drought functional state—is a critical metric of drought impact. Yet the factors influencing drought recovery and its spatiotemporal patterns at the global scale are largely unknown. Here we analyse three independent datasets of gross primary productivity and show that, across diverse ecosystems, drought recovery times are strongly associated with climate and carbon cycle dynamics, with biodiversity and CO2 fertilization as secondary factors. Our analysis also provides two key insights into the spatiotemporal patterns of drought recovery time: first, that recovery is longest in the tropics and high northern latitudes (both vulnerable areas of Earth’s climate system10) and second, that drought impacts11 (assessed using the area of ecosystems actively recovering and time to recovery) have increased over the twentieth century. If droughts become more frequent, as expected, the time between droughts may become shorter than drought recovery time, leading to permanently damaged ecosystems and widespread degradation of the land carbon sink.

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

Access options

Rent or buy this article

Prices vary by article type

from$1.95

to$39.95

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

Figure 1: Response functions for drought recovery time.
Figure 2: Spatial pattern of drought recovery time.
Figure 3: Decadal changes in drought recovery.

References

  1. Haines, A., Kovats, R. S., Campbell-Lendrum, D. & Corvalán, C. Climate change and human health: impacts, vulnerability and public health. Public Health 120, 585–596 (2006)

    Article  CAS  PubMed  Google Scholar 

  2. Schwalm, C. R. et al. Reduction in carbon uptake during turn of the century drought in western North America. Nat. Geosci. 5, 551–556 (2012)

    Article  ADS  CAS  Google Scholar 

  3. Touma, D., Ashfaq, M., Nayak, M. A., Kao, S. C. & Diffenbaugh, N. S. A multi-model and multi-index evaluation of drought characteristics in the 21st century. J. Hydrol. 526, 196–207 (2015)

    Article  ADS  Google Scholar 

  4. Reichstein, M. et al. Climate extremes and the carbon cycle. Nature 500, 287–295 (2013)

    Article  ADS  CAS  PubMed  Google Scholar 

  5. Trenberth, K. E. et al. Global warming and changes in drought. Nat. Clim. Chang. 4, 17–22 (2014)

    Article  ADS  Google Scholar 

  6. Sheffield, J., Wood, E. F. & Roderick, M. L. Little change in global drought over the past 60 years. Nature 491, 435–438 (2012)

    Article  ADS  CAS  PubMed  Google Scholar 

  7. Seneviratne, S. I. et al. in Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation (eds Field, C. B. et al.) 109–230 (Special Report of Working Groups I and II of the Intergovernmental Panel on Climate Change, Cambridge Univ. Press, 2012)

  8. Alexander, L. V. et al. Climate extremes: progress and future directions. Int. J. Climatol. 29, 317–319 (2009)

    Article  Google Scholar 

  9. Cook, B. I., Ault, T. R. & Smerdon, J. E. Unprecedented 21st century drought risk in the American Southwest and Central Plains. Sci. Adv. 1, e1400082 (2015)

    Article  ADS  PubMed  PubMed Central  Google Scholar 

  10. Lenton, T. M. et al. Tipping elements in the Earth’s climate system. Proc. Natl Acad. Sci. USA 105, 1786–1793 (2008)

    Article  ADS  MATH  PubMed  PubMed Central  Google Scholar 

  11. Mishra, A. K. & Singh, V. P. A review of drought concepts. J. Hydrol. 391, 202–216 (2010)

    Article  ADS  Google Scholar 

  12. DeChant, C. M. & Moradkhani, H. Analyzing the sensitivity of drought recovery forecasts to land surface initial conditions. J. Hydrol. 526, 89–100 (2014)

    Article  Google Scholar 

  13. Pan, M., Yuan, X. & Wood, E. F. A probabilistic framework for assessing drought recovery. Geophys. Res. Lett. 40, 3637–3642 (2013)

    Article  ADS  Google Scholar 

  14. Martorell, S., Diaz-Espejo, A., Medrano, H., Ball, M. C. & Choat, B. Rapid hydraulic recovery in Eucalyptus pauciflora after drought: linkages between stem hydraulics and leaf gas exchange. Plant Cell Environ. 37, 617–626 (2014)

    Article  CAS  PubMed  Google Scholar 

  15. Secchi, F. & Zwieniecki, M. A. Down-regulation of plasma intrinsic protein1 aquaporin in poplar trees is detrimental to recovery from embolism. Plant Physiol. 164, 1789–1799 (2014)

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  16. Anderegg, W. R. L. et al. Pervasive drought legacies in forest ecosystems and their implications for carbon cycle models. Science 349, 528–532 (2015)

    Article  ADS  CAS  PubMed  Google Scholar 

  17. Luo, Y., Keenan, T. F. & Smith, M. Predictability of the terrestrial carbon cycle. Glob. Change Biol. 21, 1737–1751 (2015)

    Article  ADS  Google Scholar 

  18. Nepstad, D. C., Stickler, C. M., Soares-Filho, B. & Merry, F. Interactions among Amazon land use, forests and climate: prospects for a near-term forest tipping point. Phil. Trans. R. Soc. Lond. B 363, 1737–1746 (2008)

    Article  Google Scholar 

  19. Banerjee, O., Bark, R., Connor, J. & Crossman, N. D. An ecosystem services approach to estimating economic losses associated with drought. Ecol. Econ. 91, 19–27 (2013)

    Article  Google Scholar 

  20. Zhao, M. & Running, S. W. Drought-induced reduction in global terrestrial net primary production from 2000 through 2009. Science 329, 940–943 (2010)

    Article  ADS  CAS  PubMed  Google Scholar 

  21. Jung, M. et al. Global patterns of land-atmosphere fluxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and meteorological observations. J. Geophys. Res. Biogeosci. 116, G00J07 (2011)

    Article  Google Scholar 

  22. Huntzinger, D. N. et al. The North American Carbon Program Multi-scale synthesis and Terrestrial Model Intercomparison Project—Part 1: Overview and experimental design. Geosci. Model Dev. 6, 2121–2133 (2013)

    Article  ADS  Google Scholar 

  23. Beer, C. et al. Terrestrial gross carbon dioxide uptake: global distribution and covariation with climate. Science 329, 834–838 (2010)

    Article  ADS  CAS  PubMed  Google Scholar 

  24. Beguería, S., Vicente-Serrano, S. M., Reig, F. & Latorre, B. Standardized precipitation evapotranspiration index (SPEI) revisited: parameter fitting, evapotranspiration models, tools, datasets and drought monitoring. Int. J. Climatol. 34, 3001–3023 (2014)

    Article  Google Scholar 

  25. Tobner, C. M., Paquette, A., Reich, P. B., Gravel, D. & Messier, C. Advancing biodiversity–ecosystem functioning science using high-density tree-based experiments over functional diversity gradients. Oecologia 174, 609–621 (2014)

    Article  ADS  PubMed  Google Scholar 

  26. Grossiord, C. et al. Tree diversity does not always improve resistance of forest ecosystems to drought. Proc. Natl Acad. Sci. USA 111, 14812–14815 (2014)

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  27. Grossiord, C., Granier, A., Gessler, A., Jucker, T. & Bonal, D. Does drought influence the relationship between biodiversity and ecosystem functioning in boreal forests? Ecosystems 17, 394–404 (2014)

    Article  CAS  Google Scholar 

  28. Engelbrecht, B. M. et al. Drought sensitivity shapes species distribution patterns in tropical forests. Nature 447, 80–82 (2007)

    Article  ADS  CAS  PubMed  Google Scholar 

  29. Hacke, U. G., Stiller, V., Sperry, J. S., Pittermann, J. & McCulloh, K. A. Cavitation fatigue. Embolism and refilling cycles can weaken the cavitation resistance of xylem. Plant Physiol. 125, 779–786 (2001)

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. Saatchi, S. et al. Persistent effects of a severe drought on Amazonian forest canopy. Proc. Natl Acad. Sci. USA 110, 565–570 (2013)

    Article  ADS  PubMed  Google Scholar 

  31. Harris, I., Jones, P. D., Osborn, T. J. & Lister, D. H. Updated high-resolution grids of monthly climatic observations—the CRU TS3.10 dataset. Int. J. Climatol. 34, 623–642 (2014)

    Article  Google Scholar 

  32. Kanamitsu, M. et al. NCEP-DOE AMIP-II Reanalysis (R-2). Bull. Am. Meteorol. Soc. 83, 1631–1643 (2002)

    Article  ADS  Google Scholar 

  33. Wei, Y. et al. The North American Carbon Program Multi-scale Synthesis and Terrestrial Model Intercomparison Project—Part 2: Environmental driver data. Geosci. Model Dev. 7, 2875–2893 (2014)

    Article  ADS  Google Scholar 

  34. Mao, J., Thornton, P. E., Shi, X., Zhao, M. & Post, W. M. Remote sensing evaluation of CLM4 GPP for the period 2000-09. J. Clim. 25, 5327–5342 (2012)

    Article  ADS  Google Scholar 

  35. Lei, H. et al. Sensitivity of global terrestrial gross primary production to hydrologic states simulated by the Community Land Model using two runoff parameterizations. J. Adv. Model. Earth Syst. 6, 658–679 (2014)

    Article  ADS  Google Scholar 

  36. Tian, H. Q. et al. Century-scale response of ecosystem carbon storage to multifactorial global change in the Southern United States. Ecosystems 15, 674–694 (2012)

    Article  CAS  Google Scholar 

  37. Jain, A. K. & Yang, X. Modeling the effects of two different land cover change data sets on the carbon stocks of plants and soils in concert with CO2 and climate change. Glob. Biogeochem. Cycles 19, GB2015 (2005)

  38. Hayes, D. J. et al. Is the northern high-latitude land-based CO2 sink weakening? Glob. Biogeochem. Cycles 25, GB3018 (2011)

  39. Liaw, A. & Wiener, M. Classification and regression by randomForest. R News 2, 18–22 (2002)

    Google Scholar 

  40. Ellis, E. C., Antill, E. C. & Kreft, H. All is not loss: plant biodiversity in the Anthropocene. PLoS One 7, e30535 (2012)

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  41. Hurtt, G. C. et al. The underpinnings of land-use history: three centuries of global gridded land-use transitions, wood-harvest activity, and resulting secondary lands. Glob. Change Biol. 12, 1208–1229 (2006)

    Article  ADS  Google Scholar 

  42. Meiyappan, P. & Jain, A. K. Three distinct global estimates of historical land-cover change and land-use conversions for over 200 years. Front. Earth Sci. 6, 122–139 (2012)

    Article  ADS  Google Scholar 

  43. Archibald, S., Lehmann, C. E., Gómez-Dans, J. L. & Bradstock, R. A. Defining pyromes and global syndromes of fire regimes. Proc. Natl Acad. Sci. USA 110, 6442–6447 (2013)

    Article  ADS  PubMed  PubMed Central  Google Scholar 

  44. Giglio, L., Randerson, J. T. & Werf, G. R. Analysis of daily, monthly, and annual burned area using the fourth-generation global fire emissions database (GFED4). J. Geophys. Res. Biogeosci. 118, 317–328 (2013)

    Article  Google Scholar 

Download references

Acknowledgements

Funding for this research was provided by the National Science Foundation (NSF) grant DEB EF-1340270. C.R.S. was also supported by National Aeronautics and Space Administration (NASA) grants NNX12AK12G, NNX12AP74G, NNX10AG01A and NNX11AO08A. J.B.F. contributed to this paper from the Jet Propulsion Laboratory, California Institute of Technology, under a contract with NASA. Government sponsorship acknowledged. Support was provided to J.B.F. by NASA grants NNN13D504T (CARBON), NNN13D202T (INCA), and NNN13D503T (SUSMAP). Funding for the MsTMIP activity was provided through NASA grant NNX10AG01A. Data management support for preparing, documenting and distributing model driver and output data was performed by the Modeling and Synthesis Thematic Data Center at Oak Ridge National Laboratory (http://nacp.ornl.gov), with funding through NASA grant NNH10AN681. Finalized MsTMIP datasets are archived at the Oak Ridge National Laboratory Distributed Active Archive Center (http://daac.ornl.gov). This is MsTMIP contribution number 10.

Author information

Authors and Affiliations

Authors

Contributions

C.R.S. and W.R.L.A. designed the analysis. C.R.S. carried out the analysis and wrote the manuscript with contributions from all authors. W.R.L.A., A.M.M. and J.B.F. contributed to the framing of the paper. D.N.H. is the overall lead of the MsTMIP effort; R.C., J.B.F., A.M.M., K.S., C.R.S., Y.F. and Y.W. serve as the MsTMIP core team. D.H., M.H., A.J. and H.T. contributed to MsTMIP results.

Corresponding author

Correspondence to Christopher R. Schwalm.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Additional information

Reviewer Information Nature thanks M. Migliavacca and the other anonymous reviewer(s) for their contribution to the peer review of this work.

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 Figure 1 Variable importance scores.

a, Covariates by rank of variable importance. b, Covariates by value of relative variable importance. Scores allow the skill of a trained Random Forest to be decomposed and mapped to individual covariates. Box plots are based on all possible combinations of GPP dataset and integration time. The box range covers the 25th to 75th percentiles; whiskers extend from about the 0.50th to 99.50th percentiles; outlying values are given by open circles. Red crosses give the composite value from the weighted median with weights given by the length, in years, of each GPP dataset. Both panels are sorted by median value from smallest rank (highest importance) to largest rank (lowest importance). Variable importance is calculated as the total decrease in residual sum of squares, averaged over all trees, from splitting on the target covariate (see Methods). CO2 indicates CO2 fertilization (p.p.m.), LULCC is land-use and land-cover change relative to the pre-industrial baseline, MAP is mean annual precipitation 1971–2000, and MAT is mean annual temperature 1971–2000.

Extended Data Figure 2 Spatial gradients of recovery time.

a, Pre-drought GPP. b, GPP amplitude. c, MAP. d, MAT. Mapped values for a and b are based on the mean across all events (all GPP datasets, all integration times, and all time steps) by grid cell. For c and d the covariates from training the Random Forests algorithm (see Methods) are used directly. Recovery time is then projected spatially by matching response function values with each mapped factor by grid cell. To highlight spatial patterns values are expressed as difference in recovery time (∆t) relative to median recovery time by factor; red (blue) values indicate where a given factor is spatially associated with longer (shorter) drought recovery times. White areas are water; grey areas are barren or did not experience any relevant drought events.

Extended Data Figure 3 Enviroclimatic spatial gradients associated with recovery time.

a, Mask of normalized biodiversity ≤ 0.2; these areas show an increase in recovery time with decreasing biodiversity across space. b, MAP. c, MAT. d, LULCC. Values in parentheses give areal extent as percentage. Aggregated types include Barren (tundra, desert, polar desert, rock, ice), Tree/Grass (shrublands, savannah, riparian/riverine systems) and Forest (all forest types).

Extended Data Figure 4 Recovery time for moderate droughts.

a, Recovery time as a function of drought severity (mean SPEI by drought event) and return interval—calculated as the number of months between successive drought events for a given pixel but excluding the first drought event and droughts where no recovery is observed. This is a representative subsample of 5,000 events from MsTMIP using 1-month SPEI as shown. The grey surface is a smoothed surface to aid visual interpretation. b, Mean recovery time by 50 equidistant bins of return time and drought severity. c, Number of drought and recovery events for each bin in b; note log scale.

Extended Data Figure 5 Response functions of recovery time relative to burned area.

MsTMIP and upscaled FLUXNET, with 1-month SPEI integration time, for 1997–2010—overlap with GFED data44 used to calculate burned area—shown. The trained Random Forests algorithm used both fire regime43 and burned area44 in addition to the standard set of covariates. Burned area is highly skewed; its 90th percentile is 0.035, such that the data support is concentrated at smaller burned area fractions. Burned area was used in training the Random Forests algorithm, given by its maximum value, for a given pixel, during drought and subsequent recovery. Both fire regime and burned area have low importance scores and are ranked 12 and 13.5, respectively, out of 16 covariates.

Extended Data Figure 6 Spatial patterns of recovery time.

Panels show mean recovery time by grid cell for all combinations of GPP dataset and integration time. Colours denote recovery time as shown. Inset values show number of drought and recovery events, n. White areas are water, barren, or did not experience any relevant drought events.

Extended Data Figure 7 Response function similarity.

Panels a to n show similarity score (correlation) across all possible combinations of GPP dataset and integration time by covariate. Each panel gives the covariate name and median of all off-diagonal scores. Integration time and GPP dataset are abbreviated in the y-axis labels as number–letter couplets (the number indicates the integration time in months, the letter indicates the GPP dataset (M, MsTMIP; F, upscaled FLUXNET; R, remotely sensed MODIS); for example, 12R indicates MODIS using a 12-month SPEI integration time. Panel o gives the cumulative distribution function of all (n = 924) off-diagonal scores. Thin horizontal red lines show, from bottom to top, the 10th, 25th and 50th percentiles. Over 75% of all similarity scores are greater than 0.70, with an overall median of 0.90; only 7% of all scores are negative.

Extended Data Figure 8 Response function similarity.

Each colourmap pixel shows the similarity score (correlation) between the mirroring and fixed window approaches to calculate the pre-drought baseline across the common time frame for all three datasets (2000 to 2008). a, 3-month fixed window (overall similarity median of 0.97). b, 6-month fixed window (overall similarity median of 0.96). Integration time and GPP dataset are abbreviated as number–letter couplets as in Extended Data Fig. 7.

Extended Data Table 1 Skill of the trained Random Forests algorithm

PowerPoint slides

Source data

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Schwalm, C., Anderegg, W., Michalak, A. et al. Global patterns of drought recovery. Nature 548, 202–205 (2017). https://doi.org/10.1038/nature23021

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nature23021

This article is cited by

Comments

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.

Search

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