1538 Conservation Biology, Pages 1538?1542 Volume 14, No. 5, October 2000 Amazonian Tree Mortality during the 1997 El Ni?o Drought G. BRUCE WILLIAMSON,*? WILLIAM F. LAURANCE,? ALEXANDRE A. OLIVEIRA,? PATRICIA DELAM?NICA,? CLAUDE GASCON,? THOMAS E. LOVEJOY,? AND LUCIANO POHL? *Department of Biological Sciences, Louisiana State University, Baton Rouge, LA 70803, U.S.A., email btwill@lsu.edu ?Biological Dynamics of Forest Fragments Project, National Institute for Amazonian Research (INPA), C.P 478, Manaus, AM 69011?970, Brazil ?Laborat?rio de Bot?nica, Universidade Paulista, Av. Paulista 900, S?o Paulo, SP 01310?100, Brazil ?Counselor to the Secretary for Biodiversity and Conservation, Smithsonian Institution, Washington, D.C. 20560, U.S.A. Abstract: In 1997, the Amazon Basin experienced an exceptionally severe El Ni?o drought. We assessed ef- fects of this rare event on mortality rates of trees in intact rain forest based on data from permanent plots. Long-term (5- to 13-year) mortality rates averaged only 1.12% per year prior to the drought. During the drought year, annual mortality jumped to 1.91% but abruptly fell back to 1.23% in the year following El Ni?o. Trees dying during the drought did not differ significantly in size or species composition from those that died previously, and there was no detectable effect of soil texture on mortality rates. These results suggest that intact Amazonian rainforests are relatively resistant to severe El Ni?o events. Mortandad de Arboles del Amazonas Durante la Sequ?a de El Ni?o de 1997 Resumen: En 1997 la Cuenca Amaz?nica sufri? una de las sequ?as m?s severas de este siglo, asociada al fen?meno meteorol?gico El Ni?o. Usando parcelas permanentes determinamos la mortandad de ?rboles an- tes, durante, y despu?s de El Ni?o de 1997. La tasa de mortandad de largo plazo (5-13 a?os) promedi? tan solo un 1.12% anual antes de la sequ?a. Durante el a?o de la sequ?a, la mortandad anual aument? a un 1.91%, pero disminuy? abruptamente a un 1.23% al a?o siguiente de El Ni?o. No se encontraron diferencias significativas en el tama?o ni en la composici?n de especies entre los ?rboles que murieron durante la se- qu?a comparados con los ?rboles que hab?an muerto previamente. Tampoco se detect? un efecto de la tex- tura del suelo sobre la tasa de mortandad. Concluimos que los bosques primarios de la Cuenca Amaz?nica son relativamente resistentes a los efectos asociados a eventos severos de El Ni?o. Introduction In 1997?1998, the Amazon Basin experienced a severe drought, associated with what many consider the stron- gest El Ni?o?Southern Oscillation (ENSO) event of this century (Marengo & Tomasella 1998; Hammond & Steege 1998; McPhaden 1999). Severe effects of this El Ni?o have been documented in the eastern Amazon, where previously logged, secondary forests suffered incursions by fire (Cochrane et al. 1999). In contrast, large tracts of primary forests of the central Amazon were not pen- etrated by fire but nevertheless, experienced extreme drought. We present the first record of the magnitude of the drought-induced mortality for these rainforest trees. Paper submitted June 14, 1999; revised manuscript accepted Febru- ary 23, 2000. Conservation Biology Volume 14, No. 5, October 2000 Williamson et al. El Ni?o and Tree Mortality in the Amazon 1539 Methods The study area is located in the central Amazon (lat2 8 20?S, long 59 8 50?W) about 80 km north of Manaus, Brazil, and includes 12 permanent 1-ha plots in the inte- rior ( . 230 m from the nearest clearing) of a large tract of primary rainforest ( . 10 6 ha) that is part of the Biolog- ical Dynamics of Forest Fragments Project (BDFFP; Love- joy & Bierregaard 1990). The plots are in two BDFFP re- serves (1301 and 1501) separated by 12 km and located in continuous old-growth forest. With rainfall averaging over 2000 mm per year, the study area is within the ?Am? classification of K?ppen (1936), indicating a tropi- cal, humid climate with several months of excessive rainfall and 1?2 months of seasonal drought. We compared mortality rates of large ( $ 10 cm in di- ameter at breast height [dbh]) trees at the sites before, during, and after the 1997 ENSO drought. We estimated mortality rates before the drought based on surveys of each plot over one to three intervals each of 3?5 years, the total time varying from 5.2 to 13.7 years per plot. For some plots, these intervals included locally weak ENSO events in 1986?1987 and 1992?1993, neither of which produced a detectable effect on tree mortality in the reserves. We determined mortality rates for 12?16 months during the 1997 drought and for 12?13 months after the drought with identical surveys (Table 1). At each survey, we considered a tree to be dead if it lacked foliage and sap in the secondary vascular tissue. We mea- sured tree diameters at 1.3 m height except for but- tressed trees, which we measured above the buttresses to avoid overestimating diameters (Clark & Clark 1996). All trees were identified to species or morphotype. For each survey, we calculated annualized rates of tree mortality as r 5 1 2 ( N t / N 0 ) 1/ t , where N 0 is the number of trees at the beginning of the interval, t is the time in- terval, and N t is the number of trees of the original N 0 that are alive at the end of the interval (Sheil & May 1996). Because there was no significant relationship be- tween mortality rate and survey interval, no correction of the rates based on interval length was necessary ( p 5 0.84, t 5 0.21, df 5 7, paired t test, for eight plots with both short [11- to 45-month] and long [102- to 119- month] intervals). For reserve 1501, we compared daily rainfall during the 1997 drought to rainfall of the previous 7 years (1990?1996), measured with a cylindrical rain gauge. Because we expected that clay content might influence soil water-holding capacity and, consequently, tree mor- tality during the drought, we also determined clay con- tent (particles , 0.002 mm diameter) of soils for eight of the plots based on 220?260 soil surface samples ( , 20 cm depth) per plot. Results Only 230 mm of rain fell during the 1997 dry season ( June?October), whereas the average for the previous 7 years was 732 mm. Thirty-day moving averages revealed 146 consecutive days (26 June?18 November 1997) below 100 mm rainfall and 62 consecutive days (11 September? 11 November) below 40 mm (Fig. 1). The intensity of the drought increased as the dry season progressed, until it was ended abruptly by heavy rains in mid-November (Fig. 1). This pattern clearly differed from the pre-drought pe- riod, during which the 30-day moving average never fell below 100 mm and the seasonal drought intensified only into September before reversing course. In 1997 there were 623 6 23 (mean 6 SD) trees per plot. Pre-ENSO mortality averaged 1.12% per year or about 7 trees/ha (Table 1). During the drought, tree mortality increased in 11 of the 12 plots, rising signifi- cantly to a mean of 1.91% ( p 5 0.004, paired t test 5 3.56; Table 1). In the post-ENSO year, mortality declined Table 1. Survey details and annualized rates of tree mortality (10-cm diameter at breast height) in the central Amazon before, during, and after the severe 1997 El Ni?o?Southern Oscillation (ENSO) drought. Reserve Both reserves1501 1301 Survey intervals pre-ENSO 1991?1996 1983/87?1996 1997-ENSO 12/96?7/98 4/97?6/98 post-ENSO 7/98?8/99 6/98?7/99 Sample sizes 1-ha plots 3 9 12 trees 1951 5525 7476 Annualized mortality rates (%/year, 6 SD) pre-ENSO 0.94 6 0.28 1.19 6 0.67 1.12 6 0.59 1997-ENSO 1.88 6 0.56 1.92 6 0.69 1.91 6 0.64 post-ENSO 1.64 6 0.87 1.10 6 0.42 1.23 6 0.57 x 1540 El Ni?o and Tree Mortality in the Amazon Williamson et al. Conservation Biology Volume 14, No. 5, October 2000 in all 12 plots to a mean of 1.23%, which was signifi- cantly lower than the ENSO mortality ( p 5 0.001, t 5 4.58) and not different from the pre-ENSO mortality ( p 5 0.68, t 5 0.43). Standard deviations of mortality rates across the 12 plots remained nearly constant across pre-ENSO, ENSO, and post-ENSO intervals (Table 1). Mortality rates were not significantly different between the two reserves (pre-ENSO, p 5 0.55, F 1,10 5 0.38; ENSO, p 5 0.93, F 1,10 5 0.01; post-ENSO, p 5 0.17, F 1,10 5 2.24, one-way analy- sis of variance; Table 1). Mortality rates were not corre- lated significantly between the pre-drought and drought intervals ( r 5 2 0.17, p 5 0.60) or between the pre- drought and post-drought intervals ( r 5 2 0.13, p 5 0.68). There was, however, a significant, positive corre- lation between the drought and post-drought intervals ( r 5 0.65, p 5 0.02; Pearson correlations). There was no difference in the size distributions of trees dying during ( n 5 178) and before ( n 5 776) the drought ( D 5 0.04, p 5 1.00, Kolmogorov-Smirnov test; Fig. 2). Mortality during the drought was roughly pro- portional to species abundances, with the 178 tree deaths spread over 131 species. Most species had no deaths (521 species) or one death (106 species). Only 25 species had more than 1 death, and only 1 had more than 5 deaths. That species, Oenocarpus bacaba Mart., is an abundant subcanopy palm that lost 10 trees and had the only mortality rate (7.0%) significantly above the mean ENSO rate ( p 5 0.006, proportion test). Mortality rates significantly below the mean were undetectable with these sample sizes. Clay content varied from 30% to 66% across the eight plots sampled and was not correlated with mortality rates for any interval (pre-drought, r 5 2 0.21, p 5 0.60; drought, r 5 0.05, p 5 0.90; post-drought, r 5 0.20, p 5 0.63; Pearson correlations). Discussion For many Amazonian trees, the overt symptoms of water stress, such as leaf wilt and abscission, were evident dur- ing the 1997 drought. Tropical rainforests typically suf- fer drought when precipitation falls below 100 mm per month, and severe stress at ,50 mm per month ( Whit- more 1984; Richards 1996), so the recorded 30-day aver- ages of 146 consecutive days below 100 mm and 62 days below 40 mm appear severe and unusual for the central Amazon (Fig. 1). The drought-year mortality increase from 1.12% to 1.91% in our study area is modest in comparison with the findings of the few other studies of ENSO effects on tropical forests. Mortality rates in unburned rainforests in Indonesia were an order of magnitude higher during the ENSO droughts of 1997?1998 (approximately 10%; Kinnaird & O?Brien 1998) and 1982?1983 (14?24%; Leighton & Wirawan 1986). Pre-drought rates were not available for these studies, but baseline mortality is 1?3% per year in comparable Asian rainforests (Phillips et al. 1994). In Panama during the 1982?1983 ENSO, annual tree mortality ($10 cm dbh) averaged 2.75% from 1982 to 1985 compared with 1.98% for the post-drought inter- val of 1985?1990 (Condit et al. 1995). The magnitude of the mortality increase (0.77%) in Panama was similar to what we found in the central Amazon (0.79%), but the increase in Panama was probably diluted by including both drought and nondrought years in the 1982?1985 interval. If the nondrought years had typical mortality Figure 1. Thirty-day moving averages for rainfall dur- ing the 1997 El Ni?o?Southern Oscillation (ENSO) drought (heavy line) and during pre-ENSO years ( fine line). Months run from February (2) of one year to February of the next year. Figure 2. Frequency distributions of diameter at breast height (dbh) of dying trees during the 1997 El Ni?o Southern?Oscillation drought and during the pre-drought years. Conservation Biology Volume 14, No. 5, October 2000 Williamson et al. El Ni?o and Tree Mortality in the Amazon 1541 (1.98%), then the drought year would have had about 4.29% mortality to produce a mean of 2.75% from 1982? 1985. Although 4.29% mortality is much lower than that observed in Indonesia, it is more than double the rate we observed for the central Amazon. Alternatively, one might view our drought-related rise in annual mortality (from 1.12% to 1.91%) as substantial be- cause it represents a relative increase of 71%. Such an in- crease would have a significant effect on the forest if it were maintained for many years. In the year after the drought, however, mortality dropped back to 1.24%, a rate not significantly different from that before the drought. Thus, in the intact forests of our study area, the effects of the unusually severe 1997 ENSO were apparently short- term in nature. Severe El Ni?o droughts have occurred at most four times in the Amazon this century (Marengo & Tomasella 1998); so they are not common events. Other attributes of the 1997 drought suggest that there was little pervasive change in the forest. First, mortality rates in 1997 were not significantly correlated with those during the pre-drought interval. The absence of such cor- relations between successive inventories has been a gen- eral feature of the BDFFP network of permanent plots during non-ENSO intervals, and it remained so in 1997 de- spite an overall increase in mortality. Second, there was little evidence of strong size- or species-specific mortality during the drought. In contrast, during the 1982?1983 drought, tropical trees in Panama and Borneo both exhib- ited striking differences in mortality based on tree size and growth habit, with larger canopy trees being most vulnerable (Leighton & Wirawan 1986; Condit et al. 1995). Finally, mortality during the drought was unrelated to soil clay content in our study area. In Borneo, mortality of large trees during the 1982?1983 ENSO was much higher on well-drained ridges and slopes than on adjacent alluvial soils (Leighton & Wirawan 1986), although similar results were not observed in Panama (Condit et al. 1995). In summary, the only evidence for lingering effects of the 1997 drought in the central Amazon were the exces- sive loss of one palm species and the fact that mortality rates during and after the drought were significantly cor- related, perhaps implying a residual, time-lag effect on tree survivorship. Therefore, we conclude that the 1997 drought had only a modest effect on tree mortality in our study area. Given the exceptional severity of the 1997 ENSO, it appears that continuous, old-growth for- ests of the central Amazon basin are relatively resistant to drought. Drought resistance in rainforest trees is known to be a function of stand age (Kaufman & Uhl 1990) and distance from forest edges (Laurance et al. 1997, 1998). Old-growth forests retain moisture more effectively than do fragmented or secondary forests. Thus, conservation of intact central Amazonian forests seems feasible even in the face of extreme climatic vari- ation, at least where large tracts of unlogged rain forests remain. Acknowledgments We thank the National Science Foundation (DEB 9812375) and the Biological Dynamics of Forest Fragments Project for funding and the Brazilian Fulbright Commission, Fun- da??o Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior, and Louisiana State University for sabbati- cal support for G.B.W. Additional support was provided by World Wildlife Fund-U.S., Instituto Nacional de Pesqui- sas da Amaz?nia, the Smithsonian Institution, the Mac- Arthur Foundation, and the Andrew W. Mellon Founda- tion. We thank SUFRAMA (the Manaus Free Trade Zone Authority) for the access to the sites of the Biological Dy- namics of Forest Fragments Project (BDFFP). O. Phillips, K. Ickes, D. A. Clark, and D. B. Clark provided useful com- ments on the manuscript. This is publication number 301 in the BDFFP technical series. Literature Cited Clark, D. B., and D. A. Clark. 1996. Abundance, growth and mortality of very large trees in Neotropical lowland rain forest. Forest Ecol- ogy and Management 80:235?244. Cochrane, M. A., A. Alencar, M. D. Schulze, C. M. Souza Jr., D. C. Neps- tad, P. Lefebvre, and E. A. Davidson. 1999. Positive feedbacks in the fire dynamic of closed canopy tropical forests. Science 284: 1832?1835. Condit, R., S. P. Hubbell, and R. B. Foster. 1995. Mortality rates of 205 Neotropical tree and shrub species and the impact of a severe drought. Ecological Monographs 65:419?439. Hammond, D. S., and H. ter Steege. 1998. Propensity for fire in Gui- anan rainforests. Conservation Biology 12:944?947. Kaufman, J. B., and C. Uhl. 1990. Interactions of anthropogenic activities, fire, and rain forests in the Amazon Basin. Pages 117?134 in J. Gold- hammer, editor. Fire in the tropical biota. Springer-Verlag, Berlin. Kinnaird, M. F., and T. G. O?Brien. 1998. Ecological effects of wildfire on lowland rainforest in Sumatra. Conservation Biology 12:954?956. K?ppen, W. 1936. Das geographische System der Klimate. Pages C1? C44, Vol. 1, Part C. in W. K?ppen and R. Geiger, editors. Hand- buch der Klimatologie. Gebr?der Borntr?ger, Berlin. Laurance, W. F., S. G. Laurance, L. V. Ferreira, J. M. Rankin-de Merona, C. Gascon, and T. E. Lovejoy. 1997. Biomass collapse in Amazonian forest fragments. Science 278:1117?1118. Laurance, W. F., L. V. Ferreira, J. M. Rankin-de Merona, and S. G. Lau- rance. 1998. Rain forest fragmentation and the dynamics of Amazo- nian tree communities. Ecology 79:2032?2040. Leighton, M., and N. Wirawan. 1986. Catastrophic drought and fire in Borneo tropical rain forest associated with the 1982?83 El Ni?o Southern Oscillation event. Pages 75?102 in G. T. Prance, editor. Tropical rain forests and the world atmosphere. Symposium 10. American Association for the Advancement of Science, Washing- ton, D.C. Lovejoy, T. E., and R. O. Bierregaard Jr. 1990. Central Amazonian for- ests and the minimum critical size of ecosystems project. Pages 60? 71 in A. H. Gentry, editor. Four Neotropical forests. Yale University Press, New Haven. Marengo, J. A., and J. Tomasella. 1998. Trends in streamflow and rain- fall in tropical South America: Amazonia, eastern Brazil, and north- western Peru. Journal of Geophysical Research 103:1775?1783. McPhaden, M. J. 1999. The child prodigy of 1997?98. Nature 398: 559?562. Phillips, O. L., P. Hall, A. H. Gentry, S. A. Sawyer, and R. V?squez. 1994. Dynamics and species richness of tropical rain forests. Pro- 1542 El Ni?o and Tree Mortality in the Amazon Williamson et al. Conservation Biology Volume 14, No. 5, October 2000 ceedings of the National Academy of Sciences of the United States of America 91:2805?2809. Richards, P. W. 1996. The tropical rain forest: an ecological study. 2nd edition. Cambridge University Press, Cambridge, United Kingdom. Sheil, D. D., and R. M. May. 1996. Mortality and recruitment rate evalua- tion in heterogeneous tropical forests. Journal of Ecology 84:91?100. Whitmore, T. C. 1984. Tropical rain forests of the far east. 2nd edition. Clarendon Press, Oxford, United Kingdom.