Fingerprinting the impacts of global change on tropical forestsIntroductionGeneral frameworkEnvironmental changes over the past three decadesTemperaturePrecipitationSolar radiationClimatic extremes/El NinÄo Southern OscillationCarbon dioxideNutrient depositionsOzone/acidic depositionsHuntingLand-use changeLianasHow might tropical forests change?TemperatureTemperature effects on photosynthesisTemperature effects on respirationTemperature effects on soil nutrient availabilityTemperature and ontogenetic developmentPrecipitationSolar radiationClimatic extremes/El NinÄo Southern OscillationCarbon dioxideCarbon dioxide effects on light-use efficiencyCarbon dioxide effects on water-use efficiencyCarbon dioxide effects on nutrient-use efficiencyNutrient depositionsOzone/acidic depositionsHuntingLand-use changeLianasPredictions of changes in tropical forestsTemperaturePrecipitationSolar radiationClimatic extremes/El NinÄo Southern OscillationCarbon dioxideNutrient depositionOzone/acidic depositionsHuntingLand-use changeLianasDiscussionGeneral patternsIdentifying uncertaintiesPotential drivers and forest plot dataThe futureAcknowledgementsREFERENCESGLOSSARYPublished online 11 February 2004Fingerprinting the impacts of global changeon tropical forestsSimon L. Lewis1,2*, Yadvinder Malhi2and Oliver L. Phillips11Earth and Biosphere Institute, Geography, University of Leeds, Leeds LS2 9JT, UK2School of GeoSciences, University of Edinburgh, Edinburgh EH9 3JU, UKRecent observations of widespread changes in mature tropical forests such as increasing tree growth,recruitment and mortality rates and increasing above-ground biomass suggest that ‘global change’ agentsmay be causing predictable changes in tropical forests. However, consensus over both the robustness ofthese changes and the environmental drivers that may be causing them is yet to emerge. This paperfocuses on the second part of this debate. We review (i) the evidence that the physical, chemical andbiological environment that tropical trees grow in has been altered over recent decades across large areasof the tropics, and (ii) the theoretical, experimental and observational evidence regarding the most likelyeffects of each of these changes on tropical forests. Ten potential widespread drivers of environmentalchange were identified: temperature, precipitation, solar radiation, climatic extremes (including El Nin˜o–Southern Oscillation events), atmospheric CO2concentrations, nutrient deposition, O3/acid depositions,hunting, land-use change and increasing liana numbers. We note that each of these environmental changesis expected to leave a unique ‘fingerprint’ in tropical forests, as drivers directly force different processes,have different distributions in space and time and may affect some forests more than others (e.g. dependingon soil fertility). Thus, in the third part of the paper we present testable a priori predictions of forestresponses to assist ecologists in attributing particular changes in forests to particular causes across multipledatasets. Finally, we discuss how these drivers may change in the future and the possible consequencesfor tropical forests.Keywords: review; carbon; permanent sample plot; forest dynamics; mortality; recruitment1. INTRODUCTIONOver the past century virtually all ecosystems on Earthhave come under increasing human influence. This hasbeen through direct contact and transformation (e.g. forfarming, through hunting or the use of fire), the effects ofhabitat fragmentation, the production of pollutants (e.g.tropospheric O3) or the substantial alteration of majorbiogeochemical cycles, such as the global C, water and Ncycles (Vitousek et al. 1997; Fowler et al. 1999; Prenticeet al. 2001; Ramanathan et al. 2001; Galloway et al. 2002;Barlow & Peres 2004; Laurance 2004; Malhi & Wright2004). However, for tropical forests that are far from mostdirect human impacts, the question as to whether theseecosystems have been substantially altered and what maybe causing these changes is actively debated (Clark 2004;Chambers & Silver 2004; Lewis et al. 2004; Phillips etal. 2004).Whether tropical forests are showing widespread secularchanges in dynamics, and why, is of broad interest as trop-ical forests store ca. 40% of the C residing in terrestrialvegetation and annually process about six times as muchC through photosynthesis as humans release to theatmosphere through fossil fuel combustion (Malhi &*Author and address for correspondence: Earth and Biosphere Institute,Geography, University of Leeds, Leeds LS2 9JT, UK ([email protected]).One contribution of 17 to a Theme Issue ‘Tropical forests and globalatmospheric change’.Phil. Trans. R. Soc. Lond. B (2004) 359, 437–462 437  2004 The Royal SocietyDOI 10.1098/rstb.2003.1432Grace 2000; Malhi et al. 2002a). In addition, tropical for-ests harbour more than 50% of the world’s species(Heywood 1995). Thus, relatively small yet consistentchanges within remaining tropical forests as a biome couldhave global consequences for the climate, biodiversity, theglobal C cycle, the rate of climate change and hencehuman welfare.Two widespread changes in tropical forests havereceived attention: increases in tree stem turnover andincreases in the above-ground biomass of forest stands(Phillips & Gentry 1994; Phillips et al. 1998). A case canbe made that these trends have not been caused by wide-spread changes in environmental drivers, but are the pro-duct of compilations of disparate datasets containingmethodological errors coupled with the use of inappropri-ate statistical techniques (Sheil 1995; Sheil & May 1996;Condit 1997; Clark 2002). In response, much work hasgone into expanding the datasets, particularly acrossSouth America (Malhi et al. 2002b), and addressing these‘artefactual’ explanations of the trends (Phillips 1995,1996; Phillips et al. 2002a, 2004; Baker et al. 2004; Lewiset al. 2004). These newer analyses confirm the qualitativefindings of the initial two analyses, and have shownadditional trends in South American tropical forests:(i) that stem turnover is rising owing to simultaneousincreases in both recruitment and mortality;(ii) that recruitment rates are greater than mortalityrates, causing a net increase in stem density;438 S. L. Lewis and others Fingerprinting changes in tropical forests(iii) that stand-level growth and mortality, in terms ofbiomass, have both increased; and(iv) that growth rates are greater than mortality rates,leading to the