Biol Fertil Soils (2002) 36:102–108DOI 10.1007/s00374-002-0517-xAbstract The eastern Amazon Basin may become drieras a result of less regional recirculation of water in alargely deforested landscape and because of increasedfrequency and intensity of El Niño events induced byglobal warming. Drier conditions may affect severalplant and soil microbial processes, including soil emis-sions of CO2, CH4, NO, and N2O. We report here unan-ticipated results of a pilot study that was initiated to testthe feasibility of a larger-scale throughfall exclusion ex-periment. In particular, soil drying caused a switch fromnet consumption of atmospheric CH4by soils in the con-trol plot to net CH4emission from soils in the experi-mentally dried plot. This result is surprising because pro-duction of CH4requires anaerobic microsites, which areuncommon in dry soil. A plausible explanation for in-creased CH4emissions in the dried plot is that dry soilconditions favor termite activity and increased coarseroot mortality provides them with a substrate. Anothersurprise was that both NO and N2O fluxes were elevatedseveral years after initiation of the drying experiment.Apparently, a pulse of N availability caused by experi-mental drying persisted for at least 3 years. As expected,CO2emissions were lower in the dried plots, which isconsistent with lower rates of root growth observed inroot in-growth cores placed in the dried plots. Morework is needed to test these explanations and to confirmthese phenomena, but these results demonstrate thatchanges in climate could have unanticipated effects onbiogeochemical processes in soils that we do not ade-quately understand.Keywords Carbon dioxide · Methane · Nitric oxide · Nitrous oxide · TermitesIntroductionLand-use change in the Amazon region is predicted toresult in drier conditions because of less regional recir-culation of water between the deforested biosphere andthe atmosphere (Costa and Foley 2000; Nobre et al.1991; Shukla et al. 1990). Global warming may also in-crease the frequency and intensity of El Niño events(Timmermann et al. 1999; Trenberth and Hoar 1997),which cause severe drought in the eastern Amazon Basin(Nepstad et al. 1999). Reduced precipitation in the Ama-zon Basin may have important feedback effects on cli-mate change by altering soil emissions of carbon dioxide(CO2), methane (CH4), nitrous oxide (N2O), and nitricoxide (NO), which are strongly dependent on soil watercontent (Firestone and Davidson 1989; Linn and Doran1984; Whalen and Reeburgh 1996). The direct effects ofland-use change in the Amazon region on soil emissionsof these gases have been studied (Davidson et al. 2000a;Fearnside 1996; Matson et al. 1990; Steudler et al. 1996;Verchot et al. 1999, 2000), but the possible effects ofchanges in precipitation have not been addressed.We report here the results of a pilot study conductedin the eastern Amazon region, where throughfall was ex-cluded from a 10×10-m plot by a subcanopy roof for4 years. We hypothesized that throughfall exclusionwould cause root mortality, a decrease in root produc-tion, and, therefore, a decrease in soil emissions of CO2.We hypothesized that drier soil conditions would favorNO production over N2O production (Firestone andJ.H. CattânioInstituto de Pesquisa Ambiental da Amazônia, Av. Nazaré,669 Belém, PA 66035–170, BrazilE.A. Davidson (✉) · D.C. NepstadThe Woods Hole Research Center, P.O. Box 296, Woods Hole,MA 02543, USAe-mail: [email protected]: +1-508-5409700L.V. VerchotInternational Centre for Research in Agroforestry, Nairobi, KenyaI.L. AckermanDepartment of Crop and Soil Sciences, Cornell University, Ithaca,NY 14853, USAPresent address:J.H. Cattânio, Institute for Crop Production in the Tropics, Grisebachstrasse 6, 37077 Goettingen, GermanyORIGINAL PAPERJosé H. Cattânio · Eric A. DavidsonDaniel C. Nepstad · Louis V. Verchot · Ilse L. AckermanUnexpected results of a pilot throughfall exclusion experiment on soilemissions of CO2, CH4, N2O, and NO in eastern AmazoniaReceived: 25 May 2001 / Accepted: 30 May 2002 / Published online: 25 July 2002© Springer-Verlag 2002103Davidson 1989) and would facilitate increased diffusionof atmospheric CH4into the soils, because water oftenlimits rates of CH4consumption (Dörr et al. 1993;Striegl 1993). Although this was a pilot study designedto test a concept as a basis for a larger study that is nowunderway (Nepstad et al., in press), the surprising resultsmerit consideration because they challenge, in part, ourunderstanding of factors that control emissions of thesegases from soils.Materials and methodsSite descriptionThe study was conducted in an old growth forest stand at FazendaVitória, located 6.5 km northwest of the town of Paragominas,Pará, Brazil, in eastern Amazonia (2°59′ S, 47°31′ W), an area ofextensive land-use change since the 1960s (Nepstad et al. 1991).Average annual rainfall is 1,800 mm and is highly seasonal, with<250 mm falling from July to November (Jipp et al. 1998). Theforest maintains an evergreen canopy throughout the dry seasonby extracting water from deep in the soil profile, where roots havebeen observed at 18-m depth (Nepstad et al. 1994; Jipp et al.1998). The soils were classified by Sombroek (1966) as KaoliniticYellow Latosols (Haplustox, according to USDA Soil Taxonomy).Soil temperature at 10-cm depth under the forest canopy only var-ies between 22°C and 25°C both seasonally and diurnally (Davidsonet al. 2000a).Throughfall exclusionThis experiment was begun in 1993 as a pilot study to test the fea-sibility of conducting a similar experiment on a larger scale. A10×10-m plot containing mature trees was trenched to 0.5-m deptharound the perimeter. A sub-canopy cover was constructed ofsheets of asbestos roofing, 2×0.5 m, placed in rows at about a 2-mheight in the center of the plot and draining into wooden gutters atthe plot edges. All four sides were open, allowing good ventila-tion. Because of the dense forest canopy cover, only transient sun-flecks hit the roof, and no heating was observed. Soil temperaturebelow the exclosure did not differ from soil temperature outsidethe exclosure. Openings were also made in the roofing to permittree trunks to penetrate. Stem flow collectors made of rubber wereplaced around the tree trunks and the water was emptied onto theroof. Gutters carried the captured throughfall off of the plot to an80-l barrel at a distance of 4.0 m from the plot. Litter that fell ontop of the