JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 101, NO. D19, PAGES 23,699-23,706, OCTOBER 30, 1996 Effects of moistureand burning on soil-atmosphere exchange of trace carbon gases in a southern African savanna Richard G. Zepp, William L. Miller • and Roger A. Burke Environmental Research Laboratory, U.S. Environmental Protection Agency, Athens, Georgia Dirk A. B. Parsons and Mary C. Scholes Department ofBotany, University ofWitwatersrand, Johannesburg, Republic ofSouth Africa Abstract.Soil fluxes of carbon dioxide (CO2), methane (CH4), and carbon monoxide (CO) were measured during aperiod ofextreme drought atsemi-arid savanna sites located inthe Kruger National Park (KNP), South Africa, as part ofthe SAFARI-92 experiments (Sept., 1992).Soil respiration inthis savanna was little affected by burning, but was strongly stimulated by addition of moisture. Mean soil respiration from the dry soil was 0.4gC m '2 d 4 inopen savanna plots that had been burned biennially and 0.5g C m '2 d 4 in woody savanna plots. A light natural rain (about 0.6 mm) increased the CO,• flux inthe open savanna sites by5-fold but the effect was short-lived. A simulated heavy rain (25mm ofadded distilled water) increased CO2 fluxes by over an order of magnitude inboth burned and control sites and the emissions remained over 5 times pre-wetting values during aweek ofdrying. Over 65% ofour measurements indicated no significant soil- atmosphere methane exchange; most of the few non-zero measurements indicated a small (<1 mg CH4-C m"• d 4)flux of methane to the atmosphere. Soil-atmosphere CH4 exchange was not signifi- cantly affected by either burning the grass layer or by the addition ofdistilled water to the soil. The net soil COfluxes, which generally increased with increasing soil temperature, were positive up to 356 x 109 molecules cm '2 s 4 with an average of8.8x 10 •ø molecules cm '2 s 4 for the untreated open savanna plots. After burning, the fluxes rose by over an order ofmagnitude but dropped back to pre- burn levels within afew days. Observed COfluxes were higher than those previously reported for southern Africa savannas during non-drought conditions. Added moisture had little effect on CO fluxes during the 3-week period of SAFARI-92. 1. Introduction Biomass burning represents an immediate and significant direct emission of trace gases to theatmosphere [Andreae, 1991;Crutzen andAndreae, 1990]. Firemay also have a longer term effect on soil tracegas fluxes by alteration of nutrient concentrations andcycling [Anderson and Levine,1987; Anderson et al., 1988;Levine et al. 1991 ], addition of root litter to soil, as well as changes in soil temperature andmoisture, especially when accompanied byland use change such asconversion of savanna to pasture [Johansson et ai., 1988; Sanhueza, 1991; Menaut et al., 1993; Sanhueza et al., 1994a, b; Sanhueza and Santana,1994] or intense agriculture [Wooruer, 1993]. The International Geosphere-Biosphere Programme (IGBP) projectson Global Change and Terrestrial Ecosystems and International Global Atmospheric Chemistry have both recognized the potentially important post-burn alteration of soil biogeochemistry and resulting effects on ecosystem structure andfunction [IGBP, 1990]. Notably, there are very fewstudies of such effects in tropical savannas [Scmhueza, 1991; Sanhueza et ai., 1994a,b; Sanhueza and Santana, 1994; Johansson et ai., 1988;Hao et al., 1988],although most fire activity occurs in theseregions. Indeed, few data generally are available concerning soil fluxes in tropical savannas [e.g., Sanhueza, •National Research Council Associate; now at Department of Oceanography, Dalhousie University, Halifax,Nova Scotia. Copyright 1996by theAmerican Geophysical Union. Paper number 95JD01371. 0148-0227/96/95 JD-01371 $09.00 1991; Sanhueza et aL, 1994a, b, Keller and Reiners,1994], evenfor soil respiration [Raich and Schlesinger, 1992; Scholes and Walker, 1993]. Globally, CO•emission tothe atmosphere from soil respiration has been estimated tobe 68 + 4 PgC yr 4 [Raich andSchlesinger, 1992]. Changes inthe biological activity in soils due to biomass burning and other perturbations may impact the atmospheric CO• budget, at least on a localscale. Following a fire, CO2 is released through soil respiration and decomposition of nonliving organic matter produced inthe fwe. Hao etal. [1990] have estimated that postburning release of CO•-C from decomposition of nonliving organic matter produced bythefires in nonfallow tropical forests can exceed direct emissions by an order of magnitude within a decade of theburn. Carbon dioxide is reincorporated into biomass during the post-burn recovery period, thusoffsetting itsrelease by soil respiration. During the period in which soil respiration exceeds photosynthetic uptake, however, there isa net emission of COy It is generally assumed that most of the carbonlost to the atmosphere from tropical savanna burning is accumulated again in subsequent seasons via regrowth of vegetation [Houghton, 1991 ]. Several studies, however, have demonstrated that repetitive burning can reduce the amount of soilorganic •natter (SOM) in southern Africansoils [White and Grossman, 1972; Jones et ai., 1990]. A recent study indicates that the distribution of SOMamong labile and passive fractions of basaltic (fertile) soils in the Kruger National Park (KNP), South Africa, is significantly affected by annual burning [Otter, 1992]. Moreover, possible effects of fire onsoil fertility via loss ofsoil nitrogen to the atmosphere have been postulated [Lobert et ai.,1990], and indry savannas, soil fertility has a controlling influence on the relationship between rainfall andabove-ground herbaceous 23,699