464 Ecological Applications, 8(2), 1998, pp. 464–468  1998 by the Ecological Society of America COMBUSTION IN NATURAL FIRES AND GLOBAL EMISSIONS BUDGETS S. J. MCNAUGHTON, 1 N. R. H. STRONACH, 2,3 AND N. J. GEORGIADIS 1,4 1 Biological Research Laboratories, Syracuse University, Syracuse, New York 13244-1220 USA 2 Department of Zoology, Cambridge University, Cambridge, England Abstract. Fires in tropical savannas are a principal source of emissions to the atmo- sphere, but there are few studies of retention in ash and residual plant mass following natural fires. Estimates of carbon and nitrogen emissions to the atmosphere from biomass burning have relied largely on indirect measures from laboratory incineration. Emissions from incinerated samples are then extrapolated to areally extensive field estimates of flam- mable plant biomass. In addition, both direct sampling of smoke plumes and comparison to ambient atmospheric samples, with combustion efficiency estimated by ratios of trace gases to carbon dioxide in plumes, are employed. To the extent that combustion in natural fires departs from laboratory burning or that assumed average standing crops are inaccurate, indirect estimates based on laboratory emissions will be in error. Similarly, the relation of emission ratios to natural fire intensity is poorly understood and will influence the accuracy of estimates based on plume sampling. Recently, data from the Southern African Fire– Atmosphere Research Initiative (SAFARI) has indicated that indirect methods can over- estimate the proportion of biomass combusted in natural fires. However, how combustion is related to fire temperature is quantitatively unknown. Studies of combustion efficiency in natural fires, as the proportion of initial biomass actually volatilized, are rare, and we know of no estimates of mass and N retention in residual ash and partially combusted biomass that are related to fire temperature. We burned plots instrumented with thermocouples to measure fire temperature and analyzed biomass from adjacent clipped plots, residual post-fire plant residue, and ash to determine the proportion of biomass actually burned and N volatilized. Data from experimental savanna grassland fires under natural conditions in Serengeti National Park indicate that (1) the proportions of mass and N volatilized are substantially less than 100%, (2) most fuel loads are within a range where combustion and volatilization are highly dependent on mass burned and resultant fire intensity, (3) proportionally more N than mass is volatilized, particularly as fire intensity increases, but (4) much less mass and N are volatilized in natural fires than laboratory estimates suggest. Therefore, (5) obtaining accurate estimates of emissions from natural fires will be substantially more difficult than is reflected in current methodology. Key words: biomass burning and global climate change; combustion in natural fires; global emission budgets; laboratory cf. natural combustion; N and C emissions. INTRODUCTION Increased atmospheric concentrations of gaseous forms of carbon and nitrogen are expected to affect global climate and nutrient cycling (Hansen et al. 1981, Trabalka 1985, Cofer et al. 1991), contributing to an- ticipated phenomena referred to collectively as ‘‘global change’’ (NAS 1988). These ‘‘greenhouse’’ gases are believed to be important contributors to total radiative climatic forcing (Levine 1991, Justice et al. 1996). Bio- mass burning, particularly in the tropics, is considered an important contributor to global change (Crutzen and Andreae 1990, Lobert et al. 1990, Kuhlbusch et al. 1991, Levine 1991). However, studies of natural fire Manuscript received 30 July 1996; revised 30 July 1997; accepted 14 August 1997; final version received 20 October 1997. 3 Present address: Fota Wildlife Park, Carrigtohill, County Cork, Ireland. 4 Present address: Mpala Research Centre, Nanyuki, Ke- nya. dynamics are rare (Hao et al. 1990) and there are few data on the retention of carbon and nitrogen in residual ash (Crutzen and Andreae 1990, Menaut et al. 1991, Scholes et al. 1996, Ward et al. 1996). Fires in tropical savannas are considered a principal source of emissions to the atmosphere (Crutzen and Andreae 1990, Kuhlbusch et al. 1991, Lindsay 1996). However, estimates are based principally upon apply- ing data from laboratory incineration to estimates of fuel loads in savanna grasslands. Laboratory inciner- ation leads to estimates of only 10% of the initial plant N retained in ash (Lobert et al. 1990) and these determinations have been applied to global emissions calculations (Crutzen and Andreae 1990). Smoke- plume sampling and estimates of fire intensity based on the ratios of trace-gas emission enrichment com- pared to CO 2 (CO 2 -normalized emission ratios) indicate that emission ratios are highest during smoldering com- bustion, generally thought to be the least efficient com- bustion stage, a clear discrepancy with laboratory-de- rived emission ratios (Cofer et al. 1991).