charcoal re-combustion efficiency in tropical savannas
TRANSCRIPT
Geoderma 219–220 (2014) 40–45
Contents lists available at ScienceDirect
Geoderma
j ourna l homepage: www.e lsev ie r .com/ locate /geoderma
Charcoal re-combustion efficiency in tropical savannas
Gustavo Saiz a,b,⁎, Iain Goodrick b, Christopher M. Wurster b, Michael Zimmermann c,Paul N. Nelson b, Michael I. Bird b
a Institute Meteorology and Climate Research, Karlsruhe Institute of Technology, Garmisch-Partenkirchen 82467, Germanyb Centre for Tropical Environmental and Sustainability Science, School of Earth and Environmental Sciences, James Cook University, Cairns QLD 4870, Australiac University of Natural Resources and Life Sciences Vienna, Institute of Soil Research, 1190 Vienna, Austria
⁎ Corresponding author. Tel.: +49 8821 183 288; fax:E-mail address: [email protected] (G. Saiz).
0016-7061/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.geoderma.2013.12.019
a b s t r a c t
a r t i c l e i n f oArticle history:Received 7 February 2013Received in revised form 16 December 2013Accepted 20 December 2013Available online xxxx
Keywords:SavannaFireBushfireCharcoalBlack carbonCombustion
Mass balance considerations suggest a significant gap in our understanding of the processes by which pyrogeniccarbon (PC) is re-mineralized in the environment. Re-combustion by subsequent fires has been evoked as aplausible mechanism explaining significant losses of PC in fire-prone ecosystems, a claim not yet backed upwith experimental data. In this study, four burning experiments were conducted in two northern Australianopen savanna woodlands subject to regular prescribed fires to specifically assess charcoal re-combustion.The experimental burns were designed to provide a set of biotic and abiotic conditions capable of maximizingthe re-ignition potential of surface charcoal fragments.We also testedwhether the size of charcoal fragments hada significant effect on their combustion potential.Although temperature profiles for each burn indicated that the conditions for potential combustion of the char-coal were met, out of the 264 charcoal pieces being monitored only five were totally combusted. Charcoal masslosses were independent of particle size, and averaged less than 8%. The results suggest that turnover timesfor charcoal in tropical savannas as a result of re-combustion alone are on the order of one century. The compar-atively long turnover time ascertained under experimental conditions designed tomaximize charcoal re-burningpotential shows that re-combustion is not an efficient sink for pyrogenic carbon in tropical savannas. This findingstrongly suggests that other processes must play a more substantial role in the re-mineralization of charcoal inorder to balance the PC budget in these ecosystems.
© 2013 Elsevier B.V. All rights reserved.
1. Introduction
Global Pyrogenic Carbon (PC; charcoal, black carbon) production isestimated at 50–270 Tg/yr (Kuhlbusch and Crutzen, 1995), with N90%of this remaining on the ground after the fire (Kuhlbusch et al., 1996).It has been demonstrated that, once formed, PC can persist in the envi-ronment for tens of thousands of years (Bird and Cali, 1998) and bothmodeling (Lehmann et al., 2008) and incubation studies (Kuzyakovet al., 2009) generally infer turnover times for PC of centuries tomillennia. In contrast to the view that turnover times for PC are atleast centennial, field studies have demonstrated almost complete lossof PC from tropical savanna soils protected from fire on decadal time-scales (Bird et al., 1999). In an experiment in frequently burnt savannain Kruger National Park, Kuhlbusch et al. (1996) ‘looked carefully atthe soil prior to the fire for any black carbon particles from previousfires, but could find only very few’, providing anecdotal evidence forthe rapid loss of PC from savanna landscapes. Likewise, Alexis et al.(2007) found no charcoal in surface litter samples in their field sitethat had been burnt 11 years prior to their experimental burn which
+49 8821 183 294.
ghts reserved.
produced surface litter containing 18% charcoal. Masiello and Druffel(2003) used a mass balance approach to demonstrate that a substantialfraction of the PC produced annually from biomass burning must bere-mineralized, or the accumulation of PC in the environment overtime would be sufficient to perturb global atmospheric oxygen levels.These observations suggest a ‘missing sink’ for PC and therefore a sub-stantial gap in our understanding of the processes by which PC mustbe re-mineralized in the environment.
Tropical savannas cover about one-sixth of the earth's land surfaceand account for over a quarter of primary production of terrestrialvegetation (Grace et al., 2006). Their relevance to the global PC budgetis underscored by the fact that these ecosystems are subject to regularfires of varying intensity due to the combination of sufficient seasonalrainfall to generate significant biomass, plus an extended dry seasonthat cures a high fuel load. Mouillot and Field (2005) estimate that86% of annual global burnt area is located in tropical savanna/grasslandregions and Kuhlbusch et al. (1996) estimated that savanna fires con-tribute 10–26 Tg (4–52%) to global annual PC production. Forbes et al.(2006) calculated a PC conversion rate from biomass of below 3%for savanna and grassland ecosystems. Savannas store larger amountsof recalcitrant carbon in the soil compared to tropical forest ecosystemsas a result of frequent fire events that promote the formation of PC(Saiz et al., 2012). Indeed, PC can constitute a significant fraction of
41G. Saiz et al. / Geoderma 219–220 (2014) 40–45
the total soil carbon pool in these ecosystems, and can comparewell andeven significantly exceed PC storage rates observed in non-tropicalgrasslands, boreal forests, and other fire-prone natural environments(Lehmann et al., 2008; Rodionov et al., 2010; Saiz et al., 2012;Skjemstad et al., 1999). Tropical savannas are therefore the environ-ments in which a substantial amount of biomass is regularly consumedby fire, where a substantial fraction of global PC production occurs, andwhere a substantial fraction of PC must be re-mineralized.
The production of charcoal from grass-derived biomass in tropicalsavannas may be significant given the high re-occurrence of fires con-suming this very abundant and easy-to-combust material (Grace et al.,2006). Cachier et al. (1985) have shown that about 75% of the total aero-sol particle production during combustion of tropical savanna vegeta-tion derives from C4 grass material. Moreover, there is also evidenceshowing that a significant proportion of all charcoal present in the soilof tropical savannas is finely divided (Bird et al., 1999; Skjemstadet al., 1999), which suggests that PC is mainly derived from precursorbiomass capable of resulting in small char particles (e.g. grasses) andthat large PC fragments get progressively comminuted to small sizes.A recent study by Nocentini et al. (2010) has also shown that physicallydistinct charcoal particles have different chemical composition implyingdifferent turnover behavior in soil, with fine particles being most prob-ably degraded faster than coarse ones.Moreover,finePCmaywell be re-moved from its site of production bywind orwater (Bird and Cali, 1998;Kuhlbusch and Crutzen, 1995), but this constitutes remobilization ofPC, not re-mineralization, and PC production is greater than the rate ofburial in the ocean (Hedges and Keil, 1995). Coarse PC derived fromwoody biomass represents a substantial component of total PC thatwill potentially remain in situ, and therefore must be re-mineralizedin situ. If this material is not as inert as currently suggested then,it might conceivably be re-mineralized over time by any, or all, ofmicrobial respiration, photochemical degradation, or re-combustionby subsequent fires. A fraction of this total coarse PC may also get com-minuted by physico-chemical degradation and moved as fine particlesor in solution into the soil and potentially on into rivers and ultimatelythe ocean (Bird et al., 1999).
The aim of this study is to specifically assess whether charcoal re-combustion by subsequent fires in tropical savannas represents asignificant mechanism by which PC is re-mineralized. Re-ignition ofsurface PC by subsequent fires has been suggested as one of the factorsexplaining charcoal disappearance in fire-prone boreal ecosystems andnon-tropical grasslands (Czimczik et al., 2005; Ohlson and Tryterud,2000; Preston and Schmidt, 2006; Rodionov et al., 2010). Moreover,the possibility that re-combustion may play a role in tropical systemshas also been suggested (Bird et al., 2000; Czimczik and Masiello,2007; Grace et al., 2006). However, to-date no field study has testedthe degree to which re-combustion in subsequent fires may providean explanation of the ‘missing sink’ for PC. We conducted duplicatedcontrolled fire experiments at two tropical savanna sites under condi-tions in which the intensity of the fire, and hence the likelihood for PCre-ignition and combustion were maximized. We also tested the influ-ence of the size of charcoal fragments on re-combustion probability.
2. Materials and methods
2.1. Site description
A total of four plots were laid out in two open savanna woodlandssubject to regular prescribed fires typically occurring at four-year inter-vals in Far North Queensland (Australia). Two of these plots wereestablished in Davies Creek National Park (DCR; 17.021° S, 145.584° E),while the other two were located in Brooklyn Nature Refuge (BRK;16.586° S, 145.155° E). Tree canopy cover determined by means ofsite-specific allometric equations and visual estimates was 55 and 40%for DCR and BRK respectively. The biomass present in the fire trialareas was mainly composed of a grass layer dominating over sparse
Eucalyptus and Acacia tree seedlings with variable amounts of coarsewoody debris and leaf litter. The most abundant grass species wereThemeda australis, Imperata cylindrica and Heteropogon contortus. Meanunderstorey biomass was determined by vegetation clippings and littercollection at eight one-square meter locations at each study site. Anaverage of 992.0 (284.5) and 771.3 (127.2) g dry mass/m2, determinedfrom 8 × 1 m2 plots at each site, was observed for DCR and BRK, respec-tively; errors in brackets are standard deviations of the mean.
2.2. Characteristics and set up of the charcoal fragments
Acacia aneura charcoal pieces that were produced during a pre-scribed fire to clear land for cattle in central Queensland in 1998 andsubjected to ‘in situ’ natural weathering for over ten years were usedin the experiment. Wood density of the species is 1.0 g/cm3 and the re-sultant apparent density of the charcoal fragments was ~0.5 g/cm3. Thecharcoal was produced at ~700 °C with an elemental composition of78.2% C, 2.3% H, 13.8% O, and 5.5% ash, with a single point BET surfacearea of 398 m2/g. For further information about charcoal characteristicsandmethods used for their determination, please refer to Zimmermannet al. (2012). Charcoal fragments were cleaned of any dirt or debris, andsize-classified bymeans of dry sieving.We usedmeshes of 1 and 0.5 cmwhich resulted in three separate size classes: b0.5, 0.5–1.0, and N1 cm.Charcoal pieces within each size class were handpicked and selected tobe of approximately the same average size (0.25, 0.75 and 1.5 cm) tomake results readily comparable. The charcoal fragments were placedin an oven at 105 °C for 24 h and then weighed.
Fire plots were delimited by firebreaks resulting in 10 × 15 m areasover which pieces of charcoal were carefully placed along a 0.1 × 0.1 mgrid set out at least 3 m away from any of the plot boundaries. Previousresearch has shown that in ecosystems where grass is the main fuel,the size of the plot is of minor importance with regard to maximummeasured fire temperatures (Stinson and Wright, 1969). The grid wasinitially set up on the ground so as to mark the exact potential locationswhere the fragments could be deployed. Charcoal fragments wererandomly allocated across the plot in the BRK experiment (randomcells within the grid), while in the later experiment conducted in DCR,we employed a stratified protocol ensuring each size-class of charcoalwas placed both in close proximity to, and midway between grasstussocks in order to test whether distance to the fuel was a significantfactor affecting re-combustion. We took note of all used cells withinthe grid.
We chose to deploy a greater number of smaller pieces of charcoal asthey are more abundant than larger particles in nature. Moreover, wehypothesized that smaller fragments would be more prone to combus-tion than larger fragments due to their larger surface to mass ratio.Accordingly, we allocated 1, 3 and 5 pieces of charcoal from the N1,0.5–1 and b0.5 cm size categories respectively at each selected location.A total of 264 pieces of charcoal were individually monitored over thefour burns (Table 1). Always taking the centre of the grid cell as refer-ence, individual large samples (N1 cm) were arranged in the middleof the grid location. The three medium size fragments (0.5–1 cm)were laid out as a triangle, while the five fragments corresponding tothe smallest size category (b0.5 cm) were set up displaying a cross.Separation between fragments within the same location was 2.0 cm.The geometrical layouts of the fragments prior to the fire were usedto facilitate their relocation after the fire. We made sure no nativecharcoal fragments were in the immediate vicinity to ours to avoidany post collection mixing-up. After all samples were on place, thegrid was offset by an accurate 2 m downwind distance. The exact po-sition of the grid's four corners at the offset location was distinctivelymarked using colored pegs, which was followed by the retrieval ofthe grid previous to the fire. Successive trials repositioning the gridback into the original charcoal locations, demonstrated the goodnessof such approach as a means to accurately relocate all the fragmentsafter the fire.
Table 1Mean charcoal mass loss (%) for each burn in each charcoal size category; standard errors are shown in parentheses.a
Burning experiment
Charcoal size category
Monitoredlocations
n
>1.0cm Monitoredlocations
n
0.5–1.0 cm Monitoredlocations
n
<0.5 cm
Near% (SE)
Between% (SE)
Near% (SE)
Between% (SE)
Near% (SE)
Between% (SE)
DCR 11.6 (0.2) 1.2 (0.1) 8.5 (2.3) 3.3 (0.3) 4.1 (0.9) 1.0 (0.1)
12 1.4 (0.1) x 12 5.9 (0.8) y 12 2.6 (0.3) x
DCR 23.2 (0.2) 3.1 (0.3) 3.2 (0.3) 7.5 (1.9) 6.6 (1.6) 1.5 (0.2)
12 3.1 (0.1) y 12 5.3 (0.7) y 12 4.0 (0.6) xy
BRK 1 6 6.5 (0.9) z 6 13.5 (1.8) z n/a
BRK 2 6 23.1 (5.1) z 6 9.4 (0.3) z n/a
Pooled burns 36 6.4 (0.4) 36 7.6 (0.2) 24 3.3 (0.2)
Individual piecessignificantly combusted b 1/36 3/108 2/120
a 1, 3 and 5 pieces of charcoal from the N1, 0.5–1 and b0.5 cm size categories respectively were allocated at each monitored location (see text). The terms ‘near’ and ‘between’ makereference to the relative position of charcoal fragments with regard to grass tussocks. The number of monitored locations for ‘near’ and ‘between’ was equally split in each sizecategory (n = 6). Independent non-parametric Mann–Whitney U tests conducted at each size category showed no significant differences between these two locations, and consequentlyvalueswere pooled. In order to allow for comparison between burning experiments and size categories non-parametricMann–WhitneyU testswere conducted consideringmass loss fromthe total weight of all the individual fragments at each monitored location. Different superscripts denote significant differences in mass loss rates between burning experiments or sizecategories (P b 0.05).
b The term ‘significantly combusted’ includes charcoal pieces being totally consumed, or as for the case of larger pieces, individual fragments having lost at least half of its initial mass.
42 G. Saiz et al. / Geoderma 219–220 (2014) 40–45
2.3. Experimental burns
The experimental burns were conducted during September andOctober 2011, as this period constitutes the late dry season, character-ized by a continuous senesced grass layer and with some tree specieshaving shed their leaves, meaning maximum fuel load is available. Allthe four plots were ignited on days with no recorded rainfall for atleast two weeks and shortly after midday, the time corresponding tomaximum diurnal temperature, all of which results in prime condi-tions for the occurrence and spread of fire in these ecosystems. Atthe time of the fire, wind speeds were approximately 20 km/h in allcases, air temperatures ranged from 26.5 to 36.1 °C and relative hu-midity varied from 45 to 17% for fires conducted at BRK and DCRsites respectively.
Fires were ignited along a line on the upwind bottom edge of therectangular plots. The plots were randomly laid out on gentle slopes(~3%), over which the fires could advance upslope promoted by backwinds. This fire progression has been shown to enhance the preheatingof the unburned fuel preceding the flames, thus augmenting both theduration and temperature of the fires (Trollope, 1984). While there isno agreement in the literature with regard to maximum temperaturesbeing achieved through either head or backfires (Bailey and Anderson,1980; Daubenmire, 1968), we chose the former category as naturalfires spread with the prevailing wind.
Fire temperaturewas recorded bymeans of K-typewire thermocou-ples placed at 0.02 m above the ground and connected to a data logger(Simple Logger II L642, AEMC Instruments, USA) logging at 5 s intervals.The emplaced charcoal fragments were re-collected an hour after eachburn, gently cleaned from any accumulated debris with the aid of an ul-trafine brush, and individually placed in sealed containers before beingweighed again to determine any mass loss.
2.4. Statistical analyses
ANOVA statistical testswere used to detect for significant differencesbetween potential ignitable charcoal rates, defined as the temperature–time (°C s) area of each fire above 500 °C (P N 0.05). Charcoal mass
losses were tested for normality using the Shapiro–Wilk test, whichshowed that the datawas not normally distributed. Hence, independentnon-parametric Mann–Whitney U tests were employed to check forstatistically significant differences in charcoal mass losses betweendifferent particle sizes and fires at the 95% confidence level (Table 1).All statistical tests were performed with SPSS (SPSS 17.0 Inc., Chicago,IL, USA).
3. Results
3.1. Temperature profiles
Fire temperature profiles over time revealed similar patterns forthe four experimental burns with maximum temperatures reaching600–700 °C (Fig. 1). All four fires showed no significant differences inpotential ignitable charcoal rates, defined as the temperature–time(°C s) area above 500 °C, as shown by ANOVA statistical tests usingindividual 5-s readings as replicates (P N 0.05). Actual values were13.5, 14.8, 12.2, and 13.9 (×103 °C s) for BRK 1, BRK 2, DCR 1, andDCR 2 burns respectively. All the four fires combusted all of the grassbiomass andmost of the litter present in the plots, while either partiallycharring, or in a few cases, totally combusting woody biomass.
3.2. Charcoal mass losses
Out of the 264 charcoal pieces being monitored only 5 werecombusted in their totality with just one of the large individual pieceslosing more than half its initial mass (Table 1, Fig. 2). The layout of thecharcoal fragments on the ground allowed us to closely monitor theircombustion, thus ruling out any potential bias derived from completeloss via transport. Indeed, it was easy to discern any burnt piecemissingfrom a specific location as these were consistently replacedwith a char-acteristically distinctive ash mound.
There were significant differences in charcoal mass losses betweenfires in all size categories (Table 1). The BRK burning experimentsshowed consistently larger losses compared to DCR burning trials(Table 1, Fig. 2). On the other hand, despite the larger rates in charcoal
Fig. 1. Temperature profile over time for the four experimental burns. Potential ignitablecharcoal rates defined as the temperature–time (°C s) area above 500 °C are shownin gray.
43G. Saiz et al. / Geoderma 219–220 (2014) 40–45
mass loss observed near the grass tussocks in the burnings conducted inDCR, the difference between locations was not statistically significant(Table 1). Charcoal mass loss after the experimental burns were onaverage less than 8%, with no discernible trend observed over the sizegradient (Table 1, Fig. 2). These resultsmay represent an overestimationof charcoal mass loss as a result of fire, as potentially small losses mayoccur during transportation and handling of the fragments.
4. Discussion
4.1. Characteristics of the experimental burns and charcoal mass losses
The experimental burns in this study were purposefully designed tocreate a set of biotic and abiotic conditions in which the effect of fire asan oxidizing agent would be maximized. This meant achieving thehighest possible burning temperatures by utilizing favorable weatherand topographic conditions most able to promote the fire, and carefullyselecting both the type and condition of the vegetation to be burnt. Thebiomass density for these plots is at the high end of that reportedin tropical savannas (Miranda et al., 1993). The grasses making upthe understorey layer of these ecosystems are composed of species re-portedly promoting high temperatures during wildfires (Elder et al.,2011). This is especially the case for Imperata cylindrica, a perennialgrass whose chemical composition and structure has been shown to
Fig. 2. Charcoal mass losses (%) in each charcoal size category for each burning experi-ment. Note: five charcoal fragments from the intermediate and smallest categories werecompletely combusted.
result in very high temperatures when combusted, a fact that has alsobeen demonstrated by thermogravimetric analyses carried out undercontrolled conditions (Elder et al., 2011). Indeed, the seemingly largerlosses of charcoal mass observed in the burnings conducted in BRK(Table 1, Fig. 2) may be resultant from the comparatively larger pres-ence of Imperata cylindrica, whose combustion may have promotedthe slightly larger peak temperatures observed at these sites (Fig. 1).We intentionally chose to burn late in the dry season as higher temper-atures can be achieved (Grant et al., 1997;Williams et al., 2003), on dayswith no recorded rainfall for at least twoweeks, during the hottest timeof the day, and insofar possible choosing instances in which wind speedwas above the long term average recorded for each site. Therefore, allthese factors taken together are likely to represent the most favorableset of natural conditions under which fire may potentially re-ignitecharcoal fragments lying on the ground of tropical savannas. Leifeld(2007) found that charcoal will ignite at temperatures above ~500 °C,consequently the conditions for potential combustion of the charcoalin our burning experimentsweremet, but only briefly (Fig. 1). Similarly,research assessing biomass combustion in tropical forests shows thatdespite the strong radiation from the flame front the exposure to igni-tion temperatures is short-lived (Carvalho et al., 2011). Such relativelyshort exposure may result in no significant flame propagation or igni-tion occurring below the superficial layer of the solid fuel.
The burning experiments showed no discernible trend in charcoalmass loss over the size gradient (Table 1). This finding suggests thatthe re-combustion potential of surface charcoal is not significantly af-fected by the contrasting chemical composition observed in physicallydistinct charcoal particles (Nocentini et al., 2010). On the other hand,the placement strategy set out in the experimental burns carried outin DCR allowed for the assessment of differences in combustion influ-enced by the physical proximity from tussocks of grass. No significantdifference in mass loss was observed between locations (Table 1),which indicates that the likelihood for pieces of charcoal to be ignitedduring a given fire event also depends on factors other than standingbiomass and size of the fragments. Thesemay includemicro-eddies cre-ated by short-lived turbulent wind conditions and micro-topographiccharacteristics, which in turn may have a strong influence on transportof heat and oxygen on the ground surface. Williams et al. (2004) ob-served no significant differences in temperatures recorded at the baseand between grass tussocks in an Australian savanna with similar char-acteristics to our sites, suggesting temperature homogeneity at the scaleof tens of centimeters. In spite of this finding, the heterogeneity ofground level temperature in natural fires may still be significant at amicro-scale. However, we are confident that the temperature profilesregistered in this study are representative, albeit at the high end, offires occurring in environments where grass and woody vegetationcoexist (Bailey and Anderson, 1980; Miranda et al., 1993; Stinson andWright, 1969; Trollope, 1984). This is further corroborated by fireexperiments conducted at these locations in previous years (G. Saizunpublished data 2011).
4.2. Factors influencing charcoal re-combustion turnover
Fire return intervals in tropical wooded savannas are strongly influ-enced by rainfall through its effects on fuel production. Other factorsposing significant influence on fire return intervals are the flammabilityof the biomass, grazing impacts, and of course the degree of human dis-turbance (Furley et al., 2008; O'Connor et al., 2011). Model predictionsand field experiments have both shown that tree biomass often cannotbe sustained under annual fire regimes (Furley et al., 2008; Russell-Smith et al., 2002; Ryan and Williams, 2011; Saiz et al., 2012). Thus, itis not realistic to expect significant macrocharcoal production in envi-ronments subject to annual fires that are unable to sustain trees. Fire re-turn intervals capable of producing significant macrocharcoal uponburning vary widely but average ~3–5 years based on literature
44 G. Saiz et al. / Geoderma 219–220 (2014) 40–45
compilations of fire return interval data (Furley et al., 2008; O'Connoret al., 2011; Sankaran et al., 2005; van Wilgen et al., 2004).
Some wooded savannas subject to anthropogenic fire management,as it is the case of areas of northern Australia where the experimentwasconducted, burn on average as frequently as once every 2–3 years(Edwards and Russell-Smith, 2009; Edwards et al., 2001; Felderhofand Gillieson, 2006). Assuming steady state conditions at these sites,whereby the amount of charcoal produced equals the amount ofcharcoal lost in a fire, the maximum and minimum average percent-age mass loss observed across the charcoal size classes used in ourexperiment (7.6 to 3.3%; Table 1) result in turnover times of 13–30,39–91, 66–152 and 132–303 years for fire return intervals of 1, 3, 5and 10 years respectively. Therefore, a typical fire return interval of3–5 years implies a turnover time for charcoal, from re-combustionalone, of around one century. Where burning frequencies in woodedsavannas are high, fires tend to be cooler and of lower intensity dueto reduced fuel loads and intentional early dry season ignition. Suchset of conditions reduces the likelihood of charcoal re-combustion(Edwards and Russell-Smith, 2009) and increases their turnovertime. Conversely, longer fire return intervals lead to higher intensityfires through higher fuel loads (Russell-Smith and Edwards, 2006;Williams et al., 2004), thereby increasing the probability of charcoalre-combustion and reducing turnover time.
Besides the environmental conditions affecting charcoal re-combustion, it is also critical to consider the nature of the charcoal ex-posed to the fire, which will necessarily have a strong influence on itspotential to get combusted. Besides the quality of the charcoal, itsstate of aggregation will, in part, determine the oxygen supply andheat transfer during a given fire event, both known to be main factorsinfluencing oxidation in other solid organic fuels (i.e. wood) (Di Blasi,2009). Carvalho et al. (2011) report that in combustion of woodymate-rial flame stabilization in the heterogeneous phase (solid–gas interface)depends strongly on the ignition and extinction aspects of the flame inthe wood, explaining whymany logs present only a superficial burningand have an intact core, and it is likely that the same concept may applyto charcoal.
With all, the turnover time calculated in the present study is stillrelatively long and a re-combustion efficiency of b100% implies along-term accumulation of charcoal in savannas eventually leading toimplausibly high charcoal abundance, a phenomenon that is the oppo-site of that observed (Alexis et al., 2007; Kuhlbusch et al., 1996). There-fore it seems likely that other processes must play a significant rolein the re-mineralization of charcoal in the environment in order tobalance global PC production. Indeed, it is increasingly evident that de-spite the relatively high resistance of charcoal to decomposition, somecomponents of this material are subject to significant degradation innatural environments as suggested by observation (Alexis et al., 2007;Kuhlbusch et al., 1996; Zimmermann et al., 2012), directmeasurementsand mass balance calculations (Abiven et al., 2011; Ascough et al.,2011; Bird et al., 1999; Czimczik and Masiello, 2007; Masiello andDruffel, 2003; Zimmermann et al., 2012). The additional possibleprocesses that may lead to charcoal re-mineralization (rather than re-mobilization) in savanna environments include microbial and fungaldecomposition, and UV degradation.
5. Conclusions
While necessarily a very rough estimate, the results of this studysuggest that the average turnover time for charcoal in tropical savannasresulting solely from re-combustion is likely to be on the order of onecentury. It should be noted that this turnover time is likely to be a con-servative estimate given that the conditions to promote charcoal re-combustion were maximized and potentially small mass losses mighthave occurred during handling of the fragments. The actual value willvary widely depending both on interactions between climate, vegeta-tion and the balance between natural and anthropogenic fire ignition
that determine the frequency and severity of fires in the landscape.The likelihood of charcoal re-combustion is also likely to depend onthe state of aggregation and the nature of the charcoal itself, which inturn depends upon the nature of the biomass pyrolysed and the condi-tions underwhich itwas pyrolysed. These characteristicswill determineoxygen supply and heat transfer during a given fire event and may po-tentially influence its reactivity to fire.
The general conclusion from this study suggests that re-combustiondoes play a role in the re-mineralization of charcoal. However, the com-paratively long turnover time ascertained under experimental condi-tions designed to maximize charcoal re-burning potential shows thatre-combustion is not an efficient sink for pyrogenic carbon in tropicalsavannas. Therefore, it seems likely that other processes must play amore substantial role in the re-mineralization of charcoal in order tobalance the PC budget in these ecosystems.
Acknowledgments
We gratefully acknowledge Queensland Parks andWildlife staff, andespecially RobMiller for all his support and valuable advice.We are alsothankful to the Australian Wildlife Conservancy Society for allowingaccess and permits to undertake research at the North-east BrooklynSanctuary site. Funding was provided by the Australian Research Coun-cil (DP1096586).
References
Abiven, S., Hengartner, P., Schneider, M.P.W., Singh, N., Schmidt, M.W.I., 2011. Pyrogeniccarbon soluble fraction is larger and more aromatic in aged charcoal than in freshcharcoal. Soil Biol. Biochem. 43, 1615–1617.
Alexis, M.A., Rasse, D.P., Rumpel, C., Bardoux, G., Péchot, N., Schmalzer, P., Drake, B.,Mariotti, A., 2007. Fire impact on C and N losses and charcoal production in a scruboak ecosystem. Biogeochemistry 82, 201–216.
Ascough, P.L., Bird, M.I., Francis, S.M., Lebl, T., 2011. Alkali extraction of archaeological andgeological charcoal: evidence for diagenetic degradation and formation of humicacids. J. Archaeol. Sci. 38, 69–78.
Bailey, A.W., Anderson, M.L., 1980. Fire temperatures in grass, shrub and aspen forestcommunities of central Alberta. J. Range Manag. 33, 37–40.
Bird, M.I., Cali, J.A., 1998. A million-year record of fire in sub-Saharan Africa. Nature 394,767–769.
Bird, M.I., Moyo, C., Veenendaal, E.M., Lloyd, J., Frost, P., 1999. Stability of elemental carbonin a savanna soil. Global Biogeochem. Cycles 13, 923–932.
Bird, M.I., Veenendaal, E.M., Moyo, C., Lloyd, J., Frost, P., 2000. Effect of fire and soil textureon soil carbon in a sub-humid savanna (Matopos, Zimbabwe). Geoderma 94, 71–90.
Cachier, H., Buat-Menard, P., Fontugne, M., Rancher, J., 1985. Source terms and sourcestrengths of the carbonaceous aerosol in the tropics. J. Atmos. Chem. 3, 469–489.
Carvalho, J.A., Costa, F.S., Veras, C.A.G., Sandberg, D.V., Alvarado, E.C., Gielow, R., Serra,A.M., Santos, J.C., 2011. Biomass fire consumption and carbon release rates ofrainforest-clearing experiments conducted in northern Mato Grosso, Brazil.J. Geophys. Res. 106 (17), 17,877–17,888.
Czimczik, C.I., Masiello, C.A., 2007. Controls on black carbon storage in soils. GlobalBiogeochem. Cycles 21 (GB3005).
Czimczik, C.I., Schmidt, M.W.I., Schulze, E.D., 2005. Effects of increasing fire frequency onblack carbon and organic matter in Podzols of Siberian Scots pine forests. Eur. J. SoilSci. 56, 417–428.
Daubenmire, R., 1968. Ecology of fires in grasslands. In: Cragg, J.B. (Ed.), Adv. Ecol. Res. 5.Academic Press, London, pp. 209–266.
Di Blasi, C., 2009. Combustion and gasification rates of lignocellulosic chars. Prog. EnergyCombust. Sci. 35, 121–140.
Edwards, A.C., Russell-Smith, J., 2009. Ecological thresholds and the status of fire-sensitivevegetation in western Arnhem Land, northern Australia: implications for manage-ment. Int. J. Wildland Fire 18, 127–146.
Edwards, A., Hauser, P., Anderson, M., McCartney, J., Armstrong, M., Thackway, R., Allan,G., Hempel, C., Russell-Smith, J., 2001. A tale of two parks: contemporary fire regimesof Litchfield and Nitmiluk National Parks, monsoonal northern Australia. Int.J. Wildland Fire 10, 79–89.
Elder, T., Kush, S., John, S., Hermann, S.M., 2011. Thermogravimetric analysis of forestunderstory grasses. Thermochim. Acta 512, 170–177.
Felderhof, L., Gillieson, D., 2006. Comparison of fire patterns and fire frequency in twotropical savanna bioregions. Aust. J. Ecol. 31, 736–746.
Forbes, M.S., Raison, R.J., Skjemstad, J.O., 2006. Formation, transformation and transport ofblack carbon, charcoal in terrestrial and aquatic ecosystems. Sci. Total Environ. 370,190–206.
Furley, P.A., Rees, R.M., Ryan, C.M., Saiz, G., 2008. Savanna burning and the assessment oflong-term fire experiments with particular reference to Zimbabwe. Prog. Phys. Geogr.32, 611–634.
Grace, J., San Jose, J., Meir, P., Miranda, H.S., Montes, R.A., 2006. Productivity and carbonfluxes of tropical savannas. J. Biogeogr. 33, 387–400.
45G. Saiz et al. / Geoderma 219–220 (2014) 40–45
Grant, C.D., Loneragan, W.A., Koch, J.M., Bell, D.T., 1997. Fuel characteristics, vegetationstructure and fire behavior of 11–15 year-old rehabilitated bauxite mines inWesternAustralia. Aust. For. 39, 210–220.
Hedges, J.I., Keil, R.G., 1995. Sedimentary organic matter preservation—an assessment andspeculative synthesis. Mar. Chem. 49, 81–115.
Kuhlbusch, T.A.J., Crutzen, P.J., 1995. Toward a global estimate of black carbon in residuesof vegetation fires representing a sink of atmospheric CO2 and a source of O2. GlobalBiogeochem. Cycles 9, 491–501.
Kuhlbusch, T.A.J., Andreae, M.O., Cachier, H., Goldammer, J.G., Lacaux, J.P., Shea, R.,Crutzen, P.J., 1996. Black carbon formation by savanna fires: measurements and im-plications for the global carbon cycle. J. Geophys. Res. 101, 23651–23665.
Kuzyakov, Y., Subbotina, I., Chen, H., Bogomolova, I., Xu, X., 2009. Black carbon decompo-sition and incorporation into soil microbial biomass estimated by C-14 labeling. SoilBiol. Biochem. 41, 210–219.
Lehmann, J., Skjemstad, J., Sohi, S., Carter, J., Barson, M., Falloon, P., Coleman, K.,Woodbury, P., Krull, E., 2008. Australian climate-carbon cycle feedback reduced bysoil black carbon. Nat. Geosci. 1, 832–835.
Leifeld, J., 2007. Thermal stability of black carbon characterized by oxidative differentialscanning calorimetry. Org. Geochem. 38, 112–127.
Masiello, C.A., Druffel, E.R.M., 2003. Organic and black carbon C-13 and C-14 throughthe Santa Monica Basin sediment oxic–anoxic transition. Geophys. Res. Lett. 30 (Art.No. 1185).
Miranda, A.C., Sinátora, H., Oliveira, I.F., Ferreira, B., 1993. Soil and air temperatures duringprescribed cerrado fires in Central Brazil. J. Trop. Ecol. 9, 313–320.
Mouillot, F., Field, C., 2005. Fire history and the global carbon budget: a 1 degrees × 1 de-grees fire history reconstruction for the 20th century. Glob. Chang. Biol. 11, 398–420.
Nocentini, C., Certini, G., Knicker, H., Francioso, O., Rumpel, C., 2010. Nature and reactivityof charcoal produced and added to soil during wildfire are particle-size dependent.Org. Geochem. 41, 682–689.
O'Connor, T.G., Mulqueeny, C.M., Goodman, P.S., 2011. Determinants of spatial variation infire return period in a semiarid African savanna. Int. J. Wildland Fire 20, 540–549.
Ohlson, M., Tryterud, E., 2000. Interpretation of the charcoal record in forest soils: forestfires and their production and deposition of macroscopic charcoal. The Holocene10, 519–525.
Preston, C.M., Schmidt,M.W.I., 2006. Black (pyrogenic) carbon: a synthesis of current knowl-edge and uncertainties with special consideration of boreal regions. Biogeosciences 3,397–420.
Rodionov, A., Amelung, W., Peinemann, N., Haumaier, L., Zhang, X., Kleber, M., Glaser,B., Urusevskaya, I., Zech, W., 2010. Black carbon in grassland ecosystems of theworld. Global Biogeochem. Cycles 24. http://dx.doi.org/10.1029/2009GB003669(GB3013).
Russell-Smith, J., Edwards, A.C., 2006. Seasonality and fire severity in savanna landscapesof monsoonal northern Australia. Int. J. Wildland Fire 15, 1–10.
Russell-Smith, J., Ryan, P.G., Cheal, D., 2002. Fire regimes and the conservation of sandstoneheath in monsoonal northern Australia: frequency, interval, patchiness. Biol. Conserv.104, 91–106.
Ryan, C.M., Williams, M., 2011. How does fire intensity and frequency affect miombowoodland tree populations and biomass? Ecol. Appl. 21, 48–60.
Saiz, G., Bird, M.I., Domingues, T., Schrodt, F., Schwarz, M., Feldpausch, T.R., Veenendaal,E.M., Djagbletey, G., Hien, F., Compaore, H., Diallo, A., Lloyd, J., 2012. Variation insoil carbon stocks and their determinants across a precipitation gradient in WestAfrica. Global Chang. Biol. 18, 1670–1683.
Sankaran, M., Hanan, N.P., Scholes, R.J., Ratnam, J., Augustine, D.J., Cade, B.S., Gignoux, J.,Higgins, S.I., 2005. Determinants of woody cover in African savannas. Nature 438,846–849.
Skjemstad, J.O., Taylor, J.A., Smernik, R.J., 1999. Estimation of charcoal (char) in soils.Commun. Soil Sci. Plant Anal. 30, 2283–2298.
Stinson, K.J., Wright, H., 1969. Temperatures of headfires in the southern mixed prairie ofTexas. J. Range Manag. 22, 169–174.
Trollope, W.S.W., 1984. Fire in savanna, in ecological effects of fire in South AfricanEcosystems. In: Booysen, P.D.V., Tainton, N.M. (Eds.), Springer-Verlag, Berlin, Germany,pp. 199–218.
Van Wilgen, B.W., Govender, N., Biggs, H.C., Ntsala, D., Funda, X.N., 2004. Response of sa-vanna fire regimes to changing fire management policies in a large African nationalpark. Conserv. Biol. 18, 1533–1540.
Williams, P.R., Congdon, R.A., Grice, A.C., Clarke, P.J., 2003. Effect of fire regime on plantabundance in a tropical eucalypt savanna of north-eastern Australia. Aust. J. Ecol.28, 327–338.
Williams, P.R., Congdon, R.A., Grice, A.C., Clarke, P.J., 2004. Soil temperature and depthof legume germination during early and late dry season fires in a tropical eucalyptsavanna of north-eastern Australia. Aust. J. Ecol. 29, 258–263.
Zimmermann, M., Bird, M.I., Wurster, C., Saiz, G., Goodrick, I., Santruckova, H., Barta, J.,Capek, P., Smernik, R., 2012. Rapid degradation of pyrogenic carbon. Glob. Chang.Biol. 18, 3306–3316.