- Email: [email protected]
Tropical savannah woodland: effects of experimental fire on soil microorganisms and soil emissions of carbon dioxide
Soil Biology & Biochemistry 36 (2004) 849–858 www.elsevier.com/locate/soilbio
Tropical savannah woodland: effects of experimental fire on soil microorganisms and soil emissions of carbon dioxide Michael Anderssona,*, Anders Michelsenb, Michael Jensenb, Annelise Kjøllera a
Department of General Microbiology, University of Copenhagen, Sølvgade 83 H, DK-1307 Copenhagen K, Denmark b Botanical Institute, University of Copenhagen, Øster Farimagsgade 2 D, DK-1353 Copenhagen K, Denmark Received 7 March 2003; received in revised form 22 October 2003; accepted 21 January 2004
Abstract Burning of the vegetation in the African savannahs in the dry season is widespread and may have significant effects on soil chemical and biological properties. A field experiment in a full factorial randomised block design with fire, ash and extra grass biomass as main factors was carried out in savannah woodland of the Gambella region in Ethiopia. The microbial biomass C (Cmic) was 52% (fumigation – extraction) and 20% (substrate-induced respiration) higher in burned than unburned plots 12 d after burning. Both basal respiration and potential denitrification enzyme activity (PDA) immediately responded to burning and increased after treatment. However, in burned plots addition of extra biomass (fuel load) led to a reduction of Cmic and PDA due to enhanced fire temperature. Five days after burning, there was a short-lived burst in the in situ soil respiration following rainfall, with twice as high soil respiration in burned than unburned plots. In contrast, 12 d after burning soil respiration was 21% lower in the burned plots, coinciding with lower soil water content in the same plots. The fire treatment resulted in higher concentrations of dissolved organic C (24– 85%) and nitrate (47 – 76%) in the soil until 90 d after burning, while soil 2 þ NHþ 4 – N was not affected to the same extent. The increase in soil NO3 – N but not NH4 – N in the burned plots together with the well-aerated 2 soil conditions indicated that nitrifying bacteria were stimulated by fire and immediately oxidised NHþ 4 – N to NO3 – N. In the subsequent rainy season, NO2 – N and, consequently, PDA were reduced by ash deposition. Further, C was lower in burned plots at that time. 3 mic However, the fire-induced changes in microbial biomass and activity were relatively small compared to the substantial seasonal variation, suggesting transient effects of the low severity experimental fire on soil microbial functioning. q 2004 Elsevier Ltd. All rights reserved. Keywords: Ash; Basal respiration; Dissolved organic carbon; Emissions of carbon dioxide; Fuel load; Microbial biomass carbon; Nitrate; Potential denitrification enzyme activity; Savannah fire
1. Introduction Across the Sahelian-north Guinean vegetation zone in Africa, burning of the savannah vegetation is a frequent, widely distributed phenomenon in the beginning of the dry season (Cahoon et al., 1992; Barbosa et al., 1999). The fires are mainly due to human activities; the frequency of the fires at present has increased as a consequence of an expanding exploitation of natural resources caused by the high growth rate of the African human population, often resulting in annual to biennial burning of the savannah areas in this region. The predominant early burning may increase the soil organic carbon (Corg) compared to protected plots, but a paradox arises by the change of fire regimes towards higher * Corresponding author. Tel.: þ 45-3532-2054; fax: þ 45-35-32-2040. E-mail address: [email protected] (M. Andersson). 0038-0717/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.soilbio.2004.01.015
fire frequencies, which may have detrimental consequences for soil fertility (Kaiser, 1983; Frost and Robertson, 1987; Jones et al., 1990; Menaut et al., 1993). Since changes in soil fertility and soil C stocks reflect alterations in the microbial decomposer community composition and activity, a better understanding of the response of soil microbial performance to savannah fires is required. The components of fire can be separated into a direct and several indirect effects on soil microbial communities. During fire, the heating of soil may kill a fraction of the microorganisms. With fungi being more susceptible to heat than bacteria (Bollen, 1969; Dunn et al., 1985) fire may lead to a changed microbial community composition. Immediately after burning, the number of bacteria and fungi and the microbial biomass is often reduced in the subsurface soil layer (Meiklejohn, 1955; Deka and Mishra, 1983; Van Reenen et al., 1992). However, the reduced microbial
850
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
numbers have also been attributed to higher temperatures in burned plots with bare soil than in unburned plots with plant cover due to solar radiation (Kaiser, 1983). Ahlgren and Ahlgren (1965) argued, that the immediate decrease in the number of soil microorganisms after fire in pine forest was caused by the heat from the fire, while the subsequent increase in number of microorganisms after a rain event could be attributed to the ash deposition. Other studies have reported increased number of microorganisms after slash burning in virgin tropical forest, and higher microbial biomass in burned than unburned plots in woodland savannah and tallgrass prairie (Corbet, 1934; Garcia and Rice, 1994; Singh, 1994). In savannah ecosystems, some studies suggest that the responses of the soil microbial communities to fire are restricted to the uppermost centimetres of the soil layer, and effects may have disappeared at the onset of the rainy season (Raison, 1979; Adedeji, 1983; Deka and Mishra, 1983; Van Reenen et al., 1992). The microorganisms killed by the heat from the fire are a source of ammonium and dissolved organic C (DOC), which can stimulate microbial growth. Higher NHþ 4 –N concentrations following burning have been attributed to deposition of ash, heating of soil, reduced microbial N immobilisation or reduced N plant uptake (Christensen, 1973; Kovacic et al., 1986; Stock and Lewis, 1986; Anderson and Levine, 1988; Araki, 1993; Singh, 1994). The heating of soil during fire releases NHþ 4 – N from protein-like components of organomineral soil complexes and from combustion of organic matter (Raison, 1979; Kovacic et al., 1986). Further, fireinduced modifications of soil chemical properties are directly related to the intensity of soil heating and the amount of ash deposition (Raison, 1979). Basically, changes in microbial activity caused by environmental changes may be independent of possible changes in microbial biomass. Some, but not all, studies have found higher C and N mineralisation, and in situ soil respiration rates after burning in savannahs and woodland, and after burning and wetting in savannah (Adedeji, 1983; Hao et al., 1988; Singh et al., 1991; Van Reenen et al., 1992; Araki, 1993; Poth et al., 1995; Zepp et al., 1996). The fireinduced increase in microbial activity and soil respiration found in these studies have been explained by higher pH, N availability, pool of readily decomposable substrates and temperature in soil from burned than unburned plots, while the lack of responses in soil respiration have been attributed to low soil moisture. These results reflect to some extent the site specific differences in soil type, fire intensity and savannah vegetation type as well as differences in time elapsed since fire took place. In savannah ecosystems, the effects of fire on microbial biomass and C and N mineralisation rates may be minor compared to the pronounced seasonal variability (Singh et al., 1991). However, burning may strongly promote root growth and aboveground plant production in the rainy season subsequent to fire events (Menaut et al., 1993;
Dhillion and Anderson, 1994; Jensen et al., 2001), and hence have strong effects on savannah ecosystems. Soil enzymatic activities act as potentially sensitive indicators of soil quality and have been used in studies of vegetation burning in prairie and Mediterranean shrubland (Ajwa et al., 1999; Castaldi and Aragosa, 2002), but enzymes of primarily microbial origin have seldom been studied in relation to fire effects on the soil microbial communities in African savannahs. More studies are necessary to provide a thorough understanding of the function of soil microbial communities in fire-prone savannah ecosystems. Apparently, there has not been any study that has separated the effects of fire on microbial biomass and activity into the heat and ash components in a controlled field experiment in African fire-prone savannah ecosystems. The objective of this study was to investigate the effects of 2 fire intensity and ash deposition on soil NHþ 4 –N, NO3 –N, DOC, microbial biomass C, basal respiration at optimal water conditions, potential denitrification enzyme activity (PDA) and in situ soil respiration. The variables are expressed per unit Corg, since microbial biomass and activity covary with the content of Corg.
2. Materials and methods 2.1. Study site The experiment was carried out in a woodland savannah in the south-western part of Ethiopia about 12 km south of the provincial capital Gambella (88090 N, 348340 E) from November 1997 to July 1998. Annual precipitation is on average 900 mm with the major part falling from late May to October. Daily variations in temperature are from 20 8C at night to 37 8C during the day, with minor seasonal variation. At the study site, the tree canopy cover was 60% and the grass vegetation covered 88% of the ground dominated by Hyparrhenia confinis var. nudiglumis, which is an annual to semi-perennial grass (Jensen et al., 2001; Gashaw and Michelsen, 2001). The grass biomass was approximately 480 g dry weight m22 by the end of the rainy season (Jensen et al., 2001). Soil pHKCl was 5.8 to 6.6 and the treatments described below elevated the pH by less than 0.2 units, while the soil organic matter (SOM) content, total N and P was about 4, 0.10 and 0.02%, respectively, and unaffected by treatments (Jensen et al., 2001). At the study area, annual fires occur in the dry season from November to May, most frequent by the beginning of the dry season. Grazing by wild animals and cattle is insignificant. A full description of the vegetation at the study site and for the Gambella region in general is given in Jensen and Friis (2001). 2.2. Experimental design Fire effects were investigated by analysing three experimental factors: fire, grass biomass addition before
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
burning (i.e. extra fuel load) and ash addition after burning. The experiment was carried out in a full factorial randomised block design with six replicate blocks, i.e. with a total of 48 plots. The blocks were distributed randomly not more than 50 m apart. Each block consisted of eight plots of 2 £ 2 m2 that were separated by fire belts of 2 m width. In order to minimise border effects, the central 1 £ 1 m2 only was used for sampling. The standing grass biomass of 480 g dw m22 was doubled in the extra biomass (fuel load) treatments. The ash addition of approximately 15 g m22 was accomplished by burning a similar sized plot adjacent to the blocks. The total C and N, loss on ignition and pH of the ash was 9.4, 0.30, 15.8 and 11.0% in the , 500 mm fraction and 14.5, 0.56, 27.3 and 9.1% in the . 500 mm fraction, respectively, as determined by methods described in Jensen et al. (2001). The fuel load was set on fire 20 November, 1997. During burning the soil temperature at 1 cm depth increased from 25 to 37 and 48 8C in the fire treatment and in the fire plus extra grass biomass treatment 7 min after fire, respectively, showing that the fire was of relatively low intensity due to the moderate grass biomass at the site (Jensen et al., 2001). 2.3. Soil sampling and extractions Soil sampling was carried out on four occasions; at 1, 12, 90 and 210 d after initiation of the experiment. In each plot two soil samples were collected with a soil corer (5 cm dia.) from 0 to 5 cm depth and mixed and sieved (, 2 mm). The chloroform fumigation and KCl extractions were performed 1 – 4 d after sampling in the field, while subsamples for other purposes were kept at 15 8C for a maximum of 3 weeks in Ethiopia and at 4 8C when returning to the laboratory in Denmark until analysis. Seven grams of fresh soil were shaken for 1 h in 35 ml 2 M KCl. At the same time, 7 g of fresh soil were fumigated with chloroform for 24 h to release C from the microbial biomass (Jenkinson and Powlson, 1976), followed by extraction. Extracts were kept at the same temperatures as the soil samples. 2.4. Gas flux measurements Field measurements of soil emissions of CO2, predominantly originating from microbial and root respiration, were performed in areas with no aboveground plant cover using an infrared gas analyser equipped with a soil respiration chamber (EGM-1/SRC-1, PP-Systems, UK) covering 78.5 cm2 and with a volume of 1.17 l. The CO2 flux was measured in all plots immediately before commencing the experiment and these data were used as covariate in the statistical analysis of CO2 flux measured in duplicate in all treatment plots 11 –13 d after initiation of the experiment. In addition, CO2 flux was measured regularly during the first 15 d after burning in all control and burned-only plots.
851
2.5. Temperature, soil moisture and precipitation measurements Concomitantly with determination of soil emissions of CO2, temperature was measured at 5 cm depth inserting a probe horizontally into the soil. Volumetric soil water content in the 0 – 10 cm soil layer was measured using a Time Domain Reflectrometry (TDR) measuring device (TRIME-FM, IMKO, Germany) equipped with a 10 cm probe (P2 rods). For calculation of water-filled pore space (WFPS), bulk density was determined by collecting soil of a known volume, and the soil particle density was determined by displacement with water carefully eliminating any air bubbles. On the first 15 d of the experiment precipitation was measured daily in six containers placed as far as possible from tree canopies. 2.6. Laboratory analyses 2.6.1. Soil water content and organic matter Soil water content was measured gravimetrically and SOM was determined by incinerating dry soil for 6 h at 550 8C. Corg was estimated as 50% of SOM. 2.6.2. Inorganic N Whatman GF-D filtered soil extracts were analysed for 2 NHþ 4 – N and NO3 – N with the indophenol method using a Hitachi U-2000 spectrophotometer and the Cd reduction method on an Aquatec 5400 analyser, respectively. 2.6.3. Microbial biomass C and basal respiration Microbial biomass C was determined by the chloroform fumigation –extraction method (Vance et al., 1987) with some modifications (7 g fresh soil; 2 M KCl solution) and substrate-induced respiration (SIR) according to the criteria of Anderson and Domsch (1978). DOC was measured on KCl extracts of fumigated and unfumigated samples (see Section 2.3) with a Shimadzu 5000 TOC analyser, and microbial biomass C (CmicFE) was calculated as the difference between fumigated and non-fumigated extracts applying a kEC of 0.45 (Wu et al., 1990). SIR was measured in fresh soil equivalent to 10 g dw placed in 50 ml serum bottles. Glucose solution was added in droplets to 60% of water holding capacity (WHC) resulting in a final concentration of 2 mg glucose g21 dw soil and mixed thoroughly into the soil. The bottles with soil were conditioned for 1.5 h, sealed with butyl rubber stoppers, and incubated for 2 h at 25 8C. For measurements of basal respiration the soil water content was adjusted to 60% of WHC and conditioned at 25 8C overnight before sealing, and subsequently incubated for 24 h. The production of CO2 from the soil in both assays was determined by taking gas samples at the beginning and the end of the incubation for immediate analysis for CO2 on a HP 5890 GC equipped with a temperature conductivity detector (TCD) and a Hayesep Q 80-100 mesh column. Hydrogen was used
852
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
as carrier gas at 20 ml min21 and the injector, oven and detector temperatures were 25, 70 and 265 8C, respectively. SIR was converted to microbial biomass C (CmicSIR) using a conversion factor of 40 (Anderson and Domsch, 1978). The glucose concentration and the duration of the conditioning and incubation periods in the SIR assay had been determined in tests for this particular soil. The biomass specific respiration rate, also known as the metabolic quotient (qCO2) (Anderson and Domsch, 1993), was calculated by dividing basal respiration with CmicSIR. 2.6.4. Potential denitrification enzyme activity PDA was determined according to Smith and Tiedje (1979) and modifications of the method by Pell et al. (1996) in triplicate for each soil sample. The soil was transferred to 30 8C 24 h before performing the PDA assay. Thereafter, an anaerobic (N2) slurry was mixed in 118 ml bottles containing soil equal to 5 g dw and 5 ml of a solution of 1 mM KNO3 and 1 mM glucose. To inhibit the enzymatic reduction of N2O, 10 kPa C2H2 (10%) was injected. A gas sample of 2 ml was taken at t ¼ 0; the bottles were incubated at 30 8C and shaken at 250 rev min21 for 5 h. Gas samples were also taken at 1, 2, 3.5 and 5 h. The gas samples were transferred to Venojecte tubes and analysed for N2O by gas chromatography. The HP 5890 GC was equipped with an electron capture detector (ECD) and a Hayesep Q 80-100 mesh column with N2 (20 ml min21) as carrier gas and injector, oven and detector temperatures of 120, 25 and 265 8C, respectively. The PDA at the beginning of the incubation ðt ¼ 0Þ was determined by non-linear regression using the equation given by Pell et al. (1996). 2.7. Statistical analysis Data were tested statistically, separately for each sampling date, with analysis of variance (ANOVA) using the SAS procedure GLM with type III sum of squares (SAS Institute, 1997). The model used was a four way ANOVA, with fire, biomass, ash and block as main factors, and including all interactions between fire, biomass and ash. An optimisation of the model included a removal of the block factor and the three factor interaction, if non-significant at alpha ¼ 0.05. The model used in the ANOVA for the CO2 flux involved CO2 flux data prior to initiation of the experiment treated as covariate, while the block was not included. A correlation between soil respiration and water content was analysed by Pearson correlation.
3. Results 3.1. Soil moisture, DOC and inorganic N The gravimetric water content in the 0-5 cm soil profile was significantly lower in burned plots on day 12 and 90 (Fig. 1). The gravimetric water content on day 12
Fig. 1. Soil gravimetric water content in the experimental plots 1, 12, 90 and 210 d after the experiment began. The values are means ^ SE, n ¼ 6 per treatment. The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s, performed separately for each date, are presented. Significance levels: P , 0:01**, P , 0:001***.
corresponded to a WFPS of 37 and 32% in unburned and burned plots, respectively, applying a bulk density of 1.42 g dw cm23 and a soil particle density of 2.3 g dw cm23. In the middle of the dry season (90 d) WFPS was 16 and 12% in unburned and burned plots, respectively. The DOC concentration differed markedly between the sampling times and the mean content of DOC was 1.1, 0.5, 3.4 and 0.3 mg C g21 Corg at days 1, 12, 90 and 210, respectively (Fig. 2). The production of DOC was stimulated by the treatments after 1, 12 and 90 d. At day 1, DOC content in soil from plots including both the fire, extra biomass and ash addition manipulations was twice as high as the average of the other treatment plots (Fig. 2). At day 12, DOC concentration tended ðP ¼ 0:11Þ to increase, by 85%, in response to fire. In the middle of the dry season (90 d) the fire manipulation and extra biomass addition resulted in 43 and 41% higher content of DOC, respectively (Fig. 2). The soil NHþ 4 –N concentration was low, increased during the experiment and was highest in the rainy season (210 d) 21 Corg (Fig. 2). In contrast, the averaging 116 mg NHþ 4 –N g 2 NO3 – N concentrations were generally lowest in the rainy 2 season and of similar concentration as NHþ 4 – N, but NO3 –N increased during the dry season to amounts five times higher þ than that of NHþ 4 – N. The NH4 – N content was generally not different among treatments, while the soil NO2 3 –N concentration was significantly higher (69, 47 and 76%) in burned than unburned plots, both 1, 12 and 90 d after the experiment began (Fig. 2). At day 12, this was accompanied by a more
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
853
pronounced increase when extra fuel load was also present in the burned plots, as shown by the tendency towards an interaction between fire and biomass ðP ¼ 0:08Þ: In the rainy season (210 d), the NO2 3 – N concentration was significantly lower both in plots with added ash and in plots with added extra grass biomass. 3.2. Soil microbial biomass C, PDA and basal and specific respiration A lower CmicFE was found at the beginning and in the middle of the dry season (12 and 90 d) ranging from 4.9 to 9.9 mg C g21 Corg, compared to a higher biomass in the rainy season (210 d; 27 –36 mg C g21 Corg) (Fig. 3). At the beginning of the dry season (1 and 12 d), CmicSIR was lowest, averaging 21 mg C g21 Corg, and it was highest in the middle of the dry season (90 d), averaging 34 mg C g21 Corg (Fig. 3). Likewise, PDA and basal respiration were lowest at the beginning of the dry season (12 d) and highest in the middle of the dry season (90 d) averaging 0.6 and 3.0 mg N2O – N g21 Corg h21 and 14 and 38 mg CO2 – C g21 Corg h21, respectively (Figs. 4 and 5). Specific respiration averaged 1.42, 0.62, 1.12 and 0.74 mg CO2 – C mg21 CmicSIR h21 at days 1, 12, 90 and 210, respectively (Fig. 5), and this variation was similar to the temporal variation of DOC (Fig. 2). At day 1 after the onset of the experiment, CmicFE declined in response to addition of extra grass biomass, especially when burned (Fig. 3), while PDA only increased after the fire manipulation in plots without extra biomass added (Fig. 4). The combination of fire and ash seemed to stimulate basal respiration, similar to the DOC response (Fig. 2), and hence also resulted in a tendency (FxA; P ¼ 0:10) of a higher specific respiration as shown by the interactions between fire and ash (Fig. 5). At day 12, both CmicFE and CmicSIR increased, by 52 and 20%, respectively, in burned compared to unburned plots without extra grass biomass (Fig. 3). PDA and basal respiration also increased in response to fire by 80 and 25% on average, respectively, although the increase in PDA was less pronounced when the fuel load was doubled, as supported by a tendency of significance of the fuel load factor (B; P ¼ 0:10) (Figs. 4 and 5). Specific respiration was higher in burned plots with added grass biomass (Fig. 5). In the middle of the dry season (90 d), there was no difference in CmicFE and basal respiration between the treatments, while CmicSIR was 19% higher in burned plots (Figs. 3 and 5). Hence, specific respiration decreased in R
Fig. 2. Soil dissolved organic C, NH4-N and NO2 3 -N concentrations in the experimental plots 1, 12, 90 and 210 d after the experiment began. The values are means ^ SE, n ¼ 6 per treatment. The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s, performed separately for each date, are presented. Significance levels: P , 0:05* ; P , 0:01**, P , 0:001***.
854
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
Fig. 4. Potential denitrification enzyme activity (PDA) of soil from the treatment plots 1, 12, 90 and 210 d after the experiment began. The values are means ^ SE, n ¼ 6 per treatment. The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s, performed separately for each date, are presented. Significance levels: P , 0:05*, P , 0:01**.
especially lower in burned plots with both extra fuel load and ash addition (Fig. 4). 3.3. In situ soil respiration and volumetric water content
Fig. 3. Soil microbial biomass C determined by fumigation–extraction (FE) and substrate-induced respiration (SIR) in the treatment plots 1, 12, 90 and 210 d after the experiment began. The values are means ^ SE, n ¼ 6 per treatment. The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s, performed separately for each date, are presented. Significance levels: P , 0:05* ; P , 0:01**, P , 0:001***.
response to fire in plots without extra fuel load, but increased in some burned plots with extra grass biomass (Fig. 5). PDA more than doubled and increased by 65% in burned plots with ambient and extra fuel load, respectively (Fig. 4). At day 210, CmicFE was 12% lower in soil from the burned than the unburned plots, while specific respiration tended to be higher in plots also including the ash manipulation (FxA; P ¼ 0:10) (Figs. 3 and 5). Further, PDA was on average 42% lower in soil from plots with ash addition compared to plots without ash addition, and it was
Soil respiration measurements carried out in the control and fire plots during the first 15 d of the experiment covered two incidents of precipitation, which were recorded as 9 and 12 mm rain at day 4 and 8, respectively (Fig. 6). This raised the volumetric water content of the upper 10 cm of soil from 6 to 12%, which corresponds to a WFPS of 15 and 31%, respectively. The variation in soil respiration was generally larger between days than between the control and burned plots and followed the variation in volumetric soil water content in 0– 10 cm depth. Moreover, the temporal pattern of fluctuation in soil respiration was generally the same in control and burned plots. However, at a single occasion 5 d after burning and 1 d after the first incident of precipitation soil respiration was twice as high in burned than control plots. The difference dissipated on the subsequent day and on day 7 and onwards both soil water content and soil respiration tended to be lower in burned than control plots. At day 12, the volumetric soil moisture content was 8% lower in all burned than all unburned plots ðP ¼ 0:07Þ (Fig. 7). At this time, the in situ soil respiration was significantly lower (21%) in all burned than all unburned plots correlating (r ¼ 0:83; P ¼ 0:01) with the lower volumetric soil water content in the burned plots (Fig. 7). Soil temperature at 5 cm depth ranged from 24– 30 8C during measurements.
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
Fig. 5. Basal respiration and biomass specific respiration rate of soil from the treatment plots 1, 12, 90 and 210 d after the experiment began. The values are means ^ SE, n ¼ 6 per treatment. The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s, performed separately for each date, are presented. Significance levels: P , 0:05*, P , 0:01**, P , 0:001***.
855
Fig. 6. In situ soil respiration (upper panel) and volumetric water content (lower panel) measured in the control (solid line) and fire plots (dashed line) during the initial 15 d of the experiment. Values are means ^ SE ðn ¼ 6Þ: No visible bars of SE for water content means that n ¼ 1: Burning of the fire plots is indicated by arrow. Two incidents of precipitation of 9 and 12 mm in the morning at day 4 and 8, respectively, are indicated by arrows.
4. Discussion 4.1. Responses in nitrification to fire The high soil NO3 – N concentration and the significant þ increase in NO2 3 – N compared to NH4 –N in the fire plots (Fig. 2) combined with the well-aerated soil conditions indicates that nitrifying bacteria were present in the soil and were stimulated by fire. The NHþ 4 –N was immediately oxidised to NO2 – N by nitrification. Low capacity of 3 nitrification of soil has been observed in Hyparrhenia
dominated savannahs in Zimbabwe and Coˆte d’Ivoire (Meiklejohn, 1968; Lensi et al., 1992). However, low and high nitrifying zones have been identified within the same savannah type dominated by H. diplandra (Le Roux et al., 1995; Lata et al., 1999). At the present study site dominated þ by H. confinis, the soil NO2 3 -to-NH4 ratio varying from unity to five implied substantial nitrification, and potential for loss of N by denitrification and leaching in the rainy season.
856
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
Fig. 7. Relationship between in situ soil respiration and volumetric water content in all treatment plots 11–13 d after the experiment began. Burned plots are indicated with filled symbols and unburned plots with open symbols. Values are means ^ SE (n ¼ 6 per treatment). The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s are presented. Significance level: P , 0:001***.
4.2. Responses in microbial biomass and activity, PDA, DOC and NO2 3 –N to fire The influence of fire on microbial biomass and activity was determined by several factors, including: (i) availability of C and N sources; (ii) heating from fire; (iii) presumably plant root dynamics in the burned plots in the subsequent rainy season. The present study shows that a relatively high fire severity caused by extra fuel load may result in a lower soil microbial biomass, while a low severity fire may stimulate microbial biomass, as shown by the significant interactions between fire and grass biomass 12 d after onset of the fire (Fig. 3). Microorganisms are killed by heating of soil to temperatures much below those relevant for heat-induced changes in SOM (DeBano et al., 1998). Hence, the direct heat effect on microbial biomass rather than an indirect fire effect on the quality of SOM may serve as explanation of the decrease in microbial biomass. The low severity of the fire with double fuel load, however, explains the transient adverse direct effect of heating on microbial biomass. The response to fire are in accord with Ojima et al. (1994), who found increased Cmic after burning in a tallgrass prairie. In the present experiment, the higher Cmic in burned than unburned plots up to 90 d after fire may have been due to higher concentrations of DOC and NO2 3 –N after fire, reducing the limitations of labile C and N on
microbial growth. Accordingly, PDA, basal respiration and biomass specific respiration rate were higher in burned than unburned plots (Figs. 4 and 5). Hence, depending on the fire severity, Cmic decreases or increases as affected directly or indirectly by burning. However, the results obtained on the effect of burning on the microbial community are related to the method used to determine microbial biomass and activity. During the dry season 90 d after fire, DOC and NO2 3 – N concentrations were still higher in the burned plots and responses of the microbial community to these changes were evident from the estimates of CmicSIR and PDA, but not of CmicFE. The different response between CmicSIR and CmicFE may be ascribed to the fact that the estimates reflect the active and total microbial biomass, respectively. The transient direct effect of the low intensity savannah fire on the microbial community was demonstrated by a shift to indirect effects of the fire during the rainy season 210 d after the experiment began. The ash deposition had reduced the soil NO2 3 – N concentrations possibly due to microbial consumption of NO2 3 – N when decomposing the organic C contained in the ash. Note, however, that ash application did not lead to enhanced C concentration in the labile DOC fraction. The lower NO2 3 – N concentration in plots with ash explains the lower PDA in plots with ash than without ash addition, especially in plots with extra fuel load, which was burned, thereby producing more ash. The slightly higher specific respiration in plots added ash, or where ash was produced by burning at day 210, may indicate a higher C availability for the microorganisms. The higher C availability in those plots is thus indicated by a bioassay and not by measurements of DOC concentrations in the soil. The C-to-N ratio of the ash was 31 and 26 in the , 500 and . 500 mm fraction, respectively, and of the soil it was around 20, and hence, both ash and soil had a higher ratio than generally found for microbial biomass (4 – 17), implying net N immobilisation by microorganisms. Further, a higher C availability is in accordance with a higher plant productivity observed in the burned plots (Jensen et al., 2001) since it is generally believed that root growth and activity favours microbial biomass and activity by root-derived production of dead organic matter. On the other hand, this also favours microbial grazers. Hence, the lower CmicFE in burned plots in the rainy season could be associated with increased grazing on microbes stimulated by ample rainfall and plant growth. The response in PDA to the manipulations seemed to depend more strongly on changes in soil NO2 3 –N concentrations than on changes in basal respiration and Cmic. Hence in this savannah woodland, PDA might depend more upon recent NO2 3 –N availability than the activity and biomass of the entire microbial community. Seasonal differences in labile C availability may explain part of the variation in the active microbial biomass (Hassink, 1993). In the present experiment, DOC peaked in the dry season (after 90 d) corresponding to the peak in CmicSIR, i.e. determined from the active component of the microbial community. However, the total microbial
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
biomass (CmicFE) reached a peak later, in the rainy season. This suggests that the environmental factors regulating the active fraction of the microbial biomass and the total microbial biomass are different. 4.3. Responses in soil emissions of CO2 to fire Many studies have found that following wetting of dry soil there is a pronounced increase in soil respiration, which is stimulated by the release of inorganic nutrients and organic substrates available for microorganisms (Ahlgren and Ahlgren, 1965; Gupta and Singh, 1981; Hao et al., 1988; Poth et al., 1995; Zepp et al., 1996). The first precipitation after burning resulted in higher soil respiration in burned than control plots (Fig. 6), presumably due to a distinct fire-liberated decomposable C pool, i.e. a fraction of the DOC, which increased after burning. This is supported by the higher basal respiration and specific respiration in burned than unburned plots. By contrast, after a week, in situ soil respiration tended to be lower in burned than control plots, although the basal and specific respiration analysed at optimal moisture conditions did not indicate that the decomposable C pool was fully utilised at that time. However, soil moisture is an important regulating factor for soil respiration in tropical grassland (Gupta and Singh, 1981; Hao et al., 1988), as also shown by the close covariation between respiration and water content (Fig. 6) and the lower soil water content and respiration in burned than unburned plots on day 12 (Fig. 7). Poth et al. (1995) found a higher soil emission of CO2 in burned than unburned plots in Brazilian tropical savannah after wetting soils. Conversely, in situ soil respiration was not different between burned and unburned plots after wetting in savannahs in Nigeria, Venezuela and South Africa (Adedeji, 1983; Hao et al., 1988; Zepp et al., 1996). Zepp et al. (1996) proposed that most of the microbial activity in savannah on well-drained soils takes place deeper in the soil than the soil layers reached by the effects of fire, while the contribution from root respiration may be negligible due to prolonged drought conditions. The gravimetric water content at 0– 5 cm soil depth was less in the burned than unburned plots until 90 d after the experiment began when the plant cover was still very scarce in the burned plots (Jensen et al., 2001), allowing drying of the soil surface by the intense solar radiation. Consequently, soil respiration was lower in burned than unburned plots in that time, suggesting that although significant amounts of C is lost from savannah ecosystems during fire, the subsequent C emission during the dry season may be lower due to stronger drying of the soil.
5. Conclusions The field experiment in savannah woodland demonstrated that the low severity fire indirectly stimulated Cmic
857
and PDA. In contrast, when the fuel load was doubled, a direct adverse effect on Cmic was shown and the fire-induced stimulation of PDA was less pronounced. The changes of Cmic, PDA and NO2 3 –N induced by the experimental manipulations in the dry season differed from the responses recorded 7 months later, in the rainy season. By this time, Cmic was lower in soil from burned plots, while PDA and NO2 3 –N was lower in plots with ash addition. An ephemeral burst in soil respiration was observed following rainfall; higher in burned than unburned plots. However, after 12 d both water content and soil respiration were lower in the burned plots. The difference in dry season emissions of CO2 caused by more intense drying of burned plots may be of significance considering the vast areas burned annually on the African continent. The fire-induced changes in microbial biomass and activity were relatively small compared to the substantial seasonal variation suggesting transient effects of the low severity experimental savannah fire on soil microbial functioning.
Acknowledgements This work was funded by the Danish Council for Development Research as part of the Fire in Tropical Ecosystems programme. We wish to thank Sten Struwe, Ib Friis, Menassie Gashaw, Mette Nordskov Jensen, Esben Vedel Jensen, Malgorzata Sylvester and Dessalegn Dessisa for their assistance in the work and The National Herbarium, Addis Ababa University, for institutional help during the fieldwork.
References Adedeji, F.O., 1983. Effect of fire on soil microbial activity in Nigerian southern Guinea savanna. Revue d’E´cologie et de Biologie du Sol 20, 483– 492. Ahlgren, I.F., Ahlgren, C.E., 1965. Effects of prescribed burning on soil microorganisms in a Minnesota Jack pine forest. Ecology 46, 304–310. Ajwa, H.A., Dell, C.J., Rice, C.W., 1999. Changes in enzyme activities and microbial biomass of tallgrass prairie soil as related to burning and nitrogen fertilization. Soil Biology & Biochemistry 31, 769 –777. Anderson, J.P.E., Domsch, K.H., 1978. A physiological method for the quantitative measurement of microbial biomass in soil. Soil Biology & Biochemistry 10, 215–221. Anderson, I.C., Levine, J.S., 1988. Enhanced biogenic emissions of nitric and nitrous oxide following surface biomass burning. Journal of Geophysical Research D 93, 3893–3898. Anderson, T.-H., Domsch, K.H., 1993. The metabolic quotient for CO2 (qCO2) as a specific activity parameter to assess the effects of environmental conditions, such as pH, on the microbial biomass of forest soils. Soil Biology & Biochemistry 25, 393–395. Araki, S., 1993. Effect on soil organic matter and soil fertility of the chitemene slash-and-burn practice used in northern Zambia. In: Mulongoy, K., Merckx, R. (Eds.), Soil Organic Matter Dynamics and Sustainability of Tropical Agriculture, Wiley-Sayce, K.U. Leuven, pp. 367–375.
858
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
Barbosa, P.M., Stroppiana, D., Gre´goire, J.-M., Pereira, J.M.C., 1999. An assessment of vegetation fire in Africa (1981–1991): burned areas, burned biomass, and atmospheric emissions. Global Biogeochemical Cycles 13, 933 –950. Bollen, G.J., 1969. The selective effect of heat treatment on the microflora of a greenhouse soil. Netherlands Journal of Plant Pathology 75, 157– 163. Cahoon, D.R., Stocks, B.J., Levine, J.S., Cofer, W.R., O’Neill, K.P., 1992. Seasonal distribution of African savanna fires. Nature 359, 812–815. Castaldi, S., Aragosa, D., 2002. Factors influencing nitrification and denitrification variability in a natural and fire-disturbed Mediterranean shrubland. Biology and Fertility of Soils 36, 418–425. Christensen, N.L., 1973. Fire and the nitrogen cycle in California chaparral. Science 181, 66–68. Corbet, A.S., 1934. Studies on tropical soil microbiology: II. The bacterial numbers in the soil of the Malay peninsula. Soil Science 38, 407–416. DeBano, L.F., Neary, D.G., Ffolliot, P.F., 1998. Fire’s Effects on Ecosystems. Wiley, New York. Deka, H.K., Mishra, R.R., 1983. The effect of slash burning on soil microflora. Plant and Soil 73, 167 –175. Dhillion, S.S., Anderson, R.C., 1994. Root growth, and microorganisms associated with the rhizoplane and root zone soil of a native C4 grass on burned and unburned sand prairies. Biology and Fertility of Soils 17, 115– 120. Dunn, P.H., Barro, S.C., Poth, M., 1985. Soil moisture affects survival of microorganisms in heated chaparral soil. Soil Biology & Biochemistry 17, 143 –148. Frost, P.G.H., Robertson, F., 1987. The ecological effects of fire in savannas. In: Walker, B.H., (Ed.), Determinants of Tropical Savannas, IRL Press, Oxford, pp. 93–140. Garcia, F.O., Rice, C.W., 1994. Microbial biomass dynamics in tallgrass prairie. Soil Science Society America Journal 58, 816–823. Gashaw, M., Michelsen, A., 2001. Soil seed bank dynamics and aboveground cover of a dominant grass, Hyparrhenia confinis, in regularly burned savanna types in Gambella, western Ethiopia. Biologiske Skrifter 54, 389–397. Gupta, S.R., Singh, J.S., 1981. Soil respiration in a tropical grassland. Soil Biology & Biochemistry 13, 261– 268. Hao, W.M., Scharffe, D., Crutzen, P.J., Sanhueza, E., 1988. Production of N2O, CH4, and CO2 from soils in tropical savanna during the dry season. Journal of Atmospheric Chemistry 7, 93 –105. Hassink, J., 1993. Relationship between the amount and the activity of the microbial biomass in Dutch grassland soils: comparison of the fumigation-incubation method and the substrate-induced respiration method. Soil Biology & Biochemistry 25, 533–538. Jenkinson, D.S., Powlson, D.S., 1976. The effects of biocidal treatments on metabolism in soil—V. A method for measuring soil biomass. Soil Biology & Biochemistry 8, 209–213. Jensen, M., Friis, I., 2001. Fire regimes, floristics, diversity, life-forms and biomass in wooded grasslands, woodlands and dry forest in Gambella, SW Ethiopia. Biologiske Skrifter 54, 349 –387. Jensen, M., Michelsen, A., Gashaw, M., 2001. Responses in plant, soil inorganic and microbial nutrient pools to experimental fire, ash and biomass addition in a woodland savanna. Oecologia 128, 85 – 93. Jones, C.L., Smithers, N.L., Scholes, M.C., Scholes, R.J., 1990. The effect of fire frequency on the organic components of a basaltic soil in the Kruger National Park. South African Journal of Plant and Soil 7, 236– 238. Kaiser, P., 1983. The role of soil micro-organisms in savanna ecosystems. In: Bourlie`re, F., (Ed.), Tropical Savannas, Elsevier, Amsterdam, pp. 541 –557.
Kovacic, D.A.D., Swift, M., Ellis, J.E., Hakonson, T.E., 1986. Immediate effects of prescribed burning on mineral soil nitrogen in ponderosa pine of New Mexico. Soil Science 141, 71–76. Lata, J.C., Durand, J., Lensi, R., Abbadie, L., 1999. Stable coexistence of contrasted nitrification statuses in a wet tropical savanna ecosystem. Functional Ecology 13, 762 –768. Lensi, R., Domenach, A.M., Abbadie, L., 1992. Field study of nitrification and denitrification in a wet savanna of West Africa (Lamto Coˆte d’Ivoire). Plant and Soil 147, 107 –113. Le Roux, X., Abbadie, L., Lensi, R., Serc¸a, D., 1995. Emission of nitrogen monoxide from African tropical ecosystems: control of emission by soil characteristics in humid and dry savannas of West Africa. Journal of Geophysical Research D 100, 23133– 23142. Meiklejohn, J., 1955. The effect of bush burning on the microflora of a Kenya upland soil. Journal of Soil Science 6, 111 –118. Meiklejohn, J., 1968. Numbers of nitrifying bacteria in some Rhodesian soils under natural grass and improved pastures. Journal of Applied Ecology 5, 291–300. Menaut, J.-C., Abbadie, L., Vitousek, P.M., 1993. Nutrient and organic matter dynamics in tropical ecosystems. In: Crutzen, P.J., Goldammer, J.G. (Eds.), Fire in the Environment: The Ecological, Atmospheric and Climatic Importance of Vegetation Fires, Wiley, Chichester, pp. 215–231. Ojima, D.S., Schimel, D.S., Parton, W.J., Owensby, C.E., 1994. Long- and short-term effects of fire on nitrogen cycling in tallgrass prairie. Biogeochemistry 24, 67–84. Pell, M., Stenberg, B., Stenstro¨m, J., Torstensson, L., 1996. Potential denitrification activity assay in soil-with or without chloramphenicol? Soil Biology & Biochemistry 28, 393–398. Poth, M., Anderson, I.C., Miranda, H.S., Miranda, A.C., Riggan, P.J., 1995. The magnitude and persistence of soil NO, N2O, CH4, and CO2 fluxes from burned tropical savanna in Brazil. Global Biogeochemical Cycles 9, 503–513. Raison, R.J., 1979. Modification of the soil environment by vegetation fires, with particular reference to nitrogen transformations: a review. Plant and Soil 51, 73 –108. Singh, R.S., 1994. Changes in soil nutrients following burning of dry tropical savanna. International Journal of Wildland Fire 4, 187–194. Singh, R.S., Raghubanshi, A.S., Singh, J.S., 1991. Nitrogen-mineralization in dry tropical savanna: effects of burning and grazing. Soil Biology & Biochemistry 23, 269–273. Smith, M.S., Tiedje, J.M., 1979. Phases of denitrification following oxygen depletion in soil. Soil Biology & Biochemistry 11, 261– 267. Statistical Analysis Systems Institute, 1997. SAS/STAT Users Guide, Release 6.12, SAS Institute, Cary. Stock, W.D., Lewis, O.A.M., 1986. Soil nitrogen and the role of fire as a mineralizing agent in a South African coastal fynbos ecosystem. Journal of Ecology 74, 317–328. Vance, E.D., Brookes, P.C., Jenkinson, D.S., 1987. An extraction method for measuring soil microbial biomass. Soil Biology & Biochemistry 19, 703 –707. Van Reenen, C.A., Visser, G.J., Loos, M.A., 1992. Soil microorganisms and activities in relation to season, soil factors and fire. In: Van Wilgen, B.W., Richardson, D.M., Kruger, F.J., Van Hensbergen, H.J. (Eds.), Fire in South African Mountain Fynbos. Ecosystem, Community and Species Responses at Swartboskloof, Springer, Berlin, pp. 258–272. Wu, J., Joergensen, R.G., Pommerening, B., Chaussod, R., Brookes, P.C., 1990. Measurement of soil microbial biomass C by fumigation–extraction, an automated procedure. Soil Biology & Biochemistry 22, 1167–1169. Zepp, R.G., Miller, W.L., Burke, R.A., Parsons, D.A.B., Scholes, M.C., 1996. Effects of moisture and burning on soil-atmosphere exchange of trace carbon gases in a southern African savanna. Journal of Geophysical Research D 101, 23699– 23706.
Tropical savannah woodland: effects of experimental fire on soil microorganisms and soil emissions of carbon dioxide Michael Anderssona,*, Anders Michelsenb, Michael Jensenb, Annelise Kjøllera a
Department of General Microbiology, University of Copenhagen, Sølvgade 83 H, DK-1307 Copenhagen K, Denmark b Botanical Institute, University of Copenhagen, Øster Farimagsgade 2 D, DK-1353 Copenhagen K, Denmark Received 7 March 2003; received in revised form 22 October 2003; accepted 21 January 2004
Abstract Burning of the vegetation in the African savannahs in the dry season is widespread and may have significant effects on soil chemical and biological properties. A field experiment in a full factorial randomised block design with fire, ash and extra grass biomass as main factors was carried out in savannah woodland of the Gambella region in Ethiopia. The microbial biomass C (Cmic) was 52% (fumigation – extraction) and 20% (substrate-induced respiration) higher in burned than unburned plots 12 d after burning. Both basal respiration and potential denitrification enzyme activity (PDA) immediately responded to burning and increased after treatment. However, in burned plots addition of extra biomass (fuel load) led to a reduction of Cmic and PDA due to enhanced fire temperature. Five days after burning, there was a short-lived burst in the in situ soil respiration following rainfall, with twice as high soil respiration in burned than unburned plots. In contrast, 12 d after burning soil respiration was 21% lower in the burned plots, coinciding with lower soil water content in the same plots. The fire treatment resulted in higher concentrations of dissolved organic C (24– 85%) and nitrate (47 – 76%) in the soil until 90 d after burning, while soil 2 þ NHþ 4 – N was not affected to the same extent. The increase in soil NO3 – N but not NH4 – N in the burned plots together with the well-aerated 2 soil conditions indicated that nitrifying bacteria were stimulated by fire and immediately oxidised NHþ 4 – N to NO3 – N. In the subsequent rainy season, NO2 – N and, consequently, PDA were reduced by ash deposition. Further, C was lower in burned plots at that time. 3 mic However, the fire-induced changes in microbial biomass and activity were relatively small compared to the substantial seasonal variation, suggesting transient effects of the low severity experimental fire on soil microbial functioning. q 2004 Elsevier Ltd. All rights reserved. Keywords: Ash; Basal respiration; Dissolved organic carbon; Emissions of carbon dioxide; Fuel load; Microbial biomass carbon; Nitrate; Potential denitrification enzyme activity; Savannah fire
1. Introduction Across the Sahelian-north Guinean vegetation zone in Africa, burning of the savannah vegetation is a frequent, widely distributed phenomenon in the beginning of the dry season (Cahoon et al., 1992; Barbosa et al., 1999). The fires are mainly due to human activities; the frequency of the fires at present has increased as a consequence of an expanding exploitation of natural resources caused by the high growth rate of the African human population, often resulting in annual to biennial burning of the savannah areas in this region. The predominant early burning may increase the soil organic carbon (Corg) compared to protected plots, but a paradox arises by the change of fire regimes towards higher * Corresponding author. Tel.: þ 45-3532-2054; fax: þ 45-35-32-2040. E-mail address: [email protected] (M. Andersson). 0038-0717/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.soilbio.2004.01.015
fire frequencies, which may have detrimental consequences for soil fertility (Kaiser, 1983; Frost and Robertson, 1987; Jones et al., 1990; Menaut et al., 1993). Since changes in soil fertility and soil C stocks reflect alterations in the microbial decomposer community composition and activity, a better understanding of the response of soil microbial performance to savannah fires is required. The components of fire can be separated into a direct and several indirect effects on soil microbial communities. During fire, the heating of soil may kill a fraction of the microorganisms. With fungi being more susceptible to heat than bacteria (Bollen, 1969; Dunn et al., 1985) fire may lead to a changed microbial community composition. Immediately after burning, the number of bacteria and fungi and the microbial biomass is often reduced in the subsurface soil layer (Meiklejohn, 1955; Deka and Mishra, 1983; Van Reenen et al., 1992). However, the reduced microbial
850
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
numbers have also been attributed to higher temperatures in burned plots with bare soil than in unburned plots with plant cover due to solar radiation (Kaiser, 1983). Ahlgren and Ahlgren (1965) argued, that the immediate decrease in the number of soil microorganisms after fire in pine forest was caused by the heat from the fire, while the subsequent increase in number of microorganisms after a rain event could be attributed to the ash deposition. Other studies have reported increased number of microorganisms after slash burning in virgin tropical forest, and higher microbial biomass in burned than unburned plots in woodland savannah and tallgrass prairie (Corbet, 1934; Garcia and Rice, 1994; Singh, 1994). In savannah ecosystems, some studies suggest that the responses of the soil microbial communities to fire are restricted to the uppermost centimetres of the soil layer, and effects may have disappeared at the onset of the rainy season (Raison, 1979; Adedeji, 1983; Deka and Mishra, 1983; Van Reenen et al., 1992). The microorganisms killed by the heat from the fire are a source of ammonium and dissolved organic C (DOC), which can stimulate microbial growth. Higher NHþ 4 –N concentrations following burning have been attributed to deposition of ash, heating of soil, reduced microbial N immobilisation or reduced N plant uptake (Christensen, 1973; Kovacic et al., 1986; Stock and Lewis, 1986; Anderson and Levine, 1988; Araki, 1993; Singh, 1994). The heating of soil during fire releases NHþ 4 – N from protein-like components of organomineral soil complexes and from combustion of organic matter (Raison, 1979; Kovacic et al., 1986). Further, fireinduced modifications of soil chemical properties are directly related to the intensity of soil heating and the amount of ash deposition (Raison, 1979). Basically, changes in microbial activity caused by environmental changes may be independent of possible changes in microbial biomass. Some, but not all, studies have found higher C and N mineralisation, and in situ soil respiration rates after burning in savannahs and woodland, and after burning and wetting in savannah (Adedeji, 1983; Hao et al., 1988; Singh et al., 1991; Van Reenen et al., 1992; Araki, 1993; Poth et al., 1995; Zepp et al., 1996). The fireinduced increase in microbial activity and soil respiration found in these studies have been explained by higher pH, N availability, pool of readily decomposable substrates and temperature in soil from burned than unburned plots, while the lack of responses in soil respiration have been attributed to low soil moisture. These results reflect to some extent the site specific differences in soil type, fire intensity and savannah vegetation type as well as differences in time elapsed since fire took place. In savannah ecosystems, the effects of fire on microbial biomass and C and N mineralisation rates may be minor compared to the pronounced seasonal variability (Singh et al., 1991). However, burning may strongly promote root growth and aboveground plant production in the rainy season subsequent to fire events (Menaut et al., 1993;
Dhillion and Anderson, 1994; Jensen et al., 2001), and hence have strong effects on savannah ecosystems. Soil enzymatic activities act as potentially sensitive indicators of soil quality and have been used in studies of vegetation burning in prairie and Mediterranean shrubland (Ajwa et al., 1999; Castaldi and Aragosa, 2002), but enzymes of primarily microbial origin have seldom been studied in relation to fire effects on the soil microbial communities in African savannahs. More studies are necessary to provide a thorough understanding of the function of soil microbial communities in fire-prone savannah ecosystems. Apparently, there has not been any study that has separated the effects of fire on microbial biomass and activity into the heat and ash components in a controlled field experiment in African fire-prone savannah ecosystems. The objective of this study was to investigate the effects of 2 fire intensity and ash deposition on soil NHþ 4 –N, NO3 –N, DOC, microbial biomass C, basal respiration at optimal water conditions, potential denitrification enzyme activity (PDA) and in situ soil respiration. The variables are expressed per unit Corg, since microbial biomass and activity covary with the content of Corg.
2. Materials and methods 2.1. Study site The experiment was carried out in a woodland savannah in the south-western part of Ethiopia about 12 km south of the provincial capital Gambella (88090 N, 348340 E) from November 1997 to July 1998. Annual precipitation is on average 900 mm with the major part falling from late May to October. Daily variations in temperature are from 20 8C at night to 37 8C during the day, with minor seasonal variation. At the study site, the tree canopy cover was 60% and the grass vegetation covered 88% of the ground dominated by Hyparrhenia confinis var. nudiglumis, which is an annual to semi-perennial grass (Jensen et al., 2001; Gashaw and Michelsen, 2001). The grass biomass was approximately 480 g dry weight m22 by the end of the rainy season (Jensen et al., 2001). Soil pHKCl was 5.8 to 6.6 and the treatments described below elevated the pH by less than 0.2 units, while the soil organic matter (SOM) content, total N and P was about 4, 0.10 and 0.02%, respectively, and unaffected by treatments (Jensen et al., 2001). At the study area, annual fires occur in the dry season from November to May, most frequent by the beginning of the dry season. Grazing by wild animals and cattle is insignificant. A full description of the vegetation at the study site and for the Gambella region in general is given in Jensen and Friis (2001). 2.2. Experimental design Fire effects were investigated by analysing three experimental factors: fire, grass biomass addition before
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
burning (i.e. extra fuel load) and ash addition after burning. The experiment was carried out in a full factorial randomised block design with six replicate blocks, i.e. with a total of 48 plots. The blocks were distributed randomly not more than 50 m apart. Each block consisted of eight plots of 2 £ 2 m2 that were separated by fire belts of 2 m width. In order to minimise border effects, the central 1 £ 1 m2 only was used for sampling. The standing grass biomass of 480 g dw m22 was doubled in the extra biomass (fuel load) treatments. The ash addition of approximately 15 g m22 was accomplished by burning a similar sized plot adjacent to the blocks. The total C and N, loss on ignition and pH of the ash was 9.4, 0.30, 15.8 and 11.0% in the , 500 mm fraction and 14.5, 0.56, 27.3 and 9.1% in the . 500 mm fraction, respectively, as determined by methods described in Jensen et al. (2001). The fuel load was set on fire 20 November, 1997. During burning the soil temperature at 1 cm depth increased from 25 to 37 and 48 8C in the fire treatment and in the fire plus extra grass biomass treatment 7 min after fire, respectively, showing that the fire was of relatively low intensity due to the moderate grass biomass at the site (Jensen et al., 2001). 2.3. Soil sampling and extractions Soil sampling was carried out on four occasions; at 1, 12, 90 and 210 d after initiation of the experiment. In each plot two soil samples were collected with a soil corer (5 cm dia.) from 0 to 5 cm depth and mixed and sieved (, 2 mm). The chloroform fumigation and KCl extractions were performed 1 – 4 d after sampling in the field, while subsamples for other purposes were kept at 15 8C for a maximum of 3 weeks in Ethiopia and at 4 8C when returning to the laboratory in Denmark until analysis. Seven grams of fresh soil were shaken for 1 h in 35 ml 2 M KCl. At the same time, 7 g of fresh soil were fumigated with chloroform for 24 h to release C from the microbial biomass (Jenkinson and Powlson, 1976), followed by extraction. Extracts were kept at the same temperatures as the soil samples. 2.4. Gas flux measurements Field measurements of soil emissions of CO2, predominantly originating from microbial and root respiration, were performed in areas with no aboveground plant cover using an infrared gas analyser equipped with a soil respiration chamber (EGM-1/SRC-1, PP-Systems, UK) covering 78.5 cm2 and with a volume of 1.17 l. The CO2 flux was measured in all plots immediately before commencing the experiment and these data were used as covariate in the statistical analysis of CO2 flux measured in duplicate in all treatment plots 11 –13 d after initiation of the experiment. In addition, CO2 flux was measured regularly during the first 15 d after burning in all control and burned-only plots.
851
2.5. Temperature, soil moisture and precipitation measurements Concomitantly with determination of soil emissions of CO2, temperature was measured at 5 cm depth inserting a probe horizontally into the soil. Volumetric soil water content in the 0 – 10 cm soil layer was measured using a Time Domain Reflectrometry (TDR) measuring device (TRIME-FM, IMKO, Germany) equipped with a 10 cm probe (P2 rods). For calculation of water-filled pore space (WFPS), bulk density was determined by collecting soil of a known volume, and the soil particle density was determined by displacement with water carefully eliminating any air bubbles. On the first 15 d of the experiment precipitation was measured daily in six containers placed as far as possible from tree canopies. 2.6. Laboratory analyses 2.6.1. Soil water content and organic matter Soil water content was measured gravimetrically and SOM was determined by incinerating dry soil for 6 h at 550 8C. Corg was estimated as 50% of SOM. 2.6.2. Inorganic N Whatman GF-D filtered soil extracts were analysed for 2 NHþ 4 – N and NO3 – N with the indophenol method using a Hitachi U-2000 spectrophotometer and the Cd reduction method on an Aquatec 5400 analyser, respectively. 2.6.3. Microbial biomass C and basal respiration Microbial biomass C was determined by the chloroform fumigation –extraction method (Vance et al., 1987) with some modifications (7 g fresh soil; 2 M KCl solution) and substrate-induced respiration (SIR) according to the criteria of Anderson and Domsch (1978). DOC was measured on KCl extracts of fumigated and unfumigated samples (see Section 2.3) with a Shimadzu 5000 TOC analyser, and microbial biomass C (CmicFE) was calculated as the difference between fumigated and non-fumigated extracts applying a kEC of 0.45 (Wu et al., 1990). SIR was measured in fresh soil equivalent to 10 g dw placed in 50 ml serum bottles. Glucose solution was added in droplets to 60% of water holding capacity (WHC) resulting in a final concentration of 2 mg glucose g21 dw soil and mixed thoroughly into the soil. The bottles with soil were conditioned for 1.5 h, sealed with butyl rubber stoppers, and incubated for 2 h at 25 8C. For measurements of basal respiration the soil water content was adjusted to 60% of WHC and conditioned at 25 8C overnight before sealing, and subsequently incubated for 24 h. The production of CO2 from the soil in both assays was determined by taking gas samples at the beginning and the end of the incubation for immediate analysis for CO2 on a HP 5890 GC equipped with a temperature conductivity detector (TCD) and a Hayesep Q 80-100 mesh column. Hydrogen was used
852
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
as carrier gas at 20 ml min21 and the injector, oven and detector temperatures were 25, 70 and 265 8C, respectively. SIR was converted to microbial biomass C (CmicSIR) using a conversion factor of 40 (Anderson and Domsch, 1978). The glucose concentration and the duration of the conditioning and incubation periods in the SIR assay had been determined in tests for this particular soil. The biomass specific respiration rate, also known as the metabolic quotient (qCO2) (Anderson and Domsch, 1993), was calculated by dividing basal respiration with CmicSIR. 2.6.4. Potential denitrification enzyme activity PDA was determined according to Smith and Tiedje (1979) and modifications of the method by Pell et al. (1996) in triplicate for each soil sample. The soil was transferred to 30 8C 24 h before performing the PDA assay. Thereafter, an anaerobic (N2) slurry was mixed in 118 ml bottles containing soil equal to 5 g dw and 5 ml of a solution of 1 mM KNO3 and 1 mM glucose. To inhibit the enzymatic reduction of N2O, 10 kPa C2H2 (10%) was injected. A gas sample of 2 ml was taken at t ¼ 0; the bottles were incubated at 30 8C and shaken at 250 rev min21 for 5 h. Gas samples were also taken at 1, 2, 3.5 and 5 h. The gas samples were transferred to Venojecte tubes and analysed for N2O by gas chromatography. The HP 5890 GC was equipped with an electron capture detector (ECD) and a Hayesep Q 80-100 mesh column with N2 (20 ml min21) as carrier gas and injector, oven and detector temperatures of 120, 25 and 265 8C, respectively. The PDA at the beginning of the incubation ðt ¼ 0Þ was determined by non-linear regression using the equation given by Pell et al. (1996). 2.7. Statistical analysis Data were tested statistically, separately for each sampling date, with analysis of variance (ANOVA) using the SAS procedure GLM with type III sum of squares (SAS Institute, 1997). The model used was a four way ANOVA, with fire, biomass, ash and block as main factors, and including all interactions between fire, biomass and ash. An optimisation of the model included a removal of the block factor and the three factor interaction, if non-significant at alpha ¼ 0.05. The model used in the ANOVA for the CO2 flux involved CO2 flux data prior to initiation of the experiment treated as covariate, while the block was not included. A correlation between soil respiration and water content was analysed by Pearson correlation.
3. Results 3.1. Soil moisture, DOC and inorganic N The gravimetric water content in the 0-5 cm soil profile was significantly lower in burned plots on day 12 and 90 (Fig. 1). The gravimetric water content on day 12
Fig. 1. Soil gravimetric water content in the experimental plots 1, 12, 90 and 210 d after the experiment began. The values are means ^ SE, n ¼ 6 per treatment. The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s, performed separately for each date, are presented. Significance levels: P , 0:01**, P , 0:001***.
corresponded to a WFPS of 37 and 32% in unburned and burned plots, respectively, applying a bulk density of 1.42 g dw cm23 and a soil particle density of 2.3 g dw cm23. In the middle of the dry season (90 d) WFPS was 16 and 12% in unburned and burned plots, respectively. The DOC concentration differed markedly between the sampling times and the mean content of DOC was 1.1, 0.5, 3.4 and 0.3 mg C g21 Corg at days 1, 12, 90 and 210, respectively (Fig. 2). The production of DOC was stimulated by the treatments after 1, 12 and 90 d. At day 1, DOC content in soil from plots including both the fire, extra biomass and ash addition manipulations was twice as high as the average of the other treatment plots (Fig. 2). At day 12, DOC concentration tended ðP ¼ 0:11Þ to increase, by 85%, in response to fire. In the middle of the dry season (90 d) the fire manipulation and extra biomass addition resulted in 43 and 41% higher content of DOC, respectively (Fig. 2). The soil NHþ 4 –N concentration was low, increased during the experiment and was highest in the rainy season (210 d) 21 Corg (Fig. 2). In contrast, the averaging 116 mg NHþ 4 –N g 2 NO3 – N concentrations were generally lowest in the rainy 2 season and of similar concentration as NHþ 4 – N, but NO3 –N increased during the dry season to amounts five times higher þ than that of NHþ 4 – N. The NH4 – N content was generally not different among treatments, while the soil NO2 3 –N concentration was significantly higher (69, 47 and 76%) in burned than unburned plots, both 1, 12 and 90 d after the experiment began (Fig. 2). At day 12, this was accompanied by a more
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
853
pronounced increase when extra fuel load was also present in the burned plots, as shown by the tendency towards an interaction between fire and biomass ðP ¼ 0:08Þ: In the rainy season (210 d), the NO2 3 – N concentration was significantly lower both in plots with added ash and in plots with added extra grass biomass. 3.2. Soil microbial biomass C, PDA and basal and specific respiration A lower CmicFE was found at the beginning and in the middle of the dry season (12 and 90 d) ranging from 4.9 to 9.9 mg C g21 Corg, compared to a higher biomass in the rainy season (210 d; 27 –36 mg C g21 Corg) (Fig. 3). At the beginning of the dry season (1 and 12 d), CmicSIR was lowest, averaging 21 mg C g21 Corg, and it was highest in the middle of the dry season (90 d), averaging 34 mg C g21 Corg (Fig. 3). Likewise, PDA and basal respiration were lowest at the beginning of the dry season (12 d) and highest in the middle of the dry season (90 d) averaging 0.6 and 3.0 mg N2O – N g21 Corg h21 and 14 and 38 mg CO2 – C g21 Corg h21, respectively (Figs. 4 and 5). Specific respiration averaged 1.42, 0.62, 1.12 and 0.74 mg CO2 – C mg21 CmicSIR h21 at days 1, 12, 90 and 210, respectively (Fig. 5), and this variation was similar to the temporal variation of DOC (Fig. 2). At day 1 after the onset of the experiment, CmicFE declined in response to addition of extra grass biomass, especially when burned (Fig. 3), while PDA only increased after the fire manipulation in plots without extra biomass added (Fig. 4). The combination of fire and ash seemed to stimulate basal respiration, similar to the DOC response (Fig. 2), and hence also resulted in a tendency (FxA; P ¼ 0:10) of a higher specific respiration as shown by the interactions between fire and ash (Fig. 5). At day 12, both CmicFE and CmicSIR increased, by 52 and 20%, respectively, in burned compared to unburned plots without extra grass biomass (Fig. 3). PDA and basal respiration also increased in response to fire by 80 and 25% on average, respectively, although the increase in PDA was less pronounced when the fuel load was doubled, as supported by a tendency of significance of the fuel load factor (B; P ¼ 0:10) (Figs. 4 and 5). Specific respiration was higher in burned plots with added grass biomass (Fig. 5). In the middle of the dry season (90 d), there was no difference in CmicFE and basal respiration between the treatments, while CmicSIR was 19% higher in burned plots (Figs. 3 and 5). Hence, specific respiration decreased in R
Fig. 2. Soil dissolved organic C, NH4-N and NO2 3 -N concentrations in the experimental plots 1, 12, 90 and 210 d after the experiment began. The values are means ^ SE, n ¼ 6 per treatment. The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s, performed separately for each date, are presented. Significance levels: P , 0:05* ; P , 0:01**, P , 0:001***.
854
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
Fig. 4. Potential denitrification enzyme activity (PDA) of soil from the treatment plots 1, 12, 90 and 210 d after the experiment began. The values are means ^ SE, n ¼ 6 per treatment. The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s, performed separately for each date, are presented. Significance levels: P , 0:05*, P , 0:01**.
especially lower in burned plots with both extra fuel load and ash addition (Fig. 4). 3.3. In situ soil respiration and volumetric water content
Fig. 3. Soil microbial biomass C determined by fumigation–extraction (FE) and substrate-induced respiration (SIR) in the treatment plots 1, 12, 90 and 210 d after the experiment began. The values are means ^ SE, n ¼ 6 per treatment. The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s, performed separately for each date, are presented. Significance levels: P , 0:05* ; P , 0:01**, P , 0:001***.
response to fire in plots without extra fuel load, but increased in some burned plots with extra grass biomass (Fig. 5). PDA more than doubled and increased by 65% in burned plots with ambient and extra fuel load, respectively (Fig. 4). At day 210, CmicFE was 12% lower in soil from the burned than the unburned plots, while specific respiration tended to be higher in plots also including the ash manipulation (FxA; P ¼ 0:10) (Figs. 3 and 5). Further, PDA was on average 42% lower in soil from plots with ash addition compared to plots without ash addition, and it was
Soil respiration measurements carried out in the control and fire plots during the first 15 d of the experiment covered two incidents of precipitation, which were recorded as 9 and 12 mm rain at day 4 and 8, respectively (Fig. 6). This raised the volumetric water content of the upper 10 cm of soil from 6 to 12%, which corresponds to a WFPS of 15 and 31%, respectively. The variation in soil respiration was generally larger between days than between the control and burned plots and followed the variation in volumetric soil water content in 0– 10 cm depth. Moreover, the temporal pattern of fluctuation in soil respiration was generally the same in control and burned plots. However, at a single occasion 5 d after burning and 1 d after the first incident of precipitation soil respiration was twice as high in burned than control plots. The difference dissipated on the subsequent day and on day 7 and onwards both soil water content and soil respiration tended to be lower in burned than control plots. At day 12, the volumetric soil moisture content was 8% lower in all burned than all unburned plots ðP ¼ 0:07Þ (Fig. 7). At this time, the in situ soil respiration was significantly lower (21%) in all burned than all unburned plots correlating (r ¼ 0:83; P ¼ 0:01) with the lower volumetric soil water content in the burned plots (Fig. 7). Soil temperature at 5 cm depth ranged from 24– 30 8C during measurements.
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
Fig. 5. Basal respiration and biomass specific respiration rate of soil from the treatment plots 1, 12, 90 and 210 d after the experiment began. The values are means ^ SE, n ¼ 6 per treatment. The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s, performed separately for each date, are presented. Significance levels: P , 0:05*, P , 0:01**, P , 0:001***.
855
Fig. 6. In situ soil respiration (upper panel) and volumetric water content (lower panel) measured in the control (solid line) and fire plots (dashed line) during the initial 15 d of the experiment. Values are means ^ SE ðn ¼ 6Þ: No visible bars of SE for water content means that n ¼ 1: Burning of the fire plots is indicated by arrow. Two incidents of precipitation of 9 and 12 mm in the morning at day 4 and 8, respectively, are indicated by arrows.
4. Discussion 4.1. Responses in nitrification to fire The high soil NO3 – N concentration and the significant þ increase in NO2 3 – N compared to NH4 –N in the fire plots (Fig. 2) combined with the well-aerated soil conditions indicates that nitrifying bacteria were present in the soil and were stimulated by fire. The NHþ 4 –N was immediately oxidised to NO2 – N by nitrification. Low capacity of 3 nitrification of soil has been observed in Hyparrhenia
dominated savannahs in Zimbabwe and Coˆte d’Ivoire (Meiklejohn, 1968; Lensi et al., 1992). However, low and high nitrifying zones have been identified within the same savannah type dominated by H. diplandra (Le Roux et al., 1995; Lata et al., 1999). At the present study site dominated þ by H. confinis, the soil NO2 3 -to-NH4 ratio varying from unity to five implied substantial nitrification, and potential for loss of N by denitrification and leaching in the rainy season.
856
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
Fig. 7. Relationship between in situ soil respiration and volumetric water content in all treatment plots 11–13 d after the experiment began. Burned plots are indicated with filled symbols and unburned plots with open symbols. Values are means ^ SE (n ¼ 6 per treatment). The treatments are control (C), fire (F), ash addition after burning (A), doubling of the grass biomass (fuel load) (B), and combinations hereof. Only the significant main factors and interactions in the ANOVA’s are presented. Significance level: P , 0:001***.
4.2. Responses in microbial biomass and activity, PDA, DOC and NO2 3 –N to fire The influence of fire on microbial biomass and activity was determined by several factors, including: (i) availability of C and N sources; (ii) heating from fire; (iii) presumably plant root dynamics in the burned plots in the subsequent rainy season. The present study shows that a relatively high fire severity caused by extra fuel load may result in a lower soil microbial biomass, while a low severity fire may stimulate microbial biomass, as shown by the significant interactions between fire and grass biomass 12 d after onset of the fire (Fig. 3). Microorganisms are killed by heating of soil to temperatures much below those relevant for heat-induced changes in SOM (DeBano et al., 1998). Hence, the direct heat effect on microbial biomass rather than an indirect fire effect on the quality of SOM may serve as explanation of the decrease in microbial biomass. The low severity of the fire with double fuel load, however, explains the transient adverse direct effect of heating on microbial biomass. The response to fire are in accord with Ojima et al. (1994), who found increased Cmic after burning in a tallgrass prairie. In the present experiment, the higher Cmic in burned than unburned plots up to 90 d after fire may have been due to higher concentrations of DOC and NO2 3 –N after fire, reducing the limitations of labile C and N on
microbial growth. Accordingly, PDA, basal respiration and biomass specific respiration rate were higher in burned than unburned plots (Figs. 4 and 5). Hence, depending on the fire severity, Cmic decreases or increases as affected directly or indirectly by burning. However, the results obtained on the effect of burning on the microbial community are related to the method used to determine microbial biomass and activity. During the dry season 90 d after fire, DOC and NO2 3 – N concentrations were still higher in the burned plots and responses of the microbial community to these changes were evident from the estimates of CmicSIR and PDA, but not of CmicFE. The different response between CmicSIR and CmicFE may be ascribed to the fact that the estimates reflect the active and total microbial biomass, respectively. The transient direct effect of the low intensity savannah fire on the microbial community was demonstrated by a shift to indirect effects of the fire during the rainy season 210 d after the experiment began. The ash deposition had reduced the soil NO2 3 – N concentrations possibly due to microbial consumption of NO2 3 – N when decomposing the organic C contained in the ash. Note, however, that ash application did not lead to enhanced C concentration in the labile DOC fraction. The lower NO2 3 – N concentration in plots with ash explains the lower PDA in plots with ash than without ash addition, especially in plots with extra fuel load, which was burned, thereby producing more ash. The slightly higher specific respiration in plots added ash, or where ash was produced by burning at day 210, may indicate a higher C availability for the microorganisms. The higher C availability in those plots is thus indicated by a bioassay and not by measurements of DOC concentrations in the soil. The C-to-N ratio of the ash was 31 and 26 in the , 500 and . 500 mm fraction, respectively, and of the soil it was around 20, and hence, both ash and soil had a higher ratio than generally found for microbial biomass (4 – 17), implying net N immobilisation by microorganisms. Further, a higher C availability is in accordance with a higher plant productivity observed in the burned plots (Jensen et al., 2001) since it is generally believed that root growth and activity favours microbial biomass and activity by root-derived production of dead organic matter. On the other hand, this also favours microbial grazers. Hence, the lower CmicFE in burned plots in the rainy season could be associated with increased grazing on microbes stimulated by ample rainfall and plant growth. The response in PDA to the manipulations seemed to depend more strongly on changes in soil NO2 3 –N concentrations than on changes in basal respiration and Cmic. Hence in this savannah woodland, PDA might depend more upon recent NO2 3 –N availability than the activity and biomass of the entire microbial community. Seasonal differences in labile C availability may explain part of the variation in the active microbial biomass (Hassink, 1993). In the present experiment, DOC peaked in the dry season (after 90 d) corresponding to the peak in CmicSIR, i.e. determined from the active component of the microbial community. However, the total microbial
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
biomass (CmicFE) reached a peak later, in the rainy season. This suggests that the environmental factors regulating the active fraction of the microbial biomass and the total microbial biomass are different. 4.3. Responses in soil emissions of CO2 to fire Many studies have found that following wetting of dry soil there is a pronounced increase in soil respiration, which is stimulated by the release of inorganic nutrients and organic substrates available for microorganisms (Ahlgren and Ahlgren, 1965; Gupta and Singh, 1981; Hao et al., 1988; Poth et al., 1995; Zepp et al., 1996). The first precipitation after burning resulted in higher soil respiration in burned than control plots (Fig. 6), presumably due to a distinct fire-liberated decomposable C pool, i.e. a fraction of the DOC, which increased after burning. This is supported by the higher basal respiration and specific respiration in burned than unburned plots. By contrast, after a week, in situ soil respiration tended to be lower in burned than control plots, although the basal and specific respiration analysed at optimal moisture conditions did not indicate that the decomposable C pool was fully utilised at that time. However, soil moisture is an important regulating factor for soil respiration in tropical grassland (Gupta and Singh, 1981; Hao et al., 1988), as also shown by the close covariation between respiration and water content (Fig. 6) and the lower soil water content and respiration in burned than unburned plots on day 12 (Fig. 7). Poth et al. (1995) found a higher soil emission of CO2 in burned than unburned plots in Brazilian tropical savannah after wetting soils. Conversely, in situ soil respiration was not different between burned and unburned plots after wetting in savannahs in Nigeria, Venezuela and South Africa (Adedeji, 1983; Hao et al., 1988; Zepp et al., 1996). Zepp et al. (1996) proposed that most of the microbial activity in savannah on well-drained soils takes place deeper in the soil than the soil layers reached by the effects of fire, while the contribution from root respiration may be negligible due to prolonged drought conditions. The gravimetric water content at 0– 5 cm soil depth was less in the burned than unburned plots until 90 d after the experiment began when the plant cover was still very scarce in the burned plots (Jensen et al., 2001), allowing drying of the soil surface by the intense solar radiation. Consequently, soil respiration was lower in burned than unburned plots in that time, suggesting that although significant amounts of C is lost from savannah ecosystems during fire, the subsequent C emission during the dry season may be lower due to stronger drying of the soil.
5. Conclusions The field experiment in savannah woodland demonstrated that the low severity fire indirectly stimulated Cmic
857
and PDA. In contrast, when the fuel load was doubled, a direct adverse effect on Cmic was shown and the fire-induced stimulation of PDA was less pronounced. The changes of Cmic, PDA and NO2 3 –N induced by the experimental manipulations in the dry season differed from the responses recorded 7 months later, in the rainy season. By this time, Cmic was lower in soil from burned plots, while PDA and NO2 3 –N was lower in plots with ash addition. An ephemeral burst in soil respiration was observed following rainfall; higher in burned than unburned plots. However, after 12 d both water content and soil respiration were lower in the burned plots. The difference in dry season emissions of CO2 caused by more intense drying of burned plots may be of significance considering the vast areas burned annually on the African continent. The fire-induced changes in microbial biomass and activity were relatively small compared to the substantial seasonal variation suggesting transient effects of the low severity experimental savannah fire on soil microbial functioning.
Acknowledgements This work was funded by the Danish Council for Development Research as part of the Fire in Tropical Ecosystems programme. We wish to thank Sten Struwe, Ib Friis, Menassie Gashaw, Mette Nordskov Jensen, Esben Vedel Jensen, Malgorzata Sylvester and Dessalegn Dessisa for their assistance in the work and The National Herbarium, Addis Ababa University, for institutional help during the fieldwork.
References Adedeji, F.O., 1983. Effect of fire on soil microbial activity in Nigerian southern Guinea savanna. Revue d’E´cologie et de Biologie du Sol 20, 483– 492. Ahlgren, I.F., Ahlgren, C.E., 1965. Effects of prescribed burning on soil microorganisms in a Minnesota Jack pine forest. Ecology 46, 304–310. Ajwa, H.A., Dell, C.J., Rice, C.W., 1999. Changes in enzyme activities and microbial biomass of tallgrass prairie soil as related to burning and nitrogen fertilization. Soil Biology & Biochemistry 31, 769 –777. Anderson, J.P.E., Domsch, K.H., 1978. A physiological method for the quantitative measurement of microbial biomass in soil. Soil Biology & Biochemistry 10, 215–221. Anderson, I.C., Levine, J.S., 1988. Enhanced biogenic emissions of nitric and nitrous oxide following surface biomass burning. Journal of Geophysical Research D 93, 3893–3898. Anderson, T.-H., Domsch, K.H., 1993. The metabolic quotient for CO2 (qCO2) as a specific activity parameter to assess the effects of environmental conditions, such as pH, on the microbial biomass of forest soils. Soil Biology & Biochemistry 25, 393–395. Araki, S., 1993. Effect on soil organic matter and soil fertility of the chitemene slash-and-burn practice used in northern Zambia. In: Mulongoy, K., Merckx, R. (Eds.), Soil Organic Matter Dynamics and Sustainability of Tropical Agriculture, Wiley-Sayce, K.U. Leuven, pp. 367–375.
858
M. Andersson et al. / Soil Biology & Biochemistry 36 (2004) 849–858
Barbosa, P.M., Stroppiana, D., Gre´goire, J.-M., Pereira, J.M.C., 1999. An assessment of vegetation fire in Africa (1981–1991): burned areas, burned biomass, and atmospheric emissions. Global Biogeochemical Cycles 13, 933 –950. Bollen, G.J., 1969. The selective effect of heat treatment on the microflora of a greenhouse soil. Netherlands Journal of Plant Pathology 75, 157– 163. Cahoon, D.R., Stocks, B.J., Levine, J.S., Cofer, W.R., O’Neill, K.P., 1992. Seasonal distribution of African savanna fires. Nature 359, 812–815. Castaldi, S., Aragosa, D., 2002. Factors influencing nitrification and denitrification variability in a natural and fire-disturbed Mediterranean shrubland. Biology and Fertility of Soils 36, 418–425. Christensen, N.L., 1973. Fire and the nitrogen cycle in California chaparral. Science 181, 66–68. Corbet, A.S., 1934. Studies on tropical soil microbiology: II. The bacterial numbers in the soil of the Malay peninsula. Soil Science 38, 407–416. DeBano, L.F., Neary, D.G., Ffolliot, P.F., 1998. Fire’s Effects on Ecosystems. Wiley, New York. Deka, H.K., Mishra, R.R., 1983. The effect of slash burning on soil microflora. Plant and Soil 73, 167 –175. Dhillion, S.S., Anderson, R.C., 1994. Root growth, and microorganisms associated with the rhizoplane and root zone soil of a native C4 grass on burned and unburned sand prairies. Biology and Fertility of Soils 17, 115– 120. Dunn, P.H., Barro, S.C., Poth, M., 1985. Soil moisture affects survival of microorganisms in heated chaparral soil. Soil Biology & Biochemistry 17, 143 –148. Frost, P.G.H., Robertson, F., 1987. The ecological effects of fire in savannas. In: Walker, B.H., (Ed.), Determinants of Tropical Savannas, IRL Press, Oxford, pp. 93–140. Garcia, F.O., Rice, C.W., 1994. Microbial biomass dynamics in tallgrass prairie. Soil Science Society America Journal 58, 816–823. Gashaw, M., Michelsen, A., 2001. Soil seed bank dynamics and aboveground cover of a dominant grass, Hyparrhenia confinis, in regularly burned savanna types in Gambella, western Ethiopia. Biologiske Skrifter 54, 389–397. Gupta, S.R., Singh, J.S., 1981. Soil respiration in a tropical grassland. Soil Biology & Biochemistry 13, 261– 268. Hao, W.M., Scharffe, D., Crutzen, P.J., Sanhueza, E., 1988. Production of N2O, CH4, and CO2 from soils in tropical savanna during the dry season. Journal of Atmospheric Chemistry 7, 93 –105. Hassink, J., 1993. Relationship between the amount and the activity of the microbial biomass in Dutch grassland soils: comparison of the fumigation-incubation method and the substrate-induced respiration method. Soil Biology & Biochemistry 25, 533–538. Jenkinson, D.S., Powlson, D.S., 1976. The effects of biocidal treatments on metabolism in soil—V. A method for measuring soil biomass. Soil Biology & Biochemistry 8, 209–213. Jensen, M., Friis, I., 2001. Fire regimes, floristics, diversity, life-forms and biomass in wooded grasslands, woodlands and dry forest in Gambella, SW Ethiopia. Biologiske Skrifter 54, 349 –387. Jensen, M., Michelsen, A., Gashaw, M., 2001. Responses in plant, soil inorganic and microbial nutrient pools to experimental fire, ash and biomass addition in a woodland savanna. Oecologia 128, 85 – 93. Jones, C.L., Smithers, N.L., Scholes, M.C., Scholes, R.J., 1990. The effect of fire frequency on the organic components of a basaltic soil in the Kruger National Park. South African Journal of Plant and Soil 7, 236– 238. Kaiser, P., 1983. The role of soil micro-organisms in savanna ecosystems. In: Bourlie`re, F., (Ed.), Tropical Savannas, Elsevier, Amsterdam, pp. 541 –557.
Kovacic, D.A.D., Swift, M., Ellis, J.E., Hakonson, T.E., 1986. Immediate effects of prescribed burning on mineral soil nitrogen in ponderosa pine of New Mexico. Soil Science 141, 71–76. Lata, J.C., Durand, J., Lensi, R., Abbadie, L., 1999. Stable coexistence of contrasted nitrification statuses in a wet tropical savanna ecosystem. Functional Ecology 13, 762 –768. Lensi, R., Domenach, A.M., Abbadie, L., 1992. Field study of nitrification and denitrification in a wet savanna of West Africa (Lamto Coˆte d’Ivoire). Plant and Soil 147, 107 –113. Le Roux, X., Abbadie, L., Lensi, R., Serc¸a, D., 1995. Emission of nitrogen monoxide from African tropical ecosystems: control of emission by soil characteristics in humid and dry savannas of West Africa. Journal of Geophysical Research D 100, 23133– 23142. Meiklejohn, J., 1955. The effect of bush burning on the microflora of a Kenya upland soil. Journal of Soil Science 6, 111 –118. Meiklejohn, J., 1968. Numbers of nitrifying bacteria in some Rhodesian soils under natural grass and improved pastures. Journal of Applied Ecology 5, 291–300. Menaut, J.-C., Abbadie, L., Vitousek, P.M., 1993. Nutrient and organic matter dynamics in tropical ecosystems. In: Crutzen, P.J., Goldammer, J.G. (Eds.), Fire in the Environment: The Ecological, Atmospheric and Climatic Importance of Vegetation Fires, Wiley, Chichester, pp. 215–231. Ojima, D.S., Schimel, D.S., Parton, W.J., Owensby, C.E., 1994. Long- and short-term effects of fire on nitrogen cycling in tallgrass prairie. Biogeochemistry 24, 67–84. Pell, M., Stenberg, B., Stenstro¨m, J., Torstensson, L., 1996. Potential denitrification activity assay in soil-with or without chloramphenicol? Soil Biology & Biochemistry 28, 393–398. Poth, M., Anderson, I.C., Miranda, H.S., Miranda, A.C., Riggan, P.J., 1995. The magnitude and persistence of soil NO, N2O, CH4, and CO2 fluxes from burned tropical savanna in Brazil. Global Biogeochemical Cycles 9, 503–513. Raison, R.J., 1979. Modification of the soil environment by vegetation fires, with particular reference to nitrogen transformations: a review. Plant and Soil 51, 73 –108. Singh, R.S., 1994. Changes in soil nutrients following burning of dry tropical savanna. International Journal of Wildland Fire 4, 187–194. Singh, R.S., Raghubanshi, A.S., Singh, J.S., 1991. Nitrogen-mineralization in dry tropical savanna: effects of burning and grazing. Soil Biology & Biochemistry 23, 269–273. Smith, M.S., Tiedje, J.M., 1979. Phases of denitrification following oxygen depletion in soil. Soil Biology & Biochemistry 11, 261– 267. Statistical Analysis Systems Institute, 1997. SAS/STAT Users Guide, Release 6.12, SAS Institute, Cary. Stock, W.D., Lewis, O.A.M., 1986. Soil nitrogen and the role of fire as a mineralizing agent in a South African coastal fynbos ecosystem. Journal of Ecology 74, 317–328. Vance, E.D., Brookes, P.C., Jenkinson, D.S., 1987. An extraction method for measuring soil microbial biomass. Soil Biology & Biochemistry 19, 703 –707. Van Reenen, C.A., Visser, G.J., Loos, M.A., 1992. Soil microorganisms and activities in relation to season, soil factors and fire. In: Van Wilgen, B.W., Richardson, D.M., Kruger, F.J., Van Hensbergen, H.J. (Eds.), Fire in South African Mountain Fynbos. Ecosystem, Community and Species Responses at Swartboskloof, Springer, Berlin, pp. 258–272. Wu, J., Joergensen, R.G., Pommerening, B., Chaussod, R., Brookes, P.C., 1990. Measurement of soil microbial biomass C by fumigation–extraction, an automated procedure. Soil Biology & Biochemistry 22, 1167–1169. Zepp, R.G., Miller, W.L., Burke, R.A., Parsons, D.A.B., Scholes, M.C., 1996. Effects of moisture and burning on soil-atmosphere exchange of trace carbon gases in a southern African savanna. Journal of Geophysical Research D 101, 23699– 23706.