The fumigation-incubation (FI) and the fumigation-extraction (FE) ninhydrin methods for quantifying the microbial biomass pool were tested and the relationships between microbial biomass characteristics and soil organic matter fractions (separated following particle size) investigated for a range of soils representative for the West-African moist savanna zone (13 soils). Three soils from the humid forest zone were also included. Microbial C values calculated using the CO2-C production of the fumigated soils from d 10 to d 20 as control (Bio-C (II)) were better correlated with the flush of ninhydrin reactive N (NRN) during 10 d of fumigation than those calculated with the CO2-C production in the unfumigated soils from d 10 to d 20 as control (Bio-C (I)). The Bio-C (II) values also showed a more consistent range of values (22-210 mg C kg(-1) soil) than the Bio-C (I) values. Using a k(C) factor of 0.35, which was considered to be more appropriate for our soils than a k(C) factor of 0.45, k(C,NRN) could be estimated as 22 and 16, for a 5- and 10-d fumigation, respectively. For all savanna soils, the NRN flush after 5 d of fumigation was closely related to the NRN flush after 10 d, indicating that a 5-d fumigation was sufficient provided that the k-values are adapted. Mineral N flushes during incubation and incubation after fumigation were small. Although microbial N values calculated as [NH4+-N flush of the fumigated soils (0-10 d)]/k(N) with k(N) = -0.014 x (CO2-C-flush-to-NH4+-N-flush during fumigation) +0.39, which gave values of 14.0-100.7 mg N kg(-1) soil, showed the best relationship with the NRN flush after 10 d of fumigation, microbial N values calculated as [(mineral N flush of the fumigated soils after 10 d)-(mineral N flush of the unfumigated soils after 10 d)]/0.68, which gave values of 3.4-46.2 mg N kg(-1) soil, including 3 values <2 mg N kg(-1) soil and 1 negative value, yielded microbial C-to-N ratios (4.7-16.4) which were better reflecting the ratio C-flush-to-N-flush of fumigated soils (5.2-13.5). Using a k(N) value of 0.68, k(N,NRN) could be estimated as 1.5. Between 7 and 17% and between 11 and 28% of the total soil C was part of the soil litter (SL) (organic particles larger than 250 mu m) and the particulate organic matter (POM) (organic particles larger than 53 mu m), respectively. The CO2-C production of the unfumigated soils was strongly (P < 0.01) related to the SL-C content. Inclusion of the silt and clay content in a linear regression equation increased the R-2 value from 0.70 to 0.91. The Bio-C (II) content showed the best relationship with the C content of the soil particles between 53 and 20 mu m and the silt fraction. The NRN flush after 10 d showed the best relationship with the C content of the particles between 250 and 53 mu m and between 53 and 20 mu m. Maximally 48% of the variation in Bio-C (II) values was explained by the C content of the various fractions, indicating that the present methods do not exclusively measure the active microbial biomass. (C) 1999 Elsevier Science Ltd. All rights reserved.
