Data from 57 permanent monitoring sites are used to document the growth in woody vegetation and estimate the carbon sink in 27 M ha of eucalypt woodlands (savannas), contained within c. 60 M ha of grazed woodlands in Queensland (northeast Australia). The study sites are shown to be representative of the environment and structure of the eucalypt woodlands in the defined study area. Mean basal area increment for all live woody plants in 30 long‐term sites, with an average initial basal area of 11.86 ± 1.38 (SE) m2 ha−1, was 1.06 m2 ha−1 over a mean 14 years timeframe. The majority of the measurement period, commencing between 1982 and 1988, was characterized by below‐average rainfall. The increase in live tree basal area was due primarily to growth of existing trees (3.12 m2 ha−1) rather than establishment of new plants (0.25 m2 ha−1) and was partly offset by death (2.31 m2 ha−1). A simple but robust relationship between stand basal area and stand biomass of all woody species was developed for the eucalypt dominant woodlands. Analysis of above‐ground carbon stocks in live and standing dead woody plants gave a mean net above‐ground annual carbon increment for all 57 sites of 0.53 t C ha−1 y−1, similar to values estimated elsewhere in world savannas. Published root : shoot ratios were used to infer C flux in woody root systems on these sites. This results in an estimated sink in above‐ and below‐ground biomass of 18 Mt C y−1 over the eucalypt woodlands studied, and potentially up to 35 Mt C y−1 if extended to all grazed woodlands in Queensland. It is suggested that introduction of livestock grazing and altered fire regimes have triggered the change in tree‐grass dominance in these woodlands. Thus, change in carbon stocks in the grazed woodlands of Queensland is identified as an important component of human‐induced greenhouse gas flux in Australia, equivalent in magnitude to c. 25% of the most recently published (1999) total estimated national net emissions. The latter inventory takes into account emissions from land clearing, but does not include the sink identified in the present study. This sequestration also represents a small but significant contribution to the global terrestrial carbon sink.

Allometric equations are presented relating stem circumference to branch, leaf, trunk, bark, total above-ground and lignotuber biomass for Eucalyptus crebra F.Muell. (woodland trees), E. melanophloia Sol. Ex Gaerth. (both woodland and regrowth community trees) and E. populnea F.Muell. (woodland trees). There were no significant differences (P > 0.05) between the slopes of individual lognormal regression lines plotting stem circumference against total above-ground biomass for E. crebra, E. melanophloia and E. populnea. Root-to-shoot ratios and leaf area indices were also determined for the stands contributing to each regression. The regressions were then applied to measured eucalypt stems in the associated plant community to give estimates of each stand’s component (eucalypt tree fraction only) biomass per hectare. These eucalypt regressions were next applied to measured stems of each species on a total of 33 woodland sites in which these eucalypts individually contributed > 75% of total site basal area. Above-ground biomass/basal area relationships averaged 6.74 0.29 t m–2 basal area for 11 E. crebra sites, 5.11 0.28 t m–2 for 12 E. melanophloia sites and 5.81 0.11 t m–2 for 10 E. populnea sites. The mean relationship for all sites was 5.86 0.18 t m–2 basal area. The allometric relationships presented at both individual tree and stand levels, along with calculated biomass : basal area relationships, enable ready estimates to be made of above-ground biomass (carbon stocks) in woodlands dominated by these eucalypts in Queensland, assuming individual stem circumferences or community basal areas are known. However, to document changes in carbon stocks (e.g. for Greenhouse Gas Inventory or Carbon Offset trading purposes), more attention needs to be placed on monitoring fluxes in the independent variables (predictors) of these allometric equations.