Allometric models for prediction of aboveground biomas of single trees in Miombo woodlands in Tanzania

Abstract

Abstract Allometric models for prediction of above ground biomass of miombo trees was developed using data from four sites in Tanzania. A total of 167 sample trees were collected from four sites representing both wet and dry miombo woodlands in Tanzania. Ten randomly selected sample trees from each site were kept for model validation while the remaining trees were used for calibration of the biomass models. Site, miombo categories, and general models (combining all data) were developed. The Pseudo-R2 of the final site specific models was 0.96, 0.96, 0.92 and 0.95 for Babati, Lindi, Mpanda, and Tabora, respectively. The Pseudo-R2 of both the miombo categories and the general biomass model was 0.95. Site specific, miombo categories and general biomass models have been developed and tested in this study. Models validation shows that the developed biomass models can sufficiently be applied to their respective sites and miombo categories. While acceptable bias% (< 10%) was mostly found which was not significantly higher than zero (p > 0.05), unacceptable bias% (> 10%) were found when Babati site specific biomass model had applied to Lindi and wet miombo biomass model was applied to dry miombo. This leads to the conclusion that for precise biomass estimation, the site specific and miombo category biomass model may not be applied to other site and miombo category respectively. A significant fit of general biomass model were found to each separate study sites and miombo categories when tested to their respective independent data. Since the general biomass models covered both wet and dry miombo woodlands, which are the major categories of miombo woodlands, the developed general model are recommended to be applied to entire miombo woodlands of Tanzania with similar conditions as the study sites. However, for improved biomass estimates of study sites, the site specific biomass model should preferably be used than the general biomass model. In addition, it should be noted that the developed biomass models predict total above ground tree biomass where separate tree components models i.e. stem and branch, and below ground also need to be developed.