Allometric models for estimating aboveground biomass in the tropical woodlands of Ghana, West Africa

Abstract

Abstract Background Modelling aboveground biomass (AGB) in forest and woodland ecosystems is critical for accurate estimation of carbon stocks. However, scarcity of allometric models for predicting AGB remains an issue that has not been adequately addressed in Africa. In particular, locally developed models for estimating AGB in the tropical woodlands of Ghana have received little attention. In the absence of locally developed allometric models, Ghana will continue to use Tier 1 biomass data through the application of pantropic models. Without local allometric models it is not certain how Ghana would achieve Tier 2 and 3 levels under the United Nations programme for reducing emissions from deforestation and forest degradation. The objective of this study is to develop a mixed-species allometric model for use in estimating AGB for the tropical woodlands in Ghana. Destructive sampling was carried out on 745 trees (as part of charcoal production) for the development of allometric equations. Diameter at breast height (dbh, i.e. 1.3 m above ground level), total tree height (H) and wood density (ρ) were used as predictors for the models. Seven models were compared and the best model selected based on model efficiency, bias (%) and corrected Akaike Information Criterion. The best model was validated by comparing its results with those of the pantropic model developed by Chave et al. (Glob Chang Biol 20:3177–3190, 2014) using equivalence test and conventional paired t-test. Results The results revealed that the best model for estimating AGB in the tropical woodlands is AGB = 0.0580ρ((dbh)2H)0.999. The equivalence test showed that this model and the pantropic model developed by Chave et al. (Glob Chang Biol 20:3177–3190, 2014) were equivalent within ±10% of their mean predictions (p-values < 0.0001 for one-tailed t-tests for both lower and upper bounds at 5% significant level), while the paired t-test revealed that the mean (181.44 ± 18.25 kg) of the model predictions of the best model of this study was significantly (n = 745, mean diff. = 16.50 ± 2.45 kg; S.E. = 1.25 kg; p < 0.001) greater than that (164.94 ± 15.82 kg) of the pantropic model of Chave et al. (Glob Chang Biol 20:3177–3190, 2014). Conclusion The model developed in this study fills a critical gap in estimating AGB in tropical woodlands in Ghana and other West African countries with similar ecological conditions. Despite the equivalence with the pantropic model it remains superior to the model of Chave et al. (Glob Chang Biol 20:3177–3190, 2014) for the estimation of AGB in local tropical woodlands. It is a relevant tool for the attainment of Tier 2 and 3 levels for REDD+. The model is recommended for use in the tropical woodlands in Ghana and other West African countries in place of the use of pantropic models.