Regression models for estimating charcoal yield in a Eucalyptus camaldulensis plantation in the semi-arid region of northeastern Nigeria

Abstract

Stepwise multiple regression technique was used to develop logarithmic and quadratic models for estimating charcoal yield from a 25-year-old Eucalyptus camaldulensis plantation in the semi-arid region in northeastern Nigeria. Forty-eight (48) mean trees were clear-felled and immediately separated into components: bole, branches and twigs. The green weights of three replicate wood discs were were measured and were thereafter dried in the oven until constant weight was attained. The fresh : dry weight ratio of the samples were used to estimate the dry biomass of the mean trees. Charcoal yield was determined by carbonizing the oven-dried wood samples (of known weights) at 350oC for 3 hours. The total charcoal production per tree ranged from 82.62 to 290.07 kg (average; 145.86 ± 7.43 kg). The mean contribution by tree components to total yield were: bole, 84.75 ± 4.59 kg; branches, 45.42 ± 2.39 kg; and twigs, 15.59 ± 0.93 kg (i.e. 58.1%, 31.1% and 10.7% respectively). Total charcoal per tree increased as the dbh of the trees increased. The same trend was observed in the yeild from both boles and branches but was slightly different from that of twigs. The total aboveground biomass production per hectare (256,245.2 kg) culminated into 128,940.2 kg of charcoal. Simple linear, quadratic, semi-log and double log equations were fitted as appropriate, for total charcoal, bole charcoal, branch charcoal and twig charcoal as dependent variables, but the best-fit models found were semi-log and double log and quadratic models that incorporated diameter at breast height (dbh), dbh2, dbh2H, and the product of dbh and merchantable height [(dbh)MH] as independent variables. Results of residual analysis showed that the models satisfied all the assumptions of regression analysis.

Keywords: Models, charcoal production, biomass, Eucalyptus, arid, anergy, allometric

Bowen Journal of Agriculture Vol. 5 (1&2) 2008: pp. 20-31