Improving the accuracy of aboveground biomass estimates for secondary tropical dry forests with allometric models

Background/Question/Methods

Biomass in tropical forests constitutes important C reservoirs and their accurate estimation is needed for an adequate assessment of global C budgets. Biomass estimates have been obtained mainly for old-growth forests, but the accelerated expansion of secondary forests through tropical landscapes urges for the development of tools for an accurate estimation of their C storage. In this study we developed allometric models to estimate aboveground biomass (AGB) of secondary tropical dry forests (STDF).We selected 27 woody species which contribute with 75-80% of total basal area within nine STDF plots (< 30 yr-old) in Jalisco, Mexico. A total of 304 trees (10-14 per species) with diameter at breast height (DBH) ≥ 1 cm were measured for total height (H) and DBH, and harvested with chainsaw. Fresh weight of each tree was determined in the field after cutting. Subsamples of stems and branches were taken to the laboratory for dry weight and wood specific gravity (WSG) determinations. Both power and logarithmic models were fitted to relate tree AGB to one or a combination of three predictors (DBH, WSG and H).

Results/Conclusions

At the species level, simple power and log-transformed linear models with only DBH as a predictor variable, explained a similar high percentage of variance (R² >0.90). At the multispecies level, linear models had generally a better fit than the power models (R²= 0.917 – 0.975).The simple power multispecies model with DBH improved when the species were grouped into low, medium and high WSG categories (from R²=0.798 to 0.950). Our single- or multi-species power allometric models are highly recommended because they use an easy-to-measure variable (DBH), avoid the systematic bias introduced by the back-transformation of the estimated values, and explain >80% of the AGB at landscape level. Our sample of harvested trees, based on a landscape approach, allowed us to generate more accurate models applicable to Mexican STDF, which will contribute to the design and implementation of REDD+ policies. Finally, we compared the accuracy of our observed AGB values with those predicted by allometric models developed for other tropical forests. Biomass estimates varied strongly among models; therefore, better predictions of the landscape AGB can be obtained by using preferably site-specific allometric models as opposed to foreign models.