Simple allometric models can facilitate biomass estimates for "fertilizer trees" in southern Africa

Background/Question/Methods

Agroforestry – the integration of trees with crops – can potentially help to mitigate climate change while also enhancing other ecosystem services and rural livelihoods. To quantify these benefits, efficient techniques are needed to estimate agroforestry tree biomass non-destructively. This study aimed to develop biomass estimation methods for southern Africa’s “fertilizer tree” systems (in which nitrogen-fixing trees such as Gliricidia sepium and Tephrosia candida are intercropped or rotated with maize).

A field trial was conducted at Makoka, Malawi (15º30’S, 35º15’E) in 2008-09 and 2009-10. The trial included three fertilizer tree systems in 24 plots (each 7m x 8m). Height of each individual tree (Gliricidia: n = 336; Tephrosia: n= 1792) was measured several times per growing season. At the end of each season, tree biomass was partially or totally harvested depending on the system.

Some trees (Gliricidia: n = 37; Tephrosia: n= 76) were weighed individually, and these data were used to create non-linear allometric models of aboveground biomass (using JMP 10.0.0 and the corrected Akaike Information Criterion). The allometric models were then used to estimate whole-plot biomass based on the known tree heights within each plot.

Results/Conclusions

For Gliricidia, the best-fitting model relating height (h, in cm) and biomass (B, in g FW) was a simple quadratic (B = 0.51197h + 0.020243h²), whereas Tephrosia required an exponential (B = -90.18 + 53.43e^0.01654h). When using the height of each individual tree in a plot as input, these models predicted whole-plot tree biomass with great accuracy (Gliricidia: R² = 0.93, slope = 1.06; Tephrosia: R²= 0.86, slope = 1.08).

Because measuring every tree within a plot is time-consuming, we also investigated the accuracy of using subsets of the tree height measurements. For both species, using height measurements from only 10% of the trees in each plot predicted plot-level biomass with reasonable accuracy. However, as farmers’ fields are more heterogeneous than research plots, actual farms may require somewhat more sampling effort than these results suggest.

More work is needed to develop allometric models and biomass estimation methods for other agroforestry systems, but these data show that simple models and efficient sampling techniques can provide useful information about tree biomass on smallholder farms.