Estimating net biomass production and loss from repeated measurements of trees in forests and woodlands: Formulae, biases and recommendations

Export citation

There is widespread interest in ensuring that assessment and knowledge of changes in forest biomass, and associated carbon gains or losses, are accurate and unbiased. Repeated measurements of individually-marked trees in permanent plots permit the estimation of rates of biomass production by tree growth and recruitment and of loss from mortality. But there are challenges, for example, simple estimates of production rate (i.e., the sum of biomass gain by growth of surviving trees and new recruits divided by census duration) decline as the census interval increases due to unrecorded growth. Even if we allow for these unobserved changes, additional biases may arise due to the non-independence of growth and mortality and to the heterogeneity and compositional changes within the forest. Here we examine these issues and demonstrate how problems can be minimized. We provide and compare alternative approaches to estimate net biomass production and loss from tree growth and mortality. Under the assumption that specific rates of biomass production and loss, i.e., turnover, are constant over time, we derive estimates of absolute biomass turnover rates that are independent of census duration. We show census-interval dependence of simple turnover rates grows with increasing specific turnover rates. While the time-dependent bias in simple estimates has previously been suggested to increase in proportion to the square of production, we show this relationship is approximately linear. Correlations between stem growth and mortality do not influence our estimates. We account for biomass gain by recruited stems without discounting their initial biomass in production estimates. We can reduce additional biases by accounting for differences in turnover among subpopulations (such as species, sites) and changes in their abundances. We provide worked examples from four forests covering a range of conditions (in Indonesia and Japan) and show the effects of accounting for these biases. For example, over five years in an Indonesian rain forest, simple estimates and instantaneous estimates neglecting species heterogeneity underestimated production by 4.9% and 1.6%, respectively when compared to comprehensive (instantaneous species-structured) estimates.

Altmetric score:
Dimensions Citation Count:

Related publications

Get the CIFOR latest news