Additivity of biomass equations means that the different equations for biomass fractions (foliage, branches, stem, bark, etc.) should add up to the same biomass as if an equation for the total aboveground biomass is used for biomass estimation to have a consistent set of models (Kozak 1970; Cunia and Briggs 1984,1985). This consis­tency between model estimates for biomass fractions and total biomass is not trivial since regressions are not perfect and even a slight bias for a model of a biomass component will affect the sum and define a deviation from an estimation of the total model. In addition, contemporaneous correlations between equations must be taken into account. If this is not done the efficiency of the model is reduced and biased parameter estimates could be the result as shown by Parresol and Thomas (1996).

For this reason different methods have been applied to ensure additivity (Parresol 1999). The most widely applied approach is a multivariate regression procedure based on the simultaneous estimation of all equations with joint-generalized least squares (Cunia and Briggs 1984, 1985), also called seemingly unrelated regression (SUR) in the more recent literature (Parresol 2001; Saint-Andre et al. 2005; Brandeis et al. 2006). Segmented regression and penalised spline approaches have also been tested to force additivity in biomass models (Goicoa et al. 2011).