Integrative modelling approach to assess the sustainability of the Eucalyptus plantations in Congo

Since 1978, 42000 ha of clonal Eucalyptus plantations have been established in the littoral savannas of Congo, mainly for pulpwood production. These plantations were established on sandy soils, characterised by low reserves of available nutrients and a low water retention capacity. An intensive silviculture is carried out, resulting in the removal of high amounts of biomass and nutrients every 7 years. Therefore, a decision-making tool is needed (i) to assess stand productions under different silvicultural options, (ii) to evaluate the risks of nutrient deficiencies and non sustainable production, and (iii) to estimate the economic return of different harvesting strategies. For many years, field trials have been established focusing on clonal selection, silvicultural practices, stand growth and quantification of the biogeochemical cycles of nutrients. This high amount of data was used (or will be used) to elaborate a chain of models named "EUCALYPT-Dendro" that includes three main modules ; • the growth module : a single tree distance-independent model was elaborated. The dominant height growth model is similar to the equation developed by Dhote (1996). However, two main differences can be reported : (i) the model was segmented in order to take into account a change in height growth rate that occurs during stand rotation, (ii) tree spacing affected the dominant height growth and was introduced as regressor in the equation. The stand basal area increment is function of the dominant height increment. Individual tree basal area is function of a potential (given by the stand basal area increment) and a reductor (given by the size of the tree). The height of the trees is obtained from a height - girth relationship. Tree mortality is not yet taken into account (but in our case, tree mortality is limited owing to the short rotation length) ; • the tree properties module : a generic stem taper equation was constructed. It explicitly takes into account the global taper of the bole (when the trunk is assumed to be a cone), the but swell and the decrease in diameter within the crown. The equation allows accurate estimations of diameters and volumes along the bole. The wood density model was based upon a non-parametrical modelling approach but is not yet introduced within the chain of models. Allometric relationships were also fitted for evaluating the biomass of roots, branches, stem, bark and leaves throughout the whole rotation ; • the biogeochemical module : a model was built to assess the distribution of nutrient concentrations (N, P, K) in individual rings within the bole and their changes with the ring ageing. Furthermore, different allometric relationships estimated the nutrient contents within the branches, roots, leaves and bark. Input-output budgets of nutrients at the plot level were simulated during the whole planted crop rotation from (i) measurements of nutrient fluxes in a clonal stand between ages 6 and 9 years, (ii) a chronosequence approach to quantify the main fluxes of the biological cycle throughout stand rotation, and (iii) hypotheses concerning nutrient losses by deep drainage during the juvenile stand growth. Work is in progress. All these models are embedded within Capsis (for the growth modelling part) and will be introduced within Win-Epifn (for the wood properties part and logging). Key points concerning the building of the models are presented. Input-output budgets of nutrients during the planted crop rotation are simulated to assess the influence of various silvicultural practices (spacing, rotation length, harvesting options) on the amount of available nutrients in the soil. Economical aspects are not evaluated.