Sabin Guendehou (2013). Methods and tools for estimating carbon dynamics in tropical forest ecosystems in Benin (West Africa).
Supervisors: Professors Mansourou Moudachirou, Brice Sinsin, Raisa Makipää and Pasi Puttonen.
Abstract: The growing interest in tropical Africa to estimate carbon dynamics in forest ecosystems motivates the development of methods and tools and the generation of data required to estimate carbon stocks and its changes in different pools: above and below ground biomass, dead wood, litter and soils. Volume and biomass models were developed for five dominant tree species, including Afzelia africana Sm. (Leguminosae-Caesalpinioideae), Anogeissus leiocarpa (DC.) Guill. & Perr. (Combretaceae), Ceiba pentandra (L.) Gaertn. (Bombacaceae), Dialium guineense Willd. (Leguminosae-Caesalpinioideae) and Diospyros mespiliformis Hochst. ex A. DC. (Ebenaceae), in the natural forest “Lama” and for Tectona grandis L. f. (Verbenaceae) in plantations. The modelling used ground-truth observations collected on diameters, heights, and basic wood densities of trees. The best predictive model was logarithmic form using diameter at breast height (Dbh) and stem height as independent predictors; the model using only Dbh as predictor also performed well. The carbon, nitrogen, organic matter and ash contents were also determined for these tree species. With regard to soils, a litterbag experiment was conducted in the Lama forest to study the microbial decomposition of leaf litter from A. africana, A. leiocarpa, C. pentandra, D. guineense, and D. mespiliformis. The litter mass loss and the chemical composition of the decomposed litter were determined every four weeks, over a six-month period. The differences in initial litter quality across species explained the variation in decay rate and the key chemical controls of leaf decomposition were the initial concentrations of acid-hydrolysable compounds, lignin and nitrogen. Initial chemical composition was the factor controlling litter decomposition process at a local scale. These observations together with climatic data (temperature, precipitation) were used to test the validity of the dynamic soil carbon model Yasso07. As the predictive ability of Yasso07 was not good, the model was recalibrated and the resulting version was found suitable to estimate the soil carbon stocks, its changes and the CO2 emissions from heterotrophic respiration.
On remote sensing, Landsat images corresponding to path 192, row 55 were acquired for the period 1986–2011. The images were processed using available algorithms and analysed for reflectance changes. This unsupervised classification method detected changes in biomass stock associated with the reflectance changes. The method has the ability for detecting deforestation, afforestation and natural regeneration when complemented with ground-truth observations. The combination of biomass models, soil carbon model and remote sensing-based analysis would be the most feasible approach for estimating forest carbon dynamics on large scale. This is particularly relevant for the global carbon cycle analysis and the reporting under mechanisms such as the Clean Development Mechanism of the Kyoto Protocol, the Reducing Emissions from Deforestation and Forest Degradation in Developing Countries of the Climate Change Convention or under other voluntary carbon sequestration initiatives.