Determinants of Aboveground Biomass across an Afromontane Landscape Mosaic in Kenya

Abstract

Afromontane tropical forests maintain high biodiversity and provide valuable ecosystem services, such as carbon sequestration. The spatial distribution of aboveground biomass (AGB) in forest-agriculture landscape mosaics is highly variable and controlled both by physical and human factors. In this study, the objectives were (1) to generate a map of AGB for the Taita Hills, in Kenya, based on field measurements and airborne laser scanning (ALS), and (2) to examine determinants of AGB using geospatial data and statistical modelling. The study area is located in the northernmost part of the Eastern Arc Mountains, with an elevation range of approximately 600–2200 m. The field measurements were carried out in 215 plots in 2013–2015 and ALS flights conducted in 2014–2015. Multiple linear regression was used for predicting AGB at a 30 m × 30 m resolution based on canopy cover and the 25th percentile height derived from ALS returns (R2 = 0.88, RMSE = 52.9 Mg ha−1). Boosted regression trees (BRT) were used for examining the relationship between AGB and explanatory variables at a 250 m × 250 m resolution. According to the results, AGB patterns were controlled mainly by mean annual precipitation (MAP), the distribution of croplands and slope, which explained together 69.8% of the AGB variation. The highest AGB densities have been retained in the semi-natural vegetation in the higher elevations receiving more rainfall and in the steep slope, which is less suitable for agriculture. AGB was also relatively high in the eastern slopes as indicated by the strong interaction between slope and aspect. Furthermore, plantation forests, topographic position and the density of buildings had a minor influence on AGB. The findings demonstrate the utility of ALS-based AGB maps and BRT for describing AGB distributions across Afromontane landscapes, which is important for making sustainable land management decisions in the region.

Keywords

• airborne laser scanning
• boosted regression trees
• carbon
• LiDAR
• REDD+