Dataset on Woody Aboveground Biomass, Disturbance Losses, and Wood Density from an African Savanna Ecosystem

Published: 27-02-2021| Version 1 | DOI: 10.17632/3cs85wd3gb.1
Magnus Dobler,
Daniela Niedeggen,
Ezequiel Chimbioputo Fabiano,


This dataset comprises raw and processed data from two tree inventories in savanna and savanna woodland vegetation along elephant disturbance gradients in Zambezi Region, Namibia. The recorded individuals represent all age classes and damage levels (including gullivers) and were classified accordingly into nine growth classes. The raw data contains allometric/dendrometric proxies (height, canopy diameters, and stem circumferences for stems > 15cm at the base) which were recorded for n = 6,179 woody individuals in 60 plots (each 0.1 ha). Wood samples were taken to measure specific wood density (SWD) for each recorded species. SWD measurements and raw data have been used to estimate aboveground biomass (AGB) per individual via allometric models. Existing methodologies and workflows had to be improved to account for damages and diverse vegetation structure as shaped by disturbance impacts. In addition to standing biomass, our proposed method also delivered data on biomass losses to respective disturbance agents (elephants, fire, and others) for each individual. For a subset of tree individuals (n = 288), which fulfilled the minimum requirements to be assessed with conventional methods, the aboveground biomass was re-estimated with a standard procedure to allow for a method comparison. The processed data comprises living AGB on individual and unit per area basis as determined with two competing methods (conventional vs our proposed). The proposed method further delivered data on pre-disturbance AGB and AGB losses to major disturbance agents. For a description of biomass partitions also see detailed legend within the Data file. Our research hypothesis was that increasing elephant densities decrease woody aboveground biomass (AGB) and increase elephant-mediated AGB losses along the disturbance gradient. Findings from the data presented here support this hypothesis. Furthermore, increasing elephant densities seem to decrease fire disturbance impacts along the same gradient. These trends can be demonstrated for two different vegetation types (savanna & savanna woodland). A comparison between the conventional standard method and our improved proposed methodology highlights the importance of suitable sampling strategies and protocols for determining biomass and carbon storage in highly disturbed dryland ecosystems. The conventional method over-estimated biomass in large but disturbed trees, while simultaneously under-estimating total biomass on a unit per area basis through omission of all smaller trees, shrubs, and highly damaged woody individuals (gullivers).