Counting elephants in Montane forests: some sources of error

The dung count method is widely used to estimate elephant numbers in forests. It was developed in the lowland forests of Central Africa but it is also used in Montane forests in eastern Africa. Using data collected on Mount Kenya and computer simulations, this paper explores the following issues associated with dung surveys in Montane forests:

• • 

  High rainfall at 3000 m altitude on Mount Kenya was expected to accelerate dung pile decay but no significant difference was found between 3000 and 2500 m where less rain falls, possibly because high rainfall at 3000 m is counteracted by lower temperatures;
• • 

  Physical obstacles make it difficult to walk long, straight transects in Montane forests. Deviating from a straight line pushes the distribution of distance measurements from dung piles to the transect centre line (pdist) towards a negative exponential (NE), which complicates data analysis and may give inaccurate estimates. Using short transects largely alleviate this problem;
• • 

  Analysis of dung count simulations shows that the expected sightability curve of pdist pushes towards a NE with increasing numbers of obstacles blocking the view, even along perfectly straight transects;
• • 

  Extrapolating measured dung density to map area on Mount Kenya resulted in an underestimate of c. 13%. An unstratified correction of map area to ground area for Montane areas would be biased because of the strong tendency for elephants to avoid steeply sloping areas.