Retrospective versus prospective designs for studies of cropraiding by elephants in Kakum, Ghana

Abstract

Crop-raiding by elephants continues to increase as human populations in elephant ranges expand. The risk of crop losses can be reduced by identifying the most important farmland features that attract elephants. Risk factors vary from place to place and must be identified by site-specific studies. The most important risk factors include distance of farm to reserve boundary line, area under cultivation, number of crop types on farm and degree of each farm’s isolation. Here we take the data from an earlier prospective study of crop-raiding around the Kakum Conservation Area in southern Ghana to illustrate a better method of analysis using a zero-inflated Poisson model. We then use the same data set to illustrate the advantages and drawbacks of a retrospective design. With a retrospective design a raided farm is matched with one or more intact farms at the end of the growing season. This method is cost-effective for field workers whose resources are limited because it does not require repeated visits to farms to monitor raids. The optimum sample size is about 30 raided farms that are each matched with at least two intact farms.