Modelling spatial distribution of tsetse (Diptera: Glossinidae) in Masoka area, an unexplored part of Mbire District, Zimbabwe

Abstract

A study was conducted from March 2015 to December 2015 in order to model the distribution of two savannah species of tsetse (Glossina sensu stricto), Glossina morsitans morsitans and G. pallidipes in the Masoka area of the Mid-Zambezi valley in Zimbabwe. Two approaches were used. The first approach sought to model the probability of presence of both species in areas which were sampled but recorded zero tsetse catches using trap efficiency, sampling effort and suitable habitat cover. A probability threshold of 0.05 was used to distinguish areas which could be potentially infested from those that had low chances of tsetse occurrence. The resultant probability model pointed to an area of 104 km2 in size where G. m. morsitans could possibly be present (P > 0.05) whilst all areas which did not record G. pallidipes had a low probability of presence for the species (P < 0.05). This study showed that there was a high probability of tsetse presence in areas where the habitat was less degraded and low probability in settled areas where suitable tsetse habitat has been disturbed due to agricultural activities. The probability model therefore has the potential to optimize vector control strategies by streamlining areas of intervention. The second model was a predictive one built using tsetse presence-only data and climatic and environmental covariates. The model had an Area Under the Curve (AUC) of 0.80 for G. m. morsitans and 0.94 for G. pallidipes, indicating that the ability of the model to predict suitable tsetse habitat in the Masoka area was better than random (AUC = 0.5). Glossina morsitans morsitans occurrence was positively correlated to Normalised Difference Vegetation Index (NDVI), riverine forest and mopane woodlands whilst crop lands and temperature indices exhibited a strong negative correlation with its occurrence. Glossina pallidipes, on the other hand, had extremely specialised habitat requirements and was positively correlated to riverine forest. The species also had a positive correlation with NDVI but a negative correlation with mopane woodland.