Computational models for improving surveillance for the early detection of direct introduction of cassava brown streak disease in Nigeria

Abstract

Cassava is a key source of calories for smallholder farmers in sub-Saharan Africa but its role as a food security crop is threatened by the cross-continental spread of cassava brown streak disease (CBSD) that causes high yield losses. In order to mitigate the impact of CBSD, it is important to minimise the delay in first detection of CBSD after introduction to a new country or state so that interventions can be deployed more effectively. Using a computational model that combines simulations of CBSD spread at both the landscape and field scales, we model the effectiveness of different country level survey strategies in Nigeria when CBSD is directly introduced. We find that the main limitation to the rapid CBSD detection in Nigeria, using the current survey strategy, is that an insufficient number of fields are surveyed in newly infected Nigerian states, not the total number of fields surveyed across the country, nor the limitation of only surveying fields near a road. We explored different strategies for geographically selecting fields to survey and found that early and consistent CBSD detection will involve confining candidate survey fields to states where CBSD has not yet been detected and where survey locations are allocated in proportion to the density of cassava crops, detects CBSD sooner, more consistently, and when the epidemic is smaller compared with distributing surveys uniformly across Nigeria.