Modeling the global distribution of Culicoides imicola: an Ensemble approach

Samson Leta,Kebede Amenu,Haileleul Negussie,Megarsa Bedasa Jaleta,Tesfaye Mulatu,Crawford W Revie,Eyerusalem Fetene,Tariku Jibat Beyene

doi:10.1038/s41598-019-50765-1

Samson Leta, Kebede Amenu + Show 6 more

Open Access

https://doi.org/10.1038/s41598-019-50765-1

Copy DOI

Abstract

Culicoides imicola is a midge species serving as vector for a number of viral diseases of livestock, including Bluetongue, and African Horse Sickness. C. imicola is also known to transmit Schmallenberg virus experimentally. Environmental and demographic factors may impose rapid changes on the global distribution of C. imicola and aid introduction into new areas. The aim of this study is to predict the global distribution of C. imicola using an ensemble modeling approach by combining climatic, livestock distribution and land cover covariates, together with a comprehensive global dataset of geo-positioned occurrence points for C. imicola. Thirty individual models were generated by ‘biomod2’, with 21 models scoring a true skill statistic (TSS) >0.8. These 21 models incorporated weighted runs from eight of ten algorithms and were used to create a final ensemble model. The ensemble model performed very well (TSS = 0.898 and ROC = 0.991) and indicated high environmental suitability for C. imicola in the tropics and subtropics. The habitat suitability for C. imicola spans from South Africa to southern Europe and from southern USA to southern China. The distribution of C. imicola is mainly constrained by climatic factors. In the ensemble model, mean annual minimum temperature had the highest overall contribution (42.9%), followed by mean annual maximum temperature (21.1%), solar radiation (13.6%), annual precipitation (11%), livestock distribution (6.2%), vapor pressure (3.4%), wind speed (0.8%), and land cover (0.1%). The present study provides the most up-to-date predictive maps of the potential distributions of C. imicola and should be of great value for decision making at global and regional scales.

Highlights

The incidence of vectors and vector-borne diseases of livestock and humans is increasing at an alarming rate associated with changes related to factors such as climate, environment, high human mobility, unplanned urbanization and agricultural intensification[1,2]
Bluetongue is a devastating viral disease of ruminants found to be historically enzootic in tropical regions of the world; since 1998 the virus has spread across southern European countries
Among the 30 models, 24 had receiver operating characteristics (ROC) > 0.90 (ROCaverage = 0.95), considered as good accuracy based on the classification of Swets[43]

Summary

Introduction

The incidence of vectors and vector-borne diseases of livestock and humans is increasing at an alarming rate associated with changes related to factors such as climate, environment, high human mobility, unplanned urbanization and agricultural intensification[1,2]. C. imicola transmits the agents of a number of viral diseases of veterinary importance, including Bluetongue[5,6], and African Horse Sickness (AHS)[7,8]. The recent emergence of two Culicoides-borne diseases (Bluetongue and Schmallenberg) in Europe has raised concerns around the potential introduction and further spread of AHS virus into temperate parts of the world[13]. Knowledge of the potential geographical distribution of this species is important to guide surveillance of C. imicola and the diseases it transmits. The present study was initiated based on the proposition that an ensemble modeling technique could be used to predict the global distribution of C. imicola, and provide better scientific evidence with regard to the potential global distribution of the vector

Objectives

Methods

Results

Discussion

Conclusion