Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Titre : Integrating Ground Survey, Remote Sensing and Ecological Niche Modelling for Monitoring Desert Locust Schistocerca gregaria (Forskål) in Sudan

Auteur : Mahgoub Mousa Mohamed Boshara

Université de soutenance : University of Khartoum

Grade : Doctor of Philosophy in Agriculture (Entomology) 2020

Résumé partiel
This study aims to improve the desert locust monitoring and surveillance methods in Sudan using ground survey and innovative geospatial techniques. Typical breeding sites in winter (Tokar Delta and Gowb) and summer (Umm Harot and Wadi Habob) seasons were selected for field experiments during recession and outbreak periods. A ground survey approach was conducted to assess the efficiency of the traditional current (surveying some of the historically known sites), systematic (division of the survey route into fixed intervals) and a newly proposed sectoral diagonal sampling methods for determining desert locust occurrence and density. The study further compared the performance of four foot transect patterns in terms of track length, path of walking and number of surveyors for estimating adult and hopper desert locust density in different habitats. Moreover, the study assessed the possibility of predicting current and future (i.e. 2050) desert locust distribution in Sudan using remotely sensed bioclimatic variables (n = 19) and a maximum entropy (MaxEnt) ecological niche model. To conduct the MaxEnt experiment, the study collated long-term (i.e. 15 years) historical data on desert locust occurrence between 2003 and 2017 from the Reconnaissance and Monitoring System of the Environment of Schistocerca (RAMSES) database, and remotely sensed bioclimatic variables from AfriClim. In addition, the generalized linear models (GLMs) were used to estimate the desert locust density at recession and outbreak periods. Specifically, the study tested whether desert locust density can be predicted using remotely sensed derived vegetation, temperature, and rainfall variables and other field-based vegetation, habitat and soil moisture observations. On the other hand, the study investigated the influence of a multi-date remotely sensed enhanced vegetation index (EVI) as a proxy for vegetation greenness (i.e. density or productivity) on predicting the current (2017–2019) desert locust distribution (habitat suitability) using the MaxEnt modelling approach. A two-way analysis of variance (ANOVA) was conducted to compare the efficiency of the various ground survey methods. Furthermore, a root mean square error (RMSE) was calculated to evaluate the performance of the different GLMs. The results showed that the sectoral diagonal method outperformed the current and systematic methods for estimating desert locust density during the recession (41% versus 42% for the current and systematic methods, respectively) and outbreak (22% versus 80% for the current and systematic methods, respectively) seasons

Présentation

Version intégrale (4,8 Mb)