Geospatial detection of Ommatissus lybicus de Bergevin using spatial and machine learning techniques

Abstract

Dubas bug (Ommatissus lybicus de Bergevin) infestations are frequent in several Middle Eastern and North African countries, causing havoc to date palms. The key objectives of this research were (1) to examine the prevalence of the Dubas bug geographically and (2) to identify hotspots and clusters at the local levels. Normalised difference vegetation indices, kernel density estimations, spatial autocorrelations, Getis-Ord G*, Anselin spatial indicators, ordinary regression and forest-based classification and regression algorithms were used to detect Dubas bug infestations in the Al Batinah North and Al Batinah South governorates in Oman from 2017 to 2021. Images of the region, including several sub-locations with diverse levels of infestation, were collected throughout the Dubas bug spring and autumn generations during the study period. The spring generation's kappa coefficient for identifying infestation levels was 0.97, whereas the autumn generation's kappa coefficient was 0.98. Varying infestation levels were observed over the study period, with some areas experiencing fewer or more cold spots or hotspots than others. The highest Moran's I value of 0.48, along with the highest z-score of 12.9, characterised the highly clustered pattern of infestation during the 2017 spring generation, while the lowest infestation clustering occurred during the 2019 spring generation, with the lowest Moran's I of 0.23 and the highest z-score of 7.2. Despite seasonal fluctuations, most Dubas bug hotspots were located in the southern parts of the study area. Thus, the infestations caused the greatest damage to palm tree farms in interior areas between mountains, away from the coastal plains. Dubas bugs were found in high-density clusters in the south of Sohar and Al Khaburah and in the southwest of Al Rustaq during the 2017 and 2018 spring generations. By observing statistically significant changes in infection clustering patterns, entomologists can assess the prevalence of infestation using spatial information techniques.