GIS-based habitat model to predict potential areas for the upcoming occurrences of an alien invasive plant, Mimosa pigra L

• Published in: Metsanduslikud uurimused Vol. 70; no. 1; pp. 31 - 43
• Main Authors: Le, Thai Son, Thoa, Pham Thi Kim, Tuan, Nguyen Van
• Format: Journal Article
• Language: English
• Published: Tartu Sciendo 01.06.2019; De Gruyter Poland
• Subjects: Agricultural management; GIS; habitat modelling; Habitats; Harvesting; Introduced species; Invasive plants; Invasive species; Mahalanobis distance; Mimosa pigra; Model testing; Modelling; Nature reserves; Nonnative species; predictive model; Spatial resolution; Statistical analysis
• ISSN: 1736-8723; 1406-9954; 1736-8723
• DOI: 10.2478/fsmu-2019-0003

Incursions of L., a super-invasive plant, were detected in Hoa Vang district, Da Nang city, Vietnam. This invasive species posed threats to the local agricultural and natural areas, especially to Ba Na - Nui Chua Nature Reserve located in the district. In this study, a habitat model was developed to predict potential areas for the upcoming occurrences of the plant. Detected locations of the species were analyzed in association with seven environmental layers (15 m spatial resolution), which characterized the habitat conditions facilitating the plant incursion, to calculate a multivariate statistic, Mahalanobis distance ( ). Mimosa occurrences were divided into subsets of modelling (for model construction) and validating data (for selecting the best model from replicate runs). The model performance was tested using a null model of 1,000 random points and indicated a significant relationship between values and mimosa occurrence. The model performed markedly better than the random model. The null model in combination with the entire dataset of mimosa locations was also used to identify the threshold value. Using that threshold value, 99.5% of existing mimosa locations were detected and 20.3% of the study area was determined as high-risk areas for mimosa occurrence. These identified high risk areas would make an important contribution to the local alien invasive species management. Given the potential threats to these species from illegal harvesting, that information may serve as an important benchmark for future habitat and population assessments. The spatial modelling techniques in this study can easily be applied to other species and areas.