Habitat Prediction and Knowledge Extraction from Musa gracilis Holttum with Limited Data
Issued Date
2022-07-01
Resource Type
ISSN
01252526
Scopus ID
2-s2.0-85134399493
Journal Title
Chiang Mai Journal of Science
Volume
49
Issue
4
Start Page
1050
End Page
1062
Rights Holder(s)
SCOPUS
Bibliographic Citation
Chiang Mai Journal of Science Vol.49 No.4 (2022) , 1050-1062
Suggested Citation
Changruenngam T., Swangpol S.C., Tovaranonte J. Habitat Prediction and Knowledge Extraction from Musa gracilis Holttum with Limited Data. Chiang Mai Journal of Science Vol.49 No.4 (2022) , 1050-1062. 1062. doi:10.12982/CMJS.2022.076 Retrieved from: https://repository.li.mahidol.ac.th/handle/20.500.14594/83682
Title
Habitat Prediction and Knowledge Extraction from Musa gracilis Holttum with Limited Data
Author(s)
Author's Affiliation
Other Contributor(s)
Abstract
Species distribution models are a powerful tool to predict suitability map addressing ecology and conservation, especially of rare species. However, the limited occurrence data often decrease the performances of the prediction models. In this research, the Random Forest with Fuzzy selection of pseudo absence point (RFFA) method was created for habitat prediction of species with limited distribution data. In our study, Musa gracilis Holttum is naturally found only in Narathiwat, one of the southernmost provinces in Thailand. With only three collected localities, the species was used as a sample to test efficacy of the RFFA method. The comparing the model results with real data, the statistical relationship, and the feasibility assessment of the two species distribution models. MaxEnt and RFFA methods showed that the performance of the RFFA model did not differ significantly from that of MaxEnt in terms of efficiency. It can be concluded from the model using the three-occurrence data that M. gracilis distributes in approximately 7,000 square kilometers, with limited boundary in Thailand peninsular and is facing a risk of extinction in the wild.