Open AccessClimate based model in determining the distribution pattern of Cecropia peltata L across global landscape
Author(s) -
Angga Yudaputra,
Prima Wahyu Kusuma Hutabarat
Publication year - 2021
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/743/1/012018
Subject(s) - species distribution , environmental niche modelling , climate change , biodiversity , representative concentration pathways , geography , support vector machine , distribution (mathematics) , predictive modelling , ecological niche , regression , range (aeronautics) , ecology , environmental science , climate model , machine learning , computer science , biology , habitat , mathematics , statistics , mathematical analysis , materials science , composite material
Climate change becomes a major threat to the global biodiversity. It alters the ecological niche of species, even small change in temperature could have a significant impact to the distribution pattern of biodiversity. Cecropia peltata is an invasive species with wide range geographic distribution. The aim of this study is to understand the impact of climate change to the current and future distribution of invasive plant C. peltata . The Support Vector Machine (SVM) and Boosted Regression Tress (BRT) algorithm of machine learning were used to predict the current and future distribution. The occurrence records of C. peltata was obtained from Global Biodiversity Information Facility (GBIF). There were 2691 occurrence records in GBIF database. The global climatic variables with the resolution 2.5 km were used as predictors of model. VIF was used to select the multicollinearity among those variables using threshold of 0.7. The CIMP5 of Global Circulation Model (GCM) was used to understand the impact of climate change to the distribution of the plant. The future projection on year 2070 with the worst climate scenario RCP 8.5 was used on these predictive models. The SVM and BRT models were actually relevant to be used as predictive models with AUC >0.90 and categorised as excellent predictive models. The future distribution pattern was likely to be shifted compared to the current distribution prediction. The output of this study as predictive current and future distribution maps would be useful to provide an information about the potential area where the species might be invading based on the training data (observation data). Furthermore, the prediction of future distribution would be necessary to understand how the climate change literally affects the range of distribution of the invasive plant species.