Browsing University of Alaska Fairbanks by Subject "machine learning"
Now showing items 1-1 of 1
Rapid multi-nation distribution assessment of a charismatic species of conservation concern using ensemble model predictions: The Red Panda in the Hindu-Kush Himalaya regionBackground The red panda (Ailurus fulgens) is a globally threatened species living in the multi-nation Hindu-Kush Himalaya (HKH) region. It has a declining population trend due to anthropogenic pressures. Additionally, human-driven climate change is expected to have substantial impacts on the fragmented populations and the fragile habitats throughout its range. However, quantitative and transparent information on the ecological niche (potential as well as realized) of this species and the distribution across the vast and complex eight nations of the HKH region is still lacking. Such baseline information is not only crucial for identifying new populations but also for restoring locally extinct populations and for understanding its bio-geographical evolution, as well as for prioritizing regions and efficient management actions. Our study presents the first quantitative large-scale prediction of the potential ecological niche of red panda for the entire HKH. Methodology/Principal Findings We compiled, and made publicly available the best known ‘presence only’ red panda dataset with ISO compliant metadata. This was done through the International Centre for Integrated Mountain Development (ICIMOD.org) data-platform to the Global Biodiversity Information Facility (GBIF.org). We used data mining and machine learning algorithms such as high-performance commercial Classification and Regression Trees (CART), Random Forest, TreeNet, and Multivariate Adaptive Regression Splines (MARS) implementations (Salford Systems Ltd). We averaged all these models for the first produced Ensemble Model for HKH as well as for this species. Conclusions/Significance Our predictive model allows finding major drivers of the red panda ecological distribution niche, as well as to assess and fine-tune earlier habitat area estimates for management. Our models can be used by the Red Panda Recovery Team, Red Panda Action Plan etc. because they are robust, transparent, publicly available, fit for use, and have a good accuracy, as judged by several metrics e.g. Receiver Operating Characteristics (ROC-AUC) curves, expert opinion and assessed by known absence locations