Employing Machine Learning to Monitor Endangered Species and Assess the Impact of Habitat Fragmentation

Abstract

A major problem of the biodiversity is habitat fragmentation, which causes the extinction of
endangered species due to a disturbed ecosystem and decreased interconnectivity. Conventional
monitoring techniques are often limited on a scalability and precision especially in remote or the large
regions. This paper will attempt to compare the application of machine learning (ML) algorithms to
track endangered species and determine how habitat fragmentation affects their populations. To
analyze satellite images and field data on the distribution of the species and the fragmentation of their
habitats, the study used supervised learning models such as Convolutional Neural Networks (CNN)
and Random Forests (RF). The ML models have been trained on the data which includes wildlife
sightings, environmental factors, and the fragmentation indices. The predominant results have shown
that the ML-based models were more effective than the conventional approaches to detecting species
in fragmented habitats, and CNN has shown the highest detection rate (85) in categorizing habitat
patches and detecting the endangered species. Also, determined that habitat fragmentation is
negatively correlated with species diversity and increasing habitat fragmentation (p = 0.01), which
indicates a strong necessity in conservation of fragmented ecosystems. This research reveals the
possible benefits of machine learning to the improvement of biodiversity monitoring, as well as
provides useful information concerning the conservation management, especially in areas with habitat
fragmentation. It has been found that the combination of sophisticated ML tools with wildlife
surveillance could help to create highly efficient data-driven decisions in saving endangered species.