Machine Learning Evaluation of Community Conservation Effects on Native Birds

Abstract

Introduction: The native birds perform an important function in ensuring that there is stability in the ecosystem due to pollination, seed dispersal, pest control, and biodiversity conservation. The fragmentation of the habitat, changes in land use, and human disturbances have impacted bird species and ecological balance. Traditional biodiversity evaluation techniques rely on fieldwork and expert observation, which limits the ability to undertake large-scale analysis. Thus, machine learning techniques can help in understanding the complexities of ecological relationships and assessing conservation efforts in the communities.

Materials & Methods: This study designed a machine learning algorithm to investigate the effects of community-based conservation efforts on native birds' populations. Biodiversity data that comprised 2,500 bird samples from 45 different native bird species at 12 conservation sites were compiled based on bird presence, habitat, environment, and conservation variables. The biodiversity data was then filtered and normalized before being split into three categories; 70%, 15%, and 15% training, validation, and testing sets respectively. Performance of the Random Forest, Support Vector Machine, Gradient Boosting, and Artificial Neural Networks models were compared using accuracy, precision, recall, F1, and MAE scores.

Results and Discussion: The findings of the evaluation revealed that Random Forest model outperformed others with 92.6% accuracy, 92.1% precision, 91.6% recall, 91.8% F1-score, and 0.084 MAE. It was found out that habitat restoration, reduction of disturbances from humans and community involvement played an important role in increasing bird diversity and stability. It was found that conservation areas having high involvement of the community have 18.5% more birds and 22.3% better habitat indicators than others.

Conclusion: This machine learning evaluation framework is able to determine relationships between the activities for conserving and the reactions of the native birds. This framework allows for evidence-based management of biodiversity through proper evaluation of ecological impacts.