Mapping Avian Habitat Suitability Using Linear and Non-Linear Techniques in the Case of Wetland Landscapes

Abstract

Habitat quality is crucial for wildlife management that impacts the conservation of sensitive landscapes such as wetlands. With advancements in GIS, habitat modelling now effectively predicts species occurrences and habitat suitability. This study aims to model and map habitat suitability for case bird species of Kentish plover in Tuzla Lagoon using multiple techniques. Kentish plover nesting data were collected from 293 nests, and reproductive success measures such as lay date, egg volume, and nest fate were analysed. Spatial habitat modelling techniques, including regression, co-kriging, artificial neural networks, and decision trees, were used with IKONOS imagery and ground data. The overall prediction accuracies were poor for lay date across all techniques, with the decision tree being the most accurate, while egg volume was best predicted by co-kriging, egg success by linear regression, and nest fate by both binomial logistic regression and ANN with 75% accuracy.