A centered kernel alignment-based strategy for pest evolution tracing: Myiopsitta monachus case

This paper presents a novel strategy for tracing the evolution of Myiopsitta monachus populations using a centered kernel alignment (CKA)-based approach. This species is of particular interest due to having been declared a pest. The proposed method utilizes a vector representation of bird sightings in Uruguayan territory, divided into 492 cells of 24 kilometers
$\times$
24 kilometers resolution. Kernel matrices are computed using a linear kernel function, which maps the input data to a higher-dimensional feature space, producing an enhanced data representation. The CKA-based approach measures the similarity between pest distributions for different periods (non-breeding seasons), allowing for the tracking of pest mobility and evolution over time. The approach has been compared against other well-known metrics to evaluate its capacity to decode spatial-temporal patterns. The proposed strategy has the potential to be applied to other populations and can aid in the development of effective pest management techniques.