Detection of an invasive plant species using Planet Scope multispectral satellite imagery

Abstract

Invasive species are one of the main threats to biodiversity worldwide. In insular settings, the proliferation of invasive species is adversely impacting unique endemic species and disrupting delicate ecosystems. This study presents a spatio-temporal analysis designed to help us understand the spread dynamics of invasive plant species in insular ecosystems.

To conduct this analysis, we performed a time-series analysis on the distribution of the invasive blackberry plant (Rubus niveus), which is displacing endemic species in the humid highlands of an uninhabited island in the Galápagos Archipelago. We successfully identified this species by applying a Random Forest Machine Learning (ML) algorithm on high-resolution Planet Scope multispectral satellite imagery. We improved the precision in the classification of our target species by introducing separate classes for other species with similar spectral characteristics into our ML model. Our results successfully outline the distribution of our target species in our study site in 2018, 2020, and 2023, offering us an opportunity to better understand the environmental factors that may drive its spread. This complements previous studies that employed various satellite imagery and classification methods, by adding a more comprehensive spatio-temporal analysis for blackberry spread in Santiago Island. The study contributes to a deeper understanding of blackberry invasion in the Galápagos, emphasizing the importance of high-resolution satellite imagery for mapping invasive plant species and informing conservation efforts. This work will be relevant to spatial scientists, conservation scholars, and protected area managers.