Remote sensing for monitoring and assessment of invasive herbaceous plants: the case of Oenothera drummondii in coastal ecosystems

By

Hidalgo, Pablo J.
Montiel, Nuria
Ortega-Moreno, Mónica
Fernández De Villarán, Rubén
Pérez-Carral, Cristina

Type

Article

Date of Publication

2025-7-11

Original Publication

NeoBiota : advancing research on alien species and biological invasions

Volume

99

Pages

323--340

Subjects

Alien species , C5.0 classification model , dune , Management , multispectral , unmanned aerial vehicles , vegetation indices

Contributed by

Pensoft Publishers

Language

English

Published by

Sophia, Bulgaria Pensoft Publishers

Identifiers

DOI: https://doi.org/10.3897/neobiota.99.151472

 

Abstract

Invasive exotic species pose a serious threat to biodiversity, particularly in fragile and degraded habitats. This is the case with Oenothera drummondii, which significantly affects certain coastal areas of the Iberian Peninsula. This study area focuses on the Odiel Marshes Natural Area, where populations of this herbaceous plant are drastically affecting native vegetation. The objective is to develop a methodology for the automatic detection of individuals of this species with Unmanned Aerial Vehicles (UAV) equipped with multispectral cameras. The final goal is to optimize monitoring, control, and potential eradication activities.<br> A photogrammetric flight was carried out over 26.26 ha using a DJI Mavic 3Multiespectral. The model used to distinguish the invasive species from the native vegetation in the area was the C5.0 classification model. A total of 800 individuals were analysed using this model, including samples of the invasive species and three of the most abundant native plant species with a similar appearance. The classification tree results were extrapolated to the entire study area. The results indicate that the use of multispectral bands and vegetation indices allows the C5.0 model to classify the studied species with an error rate of 15.4%. The subsequent application of the classification tree obtained across the entire study area resulted in the detection of O. drummondii with an accuracy rate of 83%, demonstrating that this UAV-based technique enables the identification of exotic invasive herbaceous species. This methodology could contribute to management of the species and could be easily applied to other affected areas and species worldwide.