A geospatial approach to assess habitat diversity and inform conservation efforts using a case study from North Africa

• Published in: African journal of ecology 2023-06, Vol.61 (2), p.422-450
• Main Authors: Elbahi, Abderrafea, Dugon, Michel, Oubrou, Widade, El Bekkay, Mohammed, Hermas, Jamila, Lawton, Colin
• Format: Article
• Language: English
• Subjects: Accuracy; Arid environments; Aridity; Biodiversity; Cartography; Classification; Coastal ecology; Coastal zone; Confusion; Conservation; Data acquisition; Developing countries; Ecological studies; Geographic information systems; habitat types; Habitats; LDCs; Matrix methods; National parks; natural diversity; protected areas; Remote sensing
• ISSN: 0141-6707; 1365-2028
• DOI: 10.1111/aje.13130

Geographic Information System (GIS) combined with remote sensing and field data was used to identify and map Souss‐Massa National Park (SMNP) habitats, an exceptionally biodiverse coastal area located in Morocco. Habitats in fenced and unfenced areas were compared to assess the fencing effectiveness for conserving semi‐arid/arid environments. Sentinel‐2A level‐2A data acquired on March 25, 2021, were analysed using the Semi‐Automatic Classification Plugin in QGIS software. The supervised classification technique and the maximum likelihood classifier were used. Ground truthing consisted of training data used for the supervised classification and reference data (850 validation points) used for accuracy assessment based on the confusion matrix method. SMNP composed of 11 natural habitat classes (70.91%) and 6 disturbed habitat classes (29.09%). The overall accuracy (>96%) and kappa coefficient (0.95) showed a high classification accuracy. Fenced areas were dominated by natural habitats (>88%), whereas unfenced areas contained more disturbed habitats (>51%), indicating that fencing was an effective conservation tool. Natural habitats in unfenced areas are largely being modified and degraded, and therefore, monitoring and conservation actions should be further strengthened. The presented methodology shows a promising application potential in future and related ecological studies conducted in other regions, notably developing countries. Résumé Le Système d'Information Géographique (SIG) combiné à la télédétection et aux données de terrain a été utilisé pour identifier et cartographier les habitats du Parc National de Souss‐Massa (PNSM), une zone côtière à la biodiversité exceptionnelle située au Maroc. Les habitats dans des zones clôturées et non clôturées ont été comparés pour évaluer l'efficacité des clôtures pour la conservation des environnements semi‐arides/arides. Les données de Sentinel‐2A de niveau 2A acquises le 25 mars 2021 ont été analysées à l'aide du Plugin de Classification Semi‐Automatique du logiciel QGIS. La technique de classification supervisée et le classificateur de vraisemblance maximale ont été utilisés. La vérification au sol consistait en des données de formation utilisées pour la classification supervisée et des données de référence (850 points de validation) utilisées pour l'évaluation de la précision selon la méthode de la matrice de confusion. Le PNSM est composé de 11 classes d'habitats naturels (70,91 %) et de 6 classes d'habitats