Eelgrass (Zostera marina) and benthic habitat mapping in Atlantic Canada using high-resolution SPOT 6/7 satellite imagery

摘要

Eelgrass (Zostera marina) is the dominant perennial canopy-forming vegetation along the soft-sediment shores of the Northwest Atlantic. Eelgrass is considered an ecologically significant species in Atlantic Canada as it provides essential ecosystem functions and services and is an indicator of ecosystem health. Recent declines of eelgrass habitats highlight the need for methods quantifying the large-scale distribution of eelgrass throughout Atlantic Canada to monitor for further habitat loss. We used archived, high-resolution SPOT 6/7 satellite imagery to classify where vegetated habitats exist and if eelgrass was the dominant species. We focused on three bays in Nova Scotia: Port Mouton Bay, Port Joli Bay, and Jordan Bay. In 2015, field surveys were conducted to obtain training points, which were supplemented with visually identified points to perform a supervised classification based on the maximum likelihood classifier. We also performed an unsupervised classification, where clustering algorithms were used to build training sites for a maximum likelihood classifier without using field survey data. These two pixel-based approaches provided similar results across the different images. Regardless of classification type (supervised versus unsupervised), we found different levels of success for the three bays. In Port Joli Bay, we were able to calculate where vegetated habitats occurred and what was the dominant species. This provided bay-wide distribution maps and suggested that 8.61-11.10% of the bay was covered by eelgrass. In Port Mouton Bay, we were able to calculate vegetation presence from absence, and eelgrass habitats were qualitatively differentiated from seaweed habitats by incorporating substrate data and local ecological knowledge. In contrast, benthic habitats could not be classified in Jordan Bay, highlighting the importance of sufficient water clarity for classifying satellite imagery. Our study has implications for the monitoring, conservation and management of eelgrass and other vegetated coastal habitats in Atlantic Canada by providing bay-wide distribution maps, and a classification framework which requires no field survey points for ground truthing.