Methods for Assessing Acropora spp. Populations in the Caribbean
Since the severe decline of the Acropora populations in Bonaire in the 1980s, no assessment has characterized the distribution of remnant colonies. Because of their patchy distribution, a large sampling effort is necessary to adequately describe their occurrence. However, the spatial scale at which this assessment needs to be carried out makes this prohibitive with approaches such as transects using SCUBA gear and photogrammetry. This internship project aimed to optimize and apply a simple methodology trialed by relevant stakeholders on the island to obtain coarse but spatially explicit data with relatively low time-investment. Snorkelers utilizing a waterproof GPS and a slate to record coarse categorical data outlined patches of Acropora cervicornis and Acropora palmata in-situ. These were processed with an ArcGIS workflow to create shapefiles of coral patches as polygons joined to their corresponding data. The resulting polygons were used to describe the distribution of Acropora spp. along the leeward coast of Bonaire. Furthermore, these were used as ground-truthing data to test whether remote sensing imagery can be used to detect A. cervicornis remotely. 466 polygons along 14.5km of the coast were created, showing a patchy distribution of both species, more frequent occurrence of A. palmata in the northern leeward coast compared to the southern, and vice-versa for A. cervicornis. A multinomial logistic regression, maximum likelihood classification, and forest-based classification all showed a high accuracy in labelling A. cervicornis correctly in remote sensing data, but all showed frequent misclassification of other reef structures as A. cervicornis. The mapping approach presented in this internship could be applied to investigate fragmentation effects in Acropora populations and to gather in-situ ground-truthing data for other benthic habitats.
For full report or more information, please contact email@example.com or firstname.lastname@example.org