What's a picture really worth? On the use of drone aerial imagery to estimate intertidal rocky shore mussel demographic parameters

Abstract

Describing distribution patterns of organisms on the rocky intertidal zone is crucial to generate ecological models of broad scope and validity. Our aim was to combine 3D photogrammetric models and spectral analysis derived from aerial images, with ground-based quadrat sampling to provide realistic measurements of mussel (Mytilus galloprovincialis) density, size and reproductive output. A remotely piloted aircraft (drone) was used to conduct intertidal photographic surveys during low tides (<0.4 m) over 8 study sites along the Portuguese central west coast. At each site, high pixel resolution photographs were collected at 30 m altitude (82–247 photographs, 80% overlap, with a ground spatial resolution of approximately 0.8 cm/pixel), encompassing an average intertidal area of 15200 m2 per survey location. A mosaic was created from the images and these were georeferenced and a 3D photogrammetric model was constructed for each location. An analysis of the spectral signature for the different ground cover types was used to perform a maximum likelihood supervised classification with overall high classification accuracy (86.5% ± 4.3%, mean ± SD). Additionally, we analysed the effect of environmental variables (substrate complexity and wave exposure) on mussel density on rocky substrate and size. Maximum mussel density occurred at intermediate values of wave exposure while smaller sized mussels were found at high values of wave exposure. Density, size and reproductive output maps were predicted for the study region. This information allows the validation of metapopulation models dealing with the persistence of spatially-structured populations within patchy habitats and so enables the management of pivotal conservation areas.