Micro-mesoplankton groups imaged on the Newfoundland Shelf (Nov-Dec, 2019) aboard the RRS James Cook. Plankton are segmented from holograms collected using the HoloSeaS5 (4Deep). Raw holograms were reconstructed with 4Deep Octopus software, and in-focus plankton were detected with the companion Stingray software. Detecting in-focus plankton images- or regions-of-interest (ROIs)- depended on a pipeline for depthwise thresholding, clustering and pixel autocorrelation. All plankton were manually identified using various sources from the primary literature including several major guides (1-3)
Beginning alphabetically, this will show the large-scale taxonomic (Phylum) resolution for major groups including lower level identification where possible. All names listed are shown left to right.
Copepods and a nauplii (first-stage larval copepod).
Salpingella spp. (Image 1, 3-5), Dictyocysta spp., Condonella spp., Choreotrichia spp.
Chaetoceros spp. (Image 1-3), Thalassionema spp., Pseudo-nitchzia spp., Proboscia spp., Nitzschia spp.
Tripos fusus (Image 1-2), Tripos spp. (Image 3, 5-6), Thalassionema spp., Prorocentrum spp., Gyrodinium spp. (Image 7-8).
Labyrinthulomycetes (Image 1-2), marine snow and particulates
Acantharia (Image 1-2), Ochrophyta, Dictyocha spp. (Image 4-6).
Quantitative plankton imagery can seriously improve our in-situ sampling efforts: the high-throughput capacity of many instruments and the data-rich nature of digital images can expand our knowlegde of fine-scale biological properties, absolute community compositions, and render automation increasingly possible for major plankton groups and imaging types. Holography has some important advantages in rapidly sampling large volumes at micron-level resolution, and in the case of digital in-line holographic microscopy, it is altogether lightweight, durable and deployable. Underpinning all this is the lensless design of in-line models, which Dennis Gabor (1948) realized as the solution to spherical abberation problems of early electron microscopes developed by Ernst Ruska: the only perfect lens is the point source laser with spatially coherent light waves and perfect magnification. This idea has advanced considerably in the intervening 75 years. Barriers remain in the technically difficult physics-driven approaches for numerical hologram reconstruction and segmenting an in-focus plankter from a messy volume of water, the prospects of unifying colllection through classification with deep learning (or similarly generalizable algorithms) are improving. Overall, there are good reasons for its ascendency.
[1] Tomas, CR. Identifying Marine Phytoplankton. San Diego: Academic Press; 1997.
[2] JR Dolan, DJS Montagnes, S Agatha, DW Coats, DK Stoecker.The Biology and Ecology of Tintinnid Ciliates: Models for Marine Plankton, First Edition. Wiley & Sons Ltd; 2013.
[3] MJ Tremblay, JT Anderson. Annotated species list of marine planktonic copepods occurring on the shelf and upper slope of the northwest Atlantic (Gulf of Maine to Ungava Bay). Can. Spec. Publ. Fish. Aquat. Sci. 1984.