Publisher Correction: Combining multi-marker metabarcoding and digital holography to describe eukaryotic plankton across the Newfoundland Shelf

The planktonic diversity throughout the oceans is vital to ecosystem functioning and linked to environmental change. Plankton monitoring tools have advanced considerably with high-throughput in-situ digital cameras and genomic sequencing, opening new challenges for high-frequency observations of community composition, structure, and species discovery. Here, we combine multi-marker metabarcoding based on nuclear 18S (V4) and plastidial 16S (V4–V5) rRNA gene amplicons with a digital in-line holographic microscope to provide a synoptic diversity survey of eukaryotic plankton along the Newfoundland Shelf (Canada) during the winter transition phase of the North Atlantic bloom phenomenon. Metabarcoding revealed a rich eukaryotic diversity unidentifiable in the imaging samples, confirming the presence of ecologically important saprophytic protists which were unclassifiable in matching images, and detecting important groups unobserved or taxonomically unresolved during similar sequencing campaigns in the Northwest Atlantic Ocean. In turn, imaging analysis provided quantitative observations of widely prevalent plankton from every trophic level. Despite contrasting plankton compositions portrayed by each sampling method, both capture broad spatial differences between the northern and southern sectors of the Newfoundland Shelf and suggest complementary estimations of important features in eukaryotic assemblages. Future tasks will involve standardizing digital imaging and metabarcoding for wider use and consistent, comparable ocean observations.

Eukaryotic plankton are enormously diverse and drift with surrounding water masses, also contributing to ecosystem function through primary and secondary productivity, mediating carbon sequestration along with other oceanic elements, and containing potentially harmful blooming species 1 . Plankton monitoring tools using conventional microscopy are steadily being replaced by a new generation of optics in digital imaging instruments, equipped with algorithms to extract, store, and classify major plankton groups mostly within nano-mesoplankton (2 µm to 20 mm) size ranges and bearing morphologically distinct features 2,3 . Imaging provides quantitative observations of taxonomic and functional diversity while also capturing species behaviour with unintrusive, inflow sampling designs 4 . Together, the enhanced sampling capacity and stored digital information in image data have tremendous potential for quantitative plankton ecology, especially for morphologically diverse eukaryotes 5,6 .
Numerous imaging instruments and modes currently exist for plankton (e.g., see Table 1 in 1 ). Among these, digital holographic imaging based on wavefront diffraction has grown across various disciplines, configured with or without optical lenses and supported by increasingly efficient numerical reconstruction algorithms applied to off-axis and in-line illumination paths for high-resolution amplitude and phase images 7,8 . The ordinary holographic principle to reconstruct the wavefront diffraction of an object from a coherent light source (i.e., a laser) can also be extended to record dimensions of color from incoherent light, including fluorescence 8 . Instrument designs without objective lenses or other opto-mechanical parts are well purposed to sample a wide range of

Results
Multi-marker sequencing characterizes eukaryotic diversity. The combined set of CTD-derived physicochemical variables, quantitative imaging, and ASV richness with community composition from 18 and 16S rRNA genes is shown in Fig. 1. Combining 18S and 16S markers revealed rich eukaryotic compositions on the Newfoundland Shelf. After strict quality filtering (See Methods), 18S sequences (> 332 K) were classified to 237 ASVs and 16S sequences (> 99 K) were classified to 293 ASVs. A full list of unique genera classified from both markers is shown in Fig. S6. Heterotrophs dominated 18S rRNA sequence proportions owing to the copepod order Cyclopoida, which contributed up to 86% on the Bonavista Banks ( Fig. 1) but was unclassified to lower taxonomic ranks. Four copepod species were classified in the order Calanoida (Calocalanus curtus, Centropages hamatus, Paracalanus parvus, and Temora longicornis) throughout our samples at low proportions (< 1%), although reaching > 80% of total sequence proportions at 50 m off shelf (SEGB-19) at the SE Grand Banks. Animals were also detected belonging to the gastropods (Heterobranchia) and Appendicularia (Oikopleuridae), constituting a notable fraction of ASVs (> 40%) to the SE Grand Banks coastal station (SEGB-01). A saprophyte genus of Labyrinthulomycetes (Aplanochytrium) was detected in rare proportions (< 1%) of 18S ASVs at off shelf stations on both Bonavista Banks and the SE Grand Banks. Phytoplankton usually composed < 30% of 18S ASVs mostly belonging to four dinoflagellate genera: Tripos (Neoceratium), Protoperidinium, Biechelaria, and Pelagodinium. Diatoms contributed between < 1 and 10%, consisting of ASVs classified as Chaetoceros, Guinardia, Pseudo-nitzchia, and Thalassiosira.
Despite numerous phytoplankton classified in 18S ASVs including dinoflagellates, which were undetected in our chloroplast 16S V4-V5 targets and in other V1-V2 hypervariable regions 39 , the 16S show phytoplankton diversity is drastically underestimated (Fig. 1). Picophytoplankton classified as pelagophytes were widely distributed despite our 10 µm filter and were the main component (> 60%) on the SE Grand Banks. The diatoms were more diverse and contributed similar proportions on the Bonavista Banks, mostly belonging to Thalassiosira, Thalassionema, Rhizosolenia and Delphineis, among other rarer genera. Haptophytes contributed up to 25% of compositions as Phaeocsystis and Emiliania. Other rare phytoplankton > 10 µm were distributed across our samples including rappemonads and silicoflagellates mostly belonging to the family Dictyochophyceae.
Eukaryotic plankton across the Newfoundland Shelf gradient. The assemblages of eukaryotic plankton observed here in the early winter transition of the North Atlantic bloom phenomenon 38 displayed notable differences between north and south sections on the Newfoundland Shelf. Temperature and Salinity (T-S) transitioned at both transects from relatively cooler, fresher shelf water towards warmer, saltier water masses off shelf ( Fig. 1; Fig. S1). ASV richness was greater on the Bonavista Banks than the SE Grand Banks, highest towards the Northeast Slope (BB-11, BB-12) and at 50 m off shelf (BB-15) into the increasing gradient of temperature and salinity ( Fig. 1, Fig. S1). These samples were characterized by generally low plankton concentrations observed in holographic images indicating smaller nanoplankton contribute to this diversity where microplankton concentrations captured in our images are low ( Fig. 1). Conversely, the SE Grand Banks displayed rich phytoplankton samples near the shelf break (SEGB-12) seen in 16S ASVs coincident with high fluorescence (> 4 µg L −1 ) and concentrated diatoms (> 4500 individuals L −1 ) composed mostly of chains (Figs. 1, 3). Off-shelf of the SE Grand Banks (SEGB-17) a warm water mass contained a combined elevation of chlorophyll-a fluorescence with dinoflagellates (> 120 individuals L −1 ) and diatom (> 3000 individuals L −1 ) concentrations that steadily declined while richness increased towards the warmest off-shelf station . Low autotrophic biomass on the Bonavista Banks was evident in both fluorescence and imaged plankton concentrations although no correlation was apparent; however imaged concentrations did co-vary with fluorescence on the SE Grand Banks (R 2 = 0.76) (Fig. S7). Elevated plankton concentrations clustered on the shelf break of the SE Grand Banks containing centric (> 350 L −1 ), chain-forming (> 650 L −1 ) and rod-shaped (> 5000 L −1 ) diatoms, cooccurring with relatively high concentrations of the dinoflagellate genera Tripos (> 80 L −1 ).

Discussion
The Newfoundland Shelf displayed a physical gradient transitioning across the continental shelf ( Fig. 1; Fig. S3), capturing the temperate ocean water mass boundaries delimited by the SE Grand Banks and Bonavista Banks, with subpolar and subtropical water masses prevailing to the north and south, respectively 43 . This separation between shelf and open ocean waters is a ubiquitous feature of the Northwest Atlantic Ocean 44 and can create biologically productive boundaries where (sub)mesoscale dynamics actively modify photosynthetically available radiation, nutrients, and predator encounter frequency 45 . The work described here combines high-throughput holographic imaging and multi-marker metabarcoding to characterize rich eukaryotic plankton communities (534 unique ASVs) across the Newfoundland Shelf gradient. By complementing imaging and multi-marker metabarcoding, broad quantitative and taxonomic differences were evident between north and south sections on the Newfoundland Shelf: Bonavista Banks contained low biomass, high richness samples characterized by relative higher proportions of diatoms and haptophytes; the SE Grand Banks revealed more complex eukaryotic compositions with notably high diatom concentrations near the continental margin, likely stimulated by upwelling of deep, cold, nutrient rich waters (> − 50°W; Fig. S1) that frequently characterize this region 46 . Taken together, complementing imaging and multi-marker metabarcoding afforded broad comparisons between plankton abundance, diversity, and taxonomic composition against broadly different environmental profiles.
Seasonal planktonic diversity has recently been characterized in surface waters between 37 and 66°N off the Newfoundland Shelf using chloroplast 16S (V1-V2) amplicon sequencing with conventional and imaging flow cytometry 39,47 that provides important context to our findings. The off-shelf plankton compositions from the early winter transition phase (2015) were dominated by cyanobacteria and pico-phytoeukaryotes with large fractions of Micromonas, Bathycoccus, Ostreococcus (Clade I-II) and increasing proportions of pelagophytes northward 47 , for which we observe the opposite trend on the Newfoundland Shelf. The samples here omit most small eukaryotes due to > 10 µm filtration, although the pelagophyte picophytoplankton occurring throughout both transects were likely trapped in filter pores clogged by larger cells and in particle-associated forms, instead these results highlight that the numerically rare nano-microeukaryotes are highly diverse, especially diatoms, and patchy across the Newfoundland Shelf. We also recovered multiple dinoflagellate genera through 18S ASVs and imaging samples which were undetected with chloroplast 16S targets 39 and taxonomically unresolvable from pigments 48 . Unravelling mixotrophic feeding strategies in dinoflagellates and other eukaryotic protists (e.g., radiolarians) to estimate feeding rates and biogeochemical fluxes will depend on adequate detection and behavioural observations 49  www.nature.com/scientificreports/ approach. It is also observed that the important chain-forming diatom Chaetoceros, which rapidly contributes to phytoplankton compositions upon the initiation of the subpolar spring bloom, is undetected off-shelf in wintertime and is likely transported within water masses or experiences higher success during the winter-to-spring transition 47 ; our observations suggest Chaetoceros and other springtime blooming plankton could be advected across the Newfoundland shelf-slope boundary, implying a possible pathway within the Labrador current export along the Flemish Cap and SE Grand Banks 50 . The skewing of planktonic diversity from underlying absolute abundances in compositional sequencing data is due to several technical and biological factors. Principally, there are potentially large differences in sample sources between quantitative imaging and metabarcoding: Images collected here are mostly intact living cells, but both cellular and extracellular DNA exist in the bulk DNA pools of seawater, thus it cannot be determined what fraction of a sequenced sample is living metabolizing cells, dead cells, dormant cysts, or detritus 51,52 . For 18S compositions, arthropods likely dominated due to their multicellularity, although size fractioning through a larger pore size, e.g., > 250-500 µm as per conventional net tows 53 would largely remove copepods to mitigate their overrepresentation during the amplification stage. Assigning ASVs as unique biological organisms is also limited by incomplete reference databases and biased for organisms of historical importance 54 . Work is still needed to understand biases in each marker gene target but extending to full-length gene sequencing platforms could superiorly resolve taxonomic composition and better account for intragenomic gene copies [55][56][57] . Altogether, this suggests that until quantitative community level sequencing is achieved 58 it remains a separate, but complementary line of evidence with quantitative imaging towards plankton community composition.
Despite its limitations, metabarcoding benefitted the overall community assessment with a much deeper taxonomic resolution of the eukaryotes than paired images. Only 12 genera were identified across all imaging samples, whereas 78 unique genera were identified in the metabarcoding analysis. This resulted in important microbial eukaryotic groups undetected in our images but classified in ASVs. These include protists belonging to the major phyla Cercozoa in 18S ASVs, and haptophytes in the 16S ASVs from widely abundant coccolithophorid genera Emiliania and non-coccolithophorid genera Phaeocystis, which forms the densest colonial blooms in the North Atlantic and Southern Oceans 59 . Each group was likely unidentifiable in our holographic images due to their small cell size (< 50 µm) which has previously been difficult for recovering sharp images using the HoloSea 60 , and due to morphologically indistinct features. Metabarcoding also provided potential and validated taxonomic identities of unidentified image objects. Many detected rod-shaped and centric diatoms with distinct features remained taxonomically unresolved, however the corresponding ASVs suggest they likely belong to genera Guinardia, Actinocylcus, and Thalassiosira. The fungus-like Labyrinthulomycetes are the clearest example for improved taxonomy as they were missed during manual identification but classified as the genera Aplanochytrium in paired DNA samples; their absence in paired images directed a re-assessment of detected objects which led to 10 identified specimens. The Labyrinthulomycetes identified here belong to the order Labyrinthulida 61 , a dominant marine group also detected in the photic zone during the Tara Oceans expedition 22,61 -although the Tara expedition samples largely omit the Northwest Atlantic Ocean. A new ecological view of the Labyrinthulomycetes is growing with observations showing these protists can exceed prokaryotic biomass and frequently cooccur with fungi to saprophytically break down neutrally buoyant marine snow in the bathypelagic zone, significantly contributing to the metabolism of deep-sea ecosystems 62,63 .
Quantitative imaging by digital in-line holography captured abundant micro and mesoplankton from every trophic level using the pipeline for hologram reconstruction and object detection outlined in 60 . The HoloSea returned highly focused images for quantitative observations of major plankton clades in aquatic ecology (diatoms, dinoflagellates, ciliates, copepods, radiolarians, silicoflagellates, other large protists) but other important groups (coccolithophorids, haptophytes, other small flagellates, fish larvae) are currently omitted due to size, pixel resolution, or morphologically indistinct features. Future studies investigating plankton size structure from the HoloSea or similar holographic microscopes should implement a correction factor to account for the scattering of coherent light that biases size estimates non-linearly depending on object distance from the point source 64 . The promise for digital in-line holography is shared by digital imaging technologies more broadly, and is typically prohibitive for conventional light microscopy, which is to yield data-rich, high-frequency observations which to date have captured planktonic fish-larvae survivorship under river plume discharge regimes 65 , the evolution of particulate-mediated carbon export in rapidly changing marginal zones 66 , spatial patterns in plankton community compositions through time 67 , at regional scales 4,68 and specific taxonomic lineages at global scales 69 . Novel biotic interactions have recently been observed using in-situ imaging, including pseudopodial feeding in acantharians from the East China Sea likely missed previously due to destructive sampling by conventional plankton nets 4 and the frequent parasitization of the cosmopolitan copepod Oithona at the Scripps Pier in the Pacific Ocean 70 . Digital imaging can also capture size-structure patterns as an indicator of functional diversity 71 , which is a deeply evolutionary trait that affects ocean food webs and biogeochemistry 72,73 . The specialty of imagery for morphologically distinct eukaryotes is of particular importance because, unlike bacteria, morphology and behavior are central to eukaryotic diversity and ecology and metabarcoding is blind to these features 74 .
This work presents paired high-throughput imaging, metabarcoding, and physicochemical observations from the productive Newfoundland Shelf to show that the eukaryotic plankton fraction displays subregional variation across physical gradients, including patchy biogeography and biomass along the continental margin during the early winter bloom transition. As a geographic juncture between North Atlantic water masses and ongoing climate changes to the subarctic North Atlantic Gyre 75,76 , the Newfoundland-Labrador shelves deserve a more comprehensive study of planktonic diversity and variability using state-of-the-art molecular and optical tools to improve upon existing seasonal surveys relying on net tows, conventional microscopy, and bulk chlorophyll-a 77 or flow cytometry 78 . The high-throughput, durable, and lensless designs for digital holographic microscopes supported by increasingly automated pipelines will make suitable complements with high-sensitivity molecular Oceanographic physicochemical data. Full water column profiles were obtained by lowering a conductivity-temperature-depth (CTD) rosette equipped with a calibrated fluorometer and a dissolved oxygen sensor (Seabird SBE-9+) from surface waters to 10 m above the seafloor. All CTD files were provided by Fisheries and Oceans Canada and handled in R using the oce package (v. 1.4.0) 83 . The CTD files were filtered in a standard fashion: CTD casts were trimmed to remove upcast anomalies, interpolated using the default method at 1 m increments, smoothed with local (Boxcar) averaging and gridded into sections 83 . The chlorophyll-a fluorescence (µg L −1 ) corresponding to the Niskin samples (Table S1) was extracted and averaged within three depth bins (5-6, 20-21, 50-51 m) to approximate conditions at Niskin bottle sample depths.
Sample collection: paired imaging and filtration. Seawater was collected using 20 L Niskin bottles attached to the CTD rosette, emptied directly into 10 L cubitainers where a minimum of 2 L sample volume was allocated for paired imaging of micro and mesoplankton and filtration of eDNA. The full dataset includes 27 water samples from 15 stations (See Table S1) at 5 m, seven from the SE Grand Banks and eight from Bonavista Banks; three shelf break stations along each transect were also analyzed at 20 and 50 m. First, seawater was pumped at 150 mL min −1 using a peristaltic pump (Fisherbrand GP1000) into a digital in-line holographic microscope, the HoloSea S5 (Described below) at 10 frames s −1 and the seawater was collected in a container for further filtration for DNA extraction. The lower size limit of the object detection algorithm for the HoloSea S5 was set at 20 μm as a conventional lower size threshold to capture microplankton and larger objects 60 60 . To briefly detail the acquisition process of plankton objects, first raw holograms are reconstructed in 4-Deep Octopus software using a Helmholtz-Kirchhoff transformation and subsequently, using the 4-Deep Stingray software, biological Regions of Interest (ROIs) are detected using a global adaptive thresholding and filtered for in-focus objects in each reconstructed plane 60 . These in-focus objects compose our quantitative profiles which were manually identified to the lowest possible taxonomic rank ( 85 Table S2 Bioinformatic analysis. The ASVs were taxonomically classified using the multinomial naïve-Bayes QIIME2 q2-feature-classifier plugin 94 . The 18S sequence classifier was pre-trained in-house on the SILVA (v. 138.1) database 95 , and the chloroplast 16S sequences were classified in a two-step process: First sequences were run against SILVA (v. 138.1) database to extract chloroplast sequences to be classified from the PhytoREF database 27 . The classified ASVs were then filtered in two ways: (1) Rare ASVs were removed if they were less than 0.1% of mean sample depth in accordance with Illumina's estimation of bleed-through errors during sequencing, (2) all ASVs only classified to the kingdom level were excluded. The remaining ASVs were exported from QIIME artifacts into R (v. 3.6.2) 96 as a Phyloseq object (v. 1.3.6) 97 for further analysis. Richness (alpha diversity) was estimated using ASVs as a surrogate in the R package breakaway (v. 4.7.3) 98 . The breakaway model attempts to adjust richness estimates for unobserved taxa and debias samples from differing sequencing depths 99 . To assess differences in community composition between samples (beta diversity), all ASVs were transformed using a centered log-ratio ( clr ) of a composition x = (x 1 , . . . , x i , . . . , x D ) defined as: where each ASV ( x ) is divided by the geometric mean of all ASVs in a sample given by g(x) = D √ x i . . . x D 100 . To prevent undefined log-ratios, a pseudo-count of one was added to each zero 101 . To summarize compositional differences between transects a principal component analysis of the Euclidean (Aitchison) distances 40 was performed on log-transformed ASV counts 37 and tested for statistical significance using a permutational analysis of variance 102 in the vegan package (v. 2.5.7) 84 .

Data availability
18S and 16S sequences are available under the NCBI Sequence Read Archive BioProject PRJNA803249. Additional plankton objects which are representative of major taxa found in this study are available (https:// doi. org/ 10. 6084/ m9. figsh are. 20259 756) and the imaging datasets generated during the current study are available from the corresponding author(s) upon reasonable request. (1) clr(x) = ln x 1 g(x) , . . . , ln x i g(x) , . . . , ln x D g(x)