The Gulf of Alaska is a distinct habitat within the North Pacific Ocean, bounded by the eastward flowing North Pacific Current that separates the subarctic Gulf of Alaska from the subtropical California Current System (Sutton et al., 2017). To the north and east, the Gulf of Alaska is bordered by highly productive shelf and coastal ecosystems, while offshore waters are presumed to be less productive under a High Nutrient Low Chlorophyll (HNLC) regime (Martin et al., 1989). Within these HNLC waters lies the Gulf of Alaska Seamount Province, a region characterized by hundreds of extinct submarine volcanoes (Menard and Dietz, 1951). The Seamount Province is composed of two distinct chains, the Kodiak-Bowie Seamount Chain that spans 900 km from the Aleutian Trench off eastern Kodiak Island (USA) to Queen Charlotte Island (CAN) and, to the south, the Cobb Seamount Chain that extends from the Aleutian Trench off western Kodiak Island to the Cobb hotspot on the Juan de Fuca Ridge (Menard and Dietz, 1951; Chaytor et al., 2007).

The pelagic ecosystem from surface waters to the abyssal plains was surveyed from 22 July – 2 August 2019 aboard the R/V Sikuliaq. Sampling was conducted at six oceanographic stations, including the transition zone from the Seward Line in the Northern Gulf of Alaska (https://nga.lternet.edu/) to the Quinn and Giacomini seamounts, located along the northern extent of the Kodiak-Bowie Seamount Chain ( Figure 2 ). Surveys were conducted at the Quinn and Giacomini seamounts, sampling from surface waters to the summit (~650 m deep) and to the abyssal plain where the base of the seamounts lie between 4000 and 5000 m depth.

Zooplankton were collected using a variety of sampling techniques ( Table 1 ). Two MultiNet Plankton Samplers Type Midi (0.25 m²; Hydro-Bios) each equipped with five 150-µm mesh plankton nets were fastened together and deployed in tandem ( Figure 3 ); one MultiNet system was receiving and transmitting live feed from a 0.322 conducting wire, while the trigger depths for each net on the other MultiNet system were programmed before deployment. Securing two MultiNet systems together maximized wire time, a vital resource when conducting hours-long casts to the abyssopelagic zone. The MultiNet systems were equipped with five nets and thus required two sequential vertical casts to sample the entire water column for the following depth bins: 0 – 40 – 100 – 200 – 300 – 400 – 600 – 1200 – 2000 – 3000 – bottom ( Table 1 ). When necessary, the top two strata were combined. Net samples from one MultiNet were preserved in 10% buffered formalin-seawater for taxonomic analysis, while net samples from the second MultiNet were split with a Folsom plankton splitter. One half of the sample was preserved in 95% ethanol while the other half was used for live sorting.

A 1 m² Multiple Opening/Closing Net and Environmental Sensing System (MOCNESS; Wiebe et al., 1985) was deployed following the tandem MultiNet ( Figure 3 ). The MOCNESS was equipped with 505-µm mesh nets and sampled the following depth bins: 0 –100 – 200 – 300 – 400 – 600 – 1200 – 2000 – 3000 m ( Table 1 ). Since the MOCNESS has a larger mouth diameter than the MultiNet (1 m² vs 0.25 m²) and thus filters a larger volume of water, samples were split twice. One half of the net sample was preserved in 10% buffered formalin-seawater for taxonomic analysis whereas the second half was split again; one quarter was preserved in 95% ethanol and the other quarter was used for live sorting. All bulk ethanol plankton net samples were stored at -20 °C and complete ethanol changes were performed ~24 hours after initial preservation.

In addition to traditional plankton net sampling, the Global Explorer ROV (Oceaneering, Houston, TX) conducted pelagic dives down to ~2550 m ( Table 1 ). ROV operations focused on organisms larger than 1 cm and of the gelatinous zooplankton group (e.g., ctenophores, medusae, larvaceans; Bailey et al., 1995). Specimens were collected with a rotary carousel suction sampler (Youngbluth, 1984b) or static “D-samplers” (Youngbluth, 1984a). At all stations, ROV dives transpired during daytime operations (0800 – 2000) while plankton net deployments were conducted during nighttime operations (2000 – 0800).

All mesozooplankton were identified based on current taxonomic literature (e.g., Boxshall and Halsey, 2004). Organisms were identified at sea from live collections from both the plankton net samples and the ROV collections described above. Once on board, organisms from the ROV and plankton net tows were immediately transferred to a cold room maintained at 4 °C until sorting and identifications could be performed. Zooplankton were identified to the lowest taxonomic level feasible at sea using Leica MZ16, M165C, or M205C stereomicroscopes, catalogued into cryogenic vials, and preserved in 95% pre-chilled ethanol for DNA sequencing. For each catalogued specimen, a matching formalin voucher was created for taxonomic integrity. In instances when large, rare organisms from gelatinous zooplankton groups could not have a taxonomic voucher, tissue samples were cut off and preserved in ethanol while the remainder of the organism was preserved in formalin. For ROV samples, video imaging and photographs were captured in situ with an Ultra-high definition (UHD) 4K and stereo high definition (HD) camera systems prior to collection. The video clips (Apple ProRes 4:2:0 codec,.mov files with embedded timecode) serve as further voucher "material." Organisms from net collections were photographed using a 12MPix Spot Insight camera (uncompressed JPEG) mounted to a Leica stereomicroscope.

Organisms that could not be identified to the species-level at sea were more closely examined back at the University of Alaska Fairbanks using both regional and phylum- or family-specific taxonomic keys and original descriptions. Due to morphological complexity, incomplete taxonomic keys, damaged specimens, and the lack of descriptive studies on zooplankton from the deep ocean, some specimens were unable to be identified to species. For these instances we used Open Nomenclature (ON) qualifiers defined in Horton et al. (2021) and Sigovini et al. (2016). To the best of our knowledge, there is no qualifier for specimens assigned a taxonomic identification based on molecular data. Thus, we introduce the new ON qualifier deoxyribonucleic acid, abbreviation DNA (e.g., DNA Genus, Genus DNA species) for specimens assigned a genus- or species-level identification based on clustering (100% bootstrap support) of DNA barcode genes using phylogenetic trees against preexisting barcodes derived from expertly identified specimens, where at least one of the sequences in that clade could be referred to a physical specimen (or photographs) where identification could be corroborated through morphological analyses.

DNA barcoding was performed at the Smithsonian’s Laboratory of Analytical Biology (L.A.B) at the National Museum of Natural History (Washington D.C.) from the ethanol-preserved tissue samples catalogued at sea. Genomic DNA (gDNA) was extracted using a phenol-chloroform method on an AutoGenprep 965 Automated DNA Isolation System (AutoGen Inc., Holliston, MS, USA) following manufacturers’ protocols. For small taxa (e.g., copepods and larvaceans) DNA was extracted from the entire organism while, for larger organisms (e.g., Cnidarians, ctenophore, chaetognaths, and larger crustaceans), non-diagnostic features such as legs or antennae were removed for extraction where possible.

Extracted gDNA was quantified using a SpectraMax iD3 Multi-Mode Microplate reader (Molecular Devices). Mitochondrial COI and 16S rRNA and nuclear 18S rRNA were amplified using previously published primers (COI: Folmer et al., 1994; Geller et al., 2013; 16S: Cunningham and Buss, 1993; 18S rRNA: Hillis and Dixon, 1991; Borchiellini et al., 2001). All PCR reactions were carried out in 10 µL reaction volumes containing 1 µL DNA template, 3.2 µL PCR grade H₂O, 5.0 µL Promega GoTaq HotStart 2X Master Mix, 0.3 µL each forward and reverse primer, 0.1 µL BSA, 0.1 µL MgCl₂. PCR protocols for each amplicon were as follows: COI: initial denaturation of 95 °C 7 min; 4 cycles of 94 °C 30 sec, 50 °C 45 sec, 72 °C 1 min followed by 34 cycles of 94 °C 30 sec, 45 °C 45 sec, 72 °C 1 min; final extension of 72 °C 8 min; 16S: initial denaturation of 95 °C 2 min; 35 cycles of 95 °C 30 sec, 50 °C 30 sec, 72 °C 1 min 30 sec; final extension of 72 °C 5 min; and 18S rRNA: initial denaturation of 95 °C 2 min; 35 cycles of 95 °C 30 sec, 52 °C 30 sec, 72 °C 1 min 30 sec; final extension of 72 °C 5 min. PCR products were visualized on a 1% agarose gel. Successful PCR products were purified with ExoSAP-IT Express (37 °C 4 min and 80 °C 1 min). Cycle sequencing was carried out on purified PCR product using the Big Dye Terminator v3.1 kit (Applied Biosystems) with each primer used for PCR amplification. Due to the amplicon length of 18S, the internal primer 18SCnew (5’-CAG CCG CGG TAA TTC CAG C-3’; Miranda et al., 2016) was sequenced to bridge the gap from the 3’ end of 18SE and the 5’ end of 18SL. Cycle sequencing product was cleaned using Sephadex G-50 powder and run on an ABI PRISM 3100 Genetic Analyzer.

Data management for DNA barcodes follows the Smithsonian Institution Barcode Network guidelines (Redmond et al., 2023) that abides by the Barcode Data Standard established by the Consortium of the Barcode of Life (CBOL; Hanner, 2012). Specimen, voucher, and sequence metadata are maintained in the SI Field Information Managements System Genomic Observatories MetaDatabase (FIMS-GEOME; https://geome0db.org/) and the Laboratory Information Managements System (LIMS) databases.

For each genetic marker, sequences were assembled, trimmed, manually checked for ambiguous base calling and stop codons, and aligned using MAFFT (Katoh et al., 2002) in Geneious Prime 2023.2.1 (https://www.geneious.com). Sequence names were annotated and assigned a species name using the Biocode plugin that links sequences in Geneious Prime to the SI FIMS and LIMS databases. Sequences were run against the NCBI GenBank database using BLAST (Altschul et al., 1997) to check for accuracy of species identifications and potential contamination. Due to the lack of barcode data for species from the North Pacific deep-sea ecosystem, this QA/QC step was only informative for species with existing reads in GenBank. Therefore, neighbor-joining (NJ) trees were constructed for each genetic marker to determine clustering for all groups of zooplankton. Once identifications were checked and confirmed, sequences were submitted to the National Institute of Health National Center for Biotechnology Information (NIH NCBI) Sequence Read Archive (SRA). 18S and non-cnidarian COI reads were submitted under BioProject PRJNA971221 whereas COI and 16S reads for cnidarians were submitted under BioProject PRJNA1077907.

From this work, we generated 1462 new DNA sequences from 253 unique taxa from the Gulf of Alaska ( Table 2 and Supplementary Table S1 ). DNA sequences were generated for COI from 781 specimens belonging to 73 families (GenBank Accession Nos. OR985931 – OR986516 & PP409229 – PP409419), 16S for 33 specimens belonging to 18 families (GenBank Accession Nos. PP356923 – PP356955), and 18S from 648 specimens belonging to 42 families (GenBank Accession Nos. PP032093 – PP032748). Sequences for all three genes encompassed organisms from 7 phyla including Arthropoda (Orders Copepoda, Amphipoda, Decapoda, Euphausiacea, Lophogastrida, and Isopoda), Cnidaria (Hydrozoa and Scyphozoa), Ctenophora (Tentaculata), Chaetognatha, Mollusca, Chordata (Tunicata), and Annelida.

Phylogenetic relationships were analyzed for all three genes through Neighbor-Joining trees ( Supplementary Figures S1 – Supplementary Figures S3 ). Mitochondrial COI reads clustered into major clades with >66% bootstrap support with 76% support for the order Copepoda, 50 – 100% for the phylum Cnidaria, 72% for the phylum Chaetognatha, 70 – 83% for the order Amphipoda, 66% for the order Euphausiacea, and 96 – 100% for the order Decapoda. Monophyletic relationships were resolved at 100% bootstrap support for sequences from the same species ( Supplementary Figure S1 ). Nuclear 18S rRNA reads clustered into major clades with >89% bootstrap support ( Supplementary Figure S2 ) with subclades forming at 97 – 100% support for the order Doliolida, 100% support for family Oikopleuridae, 70 – 100% support for the phylum Chaetognatha, 50 – 100% support for the order Pteropoda, 61 – 100% for the phylum Annelida, 60 – 100% support for the order Euphausiacea, 65 – 100% for the order Decapoda, and 51 – 100% for the order Lophogastrida. The order Amphipoda clustered – at 100% bootstrap support – into three distinct subclades with 18S reads from Koroga megalops (family Uristidae) forming the first clade, reads from Scina sp. (family Scinidae) forming the second clade, and reads from Primno sp. (family Phrosinidae) and Phronima sp. (Phronimidae) forming the third clade. All reads from the order Copepoda clustered at 100% bootstrap support with subclades clustering between 51 – 100%. ( Supplementary Figure S2 ). Clustering patterns from COI barcodes were used in combination with morphological characteristics to help resolve taxonomic identifications for any unresolved COI and 18S reads, where possible. For Cnidarians, mitochondrial 16S reads clustered into major clades at >54% bootstrap support ( Supplementary Figure S3 ). Reads from scyphozoans clustered at 100% bootstrap support while hydrozoans were split into two clades, represented by organisms from the subclass Hydroidolina (orders Siphonophorae and Anthoathecata) with 64% bootstrap support), then the subclass Trachylinae (orders Trachymedusae and Narcomedusae) with 96% bootstrap support.

From the zooplankton sequenced in this work, 107 species, were previously lacking barcode data from the North Pacific or the global ocean: 57 species for COI, 6 species for 16S, and 44 species for 18S ( Table 2 ). Of the 107 species barcoded, 12 represent new species of cnidarians that are currently undergoing description. Additionally, 47 contributions were made across all markers where no sequence data were previously available for the genus level. The contributions included 11 genera of hydrozoan cnidarians: Halitrephes and Ptychogastria (Trachymedusae), Pandea (Anthoathecata), Cunina, Pegantha, Solmaris, and Solmissus (Narcomedusae), Mitrocoma and Ptychogena (Leptothecata), and Rudjakovia and Vogtia (Siphonophorae); 10 genera of copepoda: Arietellus (Arietellidae), Batheuchaeta and Pseudochirella (Aetideidae), Cephalophanes, Cornucalanus and Onchocalanus (Phaennidae), Pseudhaloptilus (Augaptilidae), Pseudoamallothrix (Scolecitrichidae), Pseudolubbockia (Lubbockiidae), and Ratania (Rataniidae); and one morphotaxon of ctenophores (Cydippida aka. "Ctenoceros"), pelagic tunicates (Doliolula), and isopods (Holophryxus).