Otolith characterization and integrative species identi ﬁ cation of adult mesopelagic ﬁ shes from the western North Atlantic Ocean

Fish diversity and ecology in the ocean ’ s mesopelagic zone are understudied compared to other marine regions despite growing interest in harvesting these potential resources. Otoliths can provide a wealth of taxonomic and life history information about ﬁ sh, which can help ﬁ ll these knowledge gaps; however, there has been relatively little research to date on the otoliths of mesopelagic species. Here, a species-speci ﬁ c image library was assembled of sagittal otoliths from 70 mesopelagic ﬁ shes belonging to 29 families collected in the western North Atlantic Ocean. Images of adult sagittal otoliths from 12 species were documented and photographed for the ﬁ rst time. The ﬁ sh were identi ﬁ ed to species with a combination of morphological characters and DNA barcoding. Regressions between otolith size and ﬁ sh length are presented for the six species with the largest sample sizes in this study. This otolith image library, coupled with otolith-length and width to ﬁ sh-length relationships, can be used for prey identi ﬁ cation and back-calculation of ﬁ sh size, making it a valuable tool for studies relating to food webs in the important yet poorly understood mesopelagic zone. In addition, the 44 ﬁ sh barcodes generated in this study highlight the bene ﬁ t of using an integrative taxonomic approach to studies of this nature, as well as add to existing public databases that enable cryptic species and metabarcoding analyses of mesopelagic species.


Introduction
Mesopelagic fishes are the most abundant fishes in the global ocean (Gjøsaeter and Kawaguchi, 1980;Irigoien et al., 2014).As such, they play a critical role in carbon and energy transfer and contribute to trophic connectivity, especially due to the vertical migration behavior performed by many species (Davison et al., 2013;St. John et al., 2016;Saba et al., 2021).In spite of their abundance, mesopelagic fish species are understudied, and gaps exist in our knowledge of many species' life history strategies (Caiger et al., 2021).Pelagic predators such as white sharks (Carcharodon carcharias), Atlantic bluefin tuna (Thunnus thynnus), and swordfish (Xiphias gladius) have been found to dive down to feed on mesopelagic fishes, highlighting the importance of the mesopelagic zone to surface systems and fisheries (Karakulak et al., 2009;Battaglia et al., 2013;Gaube et al., 2018;Braun et al., 2019;Arostegui et al., 2022).In one study, mesopelagic and benthopelagic fishes were found to be a large percentage of Atlantic bluefin tuna's diet (Battaglia et al., 2013).For analysis of digested prey, Battaglia et al. (2013) drew on taxonomic keys, including otolith atlases (Tuset et al., 2008;Battaglia et al., 2010).Mesopelagic otolith identification keys, therefore, are essential for accurate reconstruction of prey for many commercially important species, and for the study of important food web dynamics that include mesopelagic fishes.
Otoliths, or ear stones, are calcareous structures located in the otic capsule in the inner ear of teleost fishes.There are three pairs of otoliths found in most teleosts: sagittae, lapilli, and asterisci, with the sagittae being the largest structure in most fishes, with the exception of Ostariophysi (Campana and Neilson, 1985;Assis, 2003;Volpedo and Fuchs, 2010).Otolith size and shape can be used for taxonomic identification due to their high interspecific variation and intraspecific consistency (Campana, 2004;Yedier and Bostanci, 2022).Gut content analyses commonly use otoliths found in stomachs to identify prey items (Jackson et al., 1998;Brophy et al., 2009;Karakulak et al., 2009;Battaglia et al., 2013).Thus, comprehensive documentation of otoliths for species is essential to increase confidence in identification and facilitate our understanding of mesopelagic food webs.
Several otolith atlases have been constructed as taxonomic identification guides for specific geographic regions (Smale et al., 1995;Rivaton and Bourret, 1999;Campana, 2004;Tuset et al., 2008;Battaglia et al., 2010;Jones and Morales, 2014;Conversani et al., 2017;Volpedo et al., 2017).However, the otoliths for many mesopelagic species are still either undocumented or poorly documented.Furthermore, atlases often only present images from one individual per species.Additional visual documentation is therefore needed to assess not only interspecific differences in otolith morphology, but also intraspecific variation that might occur due to geographical separation and/or ontogenetic changes.
Otoliths can also provide useful ecological insights, as otolith size is correlated with fish size.Especially in gut content analyses, otoliths, along with other calcified structures such as scales, can be used to estimate the size and weight of prey consumed (Jobling and Breiby, 1986;Jackson et al., 1998;Harvey et al., 2000;Brophy et al., 2009).A simple linear regression of otolith size and fish size has been commonly used to describe these relationships for many fish species, though environmental conditions may affect the relationships between calcified structures and fish size (Harvey et al., 2000;Smoliński and Berg, 2022).There are few studies detailing the otolith size and fish size relationships of mesopelagic species, and limited work on these relationships in the western North Atlantic Ocean (Battaglia et al., 2010).
Our study focuses on documenting otoliths from mesopelagic fish species in the western North Atlantic Ocean.Our objectives were to compile species-specific otolith images of mesopelagic fishes and to describe the relationships between otolith length and width and fish standard length.Mesopelagic fishes are often very difficult to identify, due to subtle morphological differences between species, damage during the collection process, or cryptic species traits (Kenchington et al., 2017).As such, the fish specimens used in this study were identified to species with a combination of morphological analyses and DNA barcoding of the mitochondrial cytochrome c oxidase subunit I (COI) gene.The resulting species-specific otolith library and associated collection metadata and barcodes deepen our understanding of variation in the shape and size of otoliths from mesopelagic fishes, documents otoliths of species not previously examined, and facilitate future integrative taxonomic research (Dayrat, 2005).

Sample collection
Fishes were collected on six cruises between 2014 and 2019, five of which were in the Slope Sea off the northeast United States in the western North Atlantic Ocean (waters between the continental shelf and the Gulf Stream; (Csanady and Hamilton, 1988;Richardson et al., 2016), and a sixth cruise in Bahamian waters (Figure 1; see Table S1 for details on the location, and collection depths for each specimen).Three specimens were included from a cruise in Bahamian waters, and these specimens belong to species found throughout the Atlantic Ocean and overlap in distribution with the main study area in the Slope Sea.Sampling was performed with a variety of gear types, including a midwater trawl (as described in De Robertis et al., 2017, but without recapture nets, and we used a 0.64 cm (1/4") knotless cod-end liner and 3.5 m 2 NETS midwater-trawl doors), a 1m 2 and 10m 2 Multiple Opening Closing Net and Environmental Sensing System (MOCNESS, Wiebe et al., 1976), and a single 10m 2 MOCNESS net configured like a Tucker trawl.Fish were flashfrozen in liquid nitrogen or a -80°C freezer upon collection.

Species identification
Fish specimens were visually identified to the species level when unequivocal (and when morphological characters and/or meristic counts were reliably intact) with the aid of fish identification keys and other sources (e.g.Bigelow, 1965;Nafpaktitus et al., 1977;Carpenter, 2002).Individuals previously flash-frozen were photographed, and standard length (SL) and weight measurements were recorded before dissection.The length was measured to the nearest millimeter using a ruler, and weight was measured to the nearest 0.0001 g using a Mettler Toledo AB204-S scale.Muscle tissue was excised from fish specimens and stored at -80°C until DNA extraction.Heads were removed and frozen at -20°C for future otolith extraction.
DNA barcoding, utilizing the COI marker gene, was conducted for fish that were difficult to identify by morphological criteria alone.Genomic DNA was extracted from the tissue samples with DNEasy Extraction Kits (Qiagen, Germantown, MD, USA) according to the manufacturer's protocol.The COI gene was amplified with PCR with primers from either Folmer et al. (1994) or Ward et al. (2005) (Fish F1/R1 or Fish F2/R2) (Table S2).The PCR program consisted of 95°C for 3 minutes, 35 cycles of 95°C for 30 seconds, 48°C for 30 seconds, 72°C for 1 minute, and 72°C for 5 minutes.PCR products were visualized on an agarose gel with GelRed (Biotium, Hayward, CA, USA).Amplicons were purified with Qiaquick PCR purification kits (Qiagen, Germantown, MD, USA), quantified using a Nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), and sent for sequencing to Eurofins Genomics (https:// www.eurofinsgenomics.com).Chromatograms were aligned and consensus sequences were generated with Geneious version 9.0.5 (Biomatters, Inc).Sequences were compared to the GenBank database to identify taxonomic matches.All sequences were deposited in GenBank (accession numbers OP114527 -OP114570).

Otolith preparation
Sagittal otoliths from 171 adult fish were extracted from the otic capsule and placed in 95% ethanol and cleaned of any extra membranous material using forceps.Dry otoliths were placed in plastic vials and labeled by fish sample and left or right side when known.Sagittal otoliths of all samples were photographed with a Canon EOS Rebel T5i camera mounted to a Leica 10447436 1.6x camera tube on a Leica MZ16A stereomicroscope.Otoliths that were chipped or damaged badly were excluded from further analysis.Photographs of sagittal otoliths were taken with reflected light against a black background for the otolith library.The microscope magnification was recorded for each photograph.Photographs were taken at each microscope magnification with a ruler placed in the field of view to convert pixels to millimeters.
Images were analyzed and measured for otolith length (OL) and otolith width (OW).Measurements were taken by calibrating the magnification of each otolith photograph to millimeters and using the computer image analysis system ImageJ (https://imagej.nih.gov/ij/).Otolith length was measured by recording the "greatest distance from the anterior rostrum to the posterior edge, parallel to the sulcus" (sensu Harvey et al., 2000).Otolith width was measured as the distance between the dorsal and ventral edges, perpendicular to the sulcus (Battaglia et al., 2010).For more information on otolith terminology and morphology, see Tuset et al. (2008).
Adobe Photoshop was used for image processing, for instance to enhance visual contrast where necessary.All species from the study were included in the visual identification library.Due to the nature of otolith morphological change through ontogeny, only otoliths from adult specimens were chosen to provide the most representative visualization of unique otolith morphological features that develop over the course of the fish's life.Additional photographs of all individuals are publicly available through the WHOI repository (https://hdl.handle.net/1912/66723,DOI: 10.26025/1912/66723).

Otolith-length and width to fishlength relationships
Simple linear regressions were fitted to OL vs. fish SL, and OW vs. fish SL for six species.These relationships were examined for species that had seven or more individuals.Species that had fewer samples were not included in this analysis due to there being too few individuals to create an informative regression.Numerous studies have shown statistically insignificant differences between left and right otoliths for the purpose of identification and linear regressions (Harvey et al., 2000;Bilge, 2013;Jones and Morales, 2014;Battaglia et al., 2015;Dehghani et al., 2016;Wang et al., 2019), so the otolith side was chosen at random or one side was picked over the other if one was damaged or chipped.

Results
A total of 70 fish species, from 29 families, were identified.Forty-four specimens were DNA barcoded (Table S2).All but three of the barcode sequences were 99-100% identical to reference sequences on GenBank.Typically, within-species variation in COI sequences is within 1% (Ward et al., 2005;Ward et al., 2009), and 97-98% identity is used as a species-level threshold in COI metabarcoding taxonomic assignment for fish (e.g.Hatzenbuhler et al., 2017;Kimmerling et al., 2018;Duke and Burton, 2020).For nearly all specimens, our results were indicative of species identification.A few fish showed possible species-level matches to multiple species, and for these, the species assignments were based on the visual assessment.These included two specimens that were >99% identical to both Scopelogadus beanii and Scopelogadus mizolepis reference sequences, consistent with previous studies that have noted indistinguishable COI sequences from these nominally different species (Ward et al., 2009;Kenchington et al., 2017).The other specimens that had a high percent identity to multiple species were Borostomias mononema (similar to Neonesthes capensis), and Notoscopelus kroyeri (similar to Notoscopelus elongatus).The three specimens that did not show ≥99% similarity to any reference sequences were identified visually as follows: Ariomma lurida (closest GenBank match was Ariomma lurida with 97.5% identity), Paracaristius maderensis (closest GenBank match was Caristius macropus with 94.7% identity), and Astronesthes sp.(closest GenBank match was Astronesthes formosana with 94.9% identity) (Table S2).
Otolith length and width are positively correlated with fish SL (Figures 6, 7; Table 1).All species we examined show a significant relationship between OL and SL, and OW and SL (p-values less than 0.006).The correlation coefficients ranged from 0.52 to 0.99.Regressions using otolith length had higher r 2 values than regressions using otolith width for eight of the twelve regressions.

Discussion
Our photographic library, like other similar atlases (eg.Campana, 2004;Tuset et al., 2008;Battaglia et al., 2010;Jones and Morales, 2014), provides informative documentation of sagittal otolith shapes and sizes of mesopelagic fishes.The fish specimens in our library were identified with a combination of morphological criteria and DNA barcoding.
This extensive otolith atlas serves as a taxonomic guide with applications in biological, paleontological, and archaeological studies, particularly those exploring food web analyses.Otoliths recovered from gut contents can be a valuable ecological tool in reconstructing food web dynamics.Furthermore, mesopelagic fishes are understudied, and the role that mesopelagic species play in the biological carbon pump is recognized to be significant; however, major gaps in our knowledge of mesopelagic ecology exist, making it a priority for research (St. John et al., 2016;Martin et al., 2020;Saba et al., 2021) In addition to the otolith library, the DNA barcodes generated in this study add to existing barcode databases which are foundational for DNA metabarcoding (e.g.Bucklin et al., 2016).Metabarcoding is increasingly used in environmental DNA (eDNA), zooplankton, and gut content studies (e.g.Casey et al., 2019;Ruppert et al., 2019;Blanco-Bercial, 2020;Govindarajan et al., 2021).Most of our barcodes were matched to barcodes already in GenBank, but our specimens add to knowledge of intraspecific variation, which is critical for detecting cryptic diversity and biogeographical patterns (Gaither et al., 2016;Kenchington et al., 2017;Christiansen et al., 2018).Furthermore, the association of our barcodes with our otolith data will facilitate future integrative taxonomic work (i.e., that incorporates multiple lines of evidence, including both genetic and morphological) (Dayrat, 2005).Of the 12 otoliths presented here that have not been previously described and photographed, seven were identified to the species level with DNA barcoding, highlighting the importance of integrative taxonomic work.Additional documentation and analyses of both otoliths and DNA barcodes of mesopelagic fishes should be conducted in order to aid in our understanding of the ecology, trophic dynamics, and importance of these species.
Numerous considerations are important when using otoliths as a means of identifying species.Otoliths resist degradation and are stable when dry; though, calcium carbonate is soluble in acid, meaning that if they remain in the stomach for too long, otoliths, especially small ones, can dissolve and lose their species-specific morphology (Jobling and Breiby, 1986;Campana, 2004).Ontogenetic morphological change in otoliths is also important to consider, as shape and features vary throughout the course of development (de Carvalho et al., 2015;Jawad et al., 2018).Otoliths collected from young fishes may not yet have formed a species-specific shape, rendering fish identification via otolith morphology impossible (Campana, 2004;Deng et al., 2013).To account for this, we only extracted otoliths from adult specimens, based on the representative sizes of adults in the literature for each species.However, we did not determine ages, and many species had small sample sizes, so it is possible that a photograph presented may not be completely characteristic of the adult form of a species.Due to a lack of samples representing all age classes, we were unable to characterize ontogenetic changes in otolith morphology.This library does not include otolith morphological descriptions; however, there are studies that focus on otolith morphometrics that can be used in conjunction with this library for more precision in identifying closely related species (Tuset et al., 2003;S ̌keljo and Ferri, 2012;Jawad et al., 2018).Additionally, abnormalities in otolith shape have been found in species, due to genetic or environmental factors (Yedier et al., 2023).In one study examining otoliths of the benthopelagic fish, Atherinella brasiliensis, four of 45 sagittal otoliths were determined to be anomalous in shape (de Carvalho Lapuch et al., 2022).As some of the species presented here had small samples sizes, there is a possibility that the otolith presented is not characteristic of that species, but rather anomalous.Future work should focus on describing the surface morphology and characteristics of the otoliths presented here to aid in identifying species and detecting anomalous otoliths, especially for the otoliths that are documented here for the first time.
Our study found that otolith size (length and width) and fish standard length are significantly correlated in the species we examined.Both otolith length and otolith width led to informative, statistically significant regressions.However, the SL regressions had higher adjusted r 2 values when using the OL compared to OW for a majority of the regressions examined.Otolith length commonly shows a more robust correlation to fish length than otolith width (Dehghani et al., 2016).For species that have otoliths that are easily chipped like Sigmops elongatus, OW can still provide an informative regression and may be more practical in certain circumstances.This tactic could likely be employed in species such as Bathylagichthys greyae and Dolicholagus longirostris, as their pyriform otolith shape is similar to that of Sigmops elongatus.
While these regressions can be useful, otolith-size to fishlength relationships can vary in different populations of the same species or with varying growth rates (Campana, 2004).Otolith degradation from an extended period of time in the predator's stomach has been documented, which could make estimates of fish size inaccurate (Jobling and Breiby, 1986).Otolith-size to fish-size relationships have also been found to be less clear for the juvenile stages of fishes (Hare and Cowen, 1995).Additional samples per species would enable a better understanding of the otolith-size to fish-size relationships, especially for the species with small sample sizes for which we do not present regressions.A greater number of specimens could also help improve the accuracy of the regressions by broadening the range of otolith sizes used to fit otolith-size to fish-size regressions.Many other similar studies benefit from larger sample sizes (eg.Bilge, 2013;Battaglia et al., 2015).We also were unable to statistically analyze differences between left and right-side otoliths due to damage on some otoliths, so we included the left or the right otolith from each individual in our regressions.Future work should test for differences between otolith side for the species presented here.
Sagittal otoliths of a large diversity of ecologically important mesopelagic fishes are documented here in this library.Furthermore, an integrative taxonomic process including visual and genetic identification proved to be highly effective for this work, enabling the identification of species with otoliths not previously documented.Given the relative paucity of research on mesopelagic fishes, and otolith documentation of these species in particular, this work effectively builds on that limited knowledge base, deepening our understanding of the ecology and trophodynamics of fishes in the mesopelagic.However, there is still much to understand, and  Otolith length vs. fish standard length.

FIGURE 7
Otolith width vs. fish standard length.more detailed analyses of otoliths, including the lapilli and asterisci, morphological descriptions, as well as additional documentation of DNA barcodes of mesopelagic fishes would continue to further this knowledge.

FIGURE 1
FIGURE 1Map of fish specimen collection locations.Open circles indicate the locations of each tow used to collect specimens.

TABLE 1
Relationships of fish standard length (SL, mm) to otolith length (OL, mm) and otolith width (OW, mm).