Modern deep-water agglutinated foraminifera from IODP Expedition 323, Bering Sea: ecological and taxonomic implications

Despite the importance of the Bering Sea for subarctic oceanography and climate, relatively little is known of the foraminifera from the extensive Aleutian Basin. We report the occurrence of modern deep-water agglutinated foraminifera collected at seven sites cored during Integrated Ocean Drilling Program (IODP) Expedition 323 in the Bering Sea. Assemblages collected from core-top samples contained 32 genera and 50 species and are described and illustrated here for the first time. Commonly occurring species include typical deep-water Rhizammina, Reophax, Rhabdammina, Recurvoides and Nodulina. Assemblages from the northern sites also consist of accessory Cyclammina, Eggerelloides and Glaphyrammina, whilst those of the Bowers Ridge sites consist of other tubular genera and Martinottiella. Of the studied stations with the lowest dissolved oxygen concentrations, the potentially Bering Sea endemic Eggerelloides sp. 1 inhabits the northern slope, which has the highest primary productivity, and the potentially endemic Martinottiella sp. 3 inhabits Bowers Ridge, which has the lowest oxygen concentrations but relatively low annual productivity. Martinottiella sp. 3, with open pores on its test surface, has previously been reported in Pliocene to Recent material from Bowers Ridge. Despite relatively small sample sizes, ecological constraints may imply that the Bering Sea experienced high productivity and reduced oxygen at times since at least the Pliocene. We note the partially endemic nature of the agglutinated foraminiferal assemblages, which may at least in part be due to basin restriction, the geologically long time period of reduced oxygen, and high organic carbon flux. Our results indicate the importance of gathering further surface sample data from the Aleutian Basin.

The Bering Sea extends over a region comparable in size to the Mediterranean, yet the modern agglutinated foraminifera are still virtually unstudied. The interaction of strong currents, upwelling high nutrient water masses, sea ice and strong winds causes high surface water productivity which supports a diverse ecosystem (Stabeno et al. 1999) and an expanded oxygen minimum zone (OMZ). Relatively recent palaeoceanographic work indicates that the Bering Sea may have been characterized by high productivity and low oxygen since at least the Pliocene (Expedition 323 Scientists 2011; Kaminski et al. 2013) and, therefore, is an ideal place to study the long-term impact of severe hypoxia and high organic carbon flux on benthic organisms, in particular the less well-studied agglutinated foraminifera which are a diverse group particularly tolerant to ocean acidification due to their noncalcareous tests. Observational studies have recorded an expansion of tropical OMZs in the Pacific Ocean and Atlantic Ocean over the last 60 years, which is likely to continue with future increased atmospheric CO 2 emissions and oceanic sequestration (Stramma et al. 2008(Stramma et al. , 2010Hofmann & Schellnhuber 2009). Studies of OMZ benthic ecology are, therefore, of particular interest (Gooday & Jorissen 2012). Although there have been several studies of modern benthic foraminifera from within OMZs world-wide (e.g. Hermelin & Shimmield 1990;Sen Gupta & Machain-Castillo 1993;Kaminski et al. 1995;Kaiho 1999;Schumacher et al. 2007), there remains a lack of information from the Bering Sea.
On account of the Bering Sea's high sedimentation rate along the slope, restricted deep-water circulation, low oxygen conditions and its partial isolation from the Pacific by the Aleutian Islands volcanic arc, the Bering Sea slope sites may be a good modern analogue to the type of high sedimentation-rate deep-sea environments in the Cretaceous to Palaeogene Alpine-Carpathian and North Atlantic basins containing rapidly deposited orogenic-derived sediments called flysch. Under such conditions agglutinated foraminifera are an extremely important component of the benthic fauna, and fossil assemblages from the flysch basins are often comprised exclusively of agglutinated benthic foraminifera (e.g. Gradstein & Berggren 1981;Kender et al. 2005;Waskowska-Oliwa 2008;Setoyama et al. 2011).
In this study we fully document the agglutinated foraminifera in the deep (>800 m water depth) Bering Sea, in order to assess the degree of endemism in this restricted basin and to assess the possible ecological controls on agglutinated foraminiferal abundance.

Bering Sea oceanography
Approximately half of the modern Bering Sea comprises a shallow (0 -200 m) neritic environment, the remainder a vast plain c. 4 km deep broken by the Bowers and Shirshov ridges (Fig. 1). The northern continental shelf is covered seasonally by sea ice, with little ice presently being formed over the deep SW areas. The Bering Sea is one of the most highly biologically productive regions in the world, exporting some 687 000 tons of carbon per year (Sambrotto et al. 1984;Stabeno et al. 1999). 'Old' deep water, characterized by low oxygen concentrations, high nutrients (e.g. phosphate and nitrate) and high dissolved CO 2 , flows into the Bering Sea at depth from the North Pacific. It cycles counter-clockwise around the Bering Sea Basin, upwelling particularly over the continental shelf feeding the so-called 'Green Belt' (Springer et al. 1996). As large fluxes of organic carbon make their way to the seafloor in parts of the Bering Sea, particularly along the slope and over the shelf in spring (Fig. 2), intense oxygen demand expands the OMZ (Fig. 3), which impacts the composition of benthic foraminiferal communities and the chemistry of ocean water (Expedition 323 Scientists 2011). Significant exchange of Pacific deep water occurs through the Kamchatka Strait (maximum depth of 4420 m), and of low oxygen intermediate water through the Commander-Near Strait at 2000 m (Coachman et al. 1999). Very small amounts of bottom water are formed in the Bering Sea today (Warner & Roden 1995) and, as a result, the deep Bering Sea has an expanded OMZ in comparison with the northern Pacific.

Previous studies of benthic foraminifera
Modern benthic foraminifera have been reported from Rose Bengalstained core-top samples collected on the Bering Sea shelf at water depths less than 200 m (Anderson 1963). This study reported the occurrence of agglutinated foraminifera, which sometimes dominate the foraminiferal assemblages in the deeper shelf basins. Anderson (1963) reported that the proportion of agglutinated foraminifera may reach 90% of the total foraminiferal fauna on the central Bering Sea shelf. However, the modern deep-water agglutinated foraminifera from the deeper Aleutian Basin, within and below the OMZ, have yet to be documented. Khusid et al. (2006) studied the benthic foraminifera from a 660 cm long core collected at 3060 m depth on Bowers Ridge. In this core, the agglutinated foraminifera were found mainly in the core top and to a depth of 20 cm. The late Holocene agglutinated foraminifera comprised 83 -99% of the fauna at this location, and consisted of Rhabdammina, Hormosina, Ammolagena, Cribrostomoides and Karreriella. However, neither Anderson (1963) nor Khusid et al. (2006) provided any descriptions or illustrations of the agglutinated foraminifera.
The agglutinated foraminifera from the North Pacific and Siberian Arctic have been more intensively studied than the fauna from the Bering Sea. In this study we made use of the taxonomic monographs of Cushman (1910Cushman ( , 1921, Saidova (1975), Matoba & Fukusawa (1992) and Zheng & Fu (2001) on North Pacific foraminifera; the work of Vázquez Riveiros & Patterson (2007) on the foraminifera from the North Pacific Fjords; as well as studies on Arctic foraminifera by Cushman (1944), Wollenburg (1992Wollenburg ( , 1995 and Lukina (2001). The distribution of foraminifera along the North Pacific continental margins was compiled by Culver & Buzas (1985. Szarek (2001, unpublished PhD thesis, 'Biodiversity and biogeography of recent benthic foraminiferal assemblages in the south-western South China Sea (Sunda Shelf )', Christian-Albrechts University, Kiel) provides an excellent taxonomic section and useful distributional data for Bering Sea fauna Reophax bradyi and R. oviculus in the South China Sea. The current study aims to bridge a geographical gap in our knowledge of the distribution of North Pacific-Arctic agglutinated foraminifera, by providing descriptions of species recovered from the IODP Expedition 323 coring sites.

Methods and materials
Samples were collected and prepared on board the JOIDES Resolution drillship from each site (U1339-45) during IODP Expedition 323, Bering Sea, in June/July 2009. Samples (quantities of sediment) were collected from the first cores recovered at each new hole (typically several holes were cored at each IODP site, within a distance of <20 m), by emptying the core barrel of unconsolidated core-top sediment into a cleaned receptacle. Samples ranged widely in size, but were generally >100 g in weight. Sediment composition varied between sites, but was varyingly dominated by diatoms and fine clays and silts with only rare coarser sand-sized particles. Two samples were then immediately stained in a Rose Bengal solution for >24 h to ascertain the living component. Samples were carefully washed over a >63 µm mesh sieve with deionized water. Sample residues were oven dried at <40°C and agglutinated foraminifera picked out and placed in showing IODP Expedition 323 sites reported in this study, and the number of specimens (n) and number of taxa (S) in major agglutinated foraminiferal assemblage components from core-top samples. Also indicated are the major surface (red arrows) and deep ( purple arrows) ocean currents. Grey line corresponds to the section shown in Figure 3. cardboard reference slides. Specimens were imaged using a JSM-5900LV SEM at King Fahd University of Petroleum and Minerals in Dhahran, and a LEO 535VP SEM at the British Geological Survey in Keyworth. The proportion of faunal groups shown in Figure 1 was calculated for each site by combining the faunal counts of all samples from that site. Correspondence Analysis (CA), a reciprocal averaging algorithm, was carried out (using the software of Hammer et al. 2005) on the dataset to statistically ascertain the relationships between samples, species and selected environmental parameters (Figs 4 and 5), as described in Hammer & Harper (2006). CA in Figure 4 was carried out on a modified dataset, in order to incorporate environmental information with widely varying numerical values compared to species counts. Species counts were summed for each site and Site U1340 was removed because of low counts. Species that had an occurrence of <10 in the total dataset were removed to avoid artificially placing emphasis on low counts. The amount of dissolved bottom water oxygen (estimated from World Ocean Circulation Experiment data, Fig. 3) and spring primary productivity (estimated from satellite chlorophyll-a data, Fig. 2) was recorded for each site (Table 1) and included in CA. All species counts and environmental parameters were subsequently normalized, and 2 added to each value to lift all numerical values above 0 (a requirement for CA). CA in Figure 5 was carried out on an unmodified dataset of all samples separately, in order to ascertain the degree of similarity between samples of different sites.

Results
A total of 524 specimens of agglutinated benthic foraminifera were recorded, identified and described from 19 core-top samples taken at all seven IODP Expedition 323 sites (Figs 1 and 4; Table 2). The samples were dominated by calcareous benthic foraminifera, which were reported by Expedition 323 Scientists (2011) and reproduced in Table 3. Preservation of agglutinated taxa is generally excellent (see Figs 6 -10), which is in marked contrast to the agglutinatedpoor Pliocene-Recent palaeontological assemblages recovered  from down-core samples (Expedition 323 Scientists 2011; Kaminski et al. 2013), indicating post-mortem dissolution of the organic cement likely occurred. Abundance is >70 specimens at most sites, apart from Sites U1339 and U1340 where abundances are low due to the small volume of core-top samples collected. CA (Fig. 5) indicates that there is generally greater similarity between samples from one site than between samples from different sites, as the majority of samples cluster near those of the same site. Samples U1341B and U1343C plot further away, which can be explained by their particularly low abundances (see Table 2). The generally lowdiversity agglutinated assemblages (9 -35 taxa per site) are predominantly composed of tubular suspension-feeders (e.g. Rhabdammina, Rhizammina and Bathysiphon), epifaunal lituolids (e.g. Recurvoides, Cyclammina), opportunistic infauna (e.g. Reophax and Hormosinella) and infauna (e.g. Eggerelloides and Martinottiella) in varying proportions (Fig. 1). CA indicates that some species are more prevalent at certain sites (i.e. plot in close proximity on Fig. 4); that Sites U1339, U1342 and U1345 are most associated with high chlorophyll-a concentrations (a proxy for primary productivity); and that Sites U1342 and U1345 are most associated with low bottom water dissolved oxygen. Two samples were stained with Rose Bengal (at Sites U1342 and U1345). These samples contained a small proportion of living individuals (Table 2), confirming that the IODP cores recovered samples of modern/sub-modern age.

Endemism
Although a sizeable proportion of Bering Sea agglutinated foraminifera have been recorded in the Pacific Ocean (Jones 1994), there are several species in the core-top samples (this study) and in the Pliocene (Kaminski et al. 2013) that appear to be endemic (e.g. Eggerelloides sp. 1; Glaphyrammina cf. americana; Martinottiella sp. 1; Martinottiella sp. 2; Martinottiella sp. 3; Karreriella sp. 1; Bathysiphon sp., Hormosinelloides sp.) and confirm the semi-isolated nature of the microfauna in the Bering Sea. Of the 131 agglutinated species recorded by Culver & Buzas (1985) from the North Pacific Margin (at 138 localities) only 13 are present in our samples. This low number of species in common suggests that many taxa present along the Alaskan margin are excluded from our study locations in the Bering Sea. In our current study of the agglutinated foraminifera, 22% of the species are left in open nomenclature and do not yet appear to have been described. In the Pleistocene calcareous benthic assemblage studied by Setoyama & Kaminski (2015) at Site 1341, 23% of the taxa were identified tentatively or left in open nomenclature. In contrast Culver & Buzas (1985) reported only a few species in open nomenclature. Geographical barriers for faunal interchange between the Bering Sea and North Pacific include the restricted Aleutian passes (Fig. 1), although the western passes are deep (>4 km) and the majority of Bering Sea benthic species recorded in our study are cosmopolitan. It is therefore possible that environmental conditions in the isolated Bering Sea, such as high productivity and reduced bottom water oxygen, have allowed for the adaptation of certain new species or varieties.
Considering the long stratigraphic ranges of the majority of benthic foraminifera (e.g. Kaminski & Gradstein 2005;Holbourn et al. 2013), and their relatively slow genetic evolution compared with planktonics (Pawlowski et al. 1997;Gooday & Jorissen 2012), the occurrence of endemic species is consistent with a Bering Sea that may have been isolated for a considerable length of time. This is not unique for semi-isolated deep-water basins, and one such example is the high-latitude Norwegian Sea during the Eocene, when it was separated from the North Atlantic by the Greenland-Scotland Ridge. The deep-water agglutinated foraminiferal assemblages that developed during the Eocene and Oligocene in this basin contain a number of endemic species that have not been found in the northern Atlantic (Gradstein & Kaminski 1989;Kaminski & Gradstein 2005). The Oligocene deep-water agglutinated foraminiferal assemblage at Site 985A on the Iceland Plateau contains 27% endemic species (Kaminski & Austin 1999). Agglutinated Fig. 4. Correspondence Analysis (CA) for dataset (including chlorophyll and inverse oxygen estimates; stars), showing species (circles) and sample (diamonds) scores for axis 1 against axis 2. Bottom water oxygen values were inverted, so that proximal species and samples exhibit low oxygen. Only species with >10 specimens are included (see 'Methods' for further details of data analysis). Mbsl, metres below sea-level.    Table 3. Semi-quantitative abundances of calcareous foraminifera in core-top samples from IODP Expedition 323 sites  foraminifera were studied by Matoba & Fukusawa (1992) in the semi-enclosed Sea of Japan, and 26% of the reported taxa were not identified at species level.

Ecological implications
Benthic foraminifera are impacted by several ecological forcing factors, which include organic carbon type and flux, bottom water oxygenation, bottom water sediment heterogeneity and hydrodynamics, temperature and corrosiveness (Jorissen et al. 1995Levin et al. 2001). Bering Sea deep-water (>800 m water depth) ecology is primarily affected by two of these ecological forcing factors: dissolved oxygen concentrations and organic carbon flux (Expedition 323 Scientists 2011), which is related to primary productivity (quantity, type and duration) and remineralization of particulate organic carbon as it is transported to depth (Arndt et al. 2013). Other factors that might affect benthic assemblages in the Bering Sea are the high sedimentation rates along the slope (Table 1). Oxygen levels in the deep Bering Sea are very low (c. 2.0 -0.2 ml l -1 ), and primary productivity highly variable (Figs 2 and 3), which can be expected to impact upon assemblage composition (e.g. Jorissen et al. 1995Jorissen et al. , 2007Kaminski et al. 1995;Sun et al. 2006 (Tables 2 and 3) are somewhat typical for low oxygen and high productivity environments, as described from the Santa Catalina Basin (Kaminski et al. 1995), Santa Barbara Basin (Moffitt et al. 2014), the Arabian Sea Schumacher et al. 2007) and OMZs elsewhere (see Sen Gupta & Machain-Castillo 1993). Assemblages within the core of the OMZ are typically dominated by calcareous infaunal taxa that exhibit elongated tapered tests, which are predominantly Bulimina, Brizalina and Bolivina in the Bering Sea (Table 3). The relatively less-specialized agglutinated foraminifera usually occur in higher abundances above and below the core of the OMZ (Kaminski et al. 1995;Schumacher et al. 2007). Our samples are dominated by calcareous foraminifera, which is consistent with high productivity and low oxygen settings in the Okhotsk Sea (Bubenshchikova et al. 2008). The most commonly occurring cosmopolitan agglutinated species in our material ( Fig. 4) are wide-ranging and described from diverse environments. Rhizammina algaeformis and Nodulina dentaliniformis are well-known cosmopolitan species ranging from neritic to abyssal depths. Nodulina dentaliniformis has been recorded from relatively shallow water in the Arctic (Lukina 2001) and Antarctic (Majewski 2005) where it may be tolerant of changes in salinity. Reophax excentricus and Reophax bilocularis are cosmopolitan open ocean species, and were recorded as part of assemblages within the OMZ of the Santa Catalina Basin (Kaminski et al. 1995). Reophax bilocularis was also recorded in high proportions along the slope beneath the Arabian Sea OMZ associated with high sedimentation rates (Hermelin & Shimmield 1990) and occurs in relatively high abundances at Site U1343 along the slope, where sedimentation rates are higher (Table 1). Cyclammina compressa is a less well-known bathyal to abyssal species originally described from the Philippines and also recorded offshore North Carolina (Gooday et al. 2001), and has close morphological affinities with the cosmopolitan and wide-ranging C. cancellata (see Jones 1994).
Although we obtained no faunal density data (because of the limited availability of equipment onboard the JOIDES Resolution), with such strong ecological gradients between sites it is possible to speculate on the ecology of some of the more abundant key species (Fig. 4). We caution that our estimates of modern ecological parameters (i.e. dissolved bottom water oxygen and primary productivity, Figs 2 and 3; Table 1) at each site are only approximations, as there were no in situ water mass measurements made during Expedition 323. The line section of Figure 3 does, however, pass in close proximity to all sites apart from U1339 (see line in Fig. 1). It should also be considered that our benthic faunas probably represent several decades at least, and that the primary productivity proxy chlorophyll-a, with its own proxy uncertainties (see Sun et al. 2006), represents the years 1998 -2003. In addition, primary productivity has only an indirect impact on benthic faunas, as much of the organic carbon is remineralized on its way to the seafloor (see Arndt et al. 2013). However, we consider that the largest changes in these parameters between sites will be semiquantitatively resolved by our estimates. Glaphyrammina cf. americana is restricted to Site U1344 (central slope), where the highest bottom water oxygen, highest sedimentation rate and deepest water depth is recorded ( Table 1). The deep-water setting probably experiences lower organic carbon fluxes compared to the other slope sites (as organic carbon is remineralized in the water column) and so this species may be adapted to more oligotrophic environments with relatively elevated oxygen levels. At Site U1345 (northern slope), where there is the highest year-round chlorophylla concentration (and assumed organic carbon flux) and low bottom water oxygen (Figs 2 and 3), the endemic species Eggerelloides sp. 1 occurs in relative high abundance ( Fig. 4; Table 2) and therefore this species may be adapted to high organic carbon flux in low oxygen settings. The morphologically similar species from the North Pacific, Eggerelloides advenum, has been associated with intense eutrophication in Osaka Bay (Tsujimoto et al. 2006). At Sites U1342 and U1340 (Bowers Ridge), the only other sites situated in the core of the OMZ (Fig. 1), the endemic species Martinottiella sp. 3 is observed ( Fig. 4; Table 2). Due to its distribution, we speculate that this species may be adapted to low oxygen environments (e.g. below c. 1 ml l -1 ; Table 1), but not high organic carbon flux (as it does not occur along the northern slope but in the more oligotrophic south-central Bering Sea). Kaminski et al. (2013) were the first to observe the highly perforate tests of Karreriella and Martinottiella from the Bering Sea (see Fig. 10:1-10:5) and suggested this feature may have been an adaptation to severely hypoxic conditions, which is supported by the species modern distribution recorded here. Martinottiella sp. 3 is larger and more robust than the otherwise morphologically similar species Martinottiella sp. 1, recorded from the Pliocene of Bowers Ridge (Kaminski et al. 2013), which may have been its evolutionary ancestor. The modern-day distribution of the morphologically similar (although possibly lacking perforations) Martinottiella sp. (cf. M. communis) is world-wide, including within the OMZ of the Santa Catalina California Borderland basin (Kaminski et al. 1995), highly productive areas of the South China Sea (Jian et al. 1999), the OMZ of the equatorial East Pacific (Culver & Buzas 1987) and East New Zealand, South Pacific (Hayward et al. 2001). It is sometimes associated with high organic carbon flux and low oxygen settings, but its modern-day ecology is yet to be fully resolved. Martinottiella spp. is dissolution resistant and survives taphonomic loss, so this taxon can be used as a palaeoenvironmental indicator. Echols (1973), Expedition 323 Scientists (2011) and Kaminski et al. (2013) recorded Martinottiella sp. (cf. M. communis) from several locations in the Bering Sea (including Bowers Ridge) from the Pliocene to Recent. Due to the new distributional data reported here, we suggest that the significance of the Pliocene occurrences of Martinottiella may be an indication of low oxygen conditions at times since at least the Pliocene. Our study highlights, however, the importance of obtaining more bottom water and surface sample material from the Aleutian Basin for further study.
We cannot be sure of Eggerelloides sp. 1 and Martinottiella sp. 3 living depth preferences within the sediment. However, many authors have attempted to ascertain palaeoecology from ancient sediments by placing agglutinated foraminifera into groups on the basis of their morphology; these 'morphogroups' are thought to be indicative of their ecology (e.g. Nagy 1992; van den Akker et al. 2000; Kender et al. 2008a, b;Nagy et al. 2009;Nikitenko et al. 2013). These are based on studies of modern foraminifera and bottom water properties, such as living depth, productivity and ecological disturbance (Jones & Charnock 1985;Kaminski et al. 1995). Both Eggerelloides and Marttinotiella are elongated and tapered in shape and would be assigned to the 'morphogroup 4b' of Kaminski & Gradstein (2005), a group regarded as infaunal and tolerant of low oxygen conditions. This approach to reconstructing palaeoenvironments using morphogroups should be taken with caution and may be an oversimplification (e.g. Sen Gupta & Machain-Castillo 1993;Jorissen et al. 2007), particularly as Eggerelloides and Martinottiella do not co-occur in the same samples (Table 2). However, the distribution of this group within the Bering Sea OMZ sites (U1340, U1342 and U1345; see distribution in Fig. 1) supports the interpretation of an infaunal living habit.

Conclusions
We document the occurrence of 50 modern agglutinated foraminiferal taxa at IODP Expedition 323 sites in the Bering Sea, and provide the first descriptions and illustrations. The 19 core-top samples at seven sites, U1339, U1340, U1341, U1342, U1343, U1344 and U1345, contain abundant agglutinated foraminifera in varying proportions, and calcareous benthic foraminifera previously reported (Expedition 323 Scientists 2011), many of which are typical for reduced oxygen and high productivity environments. The agglutinated foraminifera consist of several abundant and ecologically wide-ranging cosmopolitan taxa, and also a number of taxa (e.g. Glaphyrammina, Martinottiella, Eggerelloides, Bathysiphon, Hormosinelloides and Karreriella) that differ in morphology from their counterparts and are here left in open nomenclature. The agglutinated foraminiferal fauna of the deep Bering Sea is thus partially endemic, suggesting that geographical restriction, combined with high productivity and low oxygen environmental conditions, may have persisted within the Bering Sea for a geologically extended period of time, considering the low evolutionary rate of the group. Three of the more abundant endemic species may be ecologically restricted. Glaphyrammina cf. americana occurs largely at the slope Site U1344, which is the deepest (with likely low organic carbon flux), highest bottom water oxygen and highest sedimentation rate site sampled. Eggerelloides sp. 1 occurs in high abundance at the northern slope Site U1345, which experiences the highest seasonal productivity (and possibly organic carbon flux) of the seven sites. Martinottiella sp. 3 is restricted to the OMZ of Bowers Ridge, where there is currently relatively low annual productivity, suggesting this species may be a good indicator for reduced oxygen conditions but not elevated organic carbon flux. The occurrence of Martinottiella throughout the Bering Sea sporadically over the past c. 4 Ma (Echols 1973;Expedition 323 Scientists 2011;Kaminski et al. 2013) therefore adds evidence to the hypothesis that the Bering Sea has had a pronounced OMZ since at least the Pliocene. However, due to the relatively low number of samples, our study highlights the importance of collecting more data from the currently undersampled deep Aleutian Basin. In addition, two species of agglutinated foraminifera (Karreriella sp. 1 and Martinottiella sp. 3) were found to contain micro-pores that are open at the test surface, a morphological feature that is possibly indicative of hypoxia (Kaminski et al. 2013).

Systematic palaeontology
In this section species are arranged in taxonomic order according to the classification of Kaminski (2014). Descriptions and comments are provided and important references for understanding each species morphology and distribution are cited. For taxonomic determinations the monographs of Cushman (1910Cushman ( , 1921Cushman ( , 1944, Cushman & McCulloch (1939), Pfleger (1952), Vilks (1969), Saidova (1975), Wollenberg (1992, 1995, Jones (1994), Kaminski & Gradstein (2005) and Vázquez Riveiros & Patterson (2007)  Remarks. A single specimen was found in Hole U1342A. It is triangular in outline, with a coarsely agglutinated wall. It corresponds well with specimens illustrated by Brady (1884, pl. 20). Gooday & Smart (2000) showed that the species has a two-layered wall and agglutinates juvenile planktonic foraminiferal tests to construct its outer layer. Our specimen from Hole U1342B only uses mineral grains. Remarks. We used this name to describe fragments of straight tubes, not precisely determined. A distinctive specimen recovered in Hole U1343A agglutinates grains of very unequal size.

Rhabdammina 'smooth spicules'
Remarks. A single specimen was found in Hole U1340A. This species has a very thick wall with sponge spicules embedded in a matrix of finer agglutinated particles. The spicules are aligned at right angles to the long axis of the test.

Bathysiphon Sars, 1872
Bathysiphon sp. 'coarse' (Fig. 6:9) Remarks. A large specimen was recovered at Site U1341. The test is over 3 mm in length, arched and the early part of the test is tapered. The wall is coarsely agglutinated and constructed of grains of uneven dimensions. It most closely resembles Bathysiphon rufus de Folin, but the species in the NE Atlantic can reach a much larger size, up to 14 mm in length (Gooday 1988a Nothia sp. 'large spicules' (Fig. 6:3)

Remarks.
A thin-walled tube constructed of fine mineral grains with agglutinated sponge spicules attached to the surface of its test. The sponge spicules project at a 45°angle away from the test. Only broken fragments were found.
?Nothia sp. 'diatoms' (Fig. 6:4) Remarks. We used this designation for fragments of a flattened thin-walled tube that contains a high proportion of diatoms incorporated into the wall. Specimens from Hole U1345B use a mixture of mineral grains and centric diatoms, while the specimens from Hole U1344D use mostly diatoms to construct their test. Fossil species of Nothia (e.g. Nothia excelsa Grzybowski) are commonly flattened, implying that their wall was flexible.
Genus Psammosiphonella Avnemelich, 1952 Psammosiphonella discreta (Brady, 1881) ( Fig. 6:11 Description. Test small, round in cross-section, occasionally branching dichotomously. Test wall is thin and comprised mostly of fine sand grains with occasional short fragments of sponge spicules or diatom frustules loosely attached to the surface of the test. Remarks. Unlike the type specimens housed at the NHM, London, the specimens from the Bering Sea do not attach any planktonic foraminifera to the exterior of their test. Instead, the species attaches occasional sponge spicules or centric diatoms to its test surface. Our specimens more closely resemble those figured by Schröder-Adams et al. (1990) from the Axel Heiberg Shelf, Arctic Ocean.
Lagenammina sp. Remarks. A specimen from Hole U1341A is comprised mostly of biogenic siliceous particles, including centric diatoms. The species of Lagenammina made of quartz grains (e.g. L. atlantica and L. arenulata), which are so common in the Arctic, have not been found in our material from the Bering Sea.

Description.
Test free or attached to a single large sand grain, varying in size, consisting of a single spherical chamber. Wall agglutinated, of a single layer of coarse sand grains, cemented together in a matrix of finer agglutinated particles, without any inner organic layer. Small pores between the loosely agglutinated sand grains serve as apertures.
Remarks. The species P. fusca uses a combination of larger and smaller agglutinated particles, sometimes selecting a single larger grain that may serve as an attachment surface. The large agglutinated grains may be angular or sub-rounded. Space between grains is filled in by a matrix of much finer agglutinated particles. The size of the recovered specimens is variable. Smaller specimens build their test out of only a few (<10) grains. Already in 1948, Cushman (1948, p. 9) noted 'All the Arctic specimens I have seen are small with rather large, angular quartz grains, and quite different from some of the specimens that are figured and referred to this species'. We observed this species at Sites U1339, U1340, U1343, and U1344.
Large agglutinated 'plate' Remarks. In Hole U1345C, platy fragments of a larger agglutinated foraminifera were found. These fragments consist of equidimensional sand grains that include some mafic grains, cemented in a groundmass of much finer quartz grains. The grains are orientated in such a manner that the flat surfaces of the grains face the interior of the chamber. The agglutinated plates show little or no curvature, which means the whole specimen must have been quite large. These specimens are similar to the taxon that was named 'Astrorhizinella' by Saidova (1970). Loeblich & Tappan (1987) included this taxon in their list of genera with uncertain status, with the remark that it could be a foraminifera, xenophyophorean or inorganic. Description. Test free, very large (>2 mm), tubular or cylindrical, open only at one end, proloculus with approximately the same diameter as the tubular chamber. The fragments of the tubular chamber are several mm in length and may display constrictions or taper slightly toward the aperture. Wall is several grains thick, of fine sand with occasional larger sand grains, occasional diatom frustules and dark mafic grains, with little cement, grey in colour. The test appears to be comprised of two layers: the interior layer is much thinner, with smaller sand grains lining the central cavity, which has a yellowish-brown inner organic lining. Aperture terminal, constricted, round in outline, may be partially obstructed by agglutinated particles.
Remarks. Three slides of type specimens of Hyperammina subnodosa are preserved in the Carpenter Collection in the NHM, London, all from Valorous Station 2, 100 fathoms. The lectotype, designated by Kaminski & Cetean (2011), is the specimen illustrated by Brady (1884) in plate 23, figure 11, and is preserved in slide BMNH 1886.4.16.94. These specimens are quite large (>1 mm) for an agglutinated foraminifera. The diameter of the tubular chamber is variable in our specimens, giving the impression of pseudochambers. These pseudochambers sometimes taper toward the aperture and may be several mm in length. However, the wall of the tubular chamber is continuous from one pseudochamber to the next; therefore, these are not true chambers. The interior layer contains smaller agglutinated grains than the outer layer, giving the interior a smooth appearance. This is the type species of the genus Archimerismus Loeblich & Tappan, 1984, which differs from Hyperammina in the partial subdivision of the test to form pseudochambers. The species is found at Sites U1344 and U1345. Gooday et al. (2005) noted mass occurrences of the species at deeper stations in the outer reaches of the West Spitsbergen fjords.
Genus Hormosinella Stschedrina, 1969 Hormosinella distans (Brady, 1881) ( Fig. 7:4 Description. Proloculus round, followed by ovoid pseudochambers connected by thin stolons. Wall thin. Chambers taper toward the aperture. Remarks. The subspecies of H. distans were discussed by Kaminski & Gradstein (2005). The type specimens are from Challenger Station 300, (33°42' S, 78°18' W), north of Juan Fernández Island, South Pacific, 1375 fathoms. A lectotype was designated by Kaminski & Gradstein (2005) and is preserved in the Brady Collection in Slide BMNH ZF 2271. Cushman (1910) recorded the species from two Albatross stations in the North Pacific and from the Bering Sea at 1771 fathoms. Fragmentary coarsely agglutinated specimens were recovered at Site U1341, while specimens with as many as five chambers were found at Site U1344.
Genus Reophanus Saidova, 1970 Reophanus oviculus (Brady, 1879) ( Fig. 7:6a- Description. Test large, exceeding 4 mm in length, uniserial, rectilinear, unilocular in appearance because the elongated ovate chambers are separated by their respective necks, each new chamber attaching to the upper margin of the previous apertural lip, so that the test is fragile and tends to break into individual chambers. Wall finely agglutinated with several layers of very fine quartz grains, with yellowish-brownish cement which is more prominent on the necks of the chambers, without an inner organic lining. Aperture rounded, terminal on a distinct neck, with somewhat flared lip. Remarks. The type specimens of Hormosina ovicula are preserved in the Brady Collection in the NHM, London. The lectotype is the specimen illustrated by Brady (1884) in plate 39, figure 7 (Kaminski & Cetean 2011). This specimen was designated the 'holotype' by Loeblich & Tappan (1987) and is preserved in slide BMNH ZF1588. The specimen is from Challenger Station 241, North Pacific at 2300 fathoms. It is the type species of the genus Reophanus Saidova, 1970. This species is common in Hole U1341A and specimens consist of up to five chambers. The chambers (especially the proloculus) in our specimens are slightly more elongated than Brady's specimens from the North Pacific, but otherwise they conform very well to the original description of the species.
Description. Test comprised of a meandering series of uniserial pseudochambers. Pseudochambers are oval, elongated in the direction of growth and are connected by wide stolons. Wall thin, medium to coarse. Aperture wide, terminal.
Remarks. This species was originally described from the Palaeogene of the Polish Carpathians (Grzybowski 1898). The lectotype was designated by Kaminski & Geroch (1993). It differs from Subreophax aduncus (Brady) in possessing a more coarsely agglutinated wall and chambers that are elongated in the direction of growth. A single specimen was found in Hole U1341A.

Genus Tolypammina Rhumbler, 1895
Tolypammina vagans (Brady, 1879) ( Fig. 7: Description. Test attached, tubular, of constant diameter, streptospirally coiled, finely agglutinated with a thin wall. Remarks. Brady (1879) originally reported this species from the South Atlantic and from the North Pacific. Brady (1884) illustrated specimens that grew free or attached to shell fragments. Cushman (1921) reported that the species often attaches itself to other agglutinated foraminifera. Schröder (1986) reported that the species grows attached at bathyal depths in the North Atlantic, but is freeliving at abyssal depths. A large specimen was recovered in Hole U1341A. Our specimen was probably free-living and most closely resembles the specimens illustrated by Cushman (1921, pl. 4) from the Philippine Seas.
Remarks. The specimens from Site U1344 are consistently made up of only few chambers and the wall is made of mineral grains as well as fragments of biosiliceous particles, such as sponge spicules and radiolarians (Fig. 7:10b). The type specimens of H. guttifer from Challenger Station 323 (South Atlantic, east of Buenos Aires at 1900 fathoms) are larger and consist of 8 or more chambers (Kaminski & Cetean 2011, pl. 2, figs 10 -12). The chambers of the type specimens in the Brady Collection are slightly more pyriform than in our specimens, but they share the habit of incorporating small biogenic particles, such as radiolarians and sponge spicules, into their wall. Our specimens more closely resemble the Arctic variety of this species, which is smaller and has fewer chambers. The specimen illustrated by Vilks (1969)  Description. Test uniserial and rectilinear, similar to Reophax but with a straighter axis, more symmetrical, regular, with gradually enlarging chambers and nearly horizontal sutures. Wall coarsely agglutinated of a single layer of quartz grains. Aperture rounded, at the end of a short tubular neck.
Remarks. Brönnimann & Whittaker (1980) selected a lectotype for the type species from the Brady Collection. The type locality is Challenger Station 300, north of Juan Fernández (1375 fathoms). This six-chambered specimen (labelled Hormosina dentaliniformis) is housed in slide BMNH 3990. Anderson (1963) listed the species in a single sample from the inner Bering Sea shelf. Our specimens from Hole U1341A possess up to five chambers and closely resemble those in the Brady Collection.
Description. Test free, uniserial, comprised of two chambers, with the second one much larger than the first. Wall very coarsely agglutinated, comprised of quartz grains with an admixture of small planktonic and benthic foraminiferal tests and siliceous sponge spicules that extend outward from the test. Aperture at the end of a tapering neck.
Remarks. Differs from Reophax bilocularis in possessing small planktonic foraminifera incorporated into its wall and its more robust neck. Anderson (1963) listed the species in a single sample from the outer Bering Sea shelf. The large specimen from Hole U1342D is the typical form.
Reophax bilocularis Flint, 1899 (Fig. 8:3) 1899 Reophax bilocularis Flint: 273, p. 17, fig. 2. 1920 Reophax bilocularis Flint;Cushman: 10, pl. 3, figs 3, 4. Description. Test free, uniserial, comprised of two chambers, with the second one much larger than the first. Wall very coarsely agglutinated, comprised of angular and rounded quartz grains with an admixture of some dark minerals, cemented with a matrix of much finer agglutinated grains. Aperture on a produced neck that consists of much finer agglutinated grains than the chamber wall.
Remarks. Several good specimens were found in Holes U1343A and U1344B. In the North Atlantic, Reophax bilocularis has a habit of picking up small planktonic foraminiferal tests in addition to mineral grains. This feature is not observed in the specimens from the Bering Sea.
Remarks. Reophax excentricus Cushman has its type locality from the stomachs of holothurians dredged at Albatross station D3603 in the Bering Sea at 1773 fathoms (Cushman 1910). This species is relatively common in the Pacific Oceanit has also been reported off Oregon and the Alaskan Peninsula (Culver & Buzas 1985) and characterizes depths of 200 -2000 m off the Pacific coast of Central America (Culver & Buzas 1987). It is regarded to be an opportunistic species (Kaminski et al. 1988).
Description. Test robust, straight or curved, with 3 to 5 rapidly enlarging chambers. Chambers are globular and only slightly envelop preceding chambers. Wall coarse, comprised of a single layer of large sand grains in a matrix of finer grains, with organic cement. Aperture a round opening, situated on a low apertural shoulder, but without a neck.
Remarks. Kaminski & Gradstein (2005) designated a lectotype from the Carpenter Collection in the NHM, London. The type locality is Porcupine Station 31 in the North Atlantic, 1360 fathoms water depth. Cushman (1921) recorded it from 13 Albatross stations in the Philippine Sea between 208 and 1560 fathoms, but noted that these specimens are smaller than Atlantic specimens and 'are not  Description. Test free, broad and flattened, thin and fragile. Early portion planispirally enrolled and partially to completely evolute, sutures poorly visible, later portion uncoiled and may have a few rectilinear chambers with horizontal depressed sutures. Wall coarsely agglutinated, with larger grains in a groundmass of smaller grains. Larger pointed quartz grains project out from the periphery in the coiled portion of the test. Aperture a terminal elongate narrow opening, extending across nearly the breadth of the chamber, not produced.
Remarks. Loeblich & Tappan (1984) remarked that sutures in Glaphyrammina may simply represent external indication of periodic growth, as the interior is hollow and undivided by septa, possibly with organic partitions. The holotype of Ammobaculites americanus Cushman, 1910 is from Albatross Station 3419, collected at 772 fathoms off the west coast of Mexico. Our specimens differ from Cushman's types in possessing a well-developed uniserial portion and a very coarsely agglutinated test with the odd larger mafic grain embedded in the wall, making it difficult to observe the sutures. This species is common in Hole U1344B.
Genus Cribrostomoides Cushman, 1910 Cribrostomoides subglobosus (Cushman, 1910) (not figured) reducing to three chambers per whorl. Chambers are somewhat inflated, sutures depressed. Wall is made of fine quartz with occasional larger grains, with orange-brown organic cement. The initial part of the test has a more intense colour owing to the abundance of organic cement. Aperture a high interiomarginal arch, umbilical in position, surrounded by a thin lip.
Remarks. We place this species into Eggerelloides based on its high, loop-shaped aperture, though its internal structure (or toothplate) was not observed. Rhumblerella differs in possessing a low interiomarginal aperture. The North Pacific form of Eggerelloides advena (Cushman), as depicted by Tsujimoto et al. (2006) and by Vázquez Riveiros & Patterson (2007), differs in possessing a slender test and more well-developed triserial part. In our samples, the species is most abundant at Site U1345.
Genus Cyclammina Brady, 1879 Cyclammina compressa Cushman, 1917 (Fig. 9:9) 1917 Description. Test small for the genus, coiling planispiral, involute, with 12 -13 chambers in the final whorl. Sutures radial, nearly straight, depressed slightly. Umbilicus flush or only depressed slightly. Periphery subacute. Wall consisting of an inner alveolar layer and a much thinner imperforate epidermal (outermost) layer, finely agglutinated with a smooth finish. Alveoles are unidimensional and equally spaced on the outer chamber wall. Aperture a broad interiomarginal slit with a thin upper lip of finely agglutinated particles. Areal supplementary apertures are not observed and, if present, are only small openings between the larger agglutinated grains in the apertural face.
Remarks. Cushman (1917Cushman ( , 1921 differentiated Cyclammina compressa from C. cancellata Brady by its smaller size, open and depressed umbilicus, evolute coiling and acute periphery. The syntypes preserved in the Cushman Collection (CC420) are from Albatross Station 5470, east of Luzon, at 540 fathoms. The specimens are over 3 mm in diameter, and have an average of 14 chambers in the final whorl. Banner (1970) regarded this species to be conspecific with C. cancellata Brady based on the structure of the hypodermis and alveoles. Kaminski & Gradstein (2005) accepted Banner's synonymy of the two species.
In the Bering Sea, typical Cyclammina cancellata has not been foundthe specimens recovered in Hole U1343A are best ascribed to C. compressa, which differs from typical Cyclammina cancellata in its smaller dimensions, more compressed lateral sides, straighter sutures and in lacking distinct supplementary areal apertures. The Bering Sea specimens are even smaller than Cushman's specimens from the Philippines, and do not have any visible areal supplementary apertures, whereas these are visible in Cushman's (1921) illustrations. Additionally, the umbilicus is not as depressed as Cushman depicted it. Cushman (1921, p. 86) remarked: 'This species seems to be in deeper and colder water than the preceding [C. cancellata]'.
Description. Test free, elongate, initially trochospiral with up to five chambers per whorl, later reduced to twisted triserial and finally becoming biserial in just the final one or two pairs of chambers. Chambers in the terminal biserial part are globular, with depressed sutures. Wall finely agglutinated, with a smooth outer surface. Aperture areal, a rounded opening slightly above the base of the apertural face in the triserial stage, becoming more areal and increasingly oval in the biserial adult stage, produced and surrounded by a distinct lip.

Remarks.
A single specimen was found in Hole U1341B. It most closely resembles the species described as 'Karreriella sp. 1' by Kaminski et al. (2013) from the Pliocene of Hole U1341B.
Genus Martinottiella Cushman, 1933 Martinottiella sp. 3 ( Fig. 10:2-10:5) Description. Test elongate, cylindrical, initial 6 -7 whorls coiled in a high trochospire with four chambers per whorl, later reduced to uniserial. Wall finely agglutinated, canaliculate with pores open to the exterior. Aperture areal in the coiled stage, terminal and central in the uniserial stage, on a short tubular neck in the centre of the flattened terminal face.
Remarks. This species is common in the samples from Site U1342. It differs from the type species M. communis (d'Orbigny) in possessing open pores on the external chamber walls and a round (rather than oval or slit-like) aperture. The open pores are concentrated in the middle of the chamber wallfewer pores are observed along the sutures (Fig. 10:3b). The apertural face in the adult stage is flat ( Fig. 10:5).