A multidisciplinary approach was employed to examine a physical-biological population hypothesis for a critical forage species, the Antarctic silverfish (Pleuragramma antarctica). A previous study had shown strong gene flow along the westward Antarctic Slope Current, in addition to spatially recurring length modes that provided evidence for episodic connectivity. In this paper, otolith nucleus chemistry from a subset of fish collected in the southern Weddell Sea as part of a hydrographic survey of the Filchner Trough system was used to test between connectivity scenarios. Nucleus chemistry, which reflects environmental exposure during early life, showed significant spatial structuring despite homogeneity in microsatellite allele frequencies. Mg×Ca^-1 and Sr×Ca^-1 differentiated length modes, and Mg×Ca^-1 showed significant contrasts between Atka Bay, Halley Bay, and Filchner Trough. Physical-biological mechanisms may help reconcile structuring shown by otolith chemistry, length, and abundance data with prior evidence of gene flow. Such mechanisms include self-recruitment shaped by circulation associated with the Filchner Trough, fluctuations in mixing between immigrant and locally-recruited fish, and feeding opportunities between inflowing Modified Warm Deep Water and outflowing Ice Shelf Water. The results illustrate how comparisons between multi-disciplinary techniques based on integrated sampling designs that incorporate hydrography can enhance understanding of population structure and connectivity around the Southern Ocean.

Field sampling

Antarctic silverfish were collected in the southern Weddell Sea from January – February 2014 as part of the research cruise PS82 (ANT-XXIX/9) by the RV Polarstern, investigating the Filchner Outflow System. For this, a Seabird 911+ CTD (Conductivity-Temperature-Depth), measuring salinity, temperature, and pressure, attached to a carousel with 24 water bottles, was deployed to investigate the physical environment in the region around the Filchner Trough, including the course of the AACC and MWDW flowing towards the Filchner Ice Shelf (Knust & Schröder, 2014). Conductivity and temperature sensors were calibrated before and after the cruise by Seabird Electronics; in addition, conductivity was corrected after calibration using salinity measurements from water samples measured by two Optimare Precision Salinometers. A total of 142 CTD profiles were obtained for oceanographic transects across and along the continental slope and shelf-break, and across the eastern shelf adjoining the trough. The distribution of bottom water masses over the slope and continental shelf were reported by Schröder et al. (2014). The CTD section at 76˚S off Coats Land was described by Ryan et al. (2017), including seasonal evolution based on a 2-year time series retrieved from three moorings.

Integrated with the physical measurements, silverfish were sampled from stations around the Filchner Trough including the adjoining eastern shelf. All sampling was undertaken during the day when fish were near the bottom during their diurnal migration, using a commercial benthic trawl with a cod-end mesh line of 20 mm. Standard length (SL) was measured for all individuals, and biomass, abundance, and size distribution determined for each haul. Catches were randomly sub-sampled for tissue and otoliths for subsequent analyses. For comparison, samples were also collected in Atka Bay and off Camp Norway, with corresponding CTD data.

Laboratory procedures

Biomass and abundance indices of Antarctic silverfish were reported by Wetjen et al. (2014a) and (Wetjen et al. 2014b) and genetic analyses were undertaken and reported by Caccavo et al. (2018). Length distributions showed two recurrent modes, of immature fish at approximately 10 cm and mature fish at approximately 15 cm SL. However, the length modes were not uniformly present at every sampling area, precluding a cross-wise experimental design with area and length group as factors. Instead, five stations were selected along the shelf in relation to the position of the AACC, the ASC, and inflows across the shelf-break. These were in Atka Bay; across the eastern shelf adjoining Filchner Trough corresponding to three CTD sections across Halley Bay, at 76˚S off Coats Land; at 77˚S just east of the trough flank; and near the shelf-break corresponding to a section across the western shelf downstream of the trough mouth. All fish in a given sampling area were pooled where only one mode was present; where both were present, fish were pooled into groups of large and small individuals. A cut-off of 13 cm was used to separate the length modes of 10 cm and 15 cm, corresponding approximately to the length at which sexual maturation begins in males and females (La Mesa & Eastman 2012). This resulted in eight groups from each of which 25 fish were randomly sub-sampled to form experimental treatments. In Halley Bay, both large and small groups at Station 126 were supplemented by randomly selected fish from nearby at Station 129. Similarly at 77˚S, six fish from Station 84 supplemented those sampled from Station 78, resulting in eight treatments: Atka Bay (AB), Halley Bay small mode (HB-S) and large mode (HB-L), Coats Land small (CL-S) and large (CL-L) mode, East Filchner Trough (EF), West Filchner Trough small (WF-S) and large (WF-L) mode.

One randomly selected otolith from each fish was used for elemental analysis. Otoliths were prepared using a standard protocol developed at the Center for Quantitative Fisheries Ecology at Old Dominion University. They were initially cleaned using glass probes and rinsed in Milli-Q water. Any remaining surface contamination was removed by 5 minutes incubation in 20% Ultra-Pure hydrogen peroxide, followed by rinsing in Milli-Q water. Otoliths were then mounted onto slides using crystal bond, and ground from the anterior side using a Crystal Master 8 Machine with 3-micron 3M™ polishing film to reveal a transverse surface just above the primordium. In a clean room, mounted otoliths were given a final polish by hand to reveal the primordium using 3-micron 3M™ polishing film, and once again rinsed with Milli-Q water and allowed to dry. Otoliths were removed from the polishing slide and mounted on petrographic slides in a randomized block design, with slide as the blocking factor. In this way, all eight treatments were represented on each petrographic slide, and the slides were then rinsed and sonicated individually for 5 minutes in Milli-Q water, before placement under a laminar-flow hood to dry.

Minor and trace elements were measured using a Finnegan Mat Element 2 double-focusing sector-field Inductively Coupled Plasma Mass Spectrometer (ICP-MS) located in the Plasma Mass Spectrometry Facility at Woods Hole Oceanographic Institution (WHOI). Samples were introduced into the ICP in an automated sequence (Chen et al. 2000) in which otolith material ablated with a 193 nm laser ablation system was combined with HNO₃ aerosol introduced by a microflow nebulizer. The subsequent mixture was then carried to the ICP torch. For quality control, we used a calcium carbonate reference produced by the United States Geological Survey (Microanalytics Carbonate Standard, MACS-3), for which elemental fractionation, mass-load, and matrix effects have been shown to be small for lithophile elements, especially when used with 193 nm laser ablation (Jochum et al. 2012). In the same way as the samples, MACS-3 material was ablated and introduced into the spray chamber as an aerosol; HNO₃ aerosol alone was also introduced into the spray chamber by the nebulizer as blanks. The randomized block design controlled for operational variability in the instrument. Blank and reference readings of count rate (count s^-1) were taken before and after each block, and every four otoliths within blocks. Readings consisted of 60 scans with 200 samples/peak and a 5% window. Based on markers identified during previous studies for the Southern Ocean, otoliths were analyzed for ⁴⁸Ca, ²⁵Mg, ⁵⁵Mn, ⁸⁸Sr, and ¹³⁸Ba and reported as ratios to ⁴⁸Ca. Background counts were subtracted from otolith counts by interpolating between the readings taken before and after every four otoliths, and the corrected otolith counts were then converted to element:Ca (Me•Ca^-1) ratios using the standards. To measure ratios indicative of conditions to which fish were exposed during early life, a grid raster type 150 µm x 200 µm was placed over the otolith nucleus corresponding to the first austral summer of growth, during which larvae are thought to remain in their natal trough (Brooks et al. 2018). The area was ablated using a 20 µm diameter laser beam at a frequency of 10 Hz and a power of 60%, traveling over the raster at 6 µm s^-1. As expected from earlier studies along the Antarctic continental shelf, Mn•Ca^-1 values were extremely low, frequently showing zero values after standard conversion. Background levels of Ba increased over the course of the analyses; however, the randomized blocks design guarded against any resulting biases.

1_LAICPMS_data.xlsx

Spreadsheet with the metadata associated with samples analyzed for trace element levels using LA-ICP-MS containing the following sheets:

• Cover sheet explains the column headings in the subsequent sheets.
• Raw Data sheet compiles the raw data produced from the LA-ICP-MS (which can be accessed in the .txt files also available in this data publication), including intensity data (AVG and STD in counts per second [cps], as well as % RSD, the average Mass Offset) for all elements analyzed.
• Intensity Data sheet includes just the average intensity data (Intensity AVG from the Raw Data sheet) for each element analyzed, as well as associated metadata.

All other .txt files

These are the text file outputs from the LA-ICP-MS, containing the raw data for each blank, standard, line (otolith edge), and raster (otolith nucleus) reading. These raw data are compiled in the Raw Data sheet of the 1_LAICPMS_data.xlsx file included in this data publication, however, for certain downstream analysis programs, these raw text files are needed.