Oxygen decline poses increasing risks to global shelf communities. This study was conducted to measure species oxygen exposures in situ, and to assess risks of low oxygen based on the hypothesis that species risk varies spatially and is dependent on the interaction of shelf oxygen dynamics with habitat pattern, species mobility and tolerance to low oxygen. Here, we report concomitant observations of oxygen, depth, and habitat for the most common benthic and epibenthic megafauna on the southern California shelf (30-125 m). The study was conducted when oxygen concentrations were the lowest observed within the previous 30 years off southern California (2012-2014), yet hypoxic conditions (<1.42 ml L^-1) were not observed. No significant oxygen exposure differences were observed among highly mobile, mobile, and sessile species due to the dynamic nature of the oxycline. Habitat affinities for the most abundant benthic megafauna (53 taxa) were discriminated among four different habitats. Highly mobile rockfish (Sebastidae) exhibited distributions suggesting less tolerance to low oxygen than most other taxa. Given the mosaic of consolidated and coarse-grained habitats that rockfish prefer, the onset of hypoxic conditions near the shelf break (~90-120 m) would likely compress local rockfish populations into less favorable habitat in a non-linear fashion.

ROV surveys of the nearshore shelf (depths ~30-125 m) off San Diego, California (USA) were conducted to yield concomitant observations of benthic and epibenthic megafauna, sediment and bottom characteristics, and dissolved oxygen concentrations. The San Diego marine shelf is complex due to tectonic faulting and transgressive sea-level forcing of sediments (Hogarth et al., 2007; Le Dantec et al., 2010). Survey areas were chosen to span a range of shelf widths and slopes, habitat types (Parnell, 2015; Switzer et al., 2016), and proximity to submarine canyons. The areas studied ranged along ~45 km of shoreline from Coronado to Del Mar (Fig. 1). The shelves off Del Mar and La Jolla were repeatedly surveyed using an ROV for surveys of squid spawning and embryo deposition (Navarro et al., 2018) and the effects of tectonically forced habitat patch structure on macro and megafaunal diversity (Switzer et al., 2016). Surveys were also conducted further south on the shelves off Pt. Loma and Coronado. Every effort was made to seasonally stratify the surveys enabling coverage of both the spring/summer upwelling season and fall/winter periods when stratification is greatly reduced. ROV surveys were conducted between August 2012 and January 2014.

The ROV used in this study was a SeaBotix LBV-150 ROV equipped with a SeaBird SBE 37 MicroCAT CTD (SBE 37SMP-IDO), 2 external LED lights (1080 lumen, 140° beam angle), and red scaling lasers (5 cm separation). Temperature, conductivity, depth and [O₂] were measured at one minute intervals. The main ROV video camera (520 line, dynamic range 90 mm to infinity, color, 0.1 lux) provided an analog composite NTSC video signal that was converted to digital using a

Canopus ADVC110 converter. A GoPro camera, housed within a Golem Gear LCD_150 housing, was mounted on the ROV and recorded high definition video (1280×960 resolution, 48 frames s^-1) to aid taxonomic identification in the main video. The ROV was deployed off of a small (~8 m) vessel; the vessel track was recorded using a Hemisphere V110 inertial GPS. Survey velocities averaged ~0.75 m s^-1. Vessel track, digital video, and CTD data were recorded onto a laptop computer.

ROV videos were analyzed by sampling still captured images at 5-second intervals to identify habitat characteristics and quantify conspicuous megabenthos. Habitat factors included 5 levels of grain size and 11 levels of bedform (see Table 1). These two habitat factors characterize important differences among benthic habitats on the San Diego shelf (Switzer et al., 2016). Bedform characteristics of unconsolidated sediments provide information on the interaction between grain size and near-benthic hydrodynamic forcing such as oscillatory flows due to surface gravity waves and bottom current orientation and strength. Hard bottom bedform included ‘outcrops’ (attached hard bottom) and ‘boulders’ (>1 m largest dimension). Bedforms of mixed hard and soft bottom included ‘sediment veneer’ (a layer of unconsolidated sediment <~1 cm thick covering outcropped features), ‘rocks with sand’ (rocks <1 m in largest dimension mixed with sand), and cobble (~2.5-25 cm). Soft bottom bedform levels included ‘parallel waves’, ‘hatched waves’, ‘parallel ripples’, ‘hatched ripples’, ‘linguloid ripples’, and ‘flat’. Sediment waves were defined as undulations having amplitudes >~5 cm and ripples were <~5 cm. Grain size levels included ‘consolidated’ for hard bottom, and ‘Shell Hash’ (shell and gravel debris ~2.5 cm to 2 mm), ‘Coarse Sand’ (~0.5 to 2 mm), ‘Fine Sand’ (~0.5 to 0.05 mm), and ‘mud’ (~<0.05 mm) for sediments. All benthic megafauna (within view of the ROV while it was <~2 m off the bottom) were enumerated to the lowest possible taxonomic level.

Habitat factors were analyzed using multiple correspondence analysis (MCA; Venables and Ripley, 2002; MASS package in R (R Core Team, 2018)) to identify and classify major habitat categories. MCA enables dimensional reduction for categorical data in a manner similar to principal components analysis for numerical data. For this application, MCA identifies combinations of the most important habitat characteristics (‘MCA factors’) that account for the greatest variability, and which are orthogonal to each other.

Oxygen was measured continuously near the bottom off La Jolla to gauge oxygen variability over longer time scales than individual ROV deployments which typically lasted <8 hours. Longer term and more continuous oxygen measurements were necessary for contextualizing the representativeness of shelf oxygen conditions during the ROV survey observations. Oxygen was recorded using a SeapHOx instrument (Martz et al., 2010) mounted ~0.5 meters from the bottom off North La Jolla at a depth of 32 m (see Fig. 1). Data were recorded at 15-minute intervals from Aug 30, 2012 to Oct 21, 2013 across 4 deployments.

Usage notes are provided in file named "ROVHypoxiaReadme.txt"

Species Metadata is listed in "SpeciesMetadata.csv"

Raw Date are provided in "ROVObservationData.csv"