Methodology:

Data from OBIS were accessed on 12-01-2023 and data from GBIF were accessed on 13-01-2023. To maintain consistency, all encounters were treated as presence only, i.e. no weighting was given to records with information on abundance, biomass or number of individuals and no absence records were included. Raw downloaded data and the contributing dataset citations are provided for both databases (see Section "Related datasets").

Subsequent data processing and cleaning steps are outlined in section "Quality".

All cleaned records, regardless of dataset origin, were matched to seafloor bathymetry using a 0.25 x 0.25 decimal degree resolution raster grid (Becker et al., 2009). Cleaned records from OBIS had fields for maximum and minimum collection depth. These data were matched to vertically resolved temperature data from the 2018 World Ocean Atlas (Locarnini et al., 2018) and seasonal sea-ice concentration data from the National Snow and Ice Data Centre (Walsh et al., 2019). Both temperature and sea ice grids had a resolution of 0.25 x 0.25 decimal degrees.

Code_frontiers_2023.html (22 MB) - HTML output from Quarto (implemented in R) that provides the annotated source code to reproduce the data cleaning, analysis and figures from this dataset has been included in the publication.

Data collection:

All analyses were carried out in R version 4.2.1.

Data quality:

Data processing steps carried out on the raw downloaded files are outlined in detail in the html file of annotated source code. In brief, records were excluded if they were erroneously located on land, in the Pacific sector or in the southern hemisphere; were not identified to species level; had missing identifiers for month or year of collection. Metadata fields for "sampling protocol" ad "lifestage" were cleaned by re-categorising entries into a condensed set of inputs. Duplicate records were removed as defined by records that had the following identical identifiers: species, longitude, latitude, year, month, day, minimum and maximum collection depth (OBIS), average collection depth (GBIF), life-stage. Lastly, cleaned OBIS and GBIF data were combined into a single dataset and duplicate occurrences between databases were removed using the same identical identifiers listed above. Records where average sample collection depth was greater than bathymetric depth were flagged and removed when creating appropriate figures.