Scientific trawl survey data
Scientific bottom trawl surveys, primarily sampling demersal commercial species, were obtained from the Northeast Pacific and North Atlantic shelf regions. All data used are publicly available and were downloaded in July 2021 (see Appendix S1 for details on data processing). We selected all scientific surveys that sampled the fish community with otter trawls. For each tow in each survey, we selected all demersal teleost and elasmobranch species and obtained species weight. We corrected these weights for differences in sampling area (in km2) and trawl gear catchability to obtain a standardized fish biomass across hauls and surveys. We estimated sampling area using information on wingspread, speed of vessel and tow duration. Weights were corrected for trawl gear catchability using information for 80 species in the Northwest Atlantic (Link et al.2008) and 128 species and 7 functional groups in the North Sea (Walkeret al. 2017). The adjustments resulted in biomass estimates per unit area in metric tonnes (1000 kg) per km2.
We compared the corrected trawl survey biomasses with available fisheries stock assessment biomasses to validate the range and distribution of the biomass estimates. To this end, we calculated spatial overlap between the surveyed area and the bounding region of all fisheries assessment areas from the RAM Legacy database (Ricard et al. 2012). For each area that overlapped at least 50% with the surveyed area, we compared biomass of each assessed stock with the gear-corrected trawl survey biomass for the corresponding species. The comparison shows that the corrected biomass has a reasonable match with the stock assessment biomass and no apparent bias, for most of the 120 stocks in the Atlantic and Pacific (Figure S1.3). This finding improves confidence that the gear-corrected trawl survey estimates, hereafter termed demersal fish biomass and/or demersal community biomass, are representative and comparable across areas and surveys.
Using the individual haul coordinates, we estimated an average demersal community biomass, in tonnes per km2, per equal area grid cell (6000 km2) and surveyed year. To reduce the effect of potential outlying biomass estimates, we removed all individual observations 1.5 times less/greater than the interquantile range per survey and year based on log-transformed biomass values (but note that the overall conclusions are robust with or without such a data removal, not shown).