The study area includes the western Barents Sea, the adjacent marginal areas of the Norwegian and Greenland Seas, and of the Central Polar Basin covered by the bottom trawl stations of the Norwegian-Russian Ecosystem Survey (eastward to 40° E) ( Figure 1 ). Since 2004, the bottom trawling within this survey has been standardized to the use of a Campelen-1800 shrimp trawl by all the participating four research vessels (RVs) (Michalsen et al., 2011; Eriksen et al., 2018; Zakharov et al., 2018). All data from the 2005–2022 period were included in the analyses, except from three RVs for 2005, where cephalopods were not identified within benthos catch. The study period was separated into two equal periods for comparative purposes (2005–2013 and 2014–2022). Overall, 2720 and 1609 bottom trawl stations were performed in the studied area during the 2005–2013 and 2014–2022 period, respectively ( Table 1 ; Supplementary Table 1 ).

All invertebrate benthos catch from the shrimp trawl surveys was identified onboard by benthic experts alongside fish experts (Zakharov et al., 2018; Jørgensen et al., 2019; Zakharov et al., 2020; Jørgensen et al., 2022). Cephalopods usually were treated as the other benthos bycatch taxa onboard (Zakharov et al., 2018; Jørgensen et al., 2019; Zakharov et al., 2020; Jørgensen et al., 2022). Since 2005, a varying number of cephalopods from the benthos catch was fixed or frozen and analysed ashore by Alexey V. Golikov (A.V.G.) and Rushan M. Sabirov (R.M.S.). Such re-identification of catches from the older surveys increased the number of species initially found. Moreover, in many cases old unfixed samples were photo-verified by A.V.G. and R.M.S. These data, when provided to onboard benthic experts, resulted in significant increase of onboard identification quality in the 2014–2022 period in comparison to the 2005–2013 period (see Results & Discussion). Outdated names such as Benthoctopus sp. and later Muusoctopus sp. were changed to Muusoctopus aegir after this species was formally described in 2023 (Golikov et al., 2023a). Rossia individuals that did not fit either R. palpebrosa or R. moelleri taxonomic descriptions were identified to Rossia megaptera after this species was found to present in the Barents Sea (Golikov et al., 2020).

In some years, a cephalopod expert was absent onboard during a particular cruise. It was never the case for all participating RVs simultaneously, if we account for ashore/photo-reidentification described above. During these years, Bathypolypus and Rossia were often not identified to the species level and labelled as “Bathypolypus sp.” and “Rossia sp.”, respectively. Also during these years, species were labelled as “Incirrata” if either Bathypolypus sp. or Muusoctopus aegir; “Octopoda” if either both of the latter or Cirroteuthis muelleri; “Ommastrephidae” if either Todaropsis eblanae, Todarodes sagittatus or unidentified Ommastrephidae; and “Cephalopoda” if nothing else was recorded/photographed to help the identification ashore. In this manuscript, the data were provided on group level as Bathypolypus spp. (= combined three species and Bathypolypus sp.), Rossia spp. (= combined three species and Rossia sp.) and Ommastrephidae (T. eblanae, To. sagittatus or unidentified Ommastrephidae), as well as on individual species level, to show both species’ patterns and patterns unbiased by identification quality.

The ratio of trawl stations with cephalopod catch to standardized total catches was recorded annually and was used as a measure of cephalopod relative abundance. Frequency of occurrence of individual species/group of cephalopods in % of all stations with cephalopods was used as an abundance proxy of particular species/group. Statistical comparisons of frequencies between the 2005–2013 and 2014–2022 period were possible where we have enough data annually (= caught for more than four years within either one of the periods). These taxa were Bathypolypus arcticus, Bathypolypus spp., Gonatus fabricii, Rossia palpebrosa, R. megaptera, Rossia spp. and unidentified Cephalopoda. The latter was done to assess if the quality of onboard identification changed from the 2005–2013 to 2014–2022 period.

Biomass density, i.e. biomass in gram per nautical mile of towing (g/n.m.) was a proxy of absolute biomass. The quantitative measures were estimated using towing speed and distance by standard IMR/PINRO protocols (Jørgensen et al., 2019; Zakharov et al., 2020). We decided not to correct density values to absolute biomass, because the Campelen-1800 shrimp trawl’s catchability for cephalopods is unknown, as revealed by previous attempts to do so (Lubin and Sabirov, 2007; Golikov et al., 2017). Statistical comparisons of biomass density between the 2005–2013 and 2014–2022 period were possible where enough data was available: Bathypolypus arcticus, B. bairdii, Bathypolypus spp., Gonatus fabricii, Rossia palpebrosa, R. megaptera and Rossia spp. Deep-sea octopods (Muusoctopus aegir and Cirroteuthis muelleri) were not temporally compared, because deep-sea slopes were sampled infrequently and differently among years, unlike the shelf areas.

To compare the biomass of species/taxa (see above) between the two groups (i.e., the 2005–2013 and 2014–2022 period), a Mann–Whitney U-test was used; and to compare among three or more groups (i.e., within Bathypolypus spp., Rossia spp. and among other taxa) we used a Kruskal–Wallis H test with a post-hoc Dunn’s Z test (Zar, 2010). Frequencies of species/taxa were compared using a χ ² test (Zar, 2010). Statistical analysis, calculations, equations and plots were performed in PAST 4.12b (Hammer et al., 2001), Statistica 12.0 (Statsoft) and MS Excel 2010. The value of α = 0.05 was considered significant in this study.

The Norwegian-Russian Ecosystem Survey recorded twelve species in the Barents Sea and adjacent areas during 2005–2022: Bathypolypus arcticus, B. bairdii, B. pugniger, Muusoctopus aegir (incirrate octopods; Octopoda Incirrata); Cirroteuthis muelleri (cirrate octopod; Octopoda Cirrata); Gonatus fabricii, Todaropsis eblanae, Todarodes sagittatus (squids; Oegopsida); and Rossia palpebrosa, R. megaptera, R. moelleri and Sepietta oweniana (bobtail squids; Sepiolida). Among those, T. eblanae, To. sagittatus and S. oweniana were boreal-subtropical migrants, and all others belonged to permanent Arctic residents. All records of the boreal-subtropical species occurred south of 75° N ( Figure 1 ). All these species were recorded during both the 2005–2013 and 2014–2022 period, except for T. eblanae, which was absent during the 2014–2022 period ( Table 1 ). The latter species, however, was found east of 40° E in the Barents Sea during this period (Golikov et al., in prep.). So, the fauna composition over the studied area seemed identical between the 2005–2013 and 2014–2022 period.

The trawl catch of cephalopods increased significantly over time ( Figure 2 , Table 1 ; Supplementary Table 1 ). Moreover, significant increase in frequency was recorded in every taxon it was checked for (i.e., Bathypolypus spp., G. fabricii, R. palpebrosa, R. megaptera, Rossia spp. and unidentified Cephalopoda), except for B. arcticus ( Table 1 ). In the latter species, significant decrease in frequency was recorded ( Table 1 ). Among the taxa where the significance of temporal frequency changes could not be checked, it was increase in B. bairdii, Bathypolypus sp., deep-sea and unidentified octopods, boreal-subtropical Ommastrephidae squids, R. moelleri and Rossia sp. and decrease in B. pugniger ( Figure 2 , Table 1 ; Supplementary Table 1 ). Boreal-subtropical S. oweniana was infrequently caught and no differences in frequency were observed over time. Aggregations of Bathypolypus sp. and B. bairdii were recorded in the south-western Barents Sea ( Figure 1 ). Both deep-sea incirrate and cirrate octopods, M. aegir and C. muelleri, were much less frequent than Bathypolypus spp. ( Figure 2 , Table 1 ). Both these two deep-sea octopods were found on the slope and in the troughs, but not on the shelf ( Figure 1 ). Unidentified Incirrata and Octopoda were also largely found over the deep-sea areas, but closer to the shelf than ‘identified’ deep-sea octopods ( Figure 1 ).

Overall, Bathypolypus spp., G. fabricii and Rossia spp. were much more frequent than the deep-sea (M. aegir and C. muelleri) and the boreal-subtropical (Ommastrephidae and S. oweniana) species, and distributed all over the study area ( Figures 1 , 2 , Table 1 ). Among the three most abundant cephalopod taxa, Rossia spp. was the most frequently recorded during both the 2005–2013 and 2014–2022 period, while Bathypolypus spp. was the least frequently recorded ( Table 2 ). Within the Bathypolypus spp. and Rossia spp., widespread boreal-arctic species B. arcticus and R. palpebrosa were the most ubiquitous ( Figures 1 , 2 , Table 1 ). While the high-boreal R. megaptera and the arctic R. moelleri preferred warmer and colder areas, respectively, both of the two high-boreal species B. bairdii and B. pugniger preferred warmer areas ( Figure 1 , Table 1 ).

While frequencies of occurrence showed significant differences over time, the biomass only showed significant decrease from the 2005–2013 and 2014–2022 period for Bathypolypus spp. ( Table 3 ). A non-significant decrease was recorded for B. arcticus, B. bairdii and Gonatus fabricii, while a non-significant increase was found in Rossia palpebrosa, R. megaptera and Rossia spp. ( Table 3 ). Among the three most abundant cephalopod taxa, the biomass density was the highest for Bathypolypus spp., medium for Rossia spp. and the lowest for G. fabricii ( Tables 3 , 4 ). Within Bathypolypus spp., no significant differences were found among the species ( Tables 3 , 4 ). Within Rossia spp., high-boreal R. megaptera had significantly lower biomass density than the two other species ( Tables 3 , 4 ).

The main changes in cephalopod fauna occurred during the mid-2000s and early 2010s, when four boreal-subtropical species appeared in the area (three of them for the first time ever, and one for the first time since the early 1980s). The timing of their occurrence (i.e., the mid-2000s and early 2010s) followed the onset of increased climate-driven warming in the Barents Sea since late 1990s/early 2000s (Swart et al., 2015; Rantanen et al., 2022). The Atlantification of the cephalopod community in the Barents Sea was evidenced by increased abundance of boreal-subtropical Ommastrephidae squids, and repeated occurrence of Ommastrephidae squids and Sepietta oweniana during two periods (2005–2013 and 2014–2022). ‘Typical’ Arctic cephalopod species (Bathypolypus spp., Gonatus fabricii and Rossia spp.) are still much more abundant in the western Barents Sea compared to the deep-sea and the boreal-subtropical cephalopod species. Moreover, the abundance of these Arctic taxa increased from the 2005–2013 to 2014–2022 period, as did the abundance of other cephalopods across the studied area. The increased abundance but reduced biomass of Bathypolypus spp. from the 2005–2013to 2014–2022 indirectly suggests a body-size reduction. This would be the first evidence of size reduction in response to ocean warming in octopods, a phenomena known for other taxa elsewhere (e.g., Gardner et al., 2011; Sheridan and Bickford, 2011; Ikpewe et al., 2021). Lastly, a significant increase of onboard identification quality followed when onboard benthic experts were educated by cephalopod taxonomic experts. Such taxonomic training increases the value of the ecosystem surveys for the future monitoring.

Our study is the first to document temporal trends in the cephalopod community of the Barents Sea, which is one of the fastest warming regions in the Arctic (Overland et al., 2014; Lind et al., 2018; Gerland et al., 2023). Before 2005 (= start of the time series used here), cephalopod biodiversity of the Barents Sea was lower than it currently is (Nesis, 1987; Golikov et al., 2013; Xavier et al., 2018) ( Table 5 ). Todarodes sagittatus has been the only boreal-subtropical cephalopod known from the Barents Sea before 2005 (Nesis, 1987; Golikov et al., 2013; Xavier et al., 2018) ( Table 5 ). It was last recorded in the Barents Sea in 1983 (Nesis, 1987), while in our studies the first records of this species start from 2010 (Golikov et al., 2013, 2014; Xavier et al., 2018). Other boreal-subtropical cephalopods that were found in the Barents Sea in 2005–2013 are Todaropsis eblanae, Teuthowenia megalops and Sepietta oweniana (Golikov et al., 2013, 2014; Xavier et al., 2018) ( Table 5 ). The deep-sea squid Te. megalops was only found once in 2009 (Golikov et al., 2013). This rare record was not from the Norwegian-Russian Ecosystem Survey area, but from the north-eastern slope of the Greenland Sea ( Figure 1 ), and taken by the same Campelen-1800 trawl as used in the Norwegian-Russian Ecosystem Survey (Golikov et al., 2013). In 2014–2022, all boreal-subtropical cephalopods were recorded in the Barents Sea again, except for Te. megalops ( Table 5 ). A gradual influx of Atlantic fauna in the Barents Sea is also well known for other invertebrates and fishes, providing a basis for ongoing Atlantification of the local ecosystems (Jørgensen et al., 2016; Frainer et al., 2017; Brandt et al., 2023). The records from our study show that the cephalopod community of the Barents Sea is subjected to the Atlantification since the 2005–2013 period. Mean annual temperatures in the Arctic, both modelled and observed, continuously increase since the late 1990s/early 2000s (Swart et al., 2015; Rantanen et al., 2022). This warming, which results from climate change, is often hypothesized as a main cause of the Atlantification (e.g., Jørgensen et al., 2016; Frainer et al., 2017; Brandt et al., 2023).

Of the twelve permanent Arctic resident cephalopod species (Xavier et al., 2018; Golikov et al., 2020; Golikov et al., 2023a), only Muusoctopus sibiricus, M. leioderma and Muusoctopus sp. were absent in the western Barents Sea and adjacent areas. These octopods live in the Siberian Seas, Beaufort and Chukchi Sea (M. sibiricus and M. leioderma), and in the northern Baffin Bay and Canadian Arctic Archipelago (Muusoctopus sp.) (Nesis, 1987; Xavier et al., 2018; Golikov et al., 2023a). The absence of published records of Rossia megaptera, Bathypolypus bairdii and B. pugniger before 2005 ( Table 5 ) could rather be caused by the lack of taxonomic studies and identification expertise. Bathypolypus bairdii and B. pugniger were only taxonomically separated from B. arcticus in 2002 (Muus, 2002). And R. megaptera is often considered as living only in North-West Atlantic (Mercer, 1968; Reid and Jereb, 2005), despite its presence was later confirmed from Iceland (Golikov et al., 2018b) and the Barents Sea (Golikov et al., 2020). When reanalyzing samples from before 2005, both B. pugniger and R. megaptera have been found in the area, in 1967 and in 2003 respectively (Golikov et al., in prep.).

Alternative methods to assess biodiversity and community composition in marine ecosystems are underwater video imagery and environmental DNA (eDNA) analyses (e.g., Merten et al., 2021; Kopp et al., 2023). Environmental DNA analysis enables detection of species based on genetic material from marine animals that is released in the environment, such as shed skin cells, mucus, gametes and faeces (Taberlet et al., 2012a, Taberlet et al., 2012b). The advantages of eDNA analysis are that it gives a larger presence/absence time frame than trawl surveys (eDNA remains suspended in the water column for up to 60 days), and it allows simultaneous identification of different taxa within the same water sample (reviews: Thomsen and Willerslev, 2015; Rourke et al., 2022). Advances and successes have been made with the application of eDNA in fisheries surveys (Thomsen and Willerslev, 2015; Rourke et al., 2022) and also in cephalopod biodiversity studies (e.g., Merten et al., 2021; Visser et al., 2021; Merten et al., 2023). While the technique is being developed to be used for abundance estimates (Thomsen and Willerslev, 2015; Rourke et al., 2022), it is currently mostly used to obtain presence-absence for species or communities. Results of eDNA metabarcoding depend on the used primers, which may be biased towards certain taxonomic groups, and sequence availability in databases. For example, the 18S rRNA primer that is used in cephalopod eDNA studies does not detect all octopods (De Jonge et al., 2021). While, arctic and boreal-subtropical cephalopods recorded in the Barents Sea and adjacent areas are better represented in GenBank with COI, than with 18S. When applying, it is 18S which is better able to detect cephalopods during eDNA metabarcoding on water samples (detailed in De Jonge et al., 2021). The collection of underwater imagery enables analyses of biodiversity and community composition, color, behavior, habitat association, and estimation of abundance and biomass density in case of standardized surveys (e.g., Robison et al., 1998; Robison, 2004; Buhl-Mortensen et al., 2015; Stratmann et al., 2022). Underwater imagery analyses has been applied in biodiversity assessments of cephalopod communities (e.g., Merten et al., 2021; Pratt et al., 2021, Pratt et al., 2023). For example, G. fabricii was recorded in the central Arctic Ocean under ice by a mooring camera system (Snoeijs-Leijonmalm et al., 2022), and Arctic cirrate octopods were documented to perform benthopelagic migrations (Golikov et al., 2023b). While no studies have attempted to quantify squid biomass from underwater imaging surveys, this approach may be challenging since some squids may alter their behavior in response to lights (avoidance or attraction behavior) (e.g., Hoving et al., 2019; Snoeijs-Leijonmalm et al., 2022). On the other hand, presumably natural cephalopod behavior has been documented repeatedly via in situ observations of remotely operated vehicles (e.g., Hoving and Robison, 2012; Hoving et al., 2013b; Hoving and Haddock, 2017). Benthic cephalopods are often hard to identify to species level from images (e.g., Pratt et al., 2021; Robinson et al., 2021; Snoeijs-Leijonmalm et al., 2022; and many others). To summarise, currently trawling, eDNA and video imagery each have their strengths and weaknesses, and the best results are rendered by combining them (e.g., Thomsen et al., 2016; Merten et al., 2021; Kopp et al., 2023).

A clear increase in the ratio of cephalopod catches to standardized total catches, as well as an increase in the frequency of occurrence was recorded for most of the studied cephalopod species/taxa in the western Barents Sea from the 2005–2013 to 2014–2022 period. These observations align with ongoing continuous increase of mean annual temperatures in the Arctic during the same timeline (Swart et al., 2015; Rantanen et al., 2022). While the expansion of cephalopods’ ranges and habitats was previously indirectly implying to their increased abundance and biomass (Golikov et al., 2012, Golikov et al., 2013; Xavier et al., 2018; Golikov et al., 2019b; Oesterwind et al., 2022), our study is the first direct evidence that indeed the abundance of cephalopods in the Arctic is increasing. Our results are in line with a global trend that shows that cephalopods’ biomass is increasing in tropical and temperate areas (Doubleday et al., 2016). Our results are also in line with the trends observed for other Arctic nekton, such as increasing abundance of boreal pelagic fishes (Frainer et al., 2017; Brandt et al., 2023).

The biomass density used here can be a proxy of absolute biomass when coupled with abundance data and distribution maps. This makes comparisons of biomass among the studies possible (even though limited by the use of different gear). The biomass of deep-sea octopods is the largest in the troughs and on the slopes of the marginal areas of the Barents Sea. Cirroteuthis muelleri specifically reaches the highest biomass density in the studied area, even though it is still about two times lower compared to this species’ hotspots in the Baffin Bay (Golikov et al., 2022). At the same time widespread Arctic taxa (Bathypolypus spp., Gonatus fabricii and Rossia spp.) have much higher abundance and ubiquitous distribution in the studied area, while their biomass is lower than of deep-sea octopods. The biomass of Bathypolypus spp. and Rossia spp. in the Baffin Bay seems comparable to our values (Frandsen and Wieland, 2004; Treble, 2007), but the trawls are different and it may flaw a direct comparison. In the Porcupine Seabight, abundance data of Bathypolypus spp. and Rossia spp. suggest similar or slightly lower biomass than in the Barents Sea, but the used trawls are also different from those used in the Barents Sea (Collins et al., 2001). When correctly estimated, absolute biomass may be a good parameter to compare among areas. Because we do not know the trawl catchability of cephalopods, absolute biomass is currently rarely used. Previous conservative estimates of absolute biomass include 6.5 thousand tonnes of R. papebrosa and 24.8 thousand tonnes of G. fabricii in the Norwegian-Russian Ecosystem Survey area in 2007 and 2011, respectively by Golikov et al. (2017). Our current study exceeds these numbers in 8–10 times.

The standardized survey (gear, time and place) demonstrating the congruent significant decrease of biomass and significant increase in individual numbers (see Results and above) indirectly suggests a reduction in body-size of Bathypolypus spp. in the western Barents Sea. There are various examples of reduction in size in aquatic invertebrates, fishes, and some seabirds in response to climate change (e.g., Gardner et al., 2011; Sheridan and Bickford, 2011; Ikpewe et al., 2021). Among cephalopods, size reduction in response to climate change has only been recorded for squids (Jackson and Domeier, 2003; Hoving et al., 2013a; Arkhipkin et al., 2015; Takahara et al., 2017).

To date, cephalopod monitoring in the Barents Sea has been performed via analysis of benthos bycatch from the Norwegian-Russian Ecosystem Survey. This survey uses the Campelen-1800 bottom trawl which is designed for demersal shrimp and fish surveys (McCallum and Walsh, 1994). It is not typically used to assess pelagic squids, as was done in this study. Still, even with a gear that is suboptimal for nekton, G. fabricii is the second most abundant cephalopod in the study area. This suggests that biomass of G. fabricii and other squids in the Barents Sea may be even higher if sampled with a pelagic trawl (Golikov et al., 2012, Golikov et al., 2017; Golikov et al., 2019a).

Cephalopod identification can be challenging for non-specialists, also because of wrong or insufficient species names in GenBank (e.g., Fernández-Álvarez et al., 2021; Katugin and Zolotova, 2023). Specifically, in Bathypolypus spp. only mature males can be reliably identified by non-specialists. Genetic barcodes from reliably identified B. bairdii and B. pugniger were only recently uploaded to GenBank (Taite et al., 2023). Muusoctopus aegir was first described in 2023 (Golikov et al., 2023a), and is in GenBank as ‘Muusoctopus sp.’ (Taite et al., 2023). Sequences of specimens that were morphologically identified as Gonatus fabricii and G. steenstrupi and originated from several studies in the North Atlantic cluster as a single species in GenBank (Lindgren et al., 2005; Lindgren, 2010; Vecchione et al., 2010; Taite et al., 2020). These Gonatus sequences were referred to as being different from ‘real’ G. fabricii from the Arctic (Taite et al., 2020; Katugin and Zolotova, 2023), which was cited as ‘Lindgren, unpublished’. To date, only G. fabricii has been recorded in the Barents Sea (Golikov et al., 2012; Xavier et al., 2018; Golikov et al., 2019a). The most northern distribution is still unknown for G. steenstrupi, and is supposed to be in the low Arctic areas, such as north off Iceland (Xavier et al., 2018; Golikov et al., 2018a). Rossia palpebrosa also has several misidentifications in GenBank (they do not even cluster as one species; A.V.G., pers. obs.), and there are no uploaded sequences for R. moelleri and R. megaptera.

In this study there was a significant increase in quality of onboard identification expertise by benthic experts over time. This is demonstrated by a three times-decrease of unidentified cephalopods from the trawl catches from the 2005–2013 period to the 2014–2022 period, and by changed ratios of taxa within Bathypolypus spp. Onboard experts recognized B. bairdii and B. pugniger more often. Still, 10% of cephalopod catches were unidentified in the 2014–2022 period, which imply necessary improvements of onboard identification.