Diet and life history reduce interspecific and intraspecific competition among three sympatric Arctic cephalopods

Trophic niche and diet comparisons among closely sympatric marine species are important to understand complex food webs, particularly in regions most affected by climate change. Using stable isotope analyses, all ontogenetic stages of three sympatric species of Arctic cephalopods (genus Rossia) were studied to assess inter- and intraspecific competition with niche and diet overlap and partitioning in West Greenland and the Barents Sea. Seven traits related to resource and habitat utilization were identified in Rossia: no trait was shared by all three species. High boreal R. megaptera and Arctic endemic R. moelleri shared three traits with each other, while both R. megaptera and R. moelleri shared only two unique traits each with widespread boreal-Arctic R. palpebrosa. Thus all traits formed fully uncrossing pattern with each species having unique strategy of resource and habitat utilization. Predicted climate changes in the Arctic would have an impact on competition among Rossia with one potential ‘winner’ (R. megaptera in the Barents Sea) but no potential ‘losers’.

. Ranges and maximum recorded sizes in the studied species of the genus Rossia, and exact sampling areas and corresponding environmental parameters (temperature and depth). Values of environmental parameters are minimum − maximum (mean ± SE). *Reasons why R. megaptera was overlooked for a long time on such huge areas and related details are work in progress (Golikov et al. in prep.). Presence of this species in Iceland was recently published 50 . a New maximum mantle length of these species, exciding previous published records (cf. 38,48,49,51 (Table 1). Rossia palpebrosa (n = 49), R. megaptera (n = 45) and R. moelleri (n = 39) were collected in July-August (see Table 2, Supplementary Tables S1-S4, for detailed information per species, area, sex and lifestage). All the studied species are known to grow continuously throughout their life cycle, while having highly variable size at maturity 49,51 (Golikov et al., unpubl.). Thus, all specimens were categorized in three (R. moelleri in four) arbitrary ontogenetic size groups: mantle length (ML) < 21 mm (small), ML 21 to 40 mm (medium) and ML > 41 mm (large), corresponding roughly to the life-stages of immature, maturing and mature specimens, respectively. In R. moelleri, large specimens were categorized as ML 41 to 60 mm, and a fourth group, very large, as ML > 61 mm: they all were mature females. Eight specimens of all groups were randomly selected for SIA but all specimens were taken if less than n = 8 existed in any group (Tables 2, 3, Supplementary Tables S1-S4). Some of R. palpebrosa samples (n = 37) were used in recent SIA study of species' stomach contents 41 . Specimens were fixed in formalin onboard. Mantle length was measured, and sex and maturity stage were assessed in fixed specimens onshore. Lower beaks were taken for SIA, as they have been repeatedly used in related studies (e.g. 41,42,[52][53][54], and their rostrums measured (n = 133).  www.nature.com/scientificreports/ Stable isotope analysis. Transparent areas of the beaks were removed before proceeding with SIA, as they have different isotopic concentrations biasing the outputs 55 . The beaks were dried at 60 °C and ground into a fine powder. Powder sub-samples were weighed (to the nearest 0.3 mg) with a micro-balance and sterile-packed in tin containers. The analyses were carried out at the Marine and Environmental Science Centre (MARE)-University of Coimbra (Portugal) with Flash EA 1112 series elemental analyzer coupled online via a Finnigan Con-Flo II interface to a Delta VS mass spectrometer (ThermoFisher Scientific) and at the Laboratory of Isotopic and Elemental Analysis-Kazan Federal University (Russia) with Flash HT series elemental analyzer coupled online via a ConFlo IV interface to a Delta V Plus mass spectrometer (ThermoFisher Scientific). No significant differences in SIA were found between the specimens of the same species and group from the same area measured in both spectrometers (n = 10, U = 23.5, p = 0.31). Stable isotope values were expressed as: δ 13 C and δ 15 N = [(R sample / R standard ) − 1] × 1000, where R = 13 C/ 12 C and 15 N/ 14 N, respectively. The isotope ratios were expressed in delta (δ) notation relative to Vienna-PDB limestone (V-PDB) for δ 13 C and atmospheric nitrogen (AIR) for δ 15 N. Replicate measurements of internal laboratory standards (acetanilide STD: Thermo Scientific PN 338 36700) in every batch (n = 14) indicated precision < 0.2‰ for both δ 13 C and δ 15 N values. Mean mass C:N ratio were 3.34 ± 0.03, 3.39 ± 0.03 and 3.49 ± 0.03 (mean ± SE) in R. palpebrosa, R. megaptera and R. moelleri, respectively, with no differences among species (H 2,133 = 21.54, p = 0.47).
Data analyses. Differences in δ 13 C and δ 15 N values, and TLs among species, sexes, geographic areas (i.e. West and East Greenland, the Barents and Kara Seas) and size groups (i.e. small, medium, large and very large) were assessed with a Kruskal-Wallis H or a Mann-Whitney U test 56 . A regression analysis was used to find equations fitting our data 56 . All tests were performed using α = 0.05. Neither ethanol nor formalin fixation significantly affects δ 13 C or δ 15 N signatures of cephalopod beaks 57 , thus no corrections were performed due to fixation. Values of δ 15 N in cephalopod beaks, in contrast to δ 13 C values, are in average 4.8‰ lower than values from muscle tissue 52,53,57,58 . Therefore, this value was subtracted from muscle δ 15 N values available in the literature to enable comparison with the data reported here. However, when estimating TL, we added 4.8‰ to raw beak δ 15 N values, as proposed by 41,42,52,54 .
Trophic level can be estimated with fixed trophic enrichment factor (TEF), either 'classical' δ 15 N = 3.4‰ 59 or ' Arctic' δ 15 N = 3.8‰ 60 , and with standard TL equation 61 , or with scaled TEF equation 62,63 , adapted for the Arctic by Linnebjerg et al. 64 . We used the latter as the most up-to-date approach. Reference values for TL = 2.0 were: Isotopic niche widths and overlap were assessed with SIBER 2.1.4 15 in R 3.6.3 68 . The standard ellipse area corrected for small sample sizes (SEA c , an ellipse that contains 40% of the data regardless of sample size) and the Layman metric of convex hull area (TA) were estimated [15][16][17] , and the Bayesian approximation of the standard ellipse area (SEA b ) was adopted to compare niche width among groups 15 . Large (n = 12) and very large (n = 6) specimens of R. moelleri were pooled in the same group (Table 3), due to the small sample size for isotopic niches' analyses 69 . The overlap interpretation followed Langton 70 , where overlap ranged from 0.0 to 0.29 indicating no overlap, from 0.30 to 0.60 indicating medium overlap, and from 0.61 to 1.00 indicating large overlap and the latter only taken as significant, i.e. potential competition.
Trophic levels were used instead of δ 15 N values (Y axis) in niche estimations. This approach improves the ecological meaning of isotopic data when comparing specimens from different areas and ecosystems due to differences in baseline δ 15 N values (e.g. 52,64 ). This approach has been repeatedly applied to cephalopods 41,54 .
The newest Bayesian mixing model, i.e. SIMMR 0.4.1 71 in R 3.6.3 68 was used to assess relative contribution of prey to the diet of Rossia. All three species were reported to eat crustaceans and fishes in Canada 48 . Stomach content analysis showed the main prey of R. palpebrosa in the Barents Sea are Crustacea, Polychaeta and fishes 41 , and these taxa were used as prey group sources in our models. The models were performed for the Barents Sea and West Greenland: mean source values are detailed in Table 4. All the source values were significantly different in at least one of the isotopes (Table 4). Values and standard deviations of TEF were taken from the only experimental study showing differences between cephalopod beaks and long-time diet composition 58 : δ 13 C = − 0.20 ± 0.55‰ and δ 15 N = 3.37 ± 0.99‰. The data fitting to selected prey source values and TEFs was checked using simulated mixing polygons 72 in R 3.6.3 68 (Supplementary Fig. S1). Only the fitting specimens were used in models (Table 4). Individual-based models were performed for all specimens fitting the model (Supplementary Fig. S2). Diet derived from the models was compared among species (overall models), sexes, geographic areas and size groups with χ 2 and Fisher's exact tests: although the latter is more adequate for small sample sizes, Fisher's exact allows comparison of only two groups 56 .
Statistical analyses were performed in R 3.6.3 68 and PAST 3.25 73 . Values are presented as mean ± SE unless otherwise stated.
Ethical approval. No ethical approval was required. Beaks were only obtained from either dead or preserved specimens. No live animals were caught specifically for this project.

Results
The known geographic ranges were expanded for R. megaptera and corrected for R. moelleri, and new maximum body sizes were recorded for all the studied species (Table 1) Table S5). The largest size group was the most different from the smallest and second-most from middle one, with no differences between the smallest and middle-sized groups ( Table 2, Supplementary Tables S2-S5).

Isotopic niches.
No differences in niche width were found between sexes in R. palpebrosa; both sexes showed a large overlap (Supplementary Table S8). However, females in R. megaptera and R. moelleri had significantly wider niche than males, with males having larger overlap with females (> 95%) than vice versa (52-55%): females had medium overlap with males (Supplementary Table S8). Significant ontogenetic decrease in niche width was found in R. moelleri, and gradual (not significant) ontogenetic decrease and increase in R. palpebrosa and R. megaptera (Fig. 1, Supplementary Table S9). Larger size groups overlapped more with smaller ones in R. palpebrosa and R. moelleri, with the opposite pattern in R. megaptera (Fig. 1, Supplementary Table S9). Large overlap was found between small and medium R. palpebrosa, and consequently in small-medium-large R. megaptera (Fig. 1, Supplementary Table S9).
Differences in niche width among species were found only in the Barents Sea (Fig. 1, Table 3). In the Barents Sea, R. megaptera had significantly narrower niche than R. palpebrosa and R. moelleri (Fig. 1, Table 3). Rossia moelleri had only small overlap with R. palpebrosa and R. megaptera in the Barents Sea, and no overlap with them   (Fig. 1, Table 3). Rossia palpebrosa and R. megaptera mostly had large overlap with each other, except for the Barents Sea, where R. palpebrosa had medium overlap with R. megaptera (Fig. 1, Table 3). Rossia palpebrosa overlapped more with R. megaptera, overall and in the Barents Sea, and the opposite in West Greenland (Fig. 1, Table 3).

Discussion
This study assessed a long time series during which the samples were collected in the Barents and Kara Seas (2003-2017). We assume the potential biases which can possibly arise have been countered: (a) changes in δ 13 C values due to oceanic Suess effect were minimal (− 0.018‰ 74 ) and already proven negligible in Arctic fishes and marine mammals 75 ; (b) to our knowledge there is a lack of long-term direct baseline variation studies in the Arctic, and the only available long-term studies for plankton and walruses Odobenus rosmarus showed no significant changes in δ 13 C and δ 15 N values over long time periods in high Arctic Canada 27,75 ; and (c) all specimens were collected in the same years and during July-August, minimizing seasonal changes. Cephalopod beaks have recently been proven to be 'chemical archives' of the individual's life [76][77][78] . The analysis of the whole beak can be thus a proxy of full ontogenesis of the specimen. Seasonal changes can be accessed either by analyzing different regions of the beaks synthesized during specific periods 77,78 or by equal sample distribution throughout the year; the 'whole-beaks approach' applied in this study is more general, and most likely the majority of the revealed relationships are for the whole life history of the animal. In some cases it is obvious how sympatric species decrease competition: e.g. when they demonstrate significant size, life style of habitat differences (e.g. 23,24,26,28-33 ). However, the three studied species of the genus Rossia Figure 2. Relative contribution of prey to the diet (mean, box 25% and 75% percentiles, whiskers 5% and 95% percentiles) of the studied species of the genus Rossia predicted by Bayesian mixing model SIMMR 0.4.1.

Scientific Reports
| (2020) 10:21506 | https://doi.org/10.1038/s41598-020-78645-z www.nature.com/scientificreports/ had largely similar body sizes, often occurred in the same trawl station, and were supposed to have similar hunting behavior, i.e. had no preliminary highlighting how they decrease competition. So, how do Rossia deal with potential competition? Using SIA and its applications to assess diet, life style and ontogeny, we were able to identify seven traits related to resource and habitat utilization in the three species of the genus Rossia: (1) R. moelleri had more pelagic life style, than initially supposed, while R. megaptera and R. palpebrosa had 'typical' life style for sepiolids; (2) R. megaptera and R. moelleri showed spatial migrations, while R. palpebrosa was presumably sedentary; (3) R. megaptera and R. moelleri had more pronounced sexual dimorphism in body size, and niche width in females was significantly larger than in males, suggesting asymmetrical competition, where large and very large females are in competitive advantage; (4) R. megaptera and R. moelleri showed a less varying diet between regions, than R. palpebrosa; (5) R. megaptera and R. palpebrosa had crustaceans as their main prey, while fishes dominated in R. moelleri; (6) R. palpebrosa and R. moelleri had ontogenetic decrease in isotopic niche width (common for cephalopods), while R. megaptera showed ontogenetic increase; and (7) R. palpebrosa and R. moelleri showed similar strategies to reduce intraspecific competition, different from R. megaptera: asymmetrical competition favours smaller-sized groups in the both former species and all stages are largely overlapping, while larger-sized groups are favoured in R. megaptera. No trait was shared by all three species, and high boreal R. megaptera and Arctic endemic R. moelleri shared three traits with each other, while both R. megaptera and R. moelleri shared only two unique traits each with widespread boreal-Arctic R. palpebrosa. Thus all traits formed fully uncrossing pattern with each species having unique strategy of resource and habitat utilization. How the diet specialization and its ontogenetic changes are a means to reducing competition? These species of the genus Rossia belong to Arctic nekto-benthic predators' trophic guild, which includes large shrimps and fishes. However, shrimps and fishes present a wider diet spectrum (often scavenge) and thus a wide range of both δ 13 C and δ 15 N values 60,64,66,67,75,79 . Westward significant increase of δ 13 C values, which is usually found in different taxa from the Arctic marine ecosystems 41,42,60,64,65,67,79 , was found in R. palpebrosa and R. moelleri, and lacked in R. megaptera. Significantly higher δ 15 N values and TLs in R. moelleri than in R. palpebrosa and R. megaptera suggested marked dietary differences, which were also highlighted by SIMMR: crustaceans were the most important group in diet of R. palpebrosa and R. megaptera, and fishes in R. moelleri. Rossia moelleri had the most different diet among Rossia, and is the only sepiolid in the world ocean whose main prey are fishes (reviews 36,44 ). Rossia palpebrosa had more varying diet between the studied areas than R. megaptera.
In general, all three species had lower δ 15 N values and TLs than North Atlantic squids, and similar or higher than octopods, cuttlefishes and sepiolids 58,76,[80][81][82] . Ontogenetic increase of δ 15 N values and TLs was significant in all three species of Rossia, with a higher steep increase in R. moelleri, followed by R. palpebrosa and R. megaptera. Generally ontogenetic increase in Rossia was lower than in squids 42,53,76,77,83,84 , but similar, or more pronounced, than in octopods 32,53,78 .
Ontogenetic isotopic niche decrease is common in cephalopods, including R. palpebrosa 32,33,41,42,84 and R. moelleri. On the other hand, R. megaptera demonstrated ontogenetic niche increase, similar to Vampyroteuthis infernalis, a deep-sea cephalopod with unique diet and life style 54 , but this is rarely found in 'typical' predatory cephalopods 32,83 . Within the Arctic, isotopic niches of all Rossia were narrower than of squid Gonatus fabricii (which was the widest among Arctic invertebrates 42 ) and of shrimp Pandalus borealis and fishes due to their higher degree of opportunism in diet 60,[64][65][66][67] .
How the life style is a means to reducing competition? Rossia beaks had high range of differences in δ 13 C values (4.1-4.6‰; Table 2), as was previously found in polar squids 42,53,77,85 , compared to warm-water ones 58,76,[81][82][83][84]86 . Contrary to majority of the studied squids and octopods with ontogenetic increase of δ 13 C values 32,33,42,53,76,84 , δ 13 C values remain the same throughout the ontogenesis in R. palpebrosa, suggesting it does not migrate during ontogenesis. On the other hand, δ 13 C values decreased in R. megaptera and R. moelleri suggesting they migrate during ontogenesis, despite a nekto-benthic life style. Significantly higher TLs in the Barents Sea than in East Greenland (R. megaptera) and in the Kara Sea (R. moelleri) further suggest these species migrate during ontogenesis: their diets were less varying between regions, than in R. palpebrosa. Differences in TLs among regions were not found in other studied Arctic cephalopods 41,42 .
As nekto-benthic species, Arctic sepiolids were supposed to have higher δ 13 C values than pelagic Arctic squid. However, Rossia moelleri, the shallowest living species, had δ 13 C values similar to the Arctic squid G. fabricii 42 , and significantly lower than R. papebrosa and R. megaptera, suggesting a different, relatively more pelagic life style.
Differences in the widths of isotopic niches between sexes were found in R. megaptera and R. moelleri: females had large niches, and niches of males were almost completely within the isotopic niche of females. However, and in accordance with Golikov et al. 41 , no differences were found in isotopic niche widths of R. palpebrosa between sexes. Rossia megaptera and R. moelleri, and squid species which demonstrated the same pattern of niche differences 83,86 all had more pronounced sexual differences in body sizes, than R. palpebrosa. However, niche overlap between sexes was decreasing during ontogenesis in squids 83,86 , unlike in the studied species of the genus Rossia.
Our data suggest that predicted climate changes in the Arctic would: (1) not significantly change the situation for R. moelleri, even if its range decreases due to its Arctic affiliation; (2) create more favourable conditions for niche width increase in R. megaptera in the Barents Sea, where it is currently in disadvantage, inhabiting only the warmer, western part, and strengthen its advantage in West Greenland; (3) not significantly decrease abundance of R. palpebrosa due to its plasticity, as this is the most widespread Rossia in the Arctic, which has the most varying diet and the widest habitable diapason of temperatures.

Conclusion
Three sympatric species of cephalopods of the genus Rossia (widespread boreal Arctic R. palpebrosa, high boreal R. megaptera and Arctic endemic R. moelleri) with seemingly similar sizes and hunting behaviour, which live together to a degree they can be sampled all together in one trawl catch, were found to have seven traits related to resource and habitat utilization: no trait was shared by all three species, and high boreal R. megaptera and Arctic endemic R. moelleri shared three traits with each other, while both R. megaptera and R. moelleri shared only two unique traits each with widespread boreal-Arctic R. palpebrosa. No crossing pattern was formed from traits with each species having unique strategy of resource and habitat utilization. Such a fine level of competition-avoidance is not easily detected, these traits were only highlighted by SIA and its applications when applied to the sample including all ontogenetic stages and both sexes in largely equal ratio and missed by 'classical' methods, such as e.g. stomach contents or distributional analyses. Further SIA studies of sympatric species based on all-ontogenetic samples with equal sex ratio are recommended to increase our understanding of inter-and intraspecific competition, and thus complex trophic webs under natural conditions.