The results show an overall picture that there are positive associations between oceanographic features (SPG and MEIW), via amount of zooplankton and forage fish to seabirds and recruitment of demersal fish. A large subpolar gyre and/​or a strong East Icelandic Current has been, therefore, beneficial through out the foodweb up to higher trophic levels. Interestingly, a combination of SPG and MEIW seemed to show a closer association with the higher trophic levels than each of them (SPG or MEIW) separately. There is probably a lag of one to three years between SPG/​MEIW and higher trophic levels.

Our results show that the occurrence of a large subpolar gyre, i.e., cool water south of Iceland and a strong East Icelandic Current (EIC) coincided with lower temperatures in NE Atlantic. The fact that a combined index of SPG and EIC seemed to be more closely associated with higher trophic levels indicates that SPG as well as EIC might act as two separate sources of nutrients/​zooplankton to the three shelves. It should be stressed that this only is meant as an inspiration for future investigations. A lag of at least one year is expected if water particles originating in the SPG are physically transplanted to the three shelves (Koul et al. 2022; Orvik 2003). A similar lag of one year is expected for a water particle to move from the core area north-east of Iceland with high concentrations of zooplankton (Kristiansen et al. 2019) to the Faroese and Norwegian shelves (Skagseth et al. 2022). A lag of two to three years between SPG/​EIC and higher trophic levels, which probably is best supported by the data (Table 1.1), indicates that nutrients are main causative agents, i.e., there is lag of at least one year for the nutrients to reach the shelves, then a lag of one year before they are expressed in the spring blooms on the shelves (Hátún et al. 2022a).

As with nutrients it seems straightforeward to interpret the increased amounts of zooplankton adjacent to the Faroese or Norwegian shelves as a result of an advection of zooplankton-rich water from the area north of Iceland. However, there appears to be a strong correlation between the total amount of zooplankton in the Norwegian basin (ICES 2024d) and the amount found close to the Faroese and Norwegian shelves. The fact that the combined index of SPG and MEIW showed closer associations with higher trophic levels indicates that zooplankton might be transferred from the SPG, in addition to EIC, to the three shelves.

By being present outside the three shelves does not guarantee that the zooplankton or nutrients will become available to the shelf ecosystems, but this apparently happens in such a degree that there are clear associations between zooplankton and forage fish or demersal fish recruitment (Table 1.1). An indication of advection of zooplankton and/​or nutrients onto the shelves may come from the growth of the long-lived clam Arctica islandica that feeds on phytoplankton that enters the bottom, although the local primary production might also affect the growth (Matras et al. 2022). In this case we have used the difference between deep and shallow shells as the measure of benthic productivity (thus cancelling potential temperature effects or other unknown effects that may be associated with the location rather than advection). The actual advection process of nutrients and zooplankton onto the shelves is a field of its own and far outside the scope of this paper.

The advection of nutrients or zooplankton onto the shelves may be spread to different kinds of forage fish such as fish larvae/​juveniles (Jacobsen et al. 2019), sandeels (Greenstreet et al. 2006), Norway pout Trisopterus esmarkii and age 0 herring (Anker-Nilssen 1992). These fish species might however react differently to SPG, EIC or temperature. For example, biomasses of adult Norway pout on Faroe Shelf tend to be negatively correlated with sandeel abundance (FaMRI, unpublished material). Also, herring recruitment is positively correlated with temperature (Toresen and Østvedt, 2000), i.e., the opposite of sandeels. Hence, 0-group of forage fish (capelin, sandeels, herring) might be somewhat heterogenious but apparently show as strong association with SPG/​MEIW as do zooplankton or seabird production. We are not able to tell whether the apparent time lag of one to three years between the Icelandic and Faroe/​Norwegian shelves with regards to forage fish is real or just an artifact (i.e. only based upon peaks that may be caused by unknown random processes) but this could be substantiated by other evidence.

Of particular importance is the timing of events and top-down effects. On the Faroe shelf the abundance of fish larvae is crucial to the zooplankton. A large number of fish larvae in May, as a consequence of high primary production, may excert such a grazing pressure on zooplankton that their abundance is low during summer (June–July) and vice versa: a low amount of fish larvae causes a high abundance of zooplankton during summer (Jacobsen et al. 2019). This apparently leads to a “compensatory” food chain that is based on zooplankton during summer and associated forage fish, which in this case is adult Norway pout. In our project we have not included such “compensatory” food chains, but this is expected to affect both demersal fish and seabirds since they prey on Norway pout (Steingrund et al. 2024).

As was the case with forage fish the assemblage of demersal fish might also be somewhat heterogenious. In our case we selected cod and haddock recruitment on the Icelandic and Faroese shelves and Norwegian coastal cod as our production indices. As described above Icelandic and Faroese cod and will most likely not have the same reaction to SPG/​MEIW. In addition, recruitment of Icelandic haddock increases with increased temperature (Figure 1.2). This might be attributed to large areas north of Iceland becoming habitable by haddock when temperature increases. This does not apply for the Faroese or Norwegian shelves. Icelandic cod is probably less affected by this mechanism, but its main nursery areas are in all years north of Iceland. Norwegian coastal cod (north of 67^oN) might be composed by different local populations that may make this production index heterogenious and influenced by local conditions. It is also important to note that there is a local phytoplankton production that, in addition to advected nutrients/​zooplankton, forms the basis for the upper trophic levels (Steingrund & Gaard 2005). Hence, it is not so surprising that the associations between SPG/​MEIW and demersal fish recruitment were weaker than for the lower trophic levels.

Although not synchronous at a year-to-year level, the large-scale pattern of trends in productivity of seabirds breeding at the three Nordic shelves was relatively similar. An extraordinary long period of poor productivity started in the early 2000s and lasted for a decade before showing signs of improvementafter 2015. Before this, multi-year peaks in productivity were observed in all areas, especially in the 1990s (Figure 1.2).

Compared with forage fish or demersal fish production, seabird (puffin) production showed the strongest association with SPG/​MEIW, especially with the combined SPG+MEIW index (Table 1.1). To some extent, this might be because the flight abilities enable seabirds to exploit a wider range of prey resources over larger areas than the less mobile forage fish and demersal fish. The zooplankton or nutrients that are either produced locally or advected onto the shelves may benefit different forage fish, depending on location. Hence, one should not always expect strong linkages between seabird productivity or diets and local forage fish species although several such relationsships have been documented, including the Røst puffins’ strong dependency on age 0 herring to breed successfully (e.g., Anker-Nilssen 1992; Sætre et al. 2002; Durant et al. 2003; Walnum 2024) and the importance of young saithe Pollachius virens for the breeding performance and diet of European shags Gulosus aristotelis (Bustnes et al. 2013; Lorentsen et al. 2018).

In Southwest Iceland, Hansen et al. (2021) showed that the annual average sea temperature explained as much as 74% of the variation in the polenetting harvest of immature puffins in Vestmannaeyar over a 130 year period since 1880. This is one of very few studies documenting the importance of climate-induced effects on the producticity of a marine top predator over such a long term. The cyclic nature of this relationship follows the Atlantic Multidecadal Oscillation (AMO) pattern and is furthermore strongly influenced by the SPG control of the inflow of warmer, less nutrient-rich Atlantic Water of tropical origin (Hátún et al. 2005). As the hunters always have avoided capturing birds carrying food loads, and immature higher air attendance than breeders in the colonies, make the harvest highly age-selective, with 75% of the birds taken belonging to three cohorts of immature birds (age 2–4) that often visit the colonies in summer. This allowed the calculation of a relative cohort stength index lagged by -3 years. As immature birds are known to move widely between regions within Iceland (and to the Faroes, Hammer et al. 2014), the index likely reflects the larger-scale production of puffins in Iceland.  

For a better representation of the Selvogsbanki shelf, the local production index based on the number of puffin chicks reported in the town of Vestmannaeyjar in 2007–2024 was proven to be highly correlated (r=0.75) with the parallel measurements of breeding success in the colony on the same island, demonstrating its value as a productivity signal. The temporal pattern in this data series since 1971 can be split into four dissimilar periods: (1) 1971–1986 had a highly variable success, but the low production years (especially 1978, 1982–1984) were considerably larger in absolute magnitude than the best years, (2) 1987–1996 was characterised by many good years, peaking in 1991–1993, (3) 1997–2014 showed a decline in productivity, followed by a decade of poor production in 2005–2014 with total breeding failure or very low breeding success in most years (parallelled by an estimated 56% drop in breeding numbers (E.S. Hansen, unpublished)), and (4) 2015–2024 when production rised to a high level, peaking in 2021 and 2024. Period (3) was concurrent with a rapid warming of the Icelandic shelf and also accompanied by a delay in puffin breeding of an unpresidented scale (18 days, or 1.5 SD higher than the long-term average in 1937–2024 (E.S. Hansen, unpublised)). This pattern strongly suggests a temporal mismatch in the food chain, and is also corroborated by a significant delay in phytoplankton bloom timing (Pétursdóttir et al. 2021), and virtual disappearance of zooplankton peaks (save 2014).

In the Faroes, kittiwake success was especially high in 1982, 1995, 2001 and 2017, whereas the success of puffins on Mykines improved after a total breeding failure in 2011–2013 and was relatively high in four of seven years from 2017. It should be noted that all harvest data from the Faroes were provided by local communities. Interpretations of such data can be a bit challenging since data quality can be affected by e.g. local decisions. The low values from 1989 to 1993 at Stakkurin on Streymoy are also affected by a hunting ban during those years and thus not only reflecting bad environmental conditions. At the other end of the scale, harvest numbers at Seyðtorvu on Viðoy from 2006 onwards are inflated since, from that year, the harvest reports also include numbers from an additinal area, which means the results overestimate the true conditions. Nevertheless, there seems to be a clear general pattern, especially during the period between 2005 and 2013 with relatively low harvest mumbers and poor breeding success of both species. This matches quite well with the development in the SPG and MEIW, as also demonstrated by Hátún et al. (2017b).

In Norway, above average success was observed for both study species in Røst in 1983–1985, in 1989-1992 and in most years between 1999–2006, but for neither species this was sufficient to maintain their populations. In 16 consequtive years (2007–2015) the Puffins experienced virtually total breeding failures at the population level (mean 0.1 chicks/​pair), which succeeded a long-lasting history of reproductive problems for this population (e.g., Anker-Nilssen & Røstad 1993; Anker-Nilssen & Aarvak 2006; Cury et al. 2011) with an average yearly breeding success of only 0.3 chicks/​pair from 1964 to date. This is far below the annual rate needed to sustain the population with sufficient self-recruitment (about 0.5 chicks/​pair; Anker-Nilssen & Aarvak 2006). As a direct consequence of failed recruitment, the breeding population has dropped by 86% from more than 1.4 million pairs in 1979 to less than 0.2 million pairs at present (Anker-Nilssen & Aarvak 2006; T. Anker-Nilssen, unpublished data). The same is the case for the Røst kittiwakes. Their average success after 2006 has been less than a third of the level needed for self-sustained level for that species (> 1 chick/pair, Frederiksen et al. 2024). Over the last four decades they have only surpassed that level twice (in 1992 and 2002) and the average productivity of the kittiwakes breeding in Røst harbour has been approximately half of what’s required to maintain the population (Figure 1.2). The largest colony on natural cliffs (Vedøy) was also harassed by white-tailed eagles Haliaeetus albicilla, which reduced the kittiwakes’ productivity further, and the colony went extinct in 2020 (Anker-Nilssen et al. 2024). As a total consquence of failed recruitment, breeding numbers of kittiwakes in Røst dropped by 97% from 25,000 pairs in 1980 to only 692 pairs remaining in 2024 (T. Anker-Nilssen, unpublished data).

Common considerations across all areas. The effects of white-taile eagles on kittiwake breeding success (Anker-Nilssen et al. 2023) exemplifies how top-down effects act to reduce the statistical relationship between marine species and their key prey. Given the high nest fidelity of most seabirds between years, insufficient recruitment rates will tend to reduce the breeding density within the colony. This will in turn lower the adults’ collective defence against predators with negative consequences for both breeding success and survival rate. Furthermore, as also shown for puffins (Fayet et al. 2021), the time spent away from the colony in search of food increases with decreasing prey availability, reducing the adults’ ability to defend their offspring even further. For open-nesting species such as kittiwakes, this helps explain why populations breeding in/near human settlements (and even on offshore oil rigs), where such predation pressure is much lower, have higher productivity and better population trends than nearby colonies on natural cliffs (e.g., Christensen-Dalsgaard et al. 2019; Anker-Nilssen et al. 2023).

In addition to such density-dependent effects of predation, the relationship between breeding success and prey availability is rarely linear but more often shows a threshold-shaped response because the life-history strategy of seabirds favours a dynamic trade-off between the individuals’ investment in reproduction vs. own survival. When food availability drops below the threshold, conditions quickly get inadequate for raising offspring. Given the long life expectancy of adults, it is then more profitable for them to abandon the breeding to maximise their own survival until the next season. When surpassing the threshold, conditions are often sufficient to secure breeding success for most individuals. Consequently, the effect of changes in prey abundance is strongest around the threshold and weak at other levels of prey abundance. Also, the extreme population declines registered for some of these seabird populations would lead to less intra- and interspecific competition for food and thereby also reduce the statistical relationships with the abundance of staple fish prey and zooplankton species. Top-down effects of seabird consumption on their key prey stocks seem to be important only when prey abundance is very low and has not yet been shown to correlate with prey dynamics (Saraux et al. 2021). When prey is plentiful, the spatial correlation between the distribution of prey patches and seabirds is also shown to be reduced (e.g., Axelsen et al. 2001; Fauchald et al. 2011).

The life-history balance between these vital rates may also vary between populations of the same species, with some showing higher survival rates and lower productivity rates than others (e.g., Frederiksen et al. 2005). Data series on survival rates are not available for puffins and kittiwakes in Iceland and the Faroes, making it difficult to calculate more exactly the short- and long-term effects of the observed variability in lower trophic levels and oceanography. Nevertheless, Harris et al. (2005) found no differences in survival rates of puffins across five colonies in the Northeast Atlantic, suggesting the levels of breeding success needed to sustain the populations based on natal recruitment alone are relatively similar for populations breeding in the Nordic countries. Even if the contribution of breeding success to the observed population trends varies between colonies (Layton-Matthews et al. 2023), the larger-scale decrease in breeding numbers of these species over the last decades were mainly caused by poor breeding success, indicating also that recruitment from other breeding areas cannot be expected to buffer the negative trends.

Temperature responses in fish and zooplankton may vary by life phases (Beaugrand et al. 2002; Barbeaux & Hollowed 2018; Kjesbu et al. 2022), and the same species can react differently across life stages due to ecological differences, leading to conflicting predictions for stock development under increasing temperatures. Thus, temperature might be a confounding factor since high temperatures might cause higher basic metabolim in zooplankton and forage fish (MacDonald et al. 2018). On the other hand, in our study low temperatures might also signal ecosystem productivity by showing how much water from the subpolar gyre or EIC is present in the vicinity of the three Nordic shelves. We have not attemped to separate these factors.

Pelagic fish graze on zooplankton (Bachiller 2018) and this could potentially confound our results. We compiled a ‘zooplankton grazing biomass of pelagic fish’ by adding the Norwegian spring spawning herring biomass of age 3+ together with the mackerel biomass of age 0+ and the biomass of age 1 blue whiting. We noted that the correlation between pelagic fish and temperature was positive and tended to be negative with forage fish and demersal fish recruitment. However, the zooplanktonvorous amphipod Themisto in the Norwegian Sea seems to be a more important predator on Calanus finmarchicus than pelagic fish (Skjoldal et al. 2004). This indicates that grazing by pelagic fish might not determine the fate of zooplankton that enters the three shelves, but may, neverthless, play a role.

Our approach has been quite simple by constructing common time series of zooplankton and forage fish that were regarded as representative for the Nordic shelves. We also compared demersal fish recruitment and seabird productivity among the Nordic shelves. To keep the approach as simple as possible, we avoided to analyse in any detail the time lags involved in the mechanistic links between different variables, even though such lags are apparent between some of the variables (Table 1.1, Figure A1.1). For example, many of the ecological effects associated with SPG/​MEIW seem to be lagged by 1–3 years, depending on the trophic levels and shelf areas involved. In general,  there is an urgent need to explore in more detail the spatial and temporal differences  in the functioning of the Nordic shelf ecosystems to better understand (and be able to account for) the mechanisms that drives the productivity of these important hotpots of marine life and human exploiation of natural resources. Since the sum of normalised SPG and MEIW indices was more closely associated with productivity of the trophic levels than each index alone, this indicates that both of them may act as sources of nutrients/​zooplankton to the Nordic shelves, but the oceanographic foundation for this needs to be investigated.

In this study we did not include any calculations of fluxes of zooplankton or nutrients onto the shelves. Neither did we include any estimate of the consumption of zooplankton by forage fish or the consumption by demersal fish or seabirds of forage fish as this would be far outside the scope of this project. When such trophic models hopefully are further developed in the future, the results from our project will, nevertheless, be of value by indicating some of the main ecological pathways.

Our results do not adress local effects on the Nordic shelves, such as the importance of the local, on-shelf primary production. This is basically because we first and foremost wanted to explore the larger-scale dynamics of the productivity in Nordic waters, which we expected to be reflected best by offshore features. Still, we do know that an inclusion of parameters of on-shelf productivity would be needed to reflect a wider range of mechanisms, for example the high amount of sandeels at the Faroes in 2017 due to a high primary production there in that year that was not seen on the Icelandic shelf.