Response: Demography affects spawning location in Northeast Arctic cod, but what affects demography?

Uni Research and Hjort Centre for Marine Ecosystem Dynamics, P.O.Box 7810, 5020 Bergen, NorwayIn a letter to the editor, Sundby (2015) expresses con-cerns about our study of changing spawning locationsof the Northeast Arctic (NEA) stock of Atlantic codover the period 1866–1969, where we identified statisti-cally significant effects of the stock’s demographywhereas various climate indices all fell below statisticalsignificance. Our conclusion on the role ascribed to cli-mate disagrees with that of Sundby & Nakken (2008),which, based on a subset of our data and without con-sidering demography, concluded that spawning wasshifted northwards in warm periods.Sundby’s (2015) criticism boils down to four pointsthat we will address in turn.A: ‘the calculated spawning migration distance (...)is based on incorrect assumptions’ (quote fromabstract, covered by points 1–3 in Sundby’s letter)In our analysis, we never did calculate spawningmigration distance because the data did not allow it.From the data, a time-series of regional statistics of thecod spawner fishery, we could only quantify spawninglocation. We thus only know the endpoint of the spawn-ing migration, and as there are no data on where in theBarents Sea migration began we could not, and did not,calculate migration distance. It is unclear to us how thismisreading of our paper originated. In fact, we explic-itly raise this issue, for example in this quote (Opdal &Jorgensen, 2015, page 1527):It is also known that younger cod tend to be dis-tributed in colder water further north and east inthe Barents Sea than older, larger cod (Ottersenet al., 1998). This has the implication that one can-not infer from spawning location how far an indi-vidual has migrated to get there.Instead of using metrics of latitude and longitude, wequantified spawning location as distance from a centralpoint in the Barents Sea. This is only a technical issue ofpresentation and does not make assumptions or infer-ences about migration distance.B. An alternative hypothesis is that feedingdistribution in the Barents Sea explains spawninglocation (point 3 in Sundby’s letter)Sundby (2015) proposes the hypothesis ‘that the spawn-ing migration distance is constant, since the distribu-tions during the feeding in the Barents Sea and at thespawning areas are fluctuating with the similar ampli-tude (i.e. 250-300 km) on decadal scale’ (see fig. 1 inSundby, 2015). He also notes that ‘data is not availableto test this’. However, Sundby (2015) also refers to sev-eral studies showing that feeding cod move furthernorth and east in the Barents Sea during warm years(e.g. Kjesbu et al., 2014). With temperature-dependentfeeding location and constant migration distance, a pre-diction from his hypothesis is thus that spawning loca-tion would shift north in warm years, which is exactlythe effect that our statistical analysis finds no supportfor.C. We are using irrelevant climate indices (point 1in Sundby’s letter)Sundby stresses that the North Atlantic Oscillation(NAO) is an atmospheric pressure index and not a mea-sure of water temperature, which is what likely wouldhave the strongest effect on fish. We are of courseaware of this, and included NAO only because directobservations of sea temperature (the Kola section) wereinitiated in 1900 and therefore not available for the first34 years of our data. Neither NAO nor Kola sea tem-perature significantly explained variations in spawninglocation in our study. The use of NAO is not completelyunwarranted, however, as NAO has been found to cor-relate with oceanography and ecology of cod in theBarents Sea (Ottersen & Stenseth, 2001). More gener-ally, pressure indices describe large-scale physical driv-ers, with influences on the dynamics of both terrestrialand marine ecosystems (Stenseth et al., 2002).