Juvenile Atlantic Cod
At a hazard rate of λ=10, seven regimes of cod catch rate were detected from 1919 to 2014 (Fig. 2a). Despite some differences in the timing of the shifts, they, as do all outputs from λ=10 to 50, illustrate patterns of substantive decline. However, there is reason to believe that a value of 10 for the hazard rate yields regime shifts early in the data series that should be interpreted with caution. The first thirty years are characterized by highly variable catch rates. Data from the late 1930s to the mid-1940s seem particularly suspect, given that catch rates between successive years would normally be expected to be positively (not negatively) autocorrelated (the juvenile cod group includes primarily ages 0+ and 1+ yr). The 1939-1945 period coincided with WWII during which fisheries research was greatly curtailed. We note that other researchers (e.g., Fromentin et al. 1998) excluded this time period from their analyses of the same beach-seine survey data because sampling stations were few in number. Although we do not wish to exclude data, we are disinclined to accept these unusual patterns of data variability from the 1920s to the mid 1940s as reflecting valid regime shifts (recall that the BOCPD algorithm recognizes changes in variability, independent of the mean, as a basis for a regime shift).
As with the NAO index (Figs. 1b,c), λ values of 20 and 25 detected identical regimes (Figs. 2b,c) for the time-series of cod catch rate. Four can be identified. The first two (1919-1952 and 1953-1974) differ only in their variability, the former being more variable than the latter. Catch rates for both regimes were ~18.5 cod per seine haul. However, catch rates during the third (1975-1998) and fourth regimes (1999-2014) averaged ~10 and ~5 cod per haul, respectively. Comparing catch rates in the most recent regime with those in the first two regimes, juvenile cod are estimated to have declined more than 70% since the early 20thCentury.