Fig 1.
NAFO areas and named features.
Designations of the NAFO zones and sub-zones on and around the Scotian shelf; 4X, 4W and 4V. The area within the contour lines marks the approximates the Scotian Shelf.
Fig 2.
The triangulation utilized for the SPDE approach with n = 548 vertices. Red dots are the data locations.
Table 1.
DIC values for the various candidate models with an AR(1) spatio-temporal covariance structure.
Fig 3.
Posterior distributions for the model parameters.
Fig 4.
Plot of the posterior mean for the years 1986 and 1992 (pre and post collapse). The lower panels show the corresponding plots of uncertainty (the response SD). The scale is log(predicted count).
Fig 5.
Relative exploitation and temperature effects.
The covariate plots show the value of the covariate on the X axis, and the impact of the covariate on the response on the Y axis and their 95% credible intervals (the dashed lines). Viewed as functions of the covariates, lower Relative exploitation and colder temperatures produce higher predicted abundance.
Fig 6.
The maximum observed count (LogY), the maximum predicted count (eta) are nearly constant, even though the stock is collapsing.
Fig 7.
Plot of the posterior mean for the years 1986 and 1992, showing only those areas where the predicted mean is > 75th percentile. Viewed as ‘core range’. The species range has contracted with the reduced abundance but maximum density in the aggregations has not changed, scale is log(abundance).
Fig 8.
Plot of the posterior mean for the years 1994 through 2000, showing only those areas where the predicted mean is > 75th percentile. Viewed as ‘core range’. The species range fails to recover abundance and the habitat remains unchanging.
Fig 9.
What the parameters were observed to do. The coloured arrows correspond to the years discussed in Figs 10 and 11.
Fig 10.
Plot of the posterior mean for the years 1973 through 1975, showing a dramatic flattening of the RF. This is indicative of circumstance in which heaving fishing was eroding the structure of the cod distribution leading to a partial collapse. In this period, ρ = 3.1 → ρ = 2.61 → ρ = 7.50 σ = 3.7 → σ = 2.8 → σ = 2.2. A simultaneous large increase in ρ and decrease in σ.
Fig 11.
Plot of the posterior mean for the years 1988 through 1990, showing a dramatic flattening of the RF. This is indicative of circumstance in which heavy fishing was effectively destroying the structure of the cod distribution leading to near total collapse. In this period the flattening is more severe than what was seen in the 1970s, ρ = 4.40 → ρ = 26.43 → ρ = 28.6 σ = 4.2 → σ = 02.1 → σ = 1.6. A simultaneous very large increase in ρ and decrease in σ. the collapse of the early 1990s is much more pronounced.
Fig 12.
Schematic of parameter behaviour.
A schematic picture of expected parameter values under different conditions in the fishery. Note that large ρ values are at the bottom of the schematic.