Chapter 17 . Classification of Fish Stock-Recruitment Relationships 345
Smirnov goodness-of-fit test. For this data, there is no violation for the
assumptions. Therefore the bootstrap residuals procedure in Section (17.3.2) is
legitimate to carry out for N = 1000 times. The bootstrap sampling distributions
for the Fuzzy- SR parameters: a J , a z ' b J and b z are illustrated in Fig. 17.4 (the first
two rows in Fig. 17.4). These bootstrap sampies can be readily used to obtain the
uncertainty estimate, such as confidence intervals and standard errors (Table
17.3). In Fig. 17.4, the 95% confidence intervals are marked by the horizontal
lines with open arrows and the sampie standard errors for the fuzzy parameters are
also listed in Table 17.2. It can be seen from Table 17.3 and also Fig. 17.4 that all
the model parameters are statistically significant.
Furthermore these boots trap sampies can be used to test the significance of the
environmental impact on SR relationships from different environmental regimes.
This is carried out by testing the difference of the Ricker a and b between the
"Warm" and "Cool" regimes. The last row in Fig. 17.4 is the distributions for a J -
a 2 and bJ-b Z marked with the 95% confidence intervals, which are (0.034, l.62) and
(-0.045, -0.004), respectively. Since both intervals do not cover zero, then the
differences for aJ-aZ and bJ-b z are statistically significantly different. Specifically
WCVI herring is more productive and less density-dependent in "Cool" regime
than in "Warm" regime (a J is significantly larger than a 2 , and b) is significantly
less than b 2 ).
17.4.2
Southeast Alaska Pink Salmon
17.4.2.1
Data Description and Preliminary Analysis
Detailed data descriptions and preliminary analyses for Southeast Alaska (SEAK)
pink salmon (Oncorhynchus gorbuscha) can be found from Quinn and Deriso
(1999, p104-123). The SR time series is reproduced in Figure 17.5. To account
for some of the unexplained variation in recruitment, Quinn and Deriso introduced
an environmental factor: average annual sea surface temperature (SST) off Sitka,
Alaska into the analysis. They found that a Ricker climatic SR model produced a
statistically significant fit to the recruitment time series.
In order to determine the impact of different environmental regimes, these SR data
are sorted into two subclasses: year- classes born in years of above average SST,
and year-classes born in years of below average SST (Fig. 17.5a). The Crisp-SR
approach is then fitted to these two data sets. It is found that there exist two
different SR relationships with the stock productivity parameter for "Warm Years"
l.49 and "Cool Years" 0.57. Also the Crisp-SR model fits better than the RickerSR model, wh ich is concluded from a decrease in the RMSE and an increase in the
correlation coefficient (r) (Table 17.2).
Smirnov goodness-of-fit test. For this data, there is no violation for the
assumptions. Therefore the bootstrap residuals procedure in Section (17.3.2) is
legitimate to carry out for N = 1000 times. The bootstrap sampling distributions
for the Fuzzy- SR parameters: a J , a z ' b J and b z are illustrated in Fig. 17.4 (the first
two rows in Fig. 17.4). These bootstrap sampies can be readily used to obtain the
uncertainty estimate, such as confidence intervals and standard errors (Table
17.3). In Fig. 17.4, the 95% confidence intervals are marked by the horizontal
lines with open arrows and the sampie standard errors for the fuzzy parameters are
also listed in Table 17.2. It can be seen from Table 17.3 and also Fig. 17.4 that all
the model parameters are statistically significant.
Furthermore these boots trap sampies can be used to test the significance of the
environmental impact on SR relationships from different environmental regimes.
This is carried out by testing the difference of the Ricker a and b between the
"Warm" and "Cool" regimes. The last row in Fig. 17.4 is the distributions for a J -
a 2 and bJ-b Z marked with the 95% confidence intervals, which are (0.034, l.62) and
(-0.045, -0.004), respectively. Since both intervals do not cover zero, then the
differences for aJ-aZ and bJ-b z are statistically significantly different. Specifically
WCVI herring is more productive and less density-dependent in "Cool" regime
than in "Warm" regime (a J is significantly larger than a 2 , and b) is significantly
less than b 2 ).
17.4.2
Southeast Alaska Pink Salmon
17.4.2.1
Data Description and Preliminary Analysis
Detailed data descriptions and preliminary analyses for Southeast Alaska (SEAK)
pink salmon (Oncorhynchus gorbuscha) can be found from Quinn and Deriso
(1999, p104-123). The SR time series is reproduced in Figure 17.5. To account
for some of the unexplained variation in recruitment, Quinn and Deriso introduced
an environmental factor: average annual sea surface temperature (SST) off Sitka,
Alaska into the analysis. They found that a Ricker climatic SR model produced a
statistically significant fit to the recruitment time series.
In order to determine the impact of different environmental regimes, these SR data
are sorted into two subclasses: year- classes born in years of above average SST,
and year-classes born in years of below average SST (Fig. 17.5a). The Crisp-SR
approach is then fitted to these two data sets. It is found that there exist two
different SR relationships with the stock productivity parameter for "Warm Years"
l.49 and "Cool Years" 0.57. Also the Crisp-SR model fits better than the RickerSR model, wh ich is concluded from a decrease in the RMSE and an increase in the
correlation coefficient (r) (Table 17.2).