Munyandorero: Climate effects on Micropogonias undulcitus 
61 
Table 2 
Results of a comparison of Bayesian state-space biomass dynamic models based on the deviance information criterion (DIC) 
and Bayes factor (BF) or log Bayes factor (LBF): without minimum winter estuarine temperature, MWET (base model 1, Ml), 
and with MWET (base model 2, M2) for the Atlantic Croaker population off the U.S. Atlantic coast, 1972-2008. Sensitivities 
to Ml involving the prior t/(0,01,1.5) for the intrinsic rate of population increase r and inclusion of the southeastern (North 
Carolina-east Florida) shrimp trawl fishery (SESTF) bycatch led to models MlrU and M1B, respectively. Sensitivities to 
M2 involving the prior 17(0,01,1.5) for the scaling parameter ro and inclusion of the SESTF bycatch led to models M2rU 
and M2B. M2N, M2rUN, and M2BN are M2, M2rU, and M2B in which a diffuse normal prior centered on zero, with a pre- 
cision of 0.02, was used as sensitivity to the prior for the coefficient controlling the effect of MWET a. D and D are mean 
model deviance and model deviance at the parameter mean, respectively; po is the effective number of parameters; ADIC is 
the difference in DIC among competing models. The models compared by using BF and LBF are the same as those models 
compared by using ADIC. 
Model 
Component 1 
D 
D 
Pd 
DIC 
ADIC BF LBF 
Ml 
NEFSC 
318.42 
308.59 
9.84 
328.26 
bt 
0.71 
0.51 
0.20 
0.90 
SEAMAP 
114.79 
110.40 
4.39 
119.17 
5.04 
5.04 
0.00 
5.04 
Total 
438.95 
424.53 
14.42 
453.37 
MlrU 
NEFSC 
314.14 
304.02 
10.12 
324.26 
bt 
1.19 
1.28 
-0.08 
1.11 
SEAMAP 
113.01 
108.53 
4.48 
117.49 
< 
5.04 
5.04 
0.00 
5.04 
Total 
433.38 
418.87 
14.52 
447.90 
Ml vs. MlrU: -5.47 1.02 0.05 
M1B 
NEFSC 
330.48 
321.10 
9.38 
339.86 
bt 
0.31 
0.10 
0.21 
0.51 
SEAMAP 
117.26 
113.65 
3.61 
120.87 
5.04 
5.04 
0.00 
5.04 
Total 
453.07 
439.88 
13.20 
466.27 
M2 
NEFSC 
318.30 
308.54 
9.76 
328.06 
bt 
0.71 
0.46 
0.25 
0.96 
SEAMAP 
115.17 
110.83 
4.34 
119.51 
a 
4.17 
4.17 
0.00 
4.17 
°p 
5.04 
5.04 
0.00 
5.04 
Total 
443.39 
429.05 
14.34 
457.73 
Ml vs. M2: 4.36 1.28 0.49 
M2rU 
NEFSC 
313.66 
303.57 
10.10 
323.76 
bt 
1.20 
1.22 
-0.02 
1.19 
SEAMAP 
113.49 
109.10 
4.39 
117.88 
a 
4.17 
4.17 
0.00 
4.17 
G P 
5.04 
5.04 
0.00 
5.04 
Total 
437.56 
423.10 
14.47 
452.03 
MlrU vs.M2rU: 4.13 0.05 -6.15 
M2 vs. M2rU: -5.70 0.04 -6.59 
The process errors from Ml and M1B increased 
with MWET at a 0.95 probability given that zero was 
outside the 95% BCIs of the mean slopes (0.133 and 
0.124) of the corresponding relationships: those 95% 
BCIs were (0.017, 0.247) and (0.035, 0.212), respective- 
ly. Here, P*> 0.98. On average, these relationships ex- 
plained 14% and 19.5% of the variation in the process 
errors. In contrast, the 95% BCIs of the slopes for the 
relationships between surplus production or instanta- 
neous surplus production and MWET included zero. 
The increase in these productivity metrics with MWET 
was therefore negligible at a 0.95 probability. 
Estimates from M2N, M2rUN, and M2BN of the 
posterior credible medians of the coefficient for MWET 
(a) were 0.42, 0.29, and 0.50, respectively, suggest- 
ing positive effects of MWET on the Atlantic Croaker 
production dynamics. However, the related 95% BCIs 
equaled (-0.57, 1.25), (-0.48, 1.22), and (-0.32, 1.19): 
these effects were negligible at a 0.95 probability. This 
result was consistent with that associated with the BF 
statistic. 
Model trends 
The biomass ratios, B t /B \ visy (Fig. 6, A and B), trended 
like the depletions (B t /B„ m , not shown), which them- 
selves tracked the variations of the NEFSC index well. 
They were characterized by low precision before 1990, 
