Munyandorero: Climate effects on Micropogonias undu/atus 
65 
issue (Hare et al., 2010). However, diversifying inves- 
tigation models was in part consistent with ASMFC’s 
assessment needs and Hare et al.’s (2010) suggestion 
for this fishery resource when faced with changes in 
anthropogenic activity (here, fishing), environmental 
forcing, and also with parameter uncertainty. BDMs 
are typically suitable when fishery data are limited to 
aggregate catch and effort or indices of stock biomass 
(Hilborn and Walters, 1992; Prager, 1994). Regardless, 
even in “data-rich” jurisdictions, various stock-assess- 
ment teams customarily implement them to support 
the results of the more sophisticated, data-hungry 
models. Jacobson et al. (2002) and MacCall (2002) rec- 
ommended their systematic use as supplemental as- 
sessment tools because, in spite of their simplicity and 
alleged lack of realism, they can be the basis of useful 
management actions (Ludwig and Walters, 1985; Laloe, 
1995). 
Investigations have focused on the alleged winter- 
temperature effects on Atlantic Croaker productivity 
that occur during the prerecruit stages of the species 
(Joseph, 1972; Norcross and Austin 2 ; Lankford and Tar- 
gett, 2001a, 2001b; Hare and Able, 2007; Hare et al., 
2010). Age- and stage-structured fisheries models are 
used to investigate environmental effects on popula- 
tion changes through the deviations from an “average” 
or “virgin” recruitment or through stock-recruitment 
models, where environmental covariates, along with 
unexplained random errors, are assumed to influence 
the recruitment processes and variability (e.g., lies and 
Beverton, 1998; Levi et al., 2003; Maunder and Watters, 
2003; Hare et al., 2010). These effects can be incorpo- 
rated into density-dependent, density-independent, or 
both types of parameters of stock-recruitment models. 
By analogy to stock-recruitment models, MWET 
was introduced into the parameter r that, in surplus- 
production models, is the counterpart of the density- 
independent parameter of stock-recruitment models, 
and process errors characterized all model parameters. 
Preference was given to the Bayesian state-space mod- 
eling framework because of its anticipated flexibility 
in addressing simultaneously various types of errors 
and parameter uncertainty and because it was deemed 
suitable for shedding light on the ability of BDMs to 
detect MWET effects. A corollary of these investiga- 
tions was the examination of the extent of such effects 
on the Atlantic Croaker stock status. 
Focusing MWET effects on r was, in conjunction with 
available fishery data (i.e., fisheries removals and sur- 
vey indices only), the simplest scenario of implemented 
BDMs. However, this procedure was also dictated by 
the need of parsimony in statistical analysis, thereby 
favoring simple models. If there were supporting data 
and evidence on changes in habitat conditions — usu- 
ally affecting in other words, the density-dependent 
parameter (e.g., Jacobson et al., 2005) — or in fisheries 
effective effort and catchability, it may have also been 
convenient to consider their effects and interactions on 
Atlantic Croaker productivity. Information about these 
factors ultimately needs to be gathered and equally ac- 
counted for in future analyses. 
The analysis led to mixed outcomes. On the one 
hand, the positive and significant correlations between 
the process errors from Ml or M1B and MWET sup- 
ported the hypothesis that MWET may be playing a 
role in biomass variability of Atlantic Croaker on the 
U.S. Atlantic coast. Increased growth or increased re- 
cruitment during years of warmer winters would there- 
fore enhance biomass production in subsequent years. 
However, such relationships were weak in that only 
14% and 19.5% of process errors were related to the 
variation in MWET. On the other hand, there were pos- 
sible positive relationships between surplus production 
or instantaneous surplus production and MWET, but 
the relationships were statistically insignificant. The 
lack of a relationship between surplus production and 
an environmental covariate, however, is not unusual. 
In contrast, it was surprising that instantaneous 
surplus production vs. MWET and surplus production 
vs. MWET exhibited similar and insignificant rela- 
tionships. Instantaneous surplus production is usu- 
ally more sensitive to environmental change than is 
the corresponding surplus production (Jacobson et 
al., 2001). Likewise, the hypothesis of MWET effects 
on the Atlantic Croaker production dynamics had no 
support of the 95% BCIs for the coefficient controlling 
MWET effects (a) upon specifying its prior as a~N( 0, 
0.02). Weakness and absence of the aforementioned 
relationships corroborated the fact that BSSBDMs in- 
corporating MWET (although conceptually interesting 
and ecologically plausible) did not statistically outper- 
form those BSSBDMs without MWET nor did it predict 
significantly different metrics of stock status. In com- 
parison with Hare et al.’s (2010) results, this study re- 
vealed that correlations between MWET and a metric 
of Atlantic Croaker productivity can appear and disap- 
pear or be weak with a modeling approach. 
Surplus production models with environmental ef- 
fects have sometimes improved understanding and de- 
scription of the performance of fished populations and 
ecosystems when all key control variables and causal 
mechanisms have been unambiguously identified, un- 
derstood, and accounted for (e.g.. Freon, 1988; Evans 
et al., 1997; Yanez et al., 2001; Jacobson et al., 2005; 
Mueter and Megrey, 2006; Thiaw et al., 2009; some con- 
tributions in Bundy et al., 2012). Exceptions to such 
favorable situations exist (Laloe, 1988; Fogarty et al., 
2012; this study). Here, BDMs failed to detect MWET 
effects adequately because of 4 possible major reasons. 
First, in the process errors-MWET relationships, the 
remaining, unexplained 81-86% of the variation in the 
process errors may be rooted in other, yet unknown 
environmental anomalies. This outcome indicated the 
possibility that MWET (inter (acted with other ecologi- 
cal factors (e.g., change in other habitat conditions). 
Second, random errors and a well-established under- 
lying environmental anomaly may not be linked lin- 
early or may even be unrelated because environmental 
