Wood and Cadrin: Mortality and movement of Limanda ferruginea tagged off New England 
283 
Observed deviance/df=6.1 1 £ 
1 o 
1 ° 
§ ° 
o 
1 
| Median c=2.12 
1.0 1.5 2.0 2.5 3.0 
Simulated c 
Figure 2 
Results from a goodness-of-fit test used to estimate 
overdispersion (c) for the general model fitted to tag- 
recovery data (time- and sex-dependent parameter- 
ization) for Yellowtail Flounder (Limanda ferruginea) 
tagged off New England from June 2003 to August 
2006. The estimates of simulated deviance c are shown 
for a range of simulated c values. The value of c, deter- 
mined through logistic regression, is the point (6.11) 
where 50% of the simulated values fall above and 50% 
fall below the observed deviance c for the general mod- 
el. df=degrees of freedom. Dev=deviance. 
movement to the southern New England-Mid-Atlan- 
tic stock area), 98% residence on the Georges Bank 
stock area (with 1% movement to the Cape Cod-Gulf 
of Maine stock area and <1% movement to the south- 
ern New England-Mid-Atlantic), and 26% residence in 
the southern New England-Mid-Atlantic stock area 
(with 63% movement to the Georges Bank stock area 
and 10% movement to the southern Cape Cod-Gulf of 
Maine stock area; Table 4). However, most movement 
from southern New England was observed for Yellow- 
tail Flounder released on Nantucket Shoals, near the 
boundary with adjacent stock areas (Fig. 5). 
Discussion 
The numbers of releases of tagged fish and tag recover- 
ies in this study were 10 times greater than the num- 
bers from any previous study of Yellowtail Flounder 
movement or mortality. The results from this study 
provide updated inferences of movement patterns and 
an independent estimate of mortality. Tag-recovery 
modeling and its application in fisheries research has 
increased in popularity over the past decade and has 
become an important tool to fisheries management 
(Pine et al., 2003). Large-scale tag-recapture studies 
provide insights into fish movement and population 
dynamics that are separate (except for their reliance 
on fishery recaptures) from the fishery-dependent and 
research survey data and methods used in conventional 
stock assessments. The results from tagging analyses 
should be of particular interest when there are sus- 
pected inconsistencies with the stock assessment data 
and analyses, as is the case for Yellowtail Flounder 
stocks off New England. 
The results from tag-recovery modeling in this study 
are consistent with the perception that the Yellowtail 
Flounder resource in New England is experiencing an 
intense rate of mortality. The total annual mortality 
estimate of 1.4 derived from patterns of tag recovery is 
similar to stock-specific, age-based mortality estimates 
from the Yellowtail Flounder assessments for 2003 to 
2006 (NEFSC 1 ; Fig. 3). The results from this study 
demonstrate that models typically used in quantita- 
tive ecology (e.g., the Brownie tag-recovery model) can 
complement conventional methods for fisheries stock 
assessment. The advantage of the Brownie model is 
that survival estimates are not conditional on an as- 
sumed natural mortality rate, and recovery rates are a 
composite of exploitation rate (i.e., natural and fishing 
Table 3 
The top 5 most highly ranked models adjusted for a c=2.12 that were fitted to tag-recovery data in 
this study of the mortality and movement of Yellowtail Flounder (Limanda ferruginea) tagged off 
New England from 2003 to 2006. Models were ranked by quasi-likelihood adjusted AIC (QAIC c ). 
Survival (S) and recovery rate (f) were estimated by month (t), for the entire time series (.), and by 
sex (g). The optimal model (in bold type) was chosen on the basis of rank and parameter estimates 
that were biologically reasonable. 
Model 
QAIC c 
Delta 
QAIC c 
QAICc 
weight 
Model 
likelihood 
Number of 
Parameters 
Qdeviance 
Sit) fi g*t) 
19765.56 
Multiple boundary estimates for survival 
Sit) fit) 
19813.70 
Multiple boundary estimates for survival 
Si.) figH) 
19825.54 
0.00 
0.73 
1.00 
111 
1181.05 
Sig) fi g*t) 
19827.51 
1.97 
0.27 
0.37 
112 
1181.01 
S(g*t) fit) 
19847.22 
21.68 
0.00 
0.00 
163 
1098.07 
