Legault and Ehrhardt: Correcting annual catches from seasonal fisheries for use in virtual population analysis 
291 
Figure 7 
Corrected catch and annual fishing mortality rate {F A ) for three scenarios of when 
catch occurs under a range of natural mortality rates. These values were com- 
puted by using an N K+1 of 10,000 fish and a catch of 5,000 fish during the month 
(or split evenly between the months). 
ing the year, then the expected amount of bias removed 
by the catch correction algorithm may in fact be incor- 
rect. The actual amount of bias between the true state 
of the population and the estimated values also depends 
upon the amount and type of error in the observed catch 
estimates and tuning indices. The reduction in bias 
produced by this algorithm may be insignificant rela- 
tive to the level of bias produced by other sources in the 
analysis, but it should at least be in the correct direction. 
The direction of change between observed and cor- 
rected catches has implications for quota projections 
and biological reference-point determination, such 
as F 0 j or F %spR . Catches from early in the year are 
corrected to lower values than those for the observed 
catch, and thus use of the observed catch in VPA re- 
sults in overestimates of the population size, if unbi- 
ased indices are used to tune the VPA. The overesti- 
mated population size would then be used to predict 
a quota for the upcoming year which would be too 
large. When this quota was filled, observed catch, 
which is larger than the corrected catch in the VPA, 
would again be used in the following year’s stock 
assessment and would thus predict a larger popula- 
tion size than was present. This feedback cycle could 
cause problems if the tuning indices do not ad- 
equately reflect the decline of the population. 
