452 
Fishery Bulletin 95(3), 1997 
SSB, and %MSP because they are based on estimated 
abundance, biomass, and egg production, respec- 
tively, which have a negative exponential relation 
with fishing mortality. Therefore, small underesti- 
mates in F can produce large overestimates of abun- 
dance. Retrospective differences may also be skewed 
because negative values of R, SSB, and %MSP are 
not possible. Despite the conclusion that log-trans- 
formed retrospective differences were not biased, the 
positive skewness of retrospective differences and 
relative differences has important implications for 
management of the fishery. A skewed distribution of 
inconsistency from converged estimates may be con- 
sidered a characteristic feature of terminal estimates 
from future menhaden VPA’s. Therefore, manage- 
ment advice should account for the equal likelihood 
of moderate underestimation and substantial over- 
estimation of R, SSB, and %MSP. 
Management decisions must consider the differ- 
ence between converged estimates, which are used 
to define overfishing thresholds, and terminal esti- 
mates, which are used to determine current status. 
Although inconsistency does not quantify uncertainty 
of estimates, the large retrospective differences re- 
