638 
Fishery Bulletin 95(4), 1997 
two states with 98% of the Atlantic catch — have de- 
clined since 1987, whereas recreational catches peaked 
in 1991 with an estimated 21 million fish (Newlin, 
1992; Speir et al., 1994). 
A lack of accurate catch and effort data from both 
the commercial and recreational fisheries makes it 
difficult to evaluate to what extent these long-term 
fluctuations represent natural changes in population 
abundance or reflect historic changes in Atlantic 
croaker exploitation. There has been a growing con- 
cern, however, that recent low landings may be re- 
lated to the large numbers of young fish killed as 
bycatch in the southern shrimp fishery and as part 
of the scrap catch in pound-net, haul-seine, and trawl 
fisheries (Speir et al., 1994; Mercer 1 ). In response to 
these concerns, the 1993 review of the Atlantic States 
Marine Fisheries Commission Fishery Management 
Plan for Atlantic croaker (Speir et al., 1994) has rec- 
ommended the use of bycatch reduction devices and 
the establishment of a coast- wide minimum size limit 
that would maximize Atlantic croaker yield-per- 
recruit. 
Yield-per-recruit models, widely used in fish popu- 
lation dynamics studies (Beverton and Holt, 1957; 
Ricker, 1975; Gulland, 1983), can be a useful tool in 
defining routine fisheries management measures 
such as minimum size limits, closed seasons, etc. 
(Gulland, 1983; Deriso, 1987). However, the only 
published application of yield-per-recruit models to 
Atlantic croaker is based on data from the northwest- 
ern Gulf of Mexico (Chittenden, 1977) and points out 
that results may or may not apply to other areas. In 
this paper we use stock assessment data from the 
Chesapeake Bay (years 1988-91; Barbieri et al., 
1994a) and from North Carolina (years 1979-81; 
Ross, 1988) to evaluate the effect of different fishing 
(-induced) and natural mortality, and age-at-first- 
capture schedules on Atlantic croaker yield-per-re- 
cruit. Implications of this analysis for management 
of Atlantic croaker are discussed. 
Methods 
Yield-per-recruit analysis 
Yield-per-recruit curves were calculated with the 
Beverton-Holt yield-per-recruit model (Beverton and 
Holt, 1957): 
3 tj -nK(t c -t 0 ) 
Y/R = Fe~ M(tc ~ tr) W 00 > (1) 
^ F + M + nK 
n = 0 
where Y/R = yield-per-recruit in weight (g); 
F = instantaneous fishing mortality coefficient; 
M = instantaneous natural mortality coefficient; 
W ^ = asymptotic weight (von Bertalanffy growth 
parameter); 
U n = summation parameter (U 0 =l, U^-3, U 2 =3, 
u 3 =- D; 
t c = mean age at first capture; 
t r = mean age (years) at recruitment to the fish- 
ing area; 
t Q = hypothetical age at which fish would have 
been zero length (von Bertalanffy growth pa- 
rameter); and 
K = the Brody growth coefficient (von Berta- 
lanffy growth parameter). 
Computations were performed with the computer 
program B-H3 available in the Basic Fisheries Sci- 
ence Programs package (Saila et al. , 1988). 
Parameter values used in simulations are summa- 
rized in Table 1. Estimates of growth parameters ( 
K, and f Q ) for Chesapeake Bay and North Carolina 
were obtained from Barbieri et al.( 1994a) and Ross 
( 1988), respectively. For both areas, was converted 
from by using an allometric length-weight rela- 
tion (6=3.23; Ross, 1988; and 6=3.30; Barbieri et al, 
1994a). One of the assumptions of the Beverton-Holt 
yield-per-recruit model is that growth is isometric — 
i.e. the coefficient 6 in the length-weight relation is 
equal to 3 (Beverton and Holt, 1957; Ricker, 1975). 
We, however, considered that departure from the as- 
sumption of isometric growth did not affect interpre- 
tation of our modeling results because the factor of 
interest in these simulations is the relative differ- 
ence in yield resulting from varying t c and F at dif- 
ferent levels of M. The relative error in such differ- 
ences, when using an incorrect 6, tends to be much 
less than that in absolute levels (Ricker, 1975). 
Estimates of t r , the mean age at recruitment to the 
fishing area, were based on Atlantic croaker life his- 
tory information (Chao and Musick, 1977; Ross, 
1988). Estimates of current t c , the mean age at first 
capture, was based on Atlantic croaker age composi- 
tions reported for the pound-net, haul-seine, and 
gill-net catches in the lower Chesapeake Bay for the 
period 1988-91 (t c = age 2; Barbieri et al., 1994a) and 
from age compositions reported for the haul-seine 
fishery in North Carolina for the period 1979-81 
(t c = age 1; Ross, 1988). Because of the uncertainty 
associated with estimates of M in fish populations 
(Vetter, 1988), simulations for both areas were con- 
ducted over a range of M values (0.20-0.35; Table 1). 
The instantaneous total annual mortality rate, Z, 
for fully recruited Atlantic croaker in North Caro- 
lina is 1.3 (Ross, 1988) and ranges from 0.55 to 0.63, 
with a mean value of 0.59 for the lower Chesapeake 
