478 
Fishery Bulletin 95(3), 1997 
species. Finally, environmental variables were tested 
for their correlation with abundance. 
Materials and methods 
Sampling location and period 
The sampling station for larval fish abundance, lo- 
cated 2 km inside of Beaufort Inlet, North Carolina 
(34°43'N, 76°40'W), was a platform attached to a 
bridge over a 6-m-deep tidal channel adjacent to the 
Beaufort Laboratory at Pivers Island and has been 
the site of weekly larval fish sampling since the 1985- 
86 larvae ingress season (Warlen, 1994). We sampled 
every night, 20 November 1991 through 15 April 
1992, a period that more or less encompasses the 
annual periods of recruitment of ocean-spawned es- 
tuarine-dependent larvae that pass through North 
Carolina inlets from autumn to spring. 
Fish and environment sampling 
Oblique tows (bottom to surface) of a 1-m diameter, 
800-p mesh net were used to sample the water col- 
umn for larvae. Three consecutive 4-min tows were 
made at 15-min intervals during the time of predicted 
maximum flood-tide current. Sampling was con- 
ducted only between dusk and dawn and about 50 
minutes later each successive night because of the 
advancing tide stage. Oblique tows through the en- 
tire water column were chosen over surface, bottom, 
or other single-depth tows to eliminate depth bias. 
Species of concern, including B. tyrannus, are re- 
ported to be distributed by depth even in shallow, 
well-mixed North Carolina inlets (Lewis and 
Wilkens, 1971; Hettler and Barker, 1993 ). 
The net was deployed by paying out the winch cable 
as the net, pulled downstream by the tidal current, 
sank to the bottom. It was then retrieved obliquely 
through the water column. A depth sounder with a 
deck readout (Standard Communications DS20) was 
attached to the net frame to indicate that the net 
had reached the bottom of the channel. Tow volume 
was measured with a General Oceanics model 2030R 
flow meter. Average tow time was 4.0 minutes. The 
target tow volume was 100 m 3 , and target net re- 
trieval speed was 1 m/sec. 
Data on several environmental variables were col- 
lected concurrently with biological sampling. Salin- 
ity and temperature measurements were taken with 
a Hydrolab H20 water quality multiprobe. Water clar- 
ity was measured with a Sea Tech 25-cm transmis- 
someter with a 660-nm filter. Tidal current speed was 
measured with a Marsh-McBirney model 201 flow 
meter. Wind speed and direction data were obtained 
from the NOAA C-MAN station at Cape Lookout, 15 
km SE of the larval sampling platform. Tidal ampli- 
tude data were obtained from a NOAA tide gauge 
located on Pivers Island. 
Processing of larvae 
After preservation in 70% ethanol, fish larvae were 
identified, counted, and up to 20 individuals of each 
species were indiscriminately selected for measure- 
ment of standard length. Ages and birth dates of all 
menhaden larvae retained for length measurements 
(1,341 individual fish) were determined by otolith 
daily increment counts (estimated age in days = in- 
crement count + 5) following the methods of Warlen 
(1992). 
Data analysis 
Abundance was calculated from the number of lar- 
vae caught per tow and water volume sampled 
(density=number x 100 m -3 ). Densities per unit vol- 
ume were calculated rather than densities per unit 
area because all published relevant abundance data 
on these species is per unit volume. Mean densities 
by species for each date sampled were determined 
by averaging the densities from the three tows taken 
on that date. Although we sampled every night, no 
data were available for 10 dates during the sampling 
period (Fig. 1, A-D). This problem is explained by 
the following example. On 14 December, sampling 
started at 2359 h and ended around 0100 h, 15 De- 
cember. The next night sampling began at 0033 h, 
16 December. Thus, sampling never began on 15 
December. Sampling on 15 December at the time of 
maximum flood tide current would have occurred 
before sampling on 14 December had ended. The 
same situation occurred on nine other dates. 
Seasonal mean densities were determined by av- 
eraging the daily densities during the interval when 
each species was caught, including dates when no 
individuals of the species were caught. Variations in 
mean daily densities and associated variance esti- 
mates were derived by “sampling” individual den- 
sity data sets for each species at intervals of 2, 3, 4, 
5, 7, 14, and 30 days and by then comparing these 
with the actual data set (1-day intervals between 
sampling). From this exercise, a mean and standard 
deviation was generated for each sampling scheme. 
A 2-day cycle, for example, beginning on 20 Novem- 
ber and continuing every other day at day 1, 3, 5 
etc., produced one daily mean and standard devia- 
tion, whereas a 2-day cycle beginning a day later on 
21 November and continuing every other day at day 
