248 
Fishery Bulletin 11 6(3-4) 
n 
0 
i i i 
1 2 3 
l l 
4 5 
n- 
0 
—1-1- 
I 2 
Length (mm) 
Length (mm) 
Figure 4 
Ratios of larval abundance observed in nets with 0.202-mm mesh to larval abundance 
observed in nets with 0.333-mm mesh, by body length, for unidentified larvae and lar¬ 
vae of the suborder Percoidei. The solid and dashed lines represent modeled ratios of 
larval abundance that represent how many larvae were extruded from the coarser-mesh 
net (power and exponential functional relationships, respectively). The thin, horizontal 
line represents 1:1 ratios of total larval abundance or the point at which the finer- and 
coarser-mesh nets retained the same number of larvae at subsequent size classes. 
bers in the net with 0.202-mm mesh than in the net 
with 0.333-mm mesh (Fig. 3). 
The ratios of larval abundance collected with 
the finer net to larval abundance collected with the 
coarser net varied among the 6 taxa and unidentified 
group that were analyzed (Figs. 4 and 5). Although no 
pattern was seen regarding extrusion rates and body 
shape (perciform versus clupeiform), greater extru¬ 
sion was suggested by the somewhat higher predict¬ 
ed abundance ratio of engraulid (clupeiform) larvae 
1-2 mm in length compared with that of similar-size 
scombrid (perciform) and sciaenid (perciform) larvae. 
Extrusion for the categories of unidentified larvae and 
Percoidei was greater than for larvae identified to the 
family level. All abundance ratios for unidentified lar¬ 
vae, both observed and modeled, were greater than 
1, indicating greater retention in the samples from 
the nets with a 0.202-mm mesh (Fig. 4). The great¬ 
est modeled abundance ratio for unidentified larvae 
was 11.4 for the 0.7-mm size class with the use of the 
power function. Modeled abundance ratios for the sub¬ 
order Percoidei were above 1.0 for all sizes under 2.4 
mm, and abundance ratios were as high as 34.0 and 
17.0 for 0.7-mm larvae, with the power and exponen¬ 
tial functions, respectively (Fig. 4). Modeled engrau¬ 
lid abundances were 4.0 and 1.9 times greater (power 
and exponential models, respectively) in samples from 
nets with 0.202-mm mesh than in samples from nets 
with 0.333-mm mesh for larvae at 1.2 mm (Fig. 5). 
Abundances of engraulid larvae in samples from nets 
with the different mesh sizes were equal for larvae 
at 4.7 mm for the power model and at 6.1 mm for 
the exponential model. Contrary to expectations, both 
the power and exponential models for Clupeidae in¬ 
dicated slight increases in abundance ratios with in¬ 
creasing size (Fig. 5). Abundance ratios were greatest 
at 10 mm, reaching 1.2 and 1.1 for the power and 
exponential models, respectively. Scombrid larvae at 
1.2 mm were retained 3.1 (power model) to 2.2 (expo¬ 
nential model) times more in the samples from nets 
with a 0.202-mm mesh than in the samples from nets 
with a 0.333-mm mesh (Fig. 5). Larval abundances in 
both mesh sizes were higher at sizes between 4.1 and 
4.5 mm. Sciaenid larvae appear to be extruded from 
0.333-mm-mesh nets at sizes less than 5.5 mm (power 
model) and 5.1 mm (exponential model; Fig. 5). Sci¬ 
aenid extrusion was greatest at the 1.0-mm size, and 
power and exponential models indicated that abun¬ 
dances were 3.1 and 2.5 greater in samples from nets 
with 0.202-mm mesh were than in samples from nets 
with 0.333-mm mesh, respectively (Fig. 5). Despite 
the high variability of abundance ratios for Lutjani- 
dae (Fig. 5), both models projected greater abundanc¬ 
es of larvae in the samples from nets with 0.202-mm 
mesh over all lengths. The power model for Lutjani- 
dae reflected little overall change in abundance ratios 
by lengths, whereas the exponential model indicated 
slightly greater abundances in finer-mesh nets as lar¬ 
val lengths increased. 
Coefficients derived from the models for both power 
and exponential functions are presented for use in fu¬ 
ture comparisons of sampling with bongo nets of differ- 
