922 
Fishery Bulletin 96(4), 1998 
1.6 
1.4 
JZ 
o 
1.2 
C3 
O 
1 
JZ 
CD 
C 
0.8 
£ 
U) 
0.6 
I 
H 
0.4 
0.2 
iri v~i iri m >n >n n ir, 
ro t" ir. vo (-•' ad o\ 
Length (mm) 
Figure 4 
Day:night (A) and twilight:night (B) catch ratios as a func- 
tion of larval length. Bars indicate 95% confidence intervals. 
retain them. This is attributable to the fact that maxi- 
mum head width is not the maximum cross-sectional 
diameter of the larva, because of the bulk of the yolk 
sac. Our method of towing (at low speed, with plastic 
codend buckets) did not seem to cause damage to yolk 
sacs and resulted in full retention of yolksac larvae by 
the 0.335-mm mesh. However, the degree of extrusion 
strongly depends on filtration velocity (Smith and 
Richardson, 1977; Colton et al., 1980). Thus, at high 
towing speeds yolk sacs may easily be crushed. 
In his study on anchovy, Lenarz (1972) used depth 
of body at the insertion of the pectoral fin as the criti- 
cal dimension to compare with mesh diagonal and 
concluded that the mesh “diagonal rule” was too con- 
servative in the case of slowly towed nylon nets. 
However, body depth is not the critical dimension 
for clupeoids (Colton et al., 1980). The latter found no 
differences related to mesh size (0.253- and 0.333-mm 
mesh nets) in the length frequency distributions and 
abundances of larval herring in their collections at 
1.5-knot towing speed. The mesh “diagonal rule” 
(with skull width as the critical dimension) was also 
conservative in their case, but for the length inter- 
val including only yolksac larvae. They found that 
most yolksac larvae caught in their slowly towed 
bongo nets, which were also equipped with plastic 
codend buckets, were undamaged. 
We feel that at a low, constant towing speed (1.5— 
2.0 knots), bongo nets equipped with a 0.335-mm 
mesh net and plastic codend buckets are very effi- 
cient in retaining larvae of clupeoids. 
Net avoidance 
Changes in catchability with varying light conditions 
and larval length are demonstrated in this study. 
These changes begin at notochord flexion. If visual 
detection of the net is the primary cue for net avoid- 
ance, flexion and postflexion European anchovy lar- 
vae show the expected relationship of night>twilight> 
day catches (Morse, 1989). 
In another study, Murphy and Clutter ( 1972) com- 
pared catches of larval Hawaiian anchovy ( Stole - 
phorus purpureus ) taken by conventional towed coni- 
cal nets with catches taken by a miniature purse 
seine constructed of the same netting. The latter was 
considered to be a more effective sampler of the full 
size range of anchovy larvae than towed nets. In a 
way similar to the present study, larvae <6 mm long 
were captured with approximately equal efficiency 
by both the towed net and the seine (i.e. these larvae 
did not seem to avoid the towed net). For larger lar- 
