Schobernd et al.: A comparison of numbers of fish larvae extruded from plankton nets of different mesh size 
249 
Table 5 
Parameters and the relative difference between the Akaike information criterion (AAIC) score 
from each model and that of the best fitting model within a group for unidentified larvae and 
larvae of 6 taxa: Percoidei, Engraulidae, Clupeidae, Scombridae, Sciaenidae, and Lutjanidae. 
The 2 models used were the power (a, b) and exponential ( d, f) model. 
Parameters 
Taxa 
Power 
Exponential 
AAIC 
a 
b 
d 
f 
Power 
Exponential 
Unidentified 
6.83 
-1.44 
13.29 
-0.76 
0 
1 
Percoidei 
12.59 
-2.80 
66.26 
-1.74 
0 
2.46 
Engraulidae 
4.79 
-1.00 
2.21 
-0.13 
0 
8.04 
Clupeidae 
0.75 
0.21 
0.92 
0.02 
0 
0.3 
Scombridae 
3.64 
-0.90 
2.96 
-0.24 
0 
1.51 
Sciaenidae 
3.10 
-0.66 
3.15 
-0.22 
0 
0.22 
Lutjanidae 
1.27 
0.00 
1.13 
0.03 
0.13 
0 
ent mesh sizes, 0.202 and 0.333 mm (Table 5). Among 
the 6 taxa and the unidentified group, extrusion was 
better described by using the power rather than the 
exponential model (Table 5). The difference between 
power and exponential model performance was great¬ 
est for Engraulidae, with the power model having the 
best fit. Differences in AIC scores obtained from the 
models, however, were relatively small for the uniden¬ 
tified group and all other taxa that were analyzed 
(AAIC<2.5), indicating that both functions are suitable 
for predicting extrusion rates for those taxa (Burnham 
and Anderson, 2002). 
Discussion 
Despite the wide range in observed larval abundance 
ratios among samples collected with the nets of differ¬ 
ent mesh sizes, functional relationships between pre¬ 
dicted larval abundance and size indicated that small 
larvae were underrepresented in samples collected 
with a standard SEAMAP survey bongo net (0.333 
mm). For smaller body lengths, predicted abundance 
ratios indicated that samples from the finer-mesh net 
contained up to 3-4 times the numbers of larval fish 
for 3 of 5 Gulf of Mexico fish families than samples 
from the coarser-mesh net. Observed abundance ratios 
indicated extrusion could be even higher, depending 
on the taxa. Underestimation of the smallest size cat¬ 
egory of sampled larvae was greatest for larvae that 
could not be identified to any taxonomic level or could 
be identified only to a suborder because the smallest 
larvae in the samples had not yet developed enough 
morphological characteristics that could be used for 
identifications to family level. Fish egg densities, how¬ 
ever, did not differ significantly between the nets with 
2 mesh types, suggesting that the standard SEAMAP 
survey net adequately samples fish eggs. This finding 
is not surprising because egg sizes of the analyzed taxa 
all exceeded the mesh size of the standard SEAMAP 
plankton net. 
Differences in larval fish retention between 
0.333-mm-mesh and 0.202-mm-mesh plankton nets 
were previously investigated in northern Gulf of Mex¬ 
ico waters by Comyns (1997) and Hernandez et al. 
(2011). Comyns (1997) found that red drum larvae 
in the smallest size group, 1.5-1.9 mm, were 5 to 8 
times more abundant in the finer than in coarser mesh 
samples. Hernandez et al. (2011) found few to no sig¬ 
nificant differences among the taxa between the sam¬ 
ples from nets with the 2 mesh sizes in either larval 
abundances or length frequencies. Mean size of larvae, 
however, was smaller in samples from the finer-mesh 
nets than in samples from the coarser-mesh nets for 
4 groups: total fish larvae (all taxa combined, exclud¬ 
ing unidentified larvae), Leptocephali and Syngnathi- 
dae (combined), Sciaenidae (the family that includes 
the red drum), and unidentified larvae. In the current 
study, mean standardized abundances varied signifi¬ 
cantly between samples from nets with the different 
mesh sizes for only 2 groups examined (unidentified 
larvae and Percoidei), whereas length-frequency distri¬ 
butions for all 6 taxa and the unidentified group that 
were examined were significantly different in samples 
taken with nets of the 2 mesh sizes. 
Study design, sampling gear, and collection protocols 
used in these prior studies differed from each other 
and from the current study, as did results, making di¬ 
rect comparisons of the 3 studies problematic. Despite 
these differences, some useful inferences can be drawn 
regarding the influence of study design, sampling gear, 
and protocols on larval fish retention in comparisons 
of mesh sizes of nets. In the Comyns study, sampling 
was conducted only during the peak month of red 
drum spawning and consisted of collections taken in 
association with a subsurface current drogue, which 
