10 
Fishery Bulletin 1 1 1 (1) 
Table 2 
Kruskal-Wallis test for comparison of size difference (in mm) by transect and sampling gear for all fishes 
combined, as well as for the 3 most dominant fish families, from this study where 2 gear types were used: 
bongo sampler and the In Situ Ichthyoplankton Imaging System (ISIIS), to sample fish larvae south 
of Woods Hole, Massachusetts, in October 2008. (Upper): comparison within gear between transects. 
(Lower): comparison between gears. Asterisks (*) denote significant differences. 
Bongo 
Transect 1 
Bongo 
Transect 2 
ISIIS 
P Transect 1 
ISIIS 
Transect 2 
P 
Mean size — all larvae 
3.514 
4.397 
0.275 
7.223 
4.959 
0.001* 
Paralichthyidae mean size 
3.809 
5.044 
0.547 
4.672 
4.122 
0.061 
Phycidae mean size 
2.335 
2.980 
0.221 
4.617 
4.295 
0.199 
Merlucciidae mean size 
3.998 
3.550 
0.783 
13.701 
12.037 
0.496 
Bongo 
ISIIS 
P 
Mean size — all larvae 
3.858 
6.468 
1.67E-12* 
Paralichthyidae mean size 
4.385 
4.488 
0.0001* 
Phycidae mean size 
2.622 
4.486 
5.41E-05* 
Merlucciidae mean size 
3.795 
13.398 
3.806E-06* 
with ISI IS will require automated computer analysis. 
Several different options may be available for address- 
ing some of these needs (e.g., Davis et ah, 2004; Hu 
and Davis, 2005; Luo et ah, 2005;; Culverhouse et ah, 
2006; Benfield et ah, 2007; Zhao et ah, 2010), although 
these alternatives have not been tested with repeti- 
tive processing of millions of images. Consequently, we 
are currently developing and testing algorithms suit- 
able for segmenting and classifying individual organ- 
isms from full image files. These algorithms must be 
capable of processing data at high speeds (or with 
multiprocessor computers) and must be able to handle 
large data sets (e.g., Tsechpenakis et ah, 2008). With 
such analysis capabilities, the typical time between re- 
search cruise and ultimate data analysis could be re- 
duced greatly. 
Conclusion 
Although ISIIS can be a powerful tool for resolving fine 
to mesoscale patchiness in both vertical and horizon- 
tal distributions of plankton, it is limited by the fact 
that it is a nondestructive sampler (i.e., it does not 
collect specimens). ISIIS will not replace nets for all 
studies. There is still a strong need for sample collec- 
tion, whether for identification verification (for larvae 
or eggs) or for more specific studies, such as projects 
on food habits, growth, and genetics, that require speci- 
mens. In addition, many nets, including bongo nets, can 
be used by a greater variety of vessels and in a wider 
range of weather conditions than the ISIIS instrument 
package. When these different tools are combined, how- 
ever, ISIIS could be used to establish the vertical and 
spatial setting of fish larvae. This information could be 
used to identify locations for targeted net samples. This 
melding of samplers also would lead to more efficient 
requirements for ship time and processing time (i.e., 
less time spent with nets and on processing the survey 
samples from areas where the targeted specimens are 
rare or absent). Therefore, ISIIS (and the technology it 
represents) is a valuable addition to both process-ori- 
ented studies and routine surveys. This technology can 
contribute both to the understanding of the relation 
between larval fishes and their biological and physi- 
cal oceanographic habitat and to the quantification of 
larval fish abundance and distribution for use in stock 
and ecosystem assessments. 
Acknowledgments 
The authors wish to acknowledge the crew and cap- 
tain of the NOAA Ship Delaware II for their support in 
deploying our instrumentation. We also acknowledge K. 
Hyde (NOAA Northeast Fisheries Science Center) for 
providing the satellite data depicted in Figure 1. We 
appreciate funding from several sources, especially the 
National Marine Fisheries Service’s Advanced Sam- 
pling Technology Working Group and the Geosciences 
Directorate of the U.S. National Science Foundation. 
This manuscript was improved by careful reading and 
comments from D. Johnson (NOAA). 
Literature cited 
Begg, G. A., J. A. Hare, and D. D. Sheehan. 
1999. The role of life history parameters as indicators of 
stock structure. Fish. Res. 43:141—163. 
