162 
Fishery Bulletin 1 14(2) 
Table 2 
Results from the analysis of the effect of temperature on egg diameter for barcoded fish 
species that were identified from at least 15 stations in the northeastern U.S. continental 
shelf between Cape Hatteras, North Carolina, and Nova Scotia, Canada. Both P-values 
and the coefficient of determinations (r 2 ) were calculated with linear regression analysis 
of SST versus measured egg diameter (in millimeters). Species that displayed a signifi- 
cant trend between SST and egg diameter (P< 0.05 or 5%) are indicated with an asterisk 
and are visually illustrated in Figure 5, A-E. 
Number of 
Species 
N 
stations 
P 
r 2 
Hippoglossina oblonga * 
184 
72 
<0.0001 
0.2335 
Merluccius bilinearis 
196 
59 
0.1901 
0.0100 
Urophycis chuss 
144 
53 
0.3175 
0.0070 
Citharichthys arctifrons * 
153 
45 
0.0438 
0.0300 
Scophthalmus aquosus 
81 
29 
0.8347 
0.0006 
Peprilus triacanthus * 
52 
22 
<0.0001 
0.5083 
Limanda ferruginea * 
80 
22 
0.0012 
0.1277 
Enchelyopus cimbrius 
35 
19 
0.1160 
0.0732 
Tautogolabrus adspersus 
45 
17 
0.2615 
0.0322 
Paralichthys dentatus 
50 
16 
0.0920 
0.0580 
Gadus morhua * 
47 
15 
<0.0001 
0.4951 
species. This combination could make this approach a 
more cost-efficient one for small-scale projects. 
Although other lower-cost techniques for molecular 
identification do exist, most of them are limited in re- 
gard to research scope. Hyde et al., (2005) successfully 
implemented multiplex PCR onboard a research vessel 
to identify fish eggs and larvae in real time; however, 
this approach is best-suited for identifications of 5-20 
species at a time. Alternatively, the use of multiplex 
suspension bead arrays for the identification of fish 
eggs is lower in cost than DNA barcoding and similar 
in its high-throughput capability, but this technique 
can identify an egg only if a probe for that species is 
included in the array (Gleason and Burton, 2012) — a 
restriction that is not encountered with DNA barcod- 
ing. Where our specific approach to molecular identifi- 
cation falls short is that it is not suitable for use with 
formalin-preserved eggs, which are easier to stage than 
ethanol-preserved eggs. Applications, such as the daily 
egg production method, that require the ability to re- 
solve egg stages, would benefit greatly with the ability 
to use formalin-preserved samples. Certain molecular 
identification approaches, targeted at specific species, 
have been successfully applied to formalin material 
(Goodsir et al., 2008), and techniques with DNA bar- 
coding on shorter fragments and formalin-preserved 
material are being developed (Zhang, 2010). 
Looking forward, a DNA-barcoding approach could 
be implemented in a wide range of ichthyoplankton 
studies in addition, and in relation, to the identifica- 
tion of fish eggs. In comparison with fish larvae, eggs 
provide a more precise documentation of spawning 
location and time. For Atlantic cod in particular, the 
potential use of DNA barcoding of fish eggs is notable. 
Atlantic cod in U.S. waters comprise at least 3 genetic 
stocks, each thought to have further substock diversity 
associated with specific spawning grounds and seasons 
(Kovach et al., 2010; Zemeckis et al., 2014). It also has 
been suggested that other spawning stocks in the east- 
ern Gulf of Maine exist, although this area has been 
so depleted that it is unknown if spawning persists 
there (Ames, 2004). We found differences in egg size 
that were associated with temperature, which in turn 
was associated with season and region of collection. A 
more extensive sampling procedure would likely reveal 
further complexity. 
Although DNA barcoding is insufficient for distin- 
guishing among stocks, the laboratory procedures do 
provide an extensive time-series archive of DNA, al- 
lowing the identification of spawning components in 
future molecular studies. In regard to indices of spawn- 
ing stock biomass, where problems have arisen in the 
past as a result of morphologically based misidentifica- 
tions, DNA barcoding could be used to broaden the use 
of existing surveys, while bypassing the assumption of 
constant egg mortality that underlies the use of larval 
abundances as indices of spawning stock biomass. 
Overall, we have shown that DNA barcoding of fish 
eggs is sufficiently advanced to be incorporated into 
long-term, regional-scale ichthyoplankton monitoring 
programs. Trial runs with unidentified, well-digested 
fishes obtained from stomach samples collected dur- 
ing an ongoing food habits monitoring program (Smith 
and Link, 2010) show similar promise (Lewis and 
