160 
Fishery Bulletin 1 14(2) 
fish eggs with previous results from morphological 
identifications. Because of several inherent differences 
in methods, we were unable to conduct a direct com- 
parison between morphological and molecular identi- 
fication from the same specimens. Morphological egg 
identifications at the NEFSC were discontinued in the 
early 2000s because the involved individuals retired. 
Moreover, these identifications were performed on for- 
malin-preserved eggs, which are less suitable for use 
in a DNA-barcoding approach. Specialized techniques 
have been developed to allow for the molecular identi- 
fication of formalin-preserved material, but these pro- 
cedures typically are more costly than our approach 
and use only very short DNA fragments. In contrast to 
formalin-preserved eggs, ethanol-preserved eggs can be 
readily identified with molecular techniques. However, 
ethanol denatures fish egg proteins, obscuring many 
of the internal morphological characters used for stag- 
ing and identification. For this reason, the only mor- 
phological character we recorded for each molecularly 
identified egg was egg diameter; other characters could 
not clearly be resolved on all specimens. For some 
species, particularly those species with a large perivi- 
telline space, such as Atlantic menhaden (Brevoortia 
tyrannus ), egg shrinkage appears to differ between 
ethanol and formalin preservation, limiting compari- 
sons of egg-size ranges among our data and histori- 
cal data. Because of the impossibility of making direct 
specimen-by-specimen comparisons, we focused our 
analyses on broader patterns of species composition 
and diversity. 
Overall, our molecular identifications represented 
a broader diversity of eggs than did the morphologi- 
cal identifications, including unexpectedly encountered 
taxa, such as white hake ( Urophycis tenuis ), which 
was not thought to spawn in the Gulf of Maine, as 
well as taxa that have previously been considered 
unidentifiable. In the past, many egg identifications 
were possible only to the genus level, whereas >93% 
of our barcoded fish eggs were identifiable to species; 
of our successfully barcoded eggs, 12 were identifiable 
only to the family level (Engraulidae) owing simply to 
the fact that this particular species has not yet been 
barcoded. Furthermore, our approach provides a tech- 
nique with high success rates of unambiguous iden- 
tifications and that is not sensitive to developmental 
egg stages. Historically, for morphological egg identi- 
fications, a large proportion of eggs was assigned to 
species in part by using nondiagnostic criteria, such 
as the presence of late-stage eggs of a species in the 
sample or the time and location of collection of the 
sample. 
