Lewis et al.: Integrating DNA barcoding of fish eggs into ichthyoplankton programs 
161 
Species 
Figure 4 
Box plot showing the measured egg diameter ranges (in millimeters) for 46 of the 50 species that were suc- 
cessfully identified by using DNA barcoding. Four species were excluded from the plot because of missing mea- 
surements of egg diameters. Egg diameters for the various identified species of anchovies (Anchoa mitchilli, 
Engraulis eurystole, and Anchoa hepsetus ) were based on long-axis measurements because these eggs are 
naturally ovoid in shape 
In general, the morphological identification of fish 
eggs has been problematic; therefore, the diversity of 
species subject to egg monitoring has been very lim- 
ited. For example, the daily egg production method has 
been used successfully to estimate the biomass of a 
number of clupeoid fishes as well as Atlantic mackerel 
(Scomber scombrus) (Alheit, 1993; Priede and Watson, 
1993; Stratoudakis et al., 2006). However, these suc- 
cesses have been limited taxonomically, and, for other 
species, substantial technical problems have emerged. 
In an effort to compare 2 independent estimates 
of seasonal egg production and mortality for Atlantic 
cod and haddock on Georges Bank, Lough et al. (2008) 
stated directly that the early egg stages of both spe- 
cies are indistinguishable from one another and, fur- 
thermore, that estimates of early-stage egg abundance 
based on extrapolations of morphological identifica- 
tions of late-stage eggs are likely inaccurate — stress- 
ing the need for improved spawning ground mapping 
with the development of DNA-based identification 
techniques. Similarly, in the Irish Sea, the use of Taq- 
Man DNA technology confirmed the overestimation of 
cod egg abundance due to an extremely high rate of 
morphology-based misidentifications of “cod-like” eggs 
(Fox et al., 2005). In lower latitudes, where species di- 
versity is much higher, the breadth of these problems 
is even more striking and the morphological identifi- 
cation of fish eggs is most often not even attempted 
(Kendall and Matarese, 1994). 
Although the cost of DNA barcoding has decreased 
in recent years (Richardson et al., 2007), the cost per 
sample still presents an issue- — especially for large- 
scale monitoring programs, such as ichthyoplankton 
surveys. For programs without any internal capabili- 
ties, the cost of having the molecular identification 
performed by an outside group is currently $14-20 per 
sample, for all steps in the process. For our program, 
performing this work internally would have required 
initial expenses to purchase equipment and set up lab- 
oratory space, reoccurring annual expenses to hire staff 
and maintain equipment, and the standard per sample 
expenses for reagents. The cost of doing the work inter- 
nally would, therefore, have far exceeded the costs of 
working with an external partner — a pattern we expect 
would hold for most monitoring programs. The sub- 
sampling procedure we have developed is one means of 
reducing total barcoding cost. The development of an 
automated egg measuring tool based on image analysis 
software allowed us to optimize our sampling. Although 
we did not target a specific taxon in our sampling, such 
a direction is possible if DNA barcoding is combined 
with an efficient subsampling tool to target specific 
