126 
Fishery Bulletin 119(2-3) 
weight (%W), mean percentage by number (%N), and fre- 
quency of occurrence (%FO) (Hyslop, 1980). Because 
samples were collected over a large spatial and temporal 
range, %W and %N analyses were modified to account 
for the patchy distribution of marine predators and their 
prey in space and time (Buckel et al., 1999). In our study, 
a cluster represented a set of traps or group of hook-and- 
line deployments consisting of various anglers and 
tackle configurations at a single location (reef patch). 
Indices of relative importance (IRI) (Cortés, 1997) were 
also calculated for each individual prey item and for 
prey aggregated into 10 broader taxonomic categories: 
amphipods, bivalves, bony fish species, cephalopods, 
copepods, crabs, polychaetes, shrimps, stomatopods, and 
tunicates. Differences in diet among size classes (300-— 
500 mm TL, 501—700 mm TL, and 701-900 mm TL) and 
depths of capture (0-30 m, 31-60 m, and >60 m) for red 
snapper were also explored by using IRI. 
A species accumulation curve was generated to eval- 
uate whether sampling intensity sufficiently character- 
ized the diet of red snapper. A linear model was used to 
assess the slope of the line fit to data for the last 5 stom- 
achs (with an asymptotic slope <0.05 indicating a satu- 
rated curve because no additional prey taxa are expected 
to be encountered with additional sampling; Bizzarro 
et al., 2009). Species richness was extrapolated by using 
a first-order jackknife estimator to predict the number of 
prey species that would completely characterize the diet 
(Heltshe and Forrester, 1983). The species accumulation 
curve and extrapolated species richness were computed in 
R, vers. 3.5.2 (R Core Team, 2018) by using the package 
vegan, vers. 2.5-2 (Oksanen et al., 2018). 
Results 
Red snapper (319-854 mm TL; Fig. 2) 
were collected throughout the sampling 
area of the Southeast Reef Fish Survey 
from 31°N to 34°N (Fig. 3) and from 
depths of 23-72 m. In total, 105 stom- 
achs from red snapper were analyzed: 86 
specimens of red snapper were collected 
by using chevron traps and 19 specimens 
were captured by using hook-and-line 
gear. Generally, fish prey items were 
digested well (Fig. 4) such that only ~13% 
were visually identifiable to at least the 
genus (digestion codes 1-2; Fig. 1), with 
the majority of fish prey items (~71%) 
assigned a digestion code of 3. A total 
of 65 fish prey items from 48 stomachs 
were designated as unidentified fish 
after visual examination. For the 53 
prey items from 30 stomachs subjected 
to molecular identification, the PCR suc- 
cess rate was ~89%. For ~87% of these 
prey items, usable sequences (mean 
length: 540 base pairs) were produced. 
A 
(oe) 
Frequency 
Most specimens for whom DNA analysis resulted in failed 
PCR and sequencing reactions were in a late stage of 
digestion (digestion code 3), and the failed reactions were 
likely due to low DNA yield as evidenced by faint bands 
on agarose gels or low peaks on chromatograms. All PCR 
controls performed as expected. 
DNA barcoding allowed 32 fish prey items that could not 
be identified visually to be identified to the species level, 6 
unidentified items to be identified to the genus level, and 
2 items to be identified to the family level (Suppl. Table) 
(online only). Therefore, by using DNA barcoding, 62% of all 
unidentified fish prey items in our study could be identified 
to at least the family level, and 58% and 49% of all uniden- 
tified fish prey items could be identified to the genus and 
species levels, respectively. Compared with the use of visual 
identification methods, use of DNA barcoding reduced the 
amount of unidentified fish prey items in the overall diet, 
with amount measured both by %FO, which decreased 
from 45.71% to 24.76%, and by %W, which decreased from 
18.75% to 9.78%. All prey identified to species shared >99% 
sequence similarity with reference sequences in either the 
GenBank or Barcode of Life Database, except for 2 items 
identified as Atlantic midshipman (Porichthys plectrodon), 
which shared 98% sequence similarity with a reference 
sequence in GenBank (accession number KF930305.1). 
Sequences from the 2 most closely related fish prey items 
that were genetically identified to the species level, both as 
species in the genus Synodus, were 19.3% dissimilar. 
Several prey items could not be identified to species. 
Specimens of prey identified as Ophidion sp. were just 
below the a priori defined species identification thresh- 
old, with 97% similarity in GenBank to the bank cusk-eel 
500 600 700 
Total length (mm) 
Figure 2 
Size—frequency distribution of red snapper (Lutjanus campechanus) collected 
along the Atlantic coast of the southeastern United States in 2017 and 2018. 
The dashed line indicates the mean total length. n=sample size. 
