Bacheler and Shertzer: Estimating relative abundance and species richness from video surveys of reef fishes 
19 
Table 1 
Mean duration and standard error of the mean (SE), measured in seconds (s), of individual fishes in video, mean number of 
individuals in videos, and mean probability that a fish species would be seen in a video segment (for those videos in which 
that species occurred) summarized by family for only those species seen in at least 10 videos from footage collected during 
the National Marine Fisheries Service’s reef fish video survey conducted in the Gulf of Mexico in 2001-2002 and 2004-2007 
as part of its Southeast Area Monitoring and Assessment Program. Mean probability of being seen in a video was calculated 
for each species as the mean proportion of videos in which a species was observed (on the basis of 25 randomly selected 
frames) over all videos in which that species was present. Note that the family names for Labridae, Serranidae, and Scaridae 
follow the Integrated Taxonomic Information System (http://www.itis.gov). Standard errors of the means (SE) are provided 
in parentheses. 
Number of 
Mean duration 
Mean number 
Probability of being 
Family 
Common name 
species 
(s) 
of individuals 
seen in video 
Opistognathidae 
jawfishes 
i 
504 
22 
1.00 
Priacanthidae 
bigeyes 
2 
921 (SE 127) 
3 (SE 1) 
0.99 (SE 0.01) 
Holocentridae 
squirrelfishes 
1 
132 
10 
0.96 
Pomacanthidae 
angelfishes 
5 
44 (SE7) 
16 (SE 9) 
0.85 (SE 0.03) 
Balistidae 
triggerfishes 
3 
66 (SE 40) 
11 (SE 4) 
0.84 (SE 0.07) 
Pomacentridae 
damselfishes 
5 
35 (SE 6) 
23 (SE 7) 
0.84 (SE 0.04) 
Labridae 
wrasses 
7 
15 (SE 2) 
31 (SE 8) 
0.80 (SE 0.06) 
Serranidae 
sea basses and groupers 
23 
30 (SE 2) 
12 (SE 2) 
0.78 (SE 0.02) 
Chaetodontidae 
butterflyfishes 
3 
31 (SE 3) 
11 (SE 2) 
0.77 (SE 0.03) 
Malacanthidae 
tilefishes 
2 
28 (SE 11) 
8 (SE 0) 
0.76 (SE 0.05) 
Sparidae 
porgies 
6 
14 (SE 2) 
22 (SE 10) 
0.75 (SE 0.03) 
Acanthuridae 
surgeonfishes 
3 
26 (SE 6) 
6 (SE 1) 
0.73 (SE 0.02) 
Haemulidae 
grunts 
4 
25 (SE 8) 
22 (SE 6) 
0.73 (SE 0.05) 
Lutjanidae 
snappers 
7 
12 (SE 1) 
37 (SE 9) 
0.73 (SE 0.04) 
Scaridae 
parrotfishes 
4 
23 (SE 2) 
13 (SE 5) 
0.71 (SE 0.05) 
Tetraodontidae 
puffers 
2 
25 (SE 7) 
4 (SE 1) 
0.64 (SE 0.13) 
Mullidae 
goatfishes 
2 
11 (SE 4) 
14 (SE 6) 
0.60 (SE 0.00) 
Muraenidae 
morays 
2 
40 (SE 4) 
4 (SE 2) 
0.60 (SE 0.07) 
Carangidae 
jacks 
6 
6 (SE 1) 
14 (SE 2) 
0.40 (SE 0.04) 
Sphyraenidae 
barracudas 
1 
18 
4 
0.34 
Scombridae 
mackerels 
1 
5 
2 
0.17 
eo segment, and probability of being observed in each 
video for each of the families of fishes included in the 
analysis described previously in this section (Table 2). 
The purpose of including this table is to inform read- 
ers working in tropical and subtropical oceans about 
those groups of species they are likely to see and those 
groups that they are likely to miss if adopting a Mean- 
Count approach where a subset of frames is read. 
Results 
MeanCount bias and precision 
The MeanCount estimator behaved similarly for the 
3 species that we used as case studies. The central 
tendency across bootstrap replicates, represented by 
mean MeanCount, converged rapidly for red snapper, 
vermilion snapper, and scamp as the number of frames 
read increased (Fig. 2). MeanCount values for scamp 
and red snapper were less variable than the results for 
vermilion snapper (on the basis of 5 th and 95 th percen- 
tiles), and variability for all 3 species decreased when 
more frames were read (Fig. 2). 
Across all sampling events (i.e., all 20-min videos 
analyzed) in which the focal species was observed, 
there were no obvious biases in MeanCount for red 
snapper, vermilion snapper, or scamp at any level of 
sampling intensity for 25 to 200 frames read (Fig. 3). 
The variance for each species decreased as the number 
of frames read increased, a finding consistent with the 
results of the individually selected video analysis pre- 
viously described (Fig. 3). Furthermore, the variance 
surrounding MeanCount was approximately 50% lower 
for scamp than for either red snapper or vermilion 
snapper (Fig. 3). 
The relative patterns of MeanCount CVs were near- 
ly identical among the 3 species (Fig. 3). As the num- 
ber of frames increased from 1 to 200, the decrease 
in CV was initially rapid and then more gradual as 
more frames were read (Fig. 3). Because of this pat- 
tern, the largest reduction in CVs for all 3 species oc- 
curred as the number of frames read increased from 1 
to 50. When frames read increased from 50 to 200 (i.e., 
