22 
Fishery Bulletin 1 13(1) 
Figure 4 
Examples of species richness (total number of species) 
observed as a function of the number of video frames 
that were read for this study of the use of video surveys 
to index abundance and diversity of reef fishes in the 
northern Gulf of Mexico in 2001-2002 and 2004-2007. 
Each panel is from a different 20-min video segment 
that was analyzed and selected to represent a relative- 
ly (A) low, (B) medium, or (C) high number of species. 
In each panel, the solid line is the median value from 
bootstrap replicates, the lower dashed line is the lower 
5 th percentile, the upper dashed line is the 95 th percen- 
tile, and the dotted horizontal line is the total number 
of species observed in the full video (across all 1201 
video frames). 
we examined the tradeoff between time spent reading 
videos and the information obtained. In this study, we 
found that reading more frames decreased variability 
surrounding MeanCount for 3 reef fish species and in- 
■o 
a> 
> 
a) 
if) 
n 
o 
in 
a> 
o 
CD 
CL 
if) 
1 . 0 - 
0 . 8 - 
0 . 6 - 
o 0.4- 
c 
o 
o 0.2- 
o 
al 
0 . 0 - 
1 1 1 1 r - 
25 50 100 150 200 
Number of frames 
Figure 5 
Proportion of species observed across all 20-min videos 
analyzed in this study as a function of the number of 
video frames that were read for this study of the use 
of video surveys to index abundance and diversity of 
reef fishes in the northern Gulf of Mexico in 2001-02 
and 2004-2007. A value from each 20-min video was 
computed as the mean estimate of species richness 
(i.e., mean of the number of species observed across 
1000 bootstrap replicates) divided by the total number 
of species known to be present in that video segment 
(i.e., observed in any of the 1201 video frames). Boxes 
represent the interquartile range, thick heavy lines 
represent medians, and whiskers extend to the most 
extreme data point within 1.5 times the interquartile 
range from the box. 
creased the total number of species observed, but bias 
was negligible even when a small number of frames 
were read (e.g., F= 25). These results will be useful to 
researchers in designing and tailoring their underwa- 
ter video surveys to incorporate MeanCount for estima- 
tion of relative abundance or species richness. 
Previous studies have shown that the number of 
taxa encountered in a wide variety of fisheries and 
wildlife monitoring studies is related to the spatial 
or temporal extent of sampling (Fuller and Langslow, 
1984; St. John et ah, 1990; Barker et ah, 1993; Gled- 
hill, 2001). We observed an inverse power relationship 
between CVs and number of frames read and an as- 
ymptotic relationship between the number of species 
observed and the number of frames read. Therefore, 
CVs decreased and the number of species observed in- 
creased dramatically as the number of frames read in- 
creased from 1 to 50, but gains in precision were much 
more modest after that point. These results are simi- 
lar to results from studies of stream fishes that have 
documented a threshold of sites sampled beyond which 
the increase in species observed was negligible (Anger- 
meier and Smogor, 1995; Cao et ah, 2001; de Freitas 
Terra et ah, 2013). The number of frames that should 
