24 
Fishery Bulletin 1 13(1) 
timation of species richness over an entire study area, 
species accumulation (i.e., rarefaction) curves may be a 
useful approach (e.g., Nichols et al., 1998; Thompson et 
al., 2003). Species accumulation curves and related ap- 
proaches (Angermeier and Smogor, 1995) may be espe- 
cially useful in diverse systems with many rare species 
(Green and Young, 1993; Gotelli and Colwell, 2001). 
Our study design included several simplifications. 
First, with our bootstrap procedure frames were se- 
lected at random for analysis. Alternative approaches 
may select frames systematically, either with fixed 
intervals (e.g., one frame every 30 s; Bacheler et ah, 
2013) or through adaptive sampling. Second, we esti- 
mated the proportion of species observed in a subset of 
frames in relation to all of the species observed in each 
20-min video segment. Ideally, our estimates would 
have been compared with the total number of species 
occupying the site, but true species richness at each 
site was unknown (Gotelli and Colwell, 2001). Third, 
we lacked information on current direction or magni- 
tude; therefore, we were unable to estimate the size or 
shape of the bait plume, information that can be impor- 
tant in determining the catch or counts of fishes made 
through the use of baited gears (Collins et al., 2002; 
Jamieson et al., 2006). Fourth, we did not account for 
temporal autocorrelation (i.e., samples taken closer in 
time are likely more similar than those taken further 
apart; Strachan and Harvey, 1996) when analyzing 
frames within a particular video. Temporal autocorre- 
lation violates the standard statistical assumption of 
independence among observations and, when present, 
may affect the estimated CVs. Temporal correlation 
is problematic for characterization of diel or seasonal 
variability but not for quantification of the density or 
number of species captured in a video. Temporal cor- 
relation could be minimized or avoided in practice by 
not choosing frames clustered in time. Fifth, our study 
would have been more informative if the costs of read- 
ing video frames were known, allowing for explicit 
cost-benefit analyses related to optimum sample sizes 
(Cochran, 1977; Thompson, 1992). However, these video 
data were recorded in a time in-time out format and 
not by individual frames, and, therefore, the costs of 
reading each frame could not be estimated. 
MeanCount, computed from a sequence of video 
frames, has been shown to track linearly with true 
abundance at a site (Conn, 2011; Schobernd et al., 
2014) — a critically important issue when standard- 
izing survey data to produce abundance indices for 
use in stock assessment models (Maunder and Punt, 
2004). Our study is the first, however, to document how 
the number of frames read can relate to CVs around 
MeanCount for reef fish species and the proportion of 
reef fish species observed at a site. Previous research 
has documented the general relationship between the 
spatial or temporal extent of sampling and CVs or 
the number of species observed (Fuller and Langslow, 
1984; St. John et al., 1990; Barker et al., 1993; Gledhill, 
2001). Similarly, we showed that the number of frames 
read was negatively related to CVs and positively re- 
lated to the proportion of species observed. More impor- 
tant, however, both relationships were nonlinear and 
indicate that the information gain slowed substantially 
after reading approximately 50 frames. Video studies 
that apply the MeanCount approach to other systems 
could use our GAM results to help broadly understand 
how many frames to read, accounting for the behaviors 
of the species of interest. 
Acknowledgments 
We thank M. Campbell, C. Gledhill, A. Pollack, and 
the Pascagoula laboratory of the NOAA Southeast 
Fisheries Science Center for providing access to the 
Gulf of Mexico reef fish video data, the staff and crew 
members who participated in data collection, and the 
Southeast Area Monitoring and Assessment Program 
for funding. We also thank M. Campbell, A. Chester, P. 
Conn, A. Hohn, T. Kellison, P. Marraro, Z. Schobernd, 
and 3 anonymous reviewers for comments on previous 
versions of this manuscript. 
Literature cited 
Al-Chokhachy, R., P. Budy, and M. Conner. 
2009. Detecting declines in the abundance of a bull trout 
( Salvelinus confluentus ) population: understanding the 
accuracy, precision, and costs of our efforts. Can. J. 
Fish. Aquat. Sci. 66:649-658. doi: 10.1139/F09-026. 
Angermeier, P. L., and R. A. Smogor. 
1995. Estimating number of species and relative abun- 
dances in stream-fish communities: effects of sampling 
effort and discontinuous spatial distributions. Can. J. 
Fish. Aquat. Sci. 52:936-949. doi: 10.1139/f95-093. 
Bacheler, N. M., C. Schobernd, Z. Schobernd, W. A. Mitchell, D. 
Berrane, G. T. Kellison, and M. J. M. Reichert. 
2013. Comparison of trap and underwater video gears 
for indexing reef fish presence and abundance in the 
southeast United States. Fish. Res. 143:81-88. doi: 
10. 1016/j.fishres.2013.01.013. 
Barker, R. J., J. R. Sauer, and W. A. Link. 
1993. Optimal allocation of point-count sampling ef- 
fort. Auk 110:752-758. 
Brooks, E. J., K. A. Sloman, D. W. Sims, and A. J. Danylchuk. 
2011. Validating the use of baited remote underwater 
video surveys for assessing the diversity, distribution 
and abundance of sharks in the Bahamas. Endang. 
Spec. Res. 13:231-243. doi: 10.3354/esr00331. 
Cao, Y., D. P. Larsen, and R. M. Hughes. 
2001. Evaluating sampling sufficiency in fish assem- 
blage surveys: a similarity-based approach. Can. J. 
Fish. Aquat. Sci. 58:1782-1793. doi: 10.1139/f01-120. 
Cappo, M., P. Speare, and G. Death. 
2004. Comparison of baited remote underwater video 
stations (BRUVS) and prawn (shrimp) trawls for as- 
sessments of fish biodiversity in inter-reefal areas of 
the Great Barrier Reef Marine Park. J. Exp. Mar. Biol. 
Ecol. 302:123-152. doi: 10. 1016/j.jembe.2003. 10.006. 
