78 
Fishery Bulletin 99(1 ) 
will be tend to be underestimated by the same amount at 
the two sites. 
For clarity of exposition, we have considered only the sim- 
plest possible study in which mean lengths of size-frequen- 
cy distributions might be compared. We would expect that, 
at least qualitatively, the power comparisons made in our 
study would apply equally to other, more complex, sampling 
designs. Likewise, the focus of our study has been on mean 
length (and thereby biomass), but we conjecture that the 
benefits of using a stereo-video system will apply equally 
well for the comparison of size-frequency distributions. 
Our calculations are based on length estimates made by 
divers and a stereo-video system in a controlled environ- 
ment with plastic silhouettes. Under field conditions with 
moving fish, variable light conditions, and with other ob- 
jects such as corals or kelps in the field of view providing a 
scale, diver and stereo-video estimates may be more or less 
accurate and precise, and thus change the power analysis. 
Conclusion 
A stereo-video system provides a new and alternative 
method for collecting length data on reef fish through 
nondestructive visual sampling. The use of a stereo-video 
system reduces measurement error substantially, thus 
increasing the statistical power of a monitoring program 
to detect changes in the mean length of a population of 
fish. When environmental managers set program goals 
of detecting small effect sizes (30% or less of the mean 
length of the target fish population) with a high statistical 
power (example 90%), the use of stereo-video technology 
