AB 11:101-112 (2010)  -  DOI: https://doi.org/10.3354/ab00292

The visual fast count method: critical examination and development for underwater video sampling

Jon Barry*, Roger Coggan

Centre for Environment, Fisheries and Aquaculture Science (Cefas), Lowestoft Laboratory, Pakefield Road, Lowestoft, Suffolk NR33 OHT, UK

ABSTRACT: The visual fast count (VFC) estimator was originally developed for censusing reef-fish communities by diver observation, but has more recently been used for estimating abundance of the benthic taxa observed in underwater video transects. It is effectively a sub-sampling method that helps estimate the abundance of visible taxa by segmenting the observation period (e.g. a dive or video record) into equal aliquots and recording the frequency of each taxon only within the first segment in which it appears. Multiplication factors are applied to these counts to estimate the total abundance for the whole observation period. We were interested in the bias that this method may have, as not all species are evenly distributed throughout the observation period. Consequently, we examined the statistical characteristics of the VFC method through simulation studies, with a view to understanding this bias and informing decisions about the use of the VFC in certain applications. We derived the theoretical expected value of the abundance estimator for a random uniform spatial pattern of individuals and also for a clustered spatial process where the number of individuals per unit area follows a negative binomial distribution. We used these results to create new VFC estimators, all of which reduced the bias inherent in the original estimator. We show that this bias is greatest when the mean number of animals is small. We then carried out a number of simulation studies representing random uniform, clustered and regular patterns and compare the standard VFC estimator to a number of other estimators, including the new ones we have developed. In some circumstances, VFC estimators can provide good estimates of species abundance. However, they are also vulnerable to bad performances—particularly if the sampling transect is not homogeneous. If VFC is to be used, we recommend a VFC method based on a method-of-moments estimator, assuming a negative binomial distribution where the order of looking at video segments is randomised.

KEY WORDS: Transect surveys · Negative binomial distribution

Full text in pdf format 
Information about this Feature Article  
Cite this article as: Barry J, Coggan R (2010) The visual fast count method: critical examination and development for underwater video sampling. Aquat Biol 11:101-112. https://doi.org/10.3354/ab00292

Export citation
Mail this link - Contents Mailing Lists - RSS
- -