Australian Journal of Ecology (1998) 23, 579 586 Comparisons of abundance of coral-reef fish: Catch and effort surveys vs visual census SEAN D. CONNELL,'* MELITA A. SAMOILYS,- MARCUS R LINCOLN SMITH' AND JOHN LEQATA' ^The Ecology Lab Pty Ltd, 14/28-34 Roseberry Street, Balgowlah, NSW2093, 'Fisheries, Queensland Department of Primary hidustries, PO Box 1085, Toivmville, Qld 4810, Australia, ""Fisheries Division, Ministry of Fisheries and Agriculture, PO Box G13, Honiara, Solomon Islands Abstract Catch per unit effort (CPUE) in fisheries science and visual counts in marine ecology are widely used to provide estimates of relative abundance. Concurrent use of these techniques therefore offers an opportunity for cross-validation. This study compares CPUE to underwater visual census (UVC) estimates of relative species abundance in a multispecies fishery: coral-reef fish in the Solomon Islands. Multivariate analyses showed large differences between CPUE and UVC estimates of abundance. The families Acanthuridae and Scaridae tended to be the primary cause of differences between the techniques when the full assemblage offish was analysed. However, the relationship between CPUF and UVC did not improve when these families were excluded from the data set and the analyses repeated on families (Serranidae, Lutjanidae, Lethrinidae) caught by the predominant gear type, handlining. This result highlights the point that the choice and use of particular methods require careful consider- ation in conjunction with the nature of factors being investigated. Many problems of sampling are specific to particular methods and some investigations may benefit from a more pluralistic approach. Key words: abundance, biomass, catch composition, coral-reefs, reef fish, sampling, underwater visual census. INTRODUCTION Knowledge of patterns of abundance is a fundamental requirement of wildlife research. Irrespective of the field of study, we are all confronted with the issue of how to quantify abundance (Caughley 1977; Southwood 1978; Andrew & Mapstone 1987; Wilson et al. 1996). It is widely accepted that estimates may not necessarily give accurate numbers of individuals, but can be used in a relative sense to establish patterns of abundance between different times or locations. Despite this, bias may not be constant among species or across time and space, hence uncritical use of meth- ods may yield unreliable descriptions of abundance. In fisheries research, catch per unit effort (CPUE) provides one of the most commonly used indices of abundance. It is assumed that CPUE is proportional to the abundance offish in a particular area (Richards & Schnute 1986). In practice, however, differences in catchability are likely to bias CPUE estimates. For ex- ample, factors such as variations m behaviour of fishes and fishers in different places and times will infiuence CPUE measurements. Serious errors in yield estimates and abundance indices based on CPUE have been *Present address: Centre for Research on Ecological Impacts of Coastal Cities, Marine Ecology Laboratories All, School of Biological Sciences, University of Sydney, Sydney, NSW 2006, Australia. Acccr" ^! tor publication March 1998. found for some fisheries and a plea has been made for validation of CPUE measurements as indices of abun- dance (Westrheim et al. 1990). Little research has assessed whether CPUE is an accurate index of abundance because CPUE is often the only practical method available for estimating abun- dance. Hence, a number of studies compare CPUF data among a number offishingmethods, but few com- pare CPUE data to fishery-independent estimates of abundance (but see Richards & Schnute 1986). Underwater visual censuses (UVC) have been used to estimate the abundance of reef fish since 1954 (Brock 1982) and became widely recognized from the 1970s (Russell et al. 1978). Comparisons of abundance obtained by UVC with complete censuses show UVC to have a good relationship with actual abundance for some species and types offish (Sale & Douglas 1981). In highly speciose areas it is generally recognized that the accuracy of counts increases when a limited number of species are surveyed (Williams 1986; Lincoln Smith 1989). Similarly, the accuracy of counts improves when the survey technique is appropriate to the group of species being counted (Lincoln Smith 1989). Therefore, it is not surprising that strong relationships between UVC counts and the abundance of fish collected by rotenone (Brock 1982; Lincoln Smith 1988), or explosives (Samoilys & Carlos 1992), have been demonstrated for certain species and UVC methods.