Marine Biology (1994) 119:179-184 9 Springer-Verlag 1994 M. S. Johnson 9 R. J. Watts 9 R. Black High levels of genetic subdivision in peripherally isolated populations of the atherinid fish Craterocephaluscapreoli in the Houtman Abrolhos Islands, Western Australia Received: 18 January 1994 ! Accepted: 19 January 1994 Abstract The atherinid fish Craterocephalus capreoli Rendahl is abundant in the Houtman Abrolhos Islands, 70 km off the Western Australia coast and =250 km south of the southern limit of the range of the species along the mainland. Electrophoretic examination of 7 allozyme loci at 17 sites in the Houtman Abrolhos revealed a substan- tially lower level of polymorphism than found in an ear- lier study of the species in its mainland distribution, with many of the uncommon alleles and some common ones missing. There is a very high degree of genetic subdivi- sion among the populations in the Houtman Abrolhos, measured by a mean FsTof 0.437 over a distance of 35 km. This Fsr (standardized variance in allelic frequencies) is six times that found previously among populations along the mainland coast over distances up to 850 kin. The sub- division of populations in the Houtman Abrolhos is simi- lar within one island group on a scale up to 12 km, and between two groups that are separated by 15 km of deep water. Significant differences in allelic frequencies were found between populations from the open shore and en- closed lagoons less than 800 m apart, but the overall lev- els of subdivision were similar for the two types of envi- ronment. Previous work had shown high levels of genetic subdivision in the Houtman Abrolhos for a gastropod with direct development. The results for C. capreoli demon- strate that the archipelago favours subdivision even for a species with potentially much greater mobility and differ- ent life history. Communicated by G. F. Humphrey, Sydney M. S. Johnson ([]) 9 R. Black Department of Zoology, The University of Western Australia, Nedlands, Western Australia 6009, Australia R. J. Watts Queensland Department of Primary Industries, Southern Fisheries Centre, RO. Box 76, Deception Bay, Queensland 4508, Australia Introduction Genetic subdivision depends on an interaction between the life history of a species and the environment. In general, marine species with inherently less potential for gene flow show a higher degree of genetic subdivision among popu- lations (e.g. Burton and Feldman 1982; Waples 1987; Ward 1990). The potential for gene flow is generally great in the marine environment, and species with planktonic larvae typically show low levels of genetic subdivision over large distances (e.g. Gyllensten 1985; Palumbi 1992). In the face of this large-scale genetic homogeneity, it is of special interest to search for conditions that favour genetic subdi- vision in marine species. Genetic studies of marine species in Western Australia are providing a comparative basis for determining these conditions. Studies of several species have confirmed that low levels of genetic divergence over distances of hundreds or even thousands of kilometres are common on this coast- line (Johnson and Black 1983; Ayre 1984; Watts et al. 1990; Johnson and Joll 1993; Holborn et al. 1994; Johnson et al. 1994). Superimposed on this basic pattern of uniformity, however, is enhanced genetic subdivision of inshore spe- cies associated with some types of environment, including estuaries and embayments (Johnson et al. 1986; Watts 1991, Ayvazian et al. 1994) and islands (Watts 1991). The Houtman Abrolhos Islands are well suited for ex- amining the effects of such environments on genetic sub- division of inshore species. The archipelago spans --70 km north to south, and consists of three islands groups which are 10 to 15 km wide and separated by deep-water chan- nels 15 to 30 km across. Each group includes many small islands protected by a bordering reef. On many of the is- lands there are enclosed lagoons (which we refer to as "lakes") (Collins et al. 1991). A study of the gastropod Bembicium vittatum Phillipi in these contrasting habitats provided a striking example of the combined effects of life history and type of environment on the extent of genetic subdivision (Johnson and Black 1991). This species lacks a planktonic larval stage, and shows correspondingly high