Structure of fish communities of French Polynesian coral reefs. I. Spatial scales

Patterns of fish distribution in French Polynesia were studied at several spatial scales: a single transect (with 280 species, 22 stations), on Moorea Island; several transects on the same island (171 species, 14 stations) and transects from 5 islands including Moorea (189 species, 10 stations). There were significant variations in abundance, richness and composition of species among sites within the single transect. There were also variations among similar habitats withln one island and among islands. Differences among similar sites around one island were, however, small compared to variation across one transect at the same island. Possible explanations for these patterns are discussed. INTRODUCTION ing the Tiahura transect) and (3) among transects located in similar habitats on 5 islands (including Problems of sampling coral reef fishes, and in parMoorea Island) in French Polynesia. ticular the biases of different techniques, have been highlighted by several workshops of Australian (GBRMPA 1978, 1979), American (Barans & Bortone MATERIALS AND METHODS 1983) and French (Harmelin-Vivien et al. 1985) scientists over the last decade. Effects of the scale of samThe Tiahura area (Galzin & Pointier 1985) is situpling have been emphasised a number of times (e.g. ated in the northwest part of Moorea Island (17"301S, Sale 1980, Ogden & Ebersole 1981) and are 149"50f W) (Fig. 1). At Tiahura, the lagoon, which is summarized well by Sale et al. (1985): 'Reef fish 840 m wide on average, is characterised by a channel ecologists need to be able to determine the appropriate 5 to 12 m deep, separating the fringing reef from the spatial scale at which to attack a particular question barrier reef. This channel transfers water flowing the frequency of sampling and the duration of expericontinually over the barrier reef to a nearby ocean mental or monitoring programs which are necessary to pass. answer specific questions effectively'. A coast-to-sea transect 1020 m long was divided into Most descriptions of reef fish communities come from 22 sampling stations (Galzin & Legendre 1987). Samthe Carribbean, Australia and the Red Sea. By compling was in bands measuring up to 200 m long and parison, the fish of Central Oceania are poorly 50 m wide, centered at each station and done using 2 described. As part of an on-going study of fish on reefs complementary techniques: visual observations and in French Polynesia (Galzin 1985), I sampled a number rotenone stations. For the first, the presence or absence of reefs to determine the spatial and temporal scales of all readily visible species was noted at all stations necessary to understand the ecology of the fish, during a haphazard swim of 45 min within each station. Descriptions of these reef systems may be found in For the second, single coral patches of about 0.5 m3 Delesalle et al. (1985). were selected within each station and poisoned with In this article spatial distributions are described at 3 rotenone. The presence/absence data (from visual surscales: (1) along a single, arbitrarily-chosen transect veys and poison stations) were analysed using the (Tiahura on Moorea Island), (2) among 5 transects 'chronological clustering' of Legendre et al. (1985). representing different places on Moorea Island (includThis method partitions multi-species data series into O Inter-Research/Printed in F. R. Germany 130 Mar. Ecol. Prog. Ser. 41: 129-136, 1987 250 350 84