larval cruising speeds range between 1-3 body lengths per second and speed increases 

 as larvae grow (Hunter, 1980; Hunter & Kimbrell, 1981). These figures provide a 

 starting point for evaluating how far away a larval neighbor might be and still 

 provide positive local enhancement. 



Larval food patches commonly occur in the open sea (Owen, 1981) and fish larvae 

 do find food patches (Sherman, et al.. , 1981). When turbulent sea conditions 

 disrupted food aggregations and diluted potential food items to below densities 

 needed for larval survival, first-year anchovy recruitment declined markedly 

 (Lasker, 1981). 



Once in a food patch, larvae tend to remain there by employing non-random search 

 patterns. When larval anchovies entered dense food patches they decreased swimming 

 speed and time spent swimming, and increased their turning probability by a factor 

 of 5-6 (Hunter, 1980). The result was a substantial increase in the probability of 

 staying within the food patch. 



The onset time of these oriented search patterns influences dispersion. The 

 sooner they begin in larval development, the less widely dispersed larvae will be at 

 the moment mechanisms for maintaining patchiness become operative. This leads to a 

 consideration of the time of first feeding by young larvae. Generally, first 

 feeding precedes or coincides with onset of swimming (Hunter & Kimbrell, 1981). 

 Temperate fishes begin to feed 2-4 days after hatching. Since development is 

 faster in tropical than in temperate waters, and armed with the above estimates of 

 onset of swimming, we can estimate that feeding in coral reef fishes begins 

 0.5-3 days after hatching. 



We see, then, that there should be strong selective pressures for eggs and 

 larvae to aggregate, as an anti-predator device, for local enhancement of food- 

 patch searching, or simply to remain with food patches once they are found. In 

 order to form or remain in patches, larvae must have appropriate motor and percep- 

 tual skills and we have seen that swimming ability and feeding emerge early in 

 larval development. 



Another mechanism by which larvae may limit dispersion is by diurnal vertical 

 migration. Surface waters in the mixed layer suffer greater turbulent diffusion 

 than deeper waters (Okubo, 1974). Particulate organic matter and many planktonic 

 organisms concentrate at the boundary between layers. While vertical stratifica- 

 tion of coral reef fish larvae has scarcely been studied (Watson & Leis, 1974), 

 larvae may find a rich source of food if they migrated to the boundary, and there 

 they would undergo less turbulent diffusion than if they remained near the surface. 



Regardless of the mechanisms, dispersion of fish larvae apparently does not 

 continue throughout the entire pelagic period. Temperate larvae disperse for a 

 limited period and then re-aggregate (Hunter, 1980; Hewitt, 1981). Thus, the 

 duration of passive dispersion may be relatively short. 



Thus far, discussion has been directed to processes influencing dispersion of a 

 single spawn of eggs or larvae. Spawning, however, generally involves all or part 

 of a local population reproducing during the same period. Turbulent diffusion 

 operating on multiple egg masses will mix offspring from adjacent spawns. The 

 larger the number of eqq masses that are mixed and the greater the thoroughness of 

 mixing, the lower will be the probability that small clumps of eggs or larvae late 

 in the pelagic phase will contain kin. 



The amount of mixing will depend on slightly different factors for pelagic and 

 demersal spawners. For pelagic spawners, critical factors will be the time and 

 distance separating successive spawning events, and current speed. For example, if 

 current speed over a reef is a modest 6.2 m/min (0.2 knots), pelagic spawnings 1-4 

 hours apart, as in some parrotfish (Clavijo, 1982), would initially produce egg 

 masses separated by 0.4-1.5 km due to current plus whatever distance separated the 

 spawning sites. With eggs concentrated 24 hours later in the center of a circle 

 700 m in diameter in a Gaussian distribution, eggs from early and late spawns would 



42 



