ABSTRACT. We investigated the influence of variations in the size of prey (Mallotus villosus) and a vertebrate predator (Gasterosteus aculeatus) on larval fish mortality rates during the period of yolk absorption using mid-size mesocosms (2.7 m". Increasing predator size increased mortality rates of capelin larvae. Variations in larval capelin size resulted in 2 distinct patterns. Between experimental trials, greater mean size of larval capelin in the mesocosm reduced mortality due to predation. Within an experimental mesocosm, larger larvae suffered higher mortality than smaller individuals. Contrasting patterns of size-dependent vulnerability to predation reflect the hierarchy of processes that determine the probability that a larval fish will be preyed upon. The broad scale response of the predator was determined by the mean relative sizes of prey and predator which govern the average probabilities of encounter, attack and capture. Within the search ambit of a predator (e.g area or volume searched within a complete diurnal foraging cycle) active prey selection for larger prey due to either greater visibility or higher energy reward was an important factor. A comparison of our results with estimated predation rates by the jellyfish Aurelia aurita indicates that at a similar size a gelatinous zooplankter consumes fewer larvae than a stickleback and is a less efficient predator as measured by the energy ingested relative to energy demands. For both vertebrate and invertebrate predators, the ratio of prey to predator lengths was a strong predictor of the daily mortality rate due to predation. Relative prey-predator sizes may provide a useful perspective to assess changes in larval fish vulnerability as they grow through a predator field. INTRODUCTIONCumulative survival during the early life history of fish is influenced by both growth and mortality rates (Houde 1987, Beyer 1989, Pepin 1991 1982, Christensen 1983, Alheit 1986, Folkvord & Hunter 1986, Pepin et al. 1987, Fuiman & Gamble 1988, Fuiman 1989 predators can feed heavily on the early life history stages of fish. Second, predation rates on fish larvae by invertebrate carnivores tend to decrease as larvae increase in size (Theilacker & Lasker 1974, Bailey 1984, Bailey & Batty 1984, Purcell et al. 1987) whereas the predation rates on fish larvae by vertebrate predators first increase as larvae increase in size (Folkvord & Hunter 1986, Pepin et al. 1987, Fuiman 1989, Fuiman & Gamble 1989) but eventually decrease as the larvae become more capable of avoiding attacks (Fuiman 1989). The pattern of vulnerability of larval fish to different predator types is O Inter-Research/Printed in Germany

We investigated the magnitude of short-term variations in the flux of several species of fish larvae in Conception Bay, Newfoundland, determined the contribution of these fluxes to changes in the abundance of these species, and assessed the factors that may be causing the variations in these fluxes. The net effect of transport ranged from a daily influx of 5.8% to a daily efflux of 6.2% from the population. Short-term variations in transport were associated with variations in wind stress. The contribution of transport to variations in population abundance ranged from 12 to more than 75% of the average daily rate of change in numbers for the different species. The contribution of physical processes to estimated mortality rates is variable and can have a significant impact on the investigation and interpretation of potential biological processes (e.g., predation and starvation) influencing the survival of larval fish. We predicted that the average correction to the instantaneous mortality rate due to advection and diffusion should be proportional to A−1/2, where A is the area of system under study. This highlights the susceptibility of small systems to estimation error of mortality rates caused by short-term variations in physical forcing.

We examined the sampling variability and capture efficiency of bongo nets and a modified Tucker trawl used in the sampling of ichthyoplankton and other macrozooplankton by taking seven replicate samples at each of two stations on two separate occasions. Sample variance was highly significantly related to sample mean for all major taxonomic categories (i.e., fish eggs, fish larvae, crustaceans, and medusae-chaetognaths). Sampling variability of the bongo nets was significantly greater than that of the Tucker trawl for both fish eggs and larvae. Sampling variability of macrozooplankton was significantly greater than that of ichthyoplankton. For larval fish of 5 mm or less, bongo nets generally yielded higher estimates of abundance than the Tucker trawl and the reverse was true for lengths above 10 mm, but there was a significant influence of species-specific morphological characters. The large sample volume of the Tucker trawl relative to the bongo nets resulted in significantly higher estimates of species diversity for fish eggs and larvae but not for crustaceans or medusae. Although bongo and Tucker samplers are efficient at catching a wide range of sizes of larval fish, the latter's lower variability may make it a more effective sampler.