Chapman, M. AnnGreen, John D.Stephens, R.T. Theo2026-04-012026-04-011984https://hdl.handle.net/10289/18171This study examines smelt (Retropinna retropinna) population biology in Lake Taupo, New Zealand, and describes the development of a model which simulated their population dynamics and interactions with rainbow trout (Salmo gairdneri), their principal predator. The model was used to predict the effects of variation in habitat features on smelt population processes and on their consumption by trout. Implications for fisheries management are discussed. The smelt population was comprised of three year classes. The juvenile 0+ year class occupied only the pelagic zone and fed exclusively on plankton whilst the adult 1+ year class also occupied the littoral zone and insects present there were included in their diet. A few smelt either failed to spawn at two years old or survived spawning. These grew rapidly, fed on larval bullies (Gobiomophus cotidianus) and spawned in late winter. Juvenile smelt reached 28-34mm FL after one year, 40-55mm FL after two years and 60-110mm FL when nearly 3 years old. Growth occurred all year but was most rapid between May and November. Spawning commenced in October when adults first became abundant in the littoral zone and continued until April. Females often recovered and spawned a second time, 6 - 8 weeks after the first spawning. Males were sexually active for an extended period. Eggs were scattered over sandy areas along beaches in water less than 3.0m deep and in the lower reaches of tributary streams where water currents were sufficient to prevent formation of algal mats and deposition of silt or detritus. Egg mortalities were correlated with water temperature but not with their density, which reached 5000m⁻² off Waihaha Beach and 40,000⁻² in the Waihora and Whanganui streams. Habitat perturbations caused by extreme weather conditions were thought to cause the most serious egg mortalities. Starvation was probably the principal source of larval mortality as there was little predation on 0+ smelt. Larger (730mm) 1+ smelt were the most important item in the diet of rainbow trout, comprising over 80% by volume of items eaten. Trout tended to select the larger smelt available and the intensity of selection was correlated with measures of smelt abundance. Trout generally gathered in places where 1+ smelt densities were high. Smelt population dynamics were simulated using length and density dependent models for movements between habitat zones, feeding, growth or starvation, breeding and predatory mortality. The form of simulated length frequency distributions largely determined the rates of these processes and was itself modified by the operation of each process. Modal groups occurred at lengths where food intake was equivalent to metabolic maintenance requirements and this was controlled by the interaction of diet breadth, the size structure and quantity of the food resource and the number of fishes sharing the same food particle size range. Growth of modal groups was controlled both by the duration of the spawning season and by seasonality in food resource quantity and size structure. Pelagic productivity and the ratio of littoral to pelagic habitat area were the most influential habitat features controlling smelt densities while densities of medium sized (30-55mm) smelt and water clarity were the most important factors controlling simulated quantities of smelt eaten by individual trout. Trout predation exerted little influence on simulated smelt population dynamics and consequently, the quantity eaten by individual trout was little affected by trout density. This result and comparison with trout stocking rates used in Canada suggest that a 1 - 2 order of magnitude increase in numbers of trout released annually into the larger Rotorua lakes would not significantly reduce their growth rates.enAll items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.Thesis