Population biology and dynamics of the gummy shark (Mustelus antarcticus) harvested off southern Australia

Deposited with permission of the author. © 2010 Dr. Terence Ivan WalkerThe aim of the present study was to determine the key parameters describing the population biology and the fishing gear selectivity of gummy shark (Mustelus antarcticus) and to develop three fishery population models that make use of these parameter estimates and available fishery monitoring data. Observational biological data and experimental fishing data were collected during 1973–76, 1986– 87, and 1990–01, and monitoring data on annual catch are available since 1932 and on fishing effort and sex- and length-frequency composition of the catch since 1970. Movement rates among various broad regions of the fishery, estimated using the Integrated Tag Model, are sufficiently low to treat M. antarcticus as several separate stocks across southern Australia for assessment purposes. The sex- and age-structured Stock Assessment Model (main model) consists of a series of equations forming ‘sub-models’, where parameters representing shark growth and reproduction and gillnet and hook selectivity were estimated independently of the model (exogenously) and parameters representing catchability and natural mortality were estimated by the model (endogenously). The model avoids the common assumption of constant annual recruitment and avoids estimating stock-recruitment parameters to incorporate density-dependent population-regulation. Instead, the model calculates the number of births each year determined from the proportion of surviving female sharks in the population breeding annually (expressed as a function of shark length by a logistic equation) and from the number of births per female at each pregnancy (expressed as a function of maternal length by an exponential equation). A ‘u-shaped’ function to vary natural mortality with age over the entire life span of the sharks replaces the usual assumption of constant natural mortality for animals recruited to the exploited phase of the life cycle. This allows natural mortality to decrease rapidly with age during the early ages of life, as the sharks grow and become better able to catch prey and to escape predators, and allows natural mortality to increase during later ages with senescence. In addition, natural mortality varies with stock biomass to allow for density-dependent regulation of the population. Density-dependent natural mortality is the principal mechanism for population regulation in M. antarcticus, evidenced from undetectable decompensation in growth rate and in reproductive rate when measured before and after major changes in population size in response to large changes in fishing mortality. The model provides forward projections of the effects of alternative harvest strategies on stock biomass, recruitment, and catch for each of the two specific regions of Bass Strait and South Australia in southern Australia. Separate estimates of growth parameters for the population of M. antarcticus in Bass Strait during 1973–76, when comparatively lightly fished, and during 1986–87, when heavily fished, indicate a reduction rather than an increase in growth rate, as expected had there been a density-dependent response in growth rate to population depletion. Similarly, estimates of reproductive rates in waters east of longitude 138°E (eastern end of Kangaroo Island) for 1973–76 and 1986–87 indicate a reduction rather than an increase in reproductive rate, as expected had there been a density-dependent response in reproductive rate to population depletion. Consistent with this response, the reproductive rate increased from 1986–87, when heavily fished, to 1998–01, when moderately fished. Hence, if density-dependent changes in growth and reproductive rates occurred in response to changes in population size caused by changes in fishing mortality, then some other stronger offsetting effect, such as the effect of length-selective fishing mortality, must have masked any measurable density-dependent response. Selectivity (expressed as a function of mesh size for gillnets or hook size and of length of shark according to the gamma probability distribution function) markedly affected parameter estimates for growth and reproduction. The von Bertalanffy and Fabens growth equations were reformulated and reparameterised with Francis parameters to provide stochastic frameworks to correct the sampling bias caused by the selectivity of gillnets used to capture the sharks and to represent variation in length-at-age to simulate the effects of length-selective fishing mortality. Comparison of mean length of shark at each age among regions and periods, detection of ‘Rosa Lee’s phenomenon’ from comparing radii of vertebral growth-increment bands among several ages, and application of the Length-Selective Simulation Model provide evidence for the ‘phenomenon of apparent change in growth rate’ caused by high length-selective fishing mortality. The ovarian and parturition cycles are mostly annual to the west, but biennial to the east of Kangaroo Island, although in both these regions, the gestation period is about 1 year and parturition occurs mostly during November– December. The difference in the reproductive cycle between west and east of Kangaroo Island is attributable to environmental differences. Differences in male and female maturity ogives and in maternity ogives among the three data collection periods east of Kangaroo Island, however, are attributed to the ‘phenomenon of apparent change in size-at-maturity’ and the ‘phenomenon of apparent change in size-at-maternity’ caused by length-selective fishing mortality. Mustelus antarcticus predominantly inhabits sandy substrates in depths mostly less than 80 m, with neonates, other juveniles and large females tending to be inshore, but the species does not have well defined nursery areas. The species has a tendency to form aggregations with animals of similar sex and size. Experimental fishing indicated that both gillnets and longlines are much more effective at catching chondrichthyan species than teleost species, and the number of animals caught for species of cephalopoda, bivalvia, gastropoda, mammalia, aves and reptilia are negligible or zero. The effect of gillnet mesh-size on catch rates is strong, whereas the effects of gillnet hanging-ratio, hook size, hook shank-length, and hook spacing are all weak. Because the effects of gillnet mesh-size on catch rates and size of shark caught are strong, it was necessary to develop an appropriate sub-model to express relative selectivity of gillnets as a function of mesh size and of length of shark for inclusion in the Stock Assessment Model. Conversely, because the effects of gillnet hanging-ratio, hook size, hook shank-length, and hook spacing are all weak, these effects do not need to be accounted for in the Stock Assessment Model. Tailoring the models, and where necessary developing new methods to account specifically for the peculiarities of shark biology and for the complexities of highly length-selective fishing gear deployed in the fishery markedly reduced risk associated with uncertainty from the lack of knowledge and has resulted in improved predictive power. There is now high confidence in the stock assessments of M. antarcticus and the fishery is now one of Australia’s most securely managed fisheries. Furthermore, the present study demonstrates that the effects of sampling and, more importantly, of intense fishing using length-selective gear produce various biases in the data and parameter estimation that are likely to occur in many other fisheries. The study also demonstrates that a shark species of medium biological productivity can be harvested sustainably, whereas earlier there were doubts about whether any shark species could be sustainably harvestable