Effects of size-selective mortality and sampling bias on estimates of growth, mortality, production and yield