This chapter considers the predictability of yield from single and multispecies fisheries. We have calculated confidence intervals for equilibrium yield and equilibrium spawning stock biomass given estimates of the variability in recruitment and estimates of other biological parameters for a range of stocks. Then, using the concept of a notional "safe" threshold level for spawning stock biomass, we calculate the probability that the stock goes below the threshold in any one year as a measure of the risk to sustainability.

In a single species concept we compare average and the coefficients of variation in yield and spawning stock biomass for eight species groups using data from the literature. The variability of yield increases with the mean, with slow-growing, long-lived species having lower yields and spawning stock biomasses, as a proportion of unexploited biomass, than fast-growing species. In other words, there is a clear tradeoff between uncertainty and average yield. This uncertainty is a measure of the predictability of the resource.

Two large marine ecosystem examples are examined in more detail, the South Georgia fishery for krill and Antarctic icefish and the Barents Sea fisheries for cod, capelin, and herring. In each case, the probability of maintaining the stock above a threshold level, the risk curve, is dependent on the harvest rates of all the species in the fishery. Predator-prey interactions in a multispecies fishery can buffer the increase in uncertainty in yield that normally accrues with increasing harvest rate.

A critical feature of predictive yield models is the quantification of both the uncertainty in yield and the risk to sustainability due to depletion of the spawning stock.