Migration or mixing of stocks may make it difficult to decide what population units we are dealing with. So far, fisheries reference points seem to have been expressed largely in terms of biomass or fishing mortality. Tag and recapture methods may be one of the few feasible ways of estimating mortality more or less directly, despite the time and cost involved, and some progress has been made in this area. It remains to be considered whether it is feasible economically to run continuous tagging activities as a means of estimating mortalities. A number of other possibilities remain that may be more feasible in the case of fisheries where relatively detailed spatial information is available. Several options seem worth exploring here:
1. It is assumed that following the start of a fishery, several stages may occur as progress is made from unfished to overfished conditions, and that this transition that might be picked up by a simple spatial index of aggregation for i = 1,2,3...N unit areas, such as that proposed by Gulland (1955).
Ig = [Sum (Ci)/Sum (fi)]/[Sum (cpuei)/N]
If this is the case, simple indices of concentration could be used to formulate LRPs designed to pick up unfavourable changes. The results of simulations might be used to specify situations where CPUE becomes low and uniform or where the area fished contracts in size with over-exploitation. The following stages might perhaps be looked for in sequence:
(a) Relatively few areas are fished, with initially high CPUE and relatively low effort.
(b) The number of areas fished increases, effort increases, and mean CPUE drops.
(c) CPUE declines and becomes more uniform with further area expansion.
(d) As fishing intensity increases further, CPUE drops again and "key areas' are 'fished out', with possible secondary areas of concentration of the fishery arising.
Figure 12 shows results of a simplistic hand-calculation designed to reproduce changes (a) to (d) and indicates that Ig may be sensitive to both changes in total area exploited and increased uniformity and decline in subregional CPUEs. It suggests that Gulland's or some similar index could be developed through simulation that could measure over-exploitation as it would show up in a spatially inhomogeneous and mobile population such as for tunas. These possible zoogeographical indicators of over-exploitation could be testable by a model of the type mentioned by Anganuzzi (1996) and could lead to some useful empirical indices of trouble on the horizon, i.e. when a geographical index based on correlation between effort and resulting CPUE suggests that the fishery is approaching stages (c) or (d) above.
2. Another option may be to use one or other variants of the mechanism of fisheries closures as an approach to setting precautionary management criteria. In this case, the proportion of the fishing grounds closed as ecological reserves might be regarded as a reference point whose effectiveness would have to be determined through a knowledge of the order of magnitude of migration or dispersal rates between unit areas. Such closure proportions used as LRPs might be determined by a combination of tagging experiments and dispersion models (Fig. 13).
3. A third possibility may be to dictate the level of fishing effort that may be exerted in biologically important individual unit areas in any given season such that local depletion is not achieved, especially where these area/season units are believed to be important for reproduction.
