Summary
A new method for estimating illegal fi shing effort is put forward. The results from this new method are similar to those of the Agnew and Kirkwood (2005) method, and this suggests that the current method is adequate under circumstances of low evasion and when good knowledge exists that zero observations refl ect zero illegal fi shing. The new method performs better in the case of zero detections and can potentially better handle the evasion of detection by illegal activity. Both the new and the current method suffer from the fact that the observation method used directly affects the system. This is the prevention/detection problem, in which the greater the number of detections for a given level of illegal fi shing, the more often the illegal fi shers will curtail their fi shing trips. This leads to a negative correlation between the amount of fi shing and the estimated amount of fi shing for a given number of illegal cruises. As the number of illegal cruises increases, both the estimate and the average amount of illegal fi shing increase. This gives some confi dence that the method can produce results that have a degree of legitimacy. However, the range of actual fi shing (in the simulation datasets) for a given estimated level of fi shing is very large. This range of uncertainty increases as the evasion rate increases. This research suggests that it would be possible to calculate a precautionary assessment of illegal fi shing such that the actual number of illegal fi shing days is less than, or equal to, the precautionary assessment with some given level of confi dence (for example 80%).