In providing forecasts of run-strength and survival for the coho of northern British Columbia for 2000, Holtby et al. (2000) concluded that modest incidental catches of coho from the upper Skeena would not pose a significant risk of irreversible damage. They emphasized that forecasts for the area had not proven sufficiently reliable to proceed with modest incremental fishing without an early in-season indicator that would warn of unforeseen survival disasters such as the one that occurred in 1996 sea-entry. The purpose of this research document is to make a preliminary examination of the utility of four possible in-season indicators of run-strength and survival for northern B.C. coho.
Four fishery-performance measures were examined: the Skeena test-fishery index, which is essentially a CPUE of a river-mouth gill-net fishery, the wild coho CPUE in the SE Alaskan Tree Point gill-net fishery, the coho CPUE in the Alaskan boundary area troll fishery, and the upper Skeena CWT catch as a proportion of CWTs released in the SE Alaskan troll fishery. To ensure that these measures would serve as ‘early-warnings’, analysis was confined to data available by week 31 or roughly Aug. 1st. Forecast models for Babine and Toboggan hatchery coho and Lachmach wild coho, total stock size of the Babine Lake coho aggregate and the total stock size of the ‘average-streams’ of Areas 3, 4L, 4U and 6 were developed where there was a useful statistical relationship.
It is important that the early-warning schemes detect a sea-entry year like 1996 (return year 1997), which saw record low survival and escapement over much of northern B.C. (Holtby et al 1999). For upper Skeena coho marine survivals in 1992 and 1995 and 1998 were also very low and given the precarious state of upper Skeena coho returning off the 1997 brood, should be detected by any early-warning scheme. There are as yet no LRPs for northern B.C. coho. However, abundance values equal to 20% of carrying capacities from stock-recruitment analyses (Holtby et al. 2000) exceed levels seen in the upper Skeena in the years of poor survival. Values of 20% of the observed mean marine survivals also seem to exceed survival realized in most of the years noted above. Consequently, the models were evaluated using ‘trigger’ points set 20% of the long-term mean survival or 20% of the estimated carrying capacity of the stock. Where data-series are too short to establish these levels then the models were evaluated using trigger points set to approximately twice the values observed for the 1996 sea-entry year.
Of the four fishery-performance variables evaluated as early-warning forecast tools the CPUE in the Alaskan boundary area troll appears to be the most promising. Models based on this CPUE were predictive for most of the variables of concern and were able to reliably detect trigger situations while not giving more than one false trigger. The second best predictor was the proportion of CWTs recovered to week 29 in the Alaskan troll. Models based on this performance measure were able to detect trigger situations for upper and lower Skeena coho but failed outside of the Skeena and notably in Area 6. Furthermore, the models signaled spurious false triggers. The remaining two performance measures, the Skeena test-fishery index and CPUE in the Alaskan Tree Point gill-net fishery seem unsuitable as early-warning indicators.