tailed significance level, Po and the probabilities 

 of each more extreme table with the same row 

 and column totals which might have occurred, 

 are summed and multiplied by 2. For example, 

 the significance calculation for the iy2-yr lag 

 table, for Oregon, is 



9 



= 0.0179 



1 



10 



10 



=0.0016 



11 







11 



12 



12 







= 0.0006 



=0.0001 



where significance level, « = 2(Po + Pi + P2 + 

 P3) = 0.0404. The significance levels for dif- 

 ferent lag times for all three states are pre- 

 sented in Table 1. (The null hypothesis of inde- 

 pendence was rejected at the 5% significance 

 level.) 



The results indicate that crab catch and up- 

 welling are related as hypothesized. The rela- 

 tionship is that summers of strong upwelling 

 generally produce a good crab catch IV2 yr later 

 along the coasts of northern California and 

 Oregon. Along the coast of Washington, sum- 

 mers of strong upwelling almost always produce 

 a good crab catch the following winter. The con- 

 verse is also true (weak upwelling results in 

 poor crab catches). It must be emphasized here 

 that a qualitative statement is all that these 

 statistics will allow. 



Discussion 



The observed lag times probably are a direct 

 result of slow energy transfer through the pe- 

 lagic environment to the benthos. The benthic 

 predators may not experience upwelling-induced 

 increases in food resources until many months 

 after the onset of upwelling. Stephens, Sheldon, 

 and Parsons (1967) determined that deposition 

 of phytogenous material did not occur until 2 

 months after commencement of the annual 

 spring phytoplankton bloom, in Departure Bay, 

 British Columbia. Several times the generation 

 time of the dominant copepods would be re- 

 quired after commencement of the bloom before 

 copious amounts of fresh copepod faecal material 

 would begin settling to the sea floor. Population 

 increases of the benthic filter feeders and de- 



posit feeders would lag additionally by the 

 generation time of each species. Some of these 

 animals (amphipods, polychaetes, clams, and 

 shrimp) are known Dungeness crab prey. Other 

 preys include small fish and small crabs (Mac- 

 Kay, 1942; Waldron, 1958). 



Two-year old and older crabs moult annually. 

 This event occurs during the summer off north- 

 ern California, late summer-early fall off Ore- 

 gon, and late fall off Washington (Poole and 

 Gotshall, 1965, for northern California; Snow, 

 pers. comm., for Oregon; Cleaver, 1949, for 

 Washington). It is hypothesized that the Cali- 

 fornia and Oregon crabs moult before benefitting 

 from increased food resources, so their response 

 to a summer of strong upwelling is not exhibited 

 until the next summer's moult. Washington 

 crabs however moult several months after the 

 end of the upwelling season, and therefore must 

 benefit from the increased production since the 

 lag between strong upwelling and a good fishery 

 is only V2 yr. 



This hypothesis implies that a summer of 

 strong upwelling off California and Oregon 

 serves to maintain a large crab population 

 throughout the coming year by both increasing 

 food availability and decreasing competition 

 for the food. Although lower food availability 

 resulting from a summer of weak upwelling 

 implies increased competition for food, increased 

 mortality of sublegal adults need not be the 

 only outcome of such competition. Another pos- 

 sibility is decreased moulting success. If a crab 

 has not accumulated enough food reserves dur- 

 ing a year of feeding, it need not shed its cara- 

 pace. Numerous workers have observed from 

 tagging studies and observations on the degree 

 of fouling of the carapaces that Dungeness crabs 

 can wait 2 yr before moulting (MacKay, 1942; 

 Cleaver, 1949; Waldron, 1958; Butler, 1961; 

 Snow and Wagner, 1965; Poole, 1967). The pre- 

 valence in the field is unknown. Poole (1967) 

 believes that as many as 10-15% of the male 

 crabs may miss moults. If moults are missed, 

 animals would not enter the fishery until an 

 additional year later, i.e.. 11/2 yr after a partic- 

 ular upwelling season in Washington and 2V2 

 yr after in Oregon and California. 



Graphically, I have shown (Figures 3, 4, 5) 

 that trends of annual variability of crab catch 



I 



908 



