LOUGH: LARVAL DYNAMICS OF DUNGENESS CRAB 



The second data set compares the sampling 

 period 29 January-2 May 1970 with 18 January- 

 14 May 1971. The period selected compares the 

 larval period prior to summer upwelling, 

 eliminating the erratic surface temperature and 

 salinity fluctuations. Most of the C. magister lar- 

 vae are megalopae by early May. 



The same full model was used in the initial run 

 for the second data set and is presented in Table 

 4. There was a significant difference (5% level) 

 between y means after being adjusted for all the 

 covariates in the model. The covariates, t, T, and 

 T X S were all significant. 



Table 4. — A comparison of the total number of Cancer magis- 

 ter larvae for 1970 and 1971 (January to May) by analysis of 

 multiple covariance (full model). 



Table 5. — A comparison of the total number of Cancer magis- 

 ter leirvae for 1970 and 1971 (January to May) by analysis 

 of multiple covariance (reduced model). 



The initial model was reduced to the following 

 form: 



Y = b + boiy) + bjt) + bHT) 

 + baiS) + b^(T X S) 



which greatly increased the significance of the 

 parameters in the final model (Table 5). A highly 

 significant difference (1% level) was found be- 

 tween _y means after being adjusted for all the 

 covariates. In explaining the difference between j 

 means of C. magister larvae, the covariate t was 

 most significant (1% level) followed by T and S, 

 and T X S at the 5% level. 



The foregoing analyses support the contention 

 that there was a significant difference between 

 the C. magister larval populations of 1970 and 

 1971. Fewer larvae appeared in 1971 and they 

 appeared in the plankton for a shorter period of 

 time suggesting widespread larval mortality. 

 This apparent larval mortality was associated by 

 these analyses with the colder surface tempera- 



tures and lower salinities that occurred during 

 the winter of 1971. The direct effects of tempera- 

 ture and salinity on larval survival will be 

 explored in the next section. 



Temperature-Salinity Tolerance of 

 Laboratory-Reared Larvae 



A laboratory study by Reed (1969) determined 

 the effects of temperature and salinity on the lar- 

 val survival of C. magister. However, it was 

 necessary to assess more thoroughly the effects of 

 these factors on survival during development and 

 to extrapolate from Reed's data in order to derive 

 better estimates of larval survival at the low 

 temperatures that occurred during the 1971 sea- 

 son. The response surface technique used in the 

 analysis of his data is not only valuable in its 

 predictive role, but also visually represents any 

 change in response at various stages of develop- 

 ment. Details of this response surface technique 

 and its application to the study of marine ecology 

 are discussed by Alderdice (1972). 



A multiple regression analysis was applied to 

 Reed's (1969) survival data of C. magister after 

 20, 30, 40, and 50 days of culture at experimental 

 conditions. The mathematical model used in the 

 analysis was of the form: 



Y = bo + biiS) + b^iT) + 63(52) 

 + 64 (T^) + 65(8 X T) 



where, Y = percentage survival, 60 = a constant, 

 S = linear effect of salinity, T = linear effect of 

 temperature, S^ = quadratic effect of salinity, T^ 

 = quadratic effect of temperature, and S x T = 

 interaction effect between salinity and tem- 

 perature. 



The 6's in the model were estimated by a step- 

 wise multiple regression computer program. 

 Further details of the regression analysis are 



361 



