regression. Independent variables included mean weekly water 

 temperature, mean wind speed and direction on the sampling 

 date, and wind speed and direction averaged over the sampling 

 date and the previous 2 d. Wind direction was treated as a 

 categorical variable with two classes (onshore and offshore). 

 Wind speed and direction data were obtained from the Na- 

 tional Weather Service Station at Warwick, R.I. Larval densi- 

 ty, water temperature, wind speed, and averaged wind speed 

 were transformed to natural logarithms prior to analysis. Two 

 variables, water temperature and wind speed on the sampling 

 date, were sufficient to provide a significant regression equa- 

 tion (Table 3) with multiple correlation coefficient of R = 

 0.784. Examination of the squared multiple correlation coeffi- 

 cient indicated that 61.5% of the variance was explained by the 

 derived equation. Inspection of the standardized residuals 

 revealed no departure from the assumption of normality and a 

 Durbin- Watson test (Neter and Wasserman 1974) indicated no 

 significant autocorrelation in the residuals. 



Table 3. — Coefficienls and associated standard errors (SE), 

 F ratios (df = 2.9). and multiple correlation coefficients (ft) for 

 stepwise regression model relating larval density to wind speed 

 and temperature. 



Variable 



Coefficient 



SE 



F 



R 



Wind speed 



3.0907 



0.8507 



13.199" 



0.649 



Temperature 



5.4328 



2.5675 



4.477* 



.783 



(Constant) 



- 19.6808 









**Significan 



at P<0.01. 









•Significant 



at P<0.05. 









The lack of a significant wind direction effect was surprising 

 since a positive relationship between onshore winds and larval 

 abundance has been previously noted (Templeman and Tibbo 

 1945; Squires 1970; Stasko 1980). In the present study, the 

 highest larval densities were generally obtained when winds 

 were onshore. However, low larval densities at the beginning 

 and end of the season, despite onshore winds, tended to 

 obscure this relationship. A vector plot of surface transport 

 was constructed for the period of high larval abundance, 20 

 June to 28 July, assuming surface drift to be 3.0% of the resul- 

 tant wind speed and at an angle of 15° to the right of wind 

 direction (Fig. 5). Prevailing winds for the period were 

 southwesterly, however, variable offshore winds dominated 

 from 27 June through 5 July, culminating in reduced larval 

 catches on this date (Fig. 5). High larval densities on 12 July, 

 despite 2 d of offshore winds, do not conform to the general 

 pattern although the effects of strong onshore winds from 6 

 through 10 July may account, in part, for this result. 



The inclusion of surface water temperature in the model 

 reflects the increasing contribution of fourth stage larvae later 

 in the season when water temperatures were also increasing. 

 Lobster larvae were collected in surface water temperatures 

 ranging from 13° to 25 °C. Modal temperatures at peak larval 

 densities for stages I and II were 14°-16°C and 17°C for third 

 stage larvae (Fig. 6). Stage IV larvae were abundant at surface 

 water temperatures over 17°C. Lund and Stewart (1970) 

 collected lobster larvae in surface waters ranging from 12.5 ° to 

 28.5 °C in Long Island Sound. Surface water temperatures 

 ranged from 13.7° to 15°C during peak larval concentrations 

 in the Gulf of Maine (Sherman and Lewis 1967). 



Figure 5. — Vector plot of wind-induced surface drift during 20 June- 28 July 1978. 

 Circled figures represent larval densities on sampling dates. Dates provided at 5-d 

 intervals for reference. 



The contribution of wind speed to the regression equation 

 may reflect wind-induced advection currents which 

 presumably served to transport larvae into the study area. 

 Wind velocities on sample dates were relatively moderate and 

 apparently did not reach levels at which surface turbulence 

 would result in reduced densities (Squires 1970). 



CONCLUSIONS 



High fourth stage larval lobster densities were obtained in 

 Block Island Sound in 1978. Adjustment for probability of 

 capture based on developmental times for each larval stage 

 did not eliminate the dominance of stage IV larvae in these 

 collections. Although first and fourth stage larvae may be 

 more vulnerable to surface gear (Templeman and Tibbo 1945), 

 high mortality rates during the pelagic larval stages (Scarratt 

 1964, 1973) should result in relatively low numbers of stage 

 IV larvae if recruitment is strictly localized. Prevailing winds 

 during the period of larval occurrence are onshore, possibly- 

 resulting in a net transport of larvae from continental shelf 

 waters. Larval recruitment from offshore locations may 

 assume particular importance in maintaining inshore popu- 

 lations which are subjected to extremely high fishing mortality 

 rates. 



Stage I larval production in Rhode Island Statistical Area 4 

 was estimated to be 2.514 x 10 6 larvae based on expansion 

 of corrected larval densities. A minimum estimate of hypo- 

 thetical larval production based on population size determined 

 by cohort analysis, sex ratio, maturity, and fecundity indicated 

 that at least 3.323 x 10 7 larvae could have been produced. 

 Nichols and Lawton (1978) reported similar underestimates of 

 larval production of Homanis americanus based on neuston 

 samples. 



Larval density was significantly correlated with wind speed 

 on the day of sampling and surface water temperature. 



ACKNOWLEDGMENTS 



We would like to express our appreciation to R. Wood, T. 

 Lynch, A. Ganz, and J. Hoenig for assistance in the field and 

 to J. Hoenig for sorting samples. We are grateful to M. Penn- 

 ington for bringing the Delta distribution method to our atten- 



27 



