518 



Fishery Bulletin 93(3), 1995 



No stage-1 individuals were collected, probably be- 

 cause their small size allowed them to pass through 

 the net. 



To determine if metamorphic stage was a more 

 useful character than SL in resolving spatial pat- 

 terns, twenty individuals from each metamorphic 

 stage of each species were randomly selected from 

 the May to June survey of 1991, and their otoliths 

 were removed. The diameter of the largest otolith, 

 the sagitta, was measured with a compound micro- 

 scope connected to a video camera, monitor, digitizer, 

 and computer. 



Data analysis 



Statistical analyses were performed by using the 

 program SAS (SAS Institute, Inc., 1988). Abundances 

 were transformed by ln(x+l) to normalize the data. 

 Mean abundances from the April surveys and for each 

 sweep of the May-June surveys were then calculated 

 for all standard stations successfully sampled to de- 

 scribe temporal differences in abundance. 



Means and ranges of SL's of the metamorphic 

 stages of both sanddab species were calculated by 

 using specimens measured from the April and May- 

 June surveys of 1990 and from the May-June sur- 

 vey of 1991. Mean SL's were calculated separately 

 for the specimens randomly selected for otolith re- 

 moval in order to compare the observed changes in 

 otolith size relative to metamorphic stage with SL. 

 Tukey's studentized range tests were performed to 

 determine if there were significant differences in SL 

 and otolith diameter with metamorphic development. 



Owing to changes in the width of the net mouth 

 with depth (width=8 m at 10 m depth, 11 m at 30 m 

 depth, and 13.5 m at 110 m depth), abundances for 

 the depth-stratified trawls were adjusted prior to 

 analysis (Lenarz et al., 1991). Abundances for 10-m 

 depth trawls were multiplied by 11/8, abundances 

 for 30-m depth trawls were not adjusted, and abun- 

 dances for 110-m depth trawls were multiplied by 

 11/13.5. Because of the large spatial variability in 

 abundances and the fact that not all depth-strati- 

 fied trawls sampled every depth (i.e. because of time 

 constraints only the 10 m and 30 m depths were 

 sampled at some stations, and only the 30 m and 

 110 m depths were sampled at others), differences 

 in abundance with depth, were evaluated by paired 

 comparison f-tests. Pairing was by station; each 

 depth pair available was considered ( 10 m versus 30 

 m, 30 m versus 110 m, and 10 m versus 110 m). When 

 both observations of a pair were equal to 0, that pair 

 was deleted from the analysis. To determine if there 

 was a seasonal change in depth distribution, analy- 

 ses were performed on all the surveys from 1987 to 



1991, the April surveys alone, and the May-June 

 surveys alone. To determine changes in vertical dis- 

 tribution with development, the paired comparisons 

 were also carried out on each of the metamorphic 

 stages, by using data from the April and May-June 

 surveys of 1990 and from the May-June survey of 199 1 . 



To determine the similarity of horizontal distribu- 

 tions among metamorphic stages, Spearman rank 

 correlation coefficients were calculated for the abun- 

 dances of different stages over standard stations 

 during the 1990 and 1991 surveys. 



CTD salinity at the surface (average salinity at 

 3-5 m depth) and at the standard mid-depth of 30 m 

 (average salinity at 28—32 m) from each sweep of the 

 May-June surveys of 1990 and 1991 was contoured 

 by using the Kriging option in the program SURFER 

 (Golden Software, Inc., 1990). Owing to problems 

 with the CTD during the second sweep of the May- 

 June survey of 1991, thermosalinometer data were 

 incorporated into the available CTD data to gener- 

 ate the surface contours. The log-transformed abun- 

 dances of metamorphic stages 2 and 5 were overlaid 

 onto the salinity contours to observe the relation (if 

 any) between the abundances of these two metamor- 

 phic stages and salinity features indicative of coastal 

 upwelling. These two stages had the greatest poten- 

 tial for differences in spatial relations with salinity 

 features, given that stage-2 individuals were not com- 

 petent for settlement, whereas stage-5 individuals 

 were relatively close to settlement. 



Thermosalinometer data were used to determine 

 trawls conducted in areas of recent upwelling. 

 Schwing et al. (1991) designated recently upwelled 

 water as having surface temperatures less than 

 10.5°C and surface salinities greater than 33.6 ppt. 

 We used surface salinities greater than 33.6 ppt but 

 surface temperatures less than 11.0°C to allow for 

 marginal surface layer warming during the daylight 

 hours prior to the nighttime trawls. Log-transformed 

 abundances obtained at standard stations from the 

 May-June surveys of 1990 and 1991 were converted 

 to standard scores (log-transformed abundances 

 rescaled for each year so that mean=0 and standard 

 deviation=l) to adjust for year effects, allowing the 

 data from both years to be combined. T-tests were 

 used to compare the mean standard scores of meta- 

 morphic stages in upwelling and non-upwelling ar- 

 eas. Separate analyses were performed for shallow- 

 depth (10-m) trawls and standard mid-depth (30-m) 

 trawls because of the possibility of a depth effect due 

 to metamorphic stage and the fact that upwelling 

 typically affects only the upper 20 m of the water 

 column (Parrish et al., 1981). To increase sample size, 

 trawls at nonstandard stations were included in the 

 analysis. 



