Stone and Jessop: Seasonal distribution of Alos3 pseudoharengus and A. aestivalis 



379 



Seasonal distributions of relative abundance were 

 obtained from plots of average catch-per-set data (in- 

 cluding zero catches) aggregated by 20-minute rec- 

 tangles of latitude and longitude. Seasonal, rather than 

 monthly, distributions gave a more complete picture 

 of offshore distribution patterns. The locations of 2, 5, 

 and 10°C isotherms, generated from plots of the mean 

 bottom-water temperature in 20-minute rectangles, 

 were superimposed on seasonal distribution plots. Dif- 

 ferences in the seasonal distribution of two size-groups 

 of river herring (i.e., <19cm and >19cmFL) were ex- 

 amined by plotting the capture locations of each group. 

 The size-at-first-spawning of Saint John River blueback 

 herring (Jessop et al. 1982), which mature at a smaller 

 size than alewife, was used as the separation criterion. 



Approximate randomization tests, with 1000 per- 

 mutations (Edgington 1987) were used to examine the 

 following relationships: (a) effects of season (spring, 

 summer, fall) and depth (<93m, 93-183m, >183m) on 

 mean catch-per-set (including sets with no catches), (b) 

 diel variability in seasonal catch-per-set (day and night, 

 based on gear deployment times in relation to monthly 

 morning and evening civil twilight times for the ap- 

 propriate geographic location), and (c) mean fork length 

 by season and depth. All comparisons used catch or 

 fork-length data from a 6-year time-series (1979-84) 

 in which annual sampling occurred during spring, sum- 

 mer, and fall. Each dependent comparison (i.e., depth 

 effect within season, season effect within depth), used 

 a Bonferroni significance level (a 0.05, divided by the 

 number of dependent comparisons) (Day and Quinn 

 1989). 



Randomization procedures were used for the analysis 

 because statistically significant heteroscedasticity in 

 the variances of (a) transformed (In X -i- 1) catch per set 

 data by season and depth (Cochran's C = 0.372, P< 

 0.0001, df 289, 9), (b) transformed On X -h 1) diel catch- 

 per-set by season (Cochran's C = 0.365, P<0.0001, df 

 359, 6), and (c) fork length by season and depth 

 (Cochran's C = 0.293, P<0.0001, df 637, 9) violates an 

 assumption of parametric statistics. This violation is 

 compounded by unequal sample sizes. 



Bycatch data 



Set locations and fork lengths of river herring bycatch 

 in foreign and domestic commercial groundfish opera- 

 tions (1980-89) obtained from the DFO International 

 Observer Program database were compared with re- 

 search survey data. Catch locations were plotted for 

 spring (February-May) and summer (June-August), 

 when fishing effort and bycatch were highest. Fork 

 length distributions were truncated at 33 cm due to 

 assumed misclassification (4% of fish measured ex- 

 ceeded this length). Annual landings of alewife in 



metric tons (t) for the Scotian Shelf-Bay of Fundy 

 region (4VWX) from 1970 to 1989 were obtained from 

 Northwest Atlantic Fisheries Organization (NAFO) 

 Statistical Bulletins and correlated (Spearman rank) 

 with catch indices from research vessel surveys. The 

 analysis was conducted to determine if survey indices 

 (i.e., mean number and mean weight per set ■ season "^ 

 • year'i) were consistent with bycatch landings as an 

 indicator of relative abundance. 



Results 



Seasonal distribution and abundance 



A total of 13,357 river herring were captured in 602 

 (15%) of 4105 bottom-trawl sets conducted between 

 1970 and 1989 (Table 1). Spring survey catches were 

 the highest and most variable, followed by summer. 

 The proportion of sets with river herring from spring 

 surveys was nearly double that from the other two 

 seasons. A maximum catch of 2292 river herring oc- 

 curred on 17 March 1980 in the Scotian Gulf region, 

 south of Emerald Basin. 



During spring surveys, river herring dominated in 

 three regions: the Scotian Gulf, southern Gulf of 

 Maine, and off southwestern Nova Scotia from the 

 Northeast Channel north to the central Bay of Fundy 

 (Fig. 2). Catches also occurred along the southern edge 

 of Georges Bank and in the canyon between Ban- 

 quereau and Sable Island Banks. Relative abundance 

 was highest in the Scotian Gulf between Emerald and 

 Western Banks, and on the southern slope of Georges 

 Bank. Most catches of river herring occurred where 

 bottom temperatures exceeded 5°C (Fig. 2), although 

 in the Bay of Fundy, captures occurred at lower 

 temperatures. 



Summer distributions of river herring were less ex- 

 tensive than in spring and were limited mainly to the 

 eastern Gulf of Maine (off southwestern Nova Scotia) 

 and the Bay of Fundy, with a few occurrences near- 

 shore in the central Shelf region (Fig. 2). Catches were 

 highest along the northern shore of the Bay of Fundy, 

 with very few fish captured in the Scotian Gulf and on 

 the eastern Scotian Shelf. Bottom temperatures ex- 

 ceeded 5°C at all capture locations. 



Fall distributions of river herring were more exten- 

 sive than in summer (Fig. 2). Moderate to large catches 

 were obtained from southwestern Nova Scotia to the 

 Bay of Fundy, the central Scotian Shelf, and Sydney 

 Bight. As in the case of spring and summer surveys, 

 very few fish were captured on the eastern half of the 

 Scotian Shelf. All catches occurred at bottom tem- 

 peratures exceeding 5°C. 



Bycatch of river herring from foreign fishing fleets 

 (1980-89) occurred mainly during spring in a narrow 



