FISHERY BULLETIN: VOL. 84, NO. 3 



the Alexander Archipelago (Auke Bay is about 80 

 nmi [148 km] by water from the open coast). 

 Although this population may contain more than one 

 spawning stock, it will be identified with Auke Bay 

 in the present study (local populations spawn with- 

 in weeks of each other and within a few nautical 

 miles). 



METHODS 



Auke Bay herring were sampled several times 

 monthly from April 1973 to March 1974; however, 

 no fish were taken in February 1974. The fish were 

 captured principally by jigging with bright hooks or 

 hooks wrapped with colored yarn. Samples were also 

 taken during spring 1973 from nearby locales in 

 southeastern Alaska, including Hood Bay (off 

 Chatham Strait, southwest of Juneau), Carroll In- 

 let (near Ketchikan), and Katlian Bay (near Sitka), 

 and also from the eastern Bering Sea west of 

 Nunivak Island. 



Auke Bay herring were usually examined fresh 

 but sometimes were frozen and examined within 1 

 wk. Lengths were originally measured as standard 

 lengths (SL, tip of upper jaw to end of hypural 

 bones) but were later converted to body length (BL, 

 tip of lower jaw to end of hypural bones) by multi- 

 plying SL by 1.0132, the average ratio in 126 

 specimens from Auke Bay. 



Body lengths were back-calculated from scales 

 taken from above the pectoral fins and posterior to 

 the opercular flap. The calculations followed the pro- 

 portional method of Whitney and Carlander (1956), 

 which should reduce the variation in BL-scale size 

 relationships because the method adjusts for possi- 

 ble differences in scale length in the same-sized fish. 

 This method requires that the regression between 

 BL and scale length be linear, which was satisfied 

 (Fig. 1). The intercept of the regression (55 mm) was 

 somewhat higher than the median BL (36.5 mm) for 

 first squamation of 16 preserved specimens; 

 however, the differences between estimates for BL 

 at first squamation are probably important only for 

 fish younger than 1 yr. The regression fit the data 

 well for herring >1 yr old (Fig. 1). I also attempted 

 to reduce variability in the back-calculations for the 

 Auke Bay fish by averaging focus-to-annulus 

 distances from left and right sides of the scales (an- 

 nuli were as well defined at the sides as in the 

 centerline of the scale), but only a centerline 

 measurement was used in samples from other 

 geographic regions. 



After the growth data were analyzed by Walford 

 graphs (Walford 1946), linear regressions (Walford 



regressions) were fit by least squares to adult sec- 

 tions of constant parameters (stanzas) that were 

 indicated on the graphs. Both the Walford regres- 

 sion and von Bertalanffy formulation are variants 

 of the constant-proportion growth model, which 

 requires that growth in one year be a constant pro- 

 portion of growth the preceding year (Ricker 1975). 

 (The slope of a Walford regression equals the von 

 Bertalanffy e~ K , and the intercept equals L (1 - 



e~ K ).) 



Annual changes in development of fat and gonads 

 were evaluated by indices that were derived from 

 total body weights, eviscerated body weights, and 

 gonad weights. I estimated unbound water in the 

 eviscerated body tissues and gonads as the percent- 

 age weight lost by drying 1 cm wide transverse body 

 sections and entire gonads in a drying oven for more 

 than 4 d at 27° C, a period that yielded weight stabil- 

 ity. Visual estimates of visceral fat used a four-point 

 scale (from none to heavy), and visual estimates of 

 maturity used a seven-point scale, as follows (Roman 

 numerals in brackets refer to a similar scale 

 developed by Hay and Outram (1981) for Pacific 

 herring): 1) Newly regenerating [VIII], 2) regen- 

 erating [III], 3) nearly mature [IV], 4) ripe [V], 5) 

 ripe and running [VI], 6) partially spawned [VII], 

 and 7) spawned out [VII]. Fresh body, eviscerated, 

 and gonad weights were regressed on body lengths 

 by least squares after logarithmic transformation 

 of variates. Statistical tests were significant when 

 P < 0.05. 



Scales of Pacific herring from southeastern Alaska 

 and the eastern Bering Sea probably have two an- 

 nuli in the first growth year. When the annulus 

 nearest to the scale focus of Auke Bay herring was 

 used for back-calculations, the BL's were much 

 smaller (average of 65 mm) for the first winter than 

 the BL's of juvenile herring (at least 80 mm) cap- 

 tured at the end of their first year in Auke Bay by 

 Jones (1978). Pacific herring in British Columbia at- 

 tain a length of at least 80 mm by their first Sep- 

 tember (Hourston 1958). Furthermore, when the 

 first annulus was used as the first year mark, Wal- 

 ford graphs of the growth data were erratic and 

 differed markedly from graphs of the same type of 

 data in the literature. When the second annulus was 

 used as the first year mark, the graphs were sim- 

 ple and corresponded to graphs of similar data from 

 the literature. 



There was no indication of Lee's phenomenon 

 (slower growth in longer lived individuals) in the 

 back-calculated BL's, but there was evidence of a 

 changing relation between growth back-calculated 

 for ages 1 and 2 and the span of years that was used 



706 



