444 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



place over a period of several days (Fraser 1922). 



Another Osmerid studied by Hart and McHugh 

 (1944) is the silver smelt (Hypomesus pretiosus). 

 These fish spawn on beaches during high tides 

 and in most months of the year. The ovaries 

 contain yolked ova of intermediate sizes. Schaefer 

 (1936) has also figured the ovarian ova diameter 

 distributions of this species. The above three 

 authors concluded that this species spawns several 

 batches per season. 



The third Osmerid for which ovarian ova di- 

 ameters are figured by Hart and McHugh (1944) 

 is the capelin (Alallotus catervarius). Of this 

 species the authors say, "Spawning takes place in 

 various localities in the strait of Georgia during late 

 September or the month of October . . . Spawn- 

 ing takes place in the evening at high tide right at 

 the water's edge . . . The size frequencies of the 

 eggs [fig. 10] suggest that the mature capelin 

 spawns more than one batch of eggs as there ap- 

 pears to be one group of small eggs becoming 

 differentiated from the general mass which com- 

 prise the residual ovarian tissue after the ripe 

 eggs are spawned out. It is not known at the 

 present time whether any such second spawning 

 occurs. It is possible that more or less nomad 

 schools of capelin move around the southern part 

 of the strait of Georgia spawning on suitable 

 beaches as the eggs ripen. There is no racial 

 evidence against such a belief but the only evidence 

 in favor of such a supposition is the observations 

 on the ovaries taken in conjunction with the lack of 

 second spawnings occurring on the same beaches." 



Thus we find three possible types of spawning 

 occurring among these shore spawning species : (1) 

 only one group of yolked ovarian ova with only 

 one known spawning (eulachon, herring); (2) more 

 than one group of j'olked ovarian ova with multiple 

 spawning (grunion, silver smelt); (3) more than 

 one group of yolked ovarian ova with only one 

 known spawning (capelin). 



The degree of correlation between the modal 

 value of the most advanced group of developing 

 ova and the modal value of the secondary group 

 of developing ova was used by Clark (1934) as an 

 indication of multiple spawning in the Pacific 

 sardine. June (1953) also used this method for 

 the yellowfin tima (Neofhunnus macropterun). By 

 this method of correlating modal values, Clark 

 (1934) obtained a coefficient of correlation of 0.70 

 for modal diameters below 0.80 mm. and 0.72 for 



modal diameters above 0.80 mm.; June (1953) 

 obtained a coefficient of correlation of 0.855 for 

 his tuna data. 



There may be some doubt concerning the inter- 

 pretation of these correlations because of the 

 mathematical restrictions imposed upon the plot- 

 ting of the data. Although there are biological 

 limits to the plotting of any biological data there 

 should be no mathematical limitations; that is, it 

 should not be mathematically impossible to plot 

 any point. In the present case the convention of 

 always plotting the larger diameter on one axis 

 and the smaller on the other limits the plots to a 

 triangular area rather than a rectangular area. 

 It can be demonstrated that while a regression 

 line based on a random plotting of points (positive 

 numbers) in a rectangular area wOl have a Y- 

 intercept (a) equal to the mean of Y, a slope ib) of 

 zero and a coefficient of correlation (/•) of zero, a 

 regression line based on a random plotting of 

 points in a traingular area with a hj-potenuse of 

 slope 1.00 passing through zero (i. e., coincidence 

 of the two modes in the present case) will have a 

 F-intercept of zero, a slope of .500 and a coeffi- 

 cient of correlation of .500. 



Regarding the "decrease, as the breeding season 

 advances in the numerical ratio between suc- 

 ceeding batches of eggs and the largest size 

 group," Clark (1934:19) states: "The change in 

 the ratio from approximately 4-1 to 2-1 suggests 

 that eacli fish may mature an average of three 

 batches of eggs, although this number may be 

 higher, for this study furnishes no data to deter- 

 mine whether growth from the immature to the 

 maturing class accompanies growth within the 

 maturing sizes." 



Her ratio data are given in table 11 along with 

 data on the numbers and percentages of fish in 

 stage G (the last abundant stage before the eggs 

 become translucent and are spawned) and in 

 stages A (resting) and L (spent). The data for 

 numbers and percentages are based on sardines 

 larger than 199 mm. in length; the ratios appar- 

 ently include a few smaller sardines in some 

 months. 



High percentages of stage G fish were shown for 

 Monterey between March 15 and May 13 in 1930 

 and between March 5 and May 2 in 1931 (the 

 1929 Monterey data are not used as they are 

 based on only 8 stage G fish for the season) , fol- 

 lowed in each case by high percentages of stages A 



