108 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



Table 23. — Calculated total length of lake herring at the end 

 of each year of life, by age group and year class, 1943-50 



[Pound-net samples only. Length in inches] 



appearing in the samples. Such discrepancies as 

 do appear, therefore, are to be attributed princi- 

 pally to factors 3 and 4. 



The inconsistencies among the calculated growth 

 histories of the different age groups of the several 

 year classes ' of the Green Bay lake herring 

 (table 23) differ from Lee's phenomenon as 

 originally described (Lee 1920). It is true that 

 the estimates of length for a particular year of life 

 did tend to decrease with increase in age of fish on 

 which estimates were based. On the other hand, 

 the size of the differences did not decrease with 

 increase in the number of years of life as is charac- 

 teristic of Lee's phenomenon. In all but one 

 comparison between age groups represented by 15 

 or more fish the estimate of first-year length 

 decreased with increase of age (the one exception 

 is in age groups IV and V of the 1944 year class). 

 The trends were similar for the second-, third-, and 

 fourth-year calculated lengths, but exceptions 

 were more numerous. 



' The most discriminating comparisons are those among different age 

 groups of the same year class, since these are not biased by annual fluctua- 

 tions in growth. 



It is believed that the discrepancies among the 

 calculated lengths of the age groups of the Green 

 Bay lake herring represent the combined effects of 

 segregation according to size within the popula- 

 tion and of selective destruction of the faster- 

 growing individuals in the fishery made possible 

 by that segregation: . Because of the connection 

 between these two factors it is difficult to judge 

 their relative importance. - In fact, an attempt to 

 separate the two is not desirable, since they are 

 essentially parts of a single process. 



In the younger age groups, only the largest fish 

 (a small percentage of the total) enter the pound- 

 net fishery. (Note the small representation of 

 age groups I and II in collections — table 17). Se- 

 lection in the gill-net fishery is similar (table 17), 

 but the effects of selective destruction probably 

 occur later in gill nets than in pound nets. This 

 biological selection (plus gear selection in gill nets) 

 leads to the overestimation of the rate of growth 

 in those age groups. At the same time, destruc- 

 tion of the larger, fast-growing fish modifies the 

 growth characteristics exhibited by the remaining 

 stock. As members of a year class grow older, 

 bias to the immediate sample resulting from the 

 selective capture of the larger fish declines, but 

 the cumulative effects of destruction of the 

 faster-growing individuals become increasingly 

 important. 



Ten populations of Leucichthys artedi, for which 

 various authors have given figures of calculated 

 growth of different age groups, have all exhibited 

 Lee's phenomenon to some degree. Disagree- 

 ments were large in only one of four cisco popula- 

 tions in northeast Wisconsin (Hile 1936). In the 

 Irondequoit Bay cisco population the growth rate 

 decreased with increased age among the younger 

 age groups, but differences were random at the 

 higher ages (Stone 1938). Fry (1937) found only 

 small discrepancies among the estimates of the 

 first-year growth of the Lake Nipissing cisco, but 

 disagreements were large in later years. 



The variation in the nature of the discrepancies 

 in calculated growth of fish of different age in the 

 several populations leads to the conclusion that 

 the causes of Lee's phenomenon are not the same 

 in all populations. Principal explanations of the 

 phenomenon in lake herring advanced by various 

 authors are — 



1 . Selective action of gill nets used in collecting 

 samples. 



