236 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



work was carried out with special reference to the 

 relation between the annual fluctuations of temper- 

 atures in June and in August and those of other 

 months. 



The lack of correlation between June tempera- 

 tures and growth appears to be somewhat anoma- 

 lous in view of the evidence of a positive correlation 

 between growth and temperatures in May and 

 July. Since the absolute temperatiire in Jime 

 normally is intermediate between those of May 

 and July a similar relation would be expected for 

 all 3 months. The coefficients of correlation be- 

 tween temperatures in June and those in certain 

 other months listed below are too small, however, 

 to support any belief that a true relation between 

 growth and June temperatures has been concealed 

 by correlations with temperatures of other periods 

 of the growing season. 



Between June temperature and temperature in — 



Correlation 



May and July 0. 291 



May and September . 214 



May, July, and September . 246 



July and September . 124 



August .083 



The high negative value of r for gi-owth and 

 August temperatures cannot be termed anomalous 

 since water temperatures reach their maximum 

 in that month in most years " and the concept that 



II This statement is supported by records of lake Erie water temperatures 

 at the intake of the Chestnut Street Water Plant at Erie, Pa. (published in 

 the Annual Reports of the Commissioner of Water Works of that city). 

 According to those records the maximum monthly average water temperature 

 occurred in August in 23 of the 25 years, 1923-47; furthermore, August temper- 

 atures of the period averaged 2.3° and 3.7° F. higher than those of July and 

 September. The crib of the intake is located 5,100 feet north of the Presque 

 Isle Peninsula and is covered by 22 feet of water at low-water level. Al- 

 though Erie is located well to the east of the centers of greatest abundance of 

 the yellow perch, water temperatures off that port may be taken to indicate 

 monthly trends. 



a high maximum might exert a depressing effect 

 on growth is not unreasonable. It was considered 

 desirable, nevertheless, to determine the possible 

 effects on the interpretation of the data of correla- 

 tions between August temperatures and those of 

 months that exhibited significant positive correla- 

 tions between temperature and growth. The 

 following coefficients, including one for May and 

 July in which temperature was not correlated 

 significantly with growth, were computed. 



Between August temperature and temperature in — 



Correlalion 



May and July —0. 086 



May and September — . 186 



July and September — . 339 



May, July, and September — . 205 



Again none of the correlations between tempera- 

 tures in different periods was sufficiently close to 

 conceal possible relations. 



The data presented in this section may be taken 

 as strong evidence that temperatures exert a 

 significant effect on the annual fluctuations in 

 growth of the yeUow perch in Lake Erie, with 

 high temperatures in May, July, and September 

 (especially September) accelerating growth, and 

 high temperatures in August retarding it. 



Any attempt at a biological interpretation of 

 the observed correlations would, with our present 

 knowledge, be of little value. Conceivably, tem- 

 peratures may affect growth dnectly, as through 

 the control of the instantaneous rate of increase 

 or of the length of the growing season, or indi- 

 rectly, as through the control of the distribution 

 or abimdance of food organisms. Until more is 

 learned of the natural history of the perch, the 

 mechanism of the apparently significant correla- 

 tion between growth and temperature must remain 

 unknown. 



LENGTH-FREQUENCY DISTRIBUTION 



The catches of impoimding nets and gill nets 

 differed in the actual form of the frequency dis- 

 tribution as well as in the size of fish taken (table 

 17). The length distribution of yellow perch 

 caught in trap nets and pound nets was unimodal 

 each year. The shoal-net collections, on the 

 other hand, showed definite bimodal length distri- 

 butions for 1927 and 1929, but gave no indication 

 of bimodality in 1928. The fairly large number 

 of small perch taken by the shoal nets during 

 1927 is probably explained by the presence of the 



abundant year class of 1926, then in their second 

 year of life (age group I). The bimodal length 

 distribution of the bull-net samples in 1927 was 

 the result of the accidental captm-e of a large 

 school of small fish on a single day. These smaller 

 individuals ordinarily were not gfiled in the true 

 sense, but rather, were captured by tangling the 

 webbing of the net in the marginal bones of the 

 mouth or in the fins. 



It will be noticed that there was considerable 

 annual variation in the length of the modal fre- 



