64 S. J. HOLT 



and in the Pacific razor clam {Siliqua patula). Holt (i959<^, i960) has argued 

 that this might be expected to be a general phenomenon, knowledge about 

 which would help in the prediction of growth rate in one area and for one 

 species from knowledge of the growth in another area, at a different 

 temperature, of the same or of a related species. Comparative study could 

 give measures of the temperature coefficient of K by use of which K values 

 determined experimentally might be used to estimate K in the wild 

 population. 



There are, however, several comphcations to this line of argument, some 

 of which will be mentioned here. Firstly, at least within populations of the 

 same species, K and L^ are inversely correlated. Thus Taylor (1959) plotted 

 K against E = KL^ in several populations of a number of fish species, and 

 obtained linear regressions. K and E are not, however, proportional; they 

 have, as von Bertalanffy himself argued on theoretical grounds, different 

 temperature coefficients, so that for example a certain change in K is not 

 associated with so great a proportionate change in E, and thus L^o decreases 

 as K increases. Some, but probably not all, of the correlation between K and 

 E, may be spurious. This is so because in the common method of determining 

 K and L^, the parameters of the linear relation between values of /^ and /^+i 

 are estimated graphically and a small chance rotation of this line about its 

 mid-point will simultaneously result in lower K and higher L^ estimates, 

 and vice versa. Further, because E is estimated as the product of K and L^o 

 chance variations in K will be reflected in both 'correlated' variables. It 

 should be mentioned that the apparent correlation between K and IJL^ 

 could, theoretically, be spurious, for similar reasons. Lastly, extrapolation of 

 a theoretical linear relation between E and K gives finite E for K = o and 

 hence infinitely high L^, and it would seem to be a more suitable procedure 

 to find the linear regressions of log K on log L^, and hence find the para- 

 meters u and V in an equation of the form 



Such an equation could not be apphed to unrelated species with an assump- 

 tion that u and v were constant. It is worth noting, however, that the 

 differences in growth rate between the sexes of a particular species seem to 

 be of the same kind as the differences between growth patterns of the same 

 species in different areas. Thus where, as in most fish, the growth patterns of 

 males and females are different, the difference is usually such that the growth 

 of males can be described by higher K and lower L^, than for females. 



The correlation of K with IJL^ might be useful in choosing the most 

 likely value for iC when two or more possible values are suggested by different 

 interpretations of the rings on hard structure. 



