five reasonably distinct populations, among 

 which little mingling would be expected. He 

 states that this conclusion was not compatible 

 with evidence from the tagging experiments 

 (Fry and Roedel 1949:25), which showed sub- 

 stantial amounts of intermingling between 

 southern California and the Soledad and Vis- 

 caino regions off Baja California. Of the fish 

 released off Soledad 6.8 percent were recovered 

 from southern California, and of those released 

 in the Viscaino area 2.4 percent were recovered 

 from southern California . These returns may 

 be ^compared with the overlap of head length 

 » * (p. 17), which for the California -Soledad 

 comparison was 90 percent and for the Califor- 

 nia -Viscaino comparison was 69 percent. In 

 both comparisons it is evident that a major part 

 of the individuals possessed identical head 

 length characteristics. 



There is also reason to suspect that the 

 samples were not adequately representing the 

 mackerel populations throughout the year in the 

 areas specified. The standard errors of esti- 

 mate were not uniform among regions (p. 19); 

 the Viscaino and Gulf regions showing unduly 

 high values suggestive of a more heterogeneous 

 population The high value for the Gulf results 

 from only two samples taken on successive days 

 in February 1941, whereas the samples for the 

 Viscaino region resulted from six subsamples, 

 most of which were taken on separate days. 

 Again, Soledad was represented by only two sub- 

 samples and California by only five subsamples. 

 Some further evidence of heterogeneity is ap- 

 parent, because the regression coefficient of 

 head length on fork length is different between 

 California -Soledad, Viscaino -Cape, and Cape - 

 Gulf. These differences are such that the small 

 fish showed little difference between the Cape 

 and the Gulf and between California and Soledad 

 and the large fish between Viscaino and the Cape. 

 Roedel further notes that the population in the 

 Viscaino area showed marked differences be- 

 tween the 1941 juveniles and older fish with 

 respect to vertebral characters. 



were released in February 1941 and in October 

 1941, with the balance scattered among several 

 other months. If, as the above evidence in- 

 dicates, this mackerel population is heterogen - 

 eous, with groups moving seasonally, and 

 perhaps independently, north or south along the 

 coast, it is obvious that fish sampled for either 

 tagging or morphometric studies at one or two 

 times during the season may not be representa- 

 tive of the population in the area during the 

 entire season. 



Therefore it is not warranted to conclude, 

 as Roedel did, that the morphological studies 

 are not compatible with the evidence from tagging. 

 The morphological studies showed that the stocks 

 within an area were heterogenous but that even 

 between areas a large proportion of the individuals 

 had identical characteristics. Neither morpho- 

 logical nor tagging studies were based on samples 

 reasonably representative of any area nor even 

 of any definite period in any area, so it is not 

 possible to estimate the amount of intermingling. 

 The tagging studies show there was some inter- 

 mingling, the morphological studies that the 

 intermingling was not complete; there the matter 

 must rest. 



Thus tagging and morphometric studies 

 may provide essentially opposite and mutually 

 supplementary estimates of intermingling. A 

 tagged fish released in population A and returned 

 from B shows that intermingling has occurred. 

 If the sampling in both populations A and B has been 

 adequate and the tagged fish are representative of 

 population A, then the proportion returned from 

 B will be an estimate of the intermingling of A 

 with B. Such conditions rarely can be satisfied 

 in practice and hence we must usually stop with 

 the estimate that some intermingling occurs. 

 On the other hand a maximum amount (but no 

 proof of occurrence) may be fixed by the amount 

 of overlap of morphometric characters. The two 

 methods may then enable us to make a statement 

 to the effect that intermingling occurs but does 

 not exceed a certain amount. 



The samples that were tagged may have 

 been even less representative, because almost 

 all of those released in the Soledad area were 

 tagged in January 1940, with a few in October 

 1941. In the Viscaino area about 80 percent 



LITERATURE CITED 



COCHRAN, W. G. 



1953. Sampling techniques . Wiley, New 



York. 330pp. 



26 



