MATERIALS AND METHODS 



Fifty-two samples of adult Atlantic men- 

 haden, comprising 6,048 specimens, were 

 collected in spring (April- May) from pound 

 net catches in the vicinity of Cape Cod, Mass., 

 and Long Island, N.Y., and in autumn (October- 

 December) from purse seine catches near 

 Long Island, N.Y., and Beaufort, N.C., from 

 1957 to 1960 (fig. 1). Collecting was limited to 

 periods when spring or autumn spawners 

 occurred in the catches in greatest nximbers. 

 Incidental catches of Atlantic nnenhaden by 

 pound nets provided the earliest and only 

 available samples in spring. Adverse weather, 

 which delayed the setting of pound nets in 

 waters north of Long Island during April and 

 May, thwarted attempts to secure a greater 

 number of samples in the Cape Cod area. 

 Although efforts were made to obtain at least 

 100 specimens in each sample, early season 

 pound net catches sometimes failed to provide 

 siifficient numbers. Also, sonne of the pre- 

 served material was lost during storage be- 

 cause of decomposition, separation of verte- 

 bral columns, and dissociation of identifying 

 labels. Individual samples available for study 

 contained from 41 to 313 specinnens. 



At the time of collection, fish were examined 

 for sex and maturity. Most gonads were 

 weighed, and, following the general nnethod of 

 Higham and Nicholson (1964), males and fe- 

 males with a "gonad index" of 4.0 or more 

 were classified as spawners and those with an 

 index of 3.0 or less as nonspawners. Fish which 

 were in spawning condition in April and May 

 were classified as "spring spawners", those 

 in spavming condition from October to Decem- 

 ber as "autunrui spawners." 



The fork length of each fish, in centimeters, 

 was recorded and a scale sannple taken, 

 following the methods described by June and 

 Reintjes (1959). The vertebral column, in- 

 cluding the skull and most of the caudal fin, 

 was removed, labeled, and preserved in 10 

 percent formalin. 



Counts of vertebral numbers were nnade 

 from radiographs (Sutherland, 1958) and in- 

 cluded all vertebrae between, but not including 

 the occipital bone and the hypurals (June, 

 1958; Sutherland, 1963). Abnormal vertebrae 

 (found in 0.4 percent of the fish examined) 

 were counted if distinguished by a suture 

 between fused centra, or by the presence of a 

 neural or haemal spine. Columns in which 

 individual vertebrae could not be distinguished 

 were discarded (three specimens). Ages were 

 determined from scales, and the fish were 

 assigned to year classes as defined by Jtine 

 and Roithmayr (1960). 



The correlation between vertebral counts 

 and fish length was tested for significance. 

 Vertebral counts were subjected to the vari- 

 ance-ratio and Bartlett's tests to test the 

 assumption of common variance, and an 



analysis of variance was used to test the 

 significance among means of sexes, age groups, 

 year classes, spawning groups in the same 

 and different localities, and juveniles and 

 adults from adjacent localities (Snedecor, 

 1956). 



Table 1 summarizes data on length, age, 

 and vertebral count of 6,048 spawners and 

 nonspawners in samples by locality and date. 

 Means and sums of squared deviations used in 

 the analyses also are given. 



RESULTS 



Variation of Vertebral Counts With Fish Length 



To determine whether vertebral counts were 

 related to fish length, a correlation analysis 

 was nnade of the individual and combined 

 samples from each locality (spawners and 

 nonspawners were treated separately). Three 

 of the correlation coefficients were of border- 

 line significance (at the 5 percent level), and 

 the maximum variation attributable to length 

 in combined samples was suggested to be less 

 than 2 percent. Accordingly, length was not 

 considered in the analyses which follow. 



Variation between Sexes 



Because of unequal numbers of nnales or 

 females in some samples, differences in the 

 mean vertebral counts between samples fronn 

 the same locality might reflect differences 

 between the sexes. However, analysis of mean 

 vertebral counts of 12 samples, connprising 

 859 individuals of 3 separate year classes 

 (1955, 1956, and 1957), showed that none of 

 the samples had a statistically significant 

 difference between sexes. It was concluded 

 that variation associated with sex was random, 

 a conclusion which agrees with the findings of 

 June (1958) and Sutherland (1963) concerning 

 juveniles. Accordingly, males and females 

 were not separated in subsequent analyses. 



Variation of a Year Class with Age 



In all localities, the proportion of fish of 

 a given year class in samples changed as 

 the year class became older. Accordingly, 

 heterogeneity between mean vertebral counts 

 could arise from the varying numbers of fish 

 at different ages in the samples. Comparisons, 

 therefore, were made of the two most abundant 

 year classes (1955 and 1956) at successive 

 ages in each locality (table 2). Of the 12 

 possible comparisons (samples which con- 

 tained no fewer than 60 fish of a given age), 

 one showed a significant difference at the 

 5 percent level, and this was confined to 

 nonspawners of the 1955 year class in the 

 Long Island samples. In this case, there was 

 a small decrease in the mean vertebral count 



