FISHERY BULLETIN: VOL. 85, NO. 3 



and among population samples. For polymorphic 

 restriction enzymes, all digestion profiles of vari- 

 ants were consistent with the hypothesis of single 

 nucleotide substitutions. No mtDNA size vari- 

 ants, resulting from additions or deletions of 

 DNA, and recently found in a number of fish 

 groups (Bermingham et al. 1986; R. W. Chap- 

 man^), were observed. 



The mean mtDNA genome size, found by aver- 

 aging the sums of all digestion patterns (Table 1), 

 was 16,990 bp (base pairs) ± 620 bp (SD). In 

 quantifying molecular size from ethidium bro- 

 mide stained agarose gels, two constraints must 

 be noted. First, variation associated with mea- 

 surement of fragment mobilities is inevitable. 

 Because of the nonlinear relationship between 

 fragment mobility and molecular size, slight 

 measurement errors can produce large variations 

 in estimated sizes, particularly for fragments 

 with low mobilities. As a consequence, ho- 

 mologous cleavage fragments (those which con- 

 sistently exhibit the same mobility on a single 



2R. W. Chapman, Chesapeake Bay Institute, The Johns Hop- 

 kins University, Shandy Side, MD 20764, pers. commun. De- 

 cember 1985. 



Table 2. — Distribution of mtDNA composite digestion patterns in 

 samples of Atlantic herring. 



1 Letters (from left to right) are digestion patterns for Apa\, BglW, 

 BstEW, EcoRI, EcoRV, Kpn\, and Xho\ (Table 1). 



agarose gel) may yield different molecular size 

 estimates, e.g., Apal fragment "a". Table 1. Sec- 

 ond, mtDNA cleavage fragments less than 500 bp 

 could not be routinely scored on ethidium bro- 

 mide gels because of their low absolute staining 

 intensities and fluorescent background in this re- 

 gion (Fig. 2). Regardless of the above constraints, 

 individual cleavage fragment phenotypes could 

 be consistently determined. 



Seven polymorphic restriction endonucleases 

 (Apal, Bglll, BstEll, EcoRI, EcoRV, Pvull, and 

 Xhol) were used to generate composite digestion 

 patterns for individual Atlantic herring. Twenty- 

 six unique composite digestion patterns were ob- 

 served in 69 completely characterized individual 

 Atlantic herring. The distribution of these com- 

 posites with respect to spawning locality is given 

 in Table 2; five common composites (nos. 1-5) 

 were observed to occur at all three spawning lo- 

 calities. 



Shared fragment similarity was calculated 

 pairwise for all composites and was used to 

 generate estimates of p, percent sequence diver- 

 gence (Upholt 1977). Estimated sequence diver- 

 gence among composites varied considerably, 

 mean = 1.66% +/- 0.91 (SD), range 0.19% - 

 4.37% (Table 3). Phenetic relationships among 

 composites were examined by UPGMA (Un- 

 weighted Pair Group Method of Arithmetic aver- 

 aging) clustering (Sneath and Sokal 1973) of se- 

 quence divergence (Fig. 3). Two major clusters 

 were noted: one involving three composites (5, 17, 

 26) and the other including all other composites. 

 Composites from both clusters were present in all 

 spawning populations. 



A network of relationships among composites 

 was constructed by connecting composites in in- 

 crements of single site gains or losses to minimize 

 the total number of restriction site changes re- 

 quired (Fig. 4). Sixteen equally parsimonious net- 

 works requiring 29 steps were generated. Com- 

 posite number 1 can be considered central 

 because it is the most common pattern observed 

 and also occurs in the Eastern Atlantic (S. M. 

 Bogdanowicz unpubl. data). 



DISCUSSION 



Based on the occurrence of geographically dis- 

 junct spawning groups and homing of some 

 tagged individuals, it has been tacitly accepted 

 that Atlantic herring stocks are reproductively 

 isolated. The basic analytical premise of this 

 study was that restriction endonuclease analysis 



564 



