SHAKLEE and SAMOLLOW: GENETIC VARIATION IN A DEEPWATER SNAPPER 



very different sizes (and therefore different ages) 

 taken over a considerable period of time may have 

 obscured important information and possible 

 heterogeneity in the data. This is particularly true 

 since the pink snapper were collected over a 2V2-yr 

 period and ranged in size from 230 to 770 mm FL 

 (fork length) — presumably representing a range 

 of ages from 2 to over 16 yr (Ralston and Miyamoto 

 1983). With this in mind, the data set was par- 

 titioned (where fish lengths were known and sam- 

 ple sizes were large enough to allow reasonable 

 statistical tests) to allow analyses of the collection 

 for 1) evidence of variation in allele frequency 

 between age classes, 2) evidence of year-to-year 

 variability in allele frequency at a location, 

 and 3) evidence of allele frequency variation 

 within a homogeneous age class among localities. 

 The samples from Maro Reef and French Frigate 

 Shoals (FFS) allowed partitioning with regard to 

 size classes. The frequency distributions charac- 

 terizing individual collections making up the 

 samples for Maro Reef and FFS as well as those for 

 the three unpartitioned areas (Necker, Molokai, 

 and Hawaii) are shown in Figure 2. Because the 

 collections from both Maro Reef and FFS con- 

 tained adequate numbers of specimens through- 

 out the size range, they were subdivided into two 

 groups. One group consisting of small, young fish 

 (300-500 mm FL; about ages 2-5 yr — see Ralston 

 and Miyamoto 1983) and one of large, old fish 

 (501-770 mm FL; about ages 5-16+ yr). The allele- 

 frequency distributions characteristic of these 

 separate groups are shown in Table 4. For Maro 

 Reef and FFS, x 2 analysis of the data for 

 goodness-of-fit to Hardy- Weinberg expectations 

 revealed no significant deviations (5 loci x 2 loca- 

 tions x 2 size groups = 20 tests), although it 

 should be emphasized that the statistical test is 

 not robust for small sample sizes (Fairbairn and 

 Roff 1980). Heterogeneity \ 2 tests between size 

 classes at both Maro Reef and FFS revealed only 

 one statistically significant difference (for Adh at 

 Maro Reef, x 2 i = 4.72; P < 0.05). A third test for 

 changes in allele frequency with size was con- 

 ducted on a pooled data set including all fish of 

 known fork length from Maro Reef, FFS, and 

 Necker. This pooled NWHI sample had consider- 

 ably larger sample sizes in each cell making the 

 statistical tests more robust. Genotype propor- 

 tions at all five loci in both size (age) groups in the 

 combined NWHI samples were in agreement with 

 Hardy- Weinberg expectations. Of the five con- 

 tingency x 2 tests between size (age) groups, only 

 that for Ldh-C was significant (x ! = 4.22; P < 



S 



E 



& - 





FORK LENGTH (mm) 



FIGURE 2. — Size-frequency histograms for eight snapper col- 

 lections. 



0.05). This significant outcome was due to an in- 

 crease in the frequency of the more common (slow) 

 allele with increased size (age). The increase in the 

 frequency of the slow allele with increasing size 

 was characteristic of the individual samples from 

 both Maro Reef and FFS. A similar trend in 

 allele-frequency change, this one involving an in- 

 crease in the frequency of the rarer (slow) allele of 

 Adh, was observed in the samples. Although the 

 trend for Adh was statistically significant in the 

 Maro Reef sample, it was not in the pooled NWHI 



707 



