Laserfiche WebLink
were generated (as in the previous program), calculating h each time. Mean diversity values <br />and 95% confidence limits were calculated as described before. <br />Results and Discussion <br />MtDNA variation in hatchery year classes.- Analysis of subsamples from DNFH yielded <br />the results in Tables 2 and 4. Unique haplotypes (those not found in the sample from Lake <br />Mohave) were present in only the 1987 and 1989 year classes (2 and 4 unique haplotypes, <br />respectively). The 1987 and 1989 year classes possessed levels of variation (relative number <br />of haplotypes and diversity estimates) comparable to those of the source (Lake Mohave), <br />whereas the 1990 year class possessed fewer haplotypes and a considerably lower diversity <br />value. <br />Resampling a subset of the Lake Mohave sample resulted in a mean diversity of 0.91 <br />with a 95% confidence interval ranging from 0.80-0.98. Only the 1990 year class (h = 0.71) <br />was significantly different from the source population, with only one of the 1000 resamples <br />exhibiting a lower estimate (Figure 4). <br />This means of testing for significant differences in diversity between the source and <br />hatchery populations is conservative because our representative sample of the source was <br />relatively small (N = 25). This reflects the fact that even if the source population ~s <br />maximally diverse, the finite number of parents forces an upper limit on h. Increased source <br />size decreases the probability of resampling the same individual, thereby decreasing variance <br />in distribution of expected h values (data not shown). If subsamples were drawn from a <br />larger source population, even the diversity values for the 1987 and 1989 year classes would <br />be significantly different from the source. <br />While our estimate of diversity measures the relative frequency of haplotyp~s within a <br />19 <br />