Lowry et al.: Phylogeny of Oreosomatidae 



703 



1)1)24 



r— N. 



034 



— IllOVl 



N. rhomboidalis 



N. helgae 

 0.216 



P. maculatus 



0095 A. niger 



N. sp 



 A. verrucosus 



O. atlanticum 



C. australis 



0.0 0.1 



0.2 0.3 0.4 



Distance from root 



0.5 



Figure 3 



Tree constructed from Rogers' distance by the distance-Wagner 

 procedure. Numbers represent relative branch lengths. 



Neither analysis with the berycid or the acanthurid 

 produced the shortest trees that resembled in any 

 way the trees produced from the phenetic analyses 

 or the cladistic analyses with either all three 

 outgroups or C. australis alone. Naso tuberosus as 

 the outgroup resulted in a single shortest tree (49 

 units) with A. verrucosus as the most divergent 

 oreosomatid and O. atlanticum grouped next to the 

 N. rhomboidalis and N. helgae cluster. Four short- 

 est trees (42 units) resulted from the analysis with 

 B. splendens, all of which, although confirming the 

 divergence of O. atlanticum, resulted in N. rhom- 

 boidalis and N. helgae diverging independently 

 from the other four species. 



Discussion 



The mean heterozygosity per locus for the seven 

 oreosomatid species ranged from 8.3% to 18.1%, 

 with an overall mean of 11.8%. These figures 

 are considerably higher than the mean of 5.1% 

 for 195 species of marine and freshwater fish 

 (Ward et al., 1992) and 5.5% for 106 species of 

 marine teleosts (Smith and Fujio, 1982). Three 

 of the seven oreosomatid species had mean het- 

 erozygosity values (12.1% to 18.1%) which ex- 

 ceeded the highest value of 11.7% reported by 

 Ward et al. (1994) from comparisons of genetic 

 diversity among populations of 57 species of 

 marine fish. This value was shown by two spe- 

 cies, Fundulus heteroclitus (Linnaeus, 1766X15 

 loci, Ropson et al., 1990) and Hoplostethus 



atlanticus (22 loci, Smith, 1986). Subsequent stud- 

 ies of additional loci in the orange roughy, H. 

 atlanticus, raised the estimate of its mean het- 

 erozygosity to 13.0% (Elliott and Ward, 1992). 



Two or more samples contributed to these het- 

 erozygosity estimates for five of the oreo species 

 (Table 1). These estimates of heterozygosity are 

 Hardy- Weinberg expected heterozygosities based 

 on pooled allele frequencies. Had there been sub- 

 stantial differentiation of allele frequencies among 

 samples, such estimates of heterozygosity would 

 have been higher than average sample heterozy- 

 gosities. In fact, the degree of inter-sample differ- 

 entiation was, with a single exception, very lim- 

 ited (Lowry et al., unpubl. data). Thus these esti- 

 mates of total heterozygosity will be very similar 

 to estimates of sample heterozygosity. The one ex- 

 ception was the locus sSOD* in N. rhomboidalis. 

 Variation at this locus was found to be depth-re- 

 lated; samples with a high frequency (>0.6 [cf. 

 <0.2]) of sSOD* 140 came from deeper water (>700 

 m). The total heterozygosity for this one locus was 

 0.500, whereas its average sample heterozygosity 

 was 0.285. Use of sample heterozygosity rather than 

 total heterozygosity would effect a small reduction 

 in the overall heterozygosity estimate for this spe- 

 cies (from 0.127 to 0.119). Note also that the species 

 with the highest degree of variation, N. sp., came from 

 a single sample. The high variability seen in the oreos 

 cannot be attributed to inter-sample differentiation. 



It is clear that oreosomatids have higher heterozy- 

 gosities than most species of teleosts. Interestingly, 

 both oreosomatids and the similarly variable orange 

 roughy occupy deepwater habitats. As speculated by 



x: 



55 



100 



(85) 



(46) 



(49) 



64 



(48) 



N. rhomboidalis 

 N. helgae 

 A. verrucosus 

 P. maculatus 

 , N. sp. 



A. niger 

 < O. atlanticum 

 C. australis 



Figure 4 



The 50% majority-rule consensus tree of the eleven shortest trees 

 produced by PAUP analysis, with numbers representing the per- 

 centage consensus measures for the eleven trees. Numbers in 

 brackets are bootstrapped values ( 100 replicates). 



