Let me rephrase - the tree they arrived at may have looked the same but their interpretation of what it indicates about how many species of Boobook there are is totally different. Have a look at
this statement by Les Christidis, one of the authors of the 1998 paper, and compare it to the treatment of the same species in the Koenig book.
Christidis & Boles (
2008) repeated a similar statement - and, yes, there is indeed an obvious difference of interpretation. (In defense of Olsen 1999, though, it could be argued that there was no real "misquotation" in HBW, because the exaggeratedly poor referencing system enforced by the editors of this series simply makes this impossible - Olsen's text alludes to "molecular differences," but there is no way to know whether these were indeed derived or not from the two Norman et al. 1998 papers...)
I'm not sure on what, exactly, the split
N. leucopsis is based, but I would agree that it has no strong molecular basis, and that the cyt-b and ND2 data in the two Norman et al. (1998) papers certainly do not make it necessary.
However, the split as implemented in HBW is another story. HBW retained
leucopsis as conspecific with
novaeseelandiae (as indeed recommended by Norman et al.), but split a
N. boobook made only of taxa from the mainland of Australia. Even if, in Norman et al's papers, "the Tasmanian Boobook Owl samples were included to represent the Australian Boobook," these papers in practice did not include any mainland Australian specimen, thus they are actually completely irrelevant to this version of the split - they can neither support, nor refute it.
In their 1998 paper on the specific status of
N. natalis (
abstract here), Norman et al. argued in favor of a three-way split of the "
N. squamipila complex" based on the finding that the genetic divergence levels between the three races they had tested were "comparable to the levels of divergence observed between obviously distinct species such as
N. rufa and
N. strenua (5.4%) and were consistently greater than observed among subspecies within the monophyletic
N. novaeseelandiae complex (1.5% to 2.3% between
novaeseelandiae,
leucopsis and
undulata)." Their phylogenetic reconstructions also placed
N. s. squamipila basal to a clade grouping
N. s. natalis,
N. s. hypogramma,
N. connivens, and the
N. novaeseelandiae complex - the support given to this clade by sequence analysis alone was admittedly low, but all of these taxa also shared a 6 bp insertion in the ND2 gene, which
squamipila and the other
Ninox did not have. I've checked a few other published ND2 sequences from GenBank: this insertion is also clearly lacking in other owls, as well as in parrots. Insertions/deletions in the ND2 gene are very rare events, and it seems very unlikely that this particular insertion could have happened twice, or that it could have been exactly reversed in
squamipila only, thus its presence adds a lot of support to the hypothesis that
squamipila falls outside this group (in terms of mtDNA, at least). Within this group, ND2 united
novaeseelandiae,
leucopsis and
undulata with some support, but the relationships between them,
connivens,
natalis and
hypogramma were otherwise very poorly resolved. The distance between
natalis and
hypogramma (4%) was also the lowest in the "
N. squamipila complex" - making a split of
natalis (from
hypogramma) actually much more poorly supported in molecular terms than that of
squamipila from these two. (Yet, oddly, many subsequent authors - including König & Weick - accepted the former but not the latter... A consequence of the title of the paper...?)
If you now look at the trees in the recent Wink paper, I would argue that the case of
N. boobook is quite similar to, and looks at least as good as, that of
N. natalis. The cytochrome b is the only gene to have been sequenced for both
N. boobook and
N. novaeseelandiae as circumscribed in HBW. This gene entirely fails to resolve
N. novaeseelandiae s.l. as a monophyletic group. The longer available cyt-b sequences (see also the attached file below) suggest the existence of a clade including four lineages -
boobook,
novaeseelandiae,
rudolfi (a bit surprisingly omitted from Wink et al's Figure 4, while they did have a sequence;
rudolfi does appear in their Figure 1, that is based on cyt-b+RAG-1), and
connivens - that are all separated by quite comparable levels of divergence. This admittedly does not
prove conclusively that
N. novaeseelandiae s.l. is not monophyletic, but neither did Norman et al. (1998)
prove conclusively that
natalis and
hypogramma are not sister taxa. The uncorrected distance between
boobook and
novaeseelandiae, based on these long cytochrome b sequences, is 4.36%. The distance between
N. rufa and
N. strenua is only 3.65%...
L -