See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/242567840

Phylogenetic analysis of the 24 named albatross taxa based on full

mitochondrial cytochrome b DNA sequences

Article in Notornis · June 2009

CITATIONS READS

32 1,396

4 authors, including:

Geoff Chambers Ryan W J Steel

Victoria University of Wellington The Walter and Eliza Hall Institute of Medical Research
299 PUBLICATIONS 7,312 CITATIONS 31 PUBLICATIONS 631 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Ryan W J Steel on 11 January 2014.

The user has requested enhancement of the downloaded file.

82
Notornis, 2009, Vol. 56: 82-94
0029-4470 © The Ornithological Society of New Zealand, Inc.

Phylogenetic analysis of the 24 named albatross taxa based on full

mitochondrial cytochrome b DNA sequences

GEOFFREY K. CHAMBERS*
CASSIDY MOEKE
RYAN STEEL
School of Biological Sciences, Victoria University of Wellington, PO Box 600 Wellington, New Zealand

JOHN W.H. TRUEMAN

School of Botany and Zoology, Australian National University, ACT 2000, Australia

Abstract A stable evidence-based taxonomy is a critical requirement for the effective future conservation of
the albatrosses. Recently published partial molecular phylogenies are in broad agreement with respect to the
structure of the evolutionary tree for most named taxa, but the analytical methods used to create them have
been seriously criticised and they must be considered provisional at best. A further problem is that their
authors reach startlingly different conclusions regarding the numbers of taxa which should be recognised
as species; 13 vs. 24. Here, we attempt to resolve this situation by supplying full length mitochondrial
cytochrome b data presently missing for 2 taxa, carrying out thorough phylogenetic analyses meeting the
requirements of published prescriptions and taking into full account other sources of new molecular data
and contemporary opinions on albatross nomenclature and the status of taxa. We provide general support
for the published trees and critically evaluate claims regarding how many taxa represent full species. Some
genetic distances between pairs of taxa are so small that considerable weight of alternative evidence is
required to support any decision leading to a recommendation to split them. We note that the empirical
boundary between consensus and controversy falls at or around 1% DNA sequence divergence and further
that few, if any, commentators recognise taxa that are separated by less than 0.1% as being valid species.
Chambers, G.K.; Moeke, C.; Steel, R.; Trueman, J.W.H. 2009. Phylogenetic analysis of the 24 named albatross taxa based
on full mitochondrial cytochrome b DNA sequences. Notornis 56(2): 82-94.

Keywords albatross; phylogeny; molecular systematics; cytochrome b DNA

INTRODUCTION on the Conservation of Albatrosses and Petrels

Albatrosses are a long-lived charismatic component (ACAP) has been set up to oversee the situation.
of the oceanic avifauna. They presently face many This body recognises that a key factor in the
threats from pollution of the marine environment effective conservation management of these species
and the activities of international fishing fleets is a good understanding of species descriptions and
(Croxall & Gales 1998, Mills & Ryan 2005). These distributions (ACAP 2005). As a first step towards
problems are compounded by their intrinsically this goal they have set up a Taxonomy Working
slow rate of reproductive increase, with most Group (TWG) to critically review the published
species raising only 1 chick every year or every 2 data and make recommendations to the ACAP via
years, and losses due to disruption of breeding pairs. the annual meeting of their Advisory Committee
The situation is dire, but not impossible, and a new (AC).
international agency, the Parties to the Agreement Albatross systematics has a long and complex
history (Christidis & Boles 1994, 2007, OSNZ
Received 22 Jul 2009; accepted 6 Nov 2009 Checklist Committee 1990, 2010). During the
Corresponding author: geoff.chambers@vuw.ac.nz last 300 years biologists have produced formal
 Albatross molecular systematics 83

descriptions of more than 80 separate taxa (ACAP and 13 taxon models simultaneously via a covert
2009). Until recently these were entirely based on ‘superspecies’ approach.
morphology, occasional reports of field observations Overall, the short comings of the present
and behavioural anecdotes. In recent times, it has situation can be summarised:
become conventional for the entire assemblage to (i) The cytochrome b dataset is incomplete with
be bundled into just 2 genera (Jouanin & Mougin, only 22 of 24 described taxa represented.
1979); Phobetria (sooty albatrosses) and Diomedea (ii) The 2 leading published phylogenetic analyses
(everything else) with variable numbers of species are congruent, but independently inadequate.
and subspecies (e.g. Sibley & Monroe 1990 list 14 (iii) There has been inappropriate and/or inconsistent
species). The application of molecular methods adoption and use of species concepts in these
began about 10 years ago with an examination of previous reports.
those on this list (Nunn et al. 1996) and later expanded (iv) Investigators have missed the opportunity to
to 22 taxa and included most of the recognised incorporate new molecular data from other
subspecies (Nunn & Stanley 1998). This work was loci – notably mtDNA control region sequences
based on examination of full length nucleotide from shy albatross (Abbot & Double 2003a,b),
sequences from mitochondrial cytochrome b genes black-browed, and wandering albatrosses (Burg
and was incorporated into a synthetic proposal for & Croxall 2001, 2004, respectively).
all 24 named taxa entitled ‘Towards a new taxonomy Here we attempt to fill the gaps in (i) by supplying
for albatrosses’ (Robertson & Nunn 1998). mitochondrial cytochrome b data for the 2 missing
The operative word in the title of this work is taxa: white-capped (T. steadi) and northern (= Pacific)
‘towards’ and reflects the intent of the authors that Buller’s albatross (T. bulleri platei; see Methods for
this be regarded as a working hypothesis, rather than recent changes to common and scientific names for
an exhaustive evaluation of all available evidence. taxa). We address (ii) by carrying out a thorough
In short, all 24 taxa were raised to species level. phylogenetic analysis of the 24 taxon dataset intended
Despite the caveat above, this suggestion has been to satisfy all previously published criticisms and
generally well received, but not universally adopted supply a full matrix of genetic distances between
in its entirety (Brooke 2004, Onley & Scofield 2007), taxa. Finally, we consider the taxonomic ranking
i.e., most authors accept most, but not all, of the splits of the terminal branches in our tree(s) to the extent
while keeping their own counsel regarding which that this gene is informative about those rankings
ones. Perhaps the strongest feature of the Robertson and consider opinions based on the new data in (iv)
and Nunn (1998) model is their re-establishment of in the light of published commentary on (iii) from
4 genera (after Alexander et al. 1965); Phobetria and ACAP and others in the sources above.
Diomedea are retained, but the latter now restricted
to just the large royal and wandering albatrosses, METHODS
while 2 new genera Phobastria (containing all Tissue samples and DNA extraction
Northern Hemisphere taxa) and Thalassarche Extracts were isolated from frozen liver samples
(smaller albatrosses sometimes commonly known taken from by-catch autopsies and provided by
as ‘mollymawks’) are established. These generic C.J.R. Robertson with reference numbers; 54014 for
names have been widely adopted by recent authors T. steadi and 54245 for T. platei. The DNA extracts
(Tickell 2000, Brooke 2004, Christidis & Boles 2007, were originally made by van Bekkum (2004) using
Onley & Scofield 2007, OSNZ 2010). The strongest methods described in van Bekkum et al. (2004) and
critics of their model (Penhallurick & Wink 2004) stored frozen at -80oC.
accepted the 4 genera, but lumped the taxa into
just 13 species. They based this decision on their PCR amplification and DNA sequencing
interpretation of a ‘Multidimensional Species Concept’ The mitochondrial cytochrome b target was
(MDSC) after Mayr (1996), as applied to their novel amplified from template DNA (5 μL each reaction)
phylogenetic analysis of the Procellariformes. This using the L14863/H16065 primer pair under the
study (and to some extent those carried out by reaction conditions and thermal cycling regime
their predecessors) has been extensively criticised described by Nunn et al. (1996). The quality and
in turn (Rheindt & Austin 2005). One key point quantity of dsDNA amplification products were
is that the MDSC had simply been reduced to a estimated by agarose gel electrophoresis. Products
‘barcoding’ (see Methods) decision process based on from reaction giving clean single bands of the
uncorrected nucleotide sequence divergence, where expected size (~1200 bp) were purified using Roche
only taxa separated by greater than 1% divergence High Pure kits according to the manufacturer’s
were recognised as species. Uncritical application instruction. Fluorescent dideoxynucleotide cycle
of this view in the mistaken belief that it reflects DNA sequencing services were provided by the
a consensus can lead to confusion. For instance, Allan Wilson Centre at Massey University, Albany
Lindsay (2008) tries to adopt (p.114) both the 24 Campus and data returned as electronic files.
84 Chambers et al.

Table 1. Albatross taxa examined in this study with names after ACAP (2009) including subspecies. * The data for these
taxa are new to this study.

Common Name Scientific Name GenBank #

Genus Phoebastria:
Short-tailed albatross P. albatrus U48952.1
Waved albatross P. irrorata U48951.1
Laysan albatross P. immutabilis U48949.1
Black-footed albatross P. nigripes U48950.1

Genus Diomedea:
Southern royal albatross D. epomophora AF076049.1
Northern royal albatross D. sanfordi U48946.1
Tristan albatross D. dabbenena U48947.1
Antipodean albatross D. a. antipodensis AF076047.1
Gibson’s albatross D. a. gibsoni AF076050.1
Amsterdam albatross D. amsterdamensis U48948.1
Wandering albatross D. exulans AF076048.1

Genus Phoebetria:
Sooty albatross P. fusca U48942.1
Light-mantled albatross P. palpebatra U48943.1

Genus Thalassarche:
Indian yellow-nosed albatross T. carteri AF076091.1
Atlantic yellow-nosed albatross T. chlororhynchos U48944.1
Grey-headed albatross T. chrysostoma U48954.1
Campbell albatross T. impavida AF076093.1
Black-browed albatross T. melanophris U48955.1
Buller’s albatross T. b. bulleri U48945.1
Pacific albatross T. b. platei EU024821*
Chatham albatross T. eremita AF076092.1
Salvin’s albatross T. salvini AF076094.1
Shy albatross T. cauta U48953.1
White-capped albatross T. steadi EU024820*

Data analysis AF076060* + U48941 for M. giganteus differ by

The DNA sequence files were checked carefully 0.88% and AP009191 + U48940* for P cinerea differ
by eye to eliminate misassigned bases and full by 0.35%. Those marked (*) were the ones used in
sequences assembled using DNAStar Inc Lasergene Penhallurick & Wink (2004). A full list of the taxa
software package. Further quality assurance checks examined is given in Table 1.
were made by translating the ORFs into protein First, we constructed a Neighbour Joining tree
sequences and making sure that they corresponded with PAUP* 4.0b10 (Swofford 2002) using all 28
to typical cytochrome b protein sequences containing sequences and assembled a matrix of uncorrected ‘p’
neither stop codons nor non-conservative amino distances followed by bootstrap repeats with a 1000
acid substitutions. These consensus sequences pseudoreplicates to obtain bootstrap support values.
were further checked for close nucleotide sequence Next, we carried out a cladistic parsimony analysis
matches to their close relatives (see below). These in PAUP* 4.0b10 with characters unordered and
2 sequences have been deposited in GenBank with using a branch and bound search which returned
acquisition numbers (EU024820 and EU024821). just 1 shortest tree. The procedure was repeated with
They were then combined into an extended data 1000 pseudoreplicates to obtain bootstrap values
set with the sequences for other taxa as used by and then rerun with tv/ts weightings of 2:1 and 3:1
Penhallurick & Wink (2004) but independently which yielded the same shortest tree. Finally, we
downloaded from the NCBI database (http://www. ran a maximum likelihood (ML) analysis in PAUP*
ncbi.nih.gov/). We also downloaded outgroup 4.0b10 using the model of nucleotide sequence
sequences for giant petrel (Macronectes giganteus) evolution (K81 + I + G) recommended by ModelTest
and grey petrel (Procellaria cinerea). However, since V 3.06 software (Posada & Crandall 1998) under the
we discovered that both of these taxa are now Aikake Information Criterion (AIC). The procedure
represented by 2 slightly divergent cytochrome b was repeated with 100 pseudoreplicates and fast
sequences we imported both for each respectively: search setting to obtain bootstrap values. We note in
 Albatross molecular systematics 85

passing that preliminary analyses using TVM + G, taxa whose COI sequences differ by more than 1%
JC and HKY85 models of evolution gave the same are good species, although this value may differ
result. Further details of analytical parameters and for some sets of taxa or when other genes (e.g.
methods are given for the tree diagram in Figure 1 cytochrome b) are used (Newton 2003).
and Table 2. Finally, we analysed the data using the This approach was recently applied to 260
GTR+I+G model with all parameters estimated from species of North American birds and found to work
the data, as implemented in the program GARLI well (Hebert et al. 2004). However, the method is not
(Zwikl 2006), to obtain both a best-fit maximum without its pitfalls and inherent limitations and has
likelihood tree and branch support values based on attracted a number of influential critics (e.g. Moritz
a 1000 pseudoreplicate bootstrap run. & Cicero 2004). For instance, Hickerson et al. (2006)
We did not perform extensive Bayesian analyses warn that DNA barcoding will fail to discover many
principally because the 3 independent methods animal species. However, the true key diagnostic
above had already returned identical trees showing criterion for declaring a ‘barcoding species’ is that
that the tree estimate for these data is insensitive the range of genetic distances between individuals
to tree estimation procedure. There are also other within the limits of the proposed species should
technical and practical reasons for this decision: be significantly smaller than those between them
1) Bayesian tree estimation with flat priors on and members of other species. This is effectively
the trees reduces to likelihood-weighted survey a form of phylogenetic species concept based on
of reasonably likely trees and as such provides phenetic distance analysis (see Discussion). It is
little information beyond the conventional ML clear that multiple taxon sampling is a minimum
procedure, and 2) a conventional ML tree can be requirement to support such claims and any wider
computed from our data in reasonable time and application of single cut off values must be taken as
hence there is no need for Bayesian tree estimation both hypothetical and provisional.
as a shortcut. However, we did run quick checks
using MrBayes (Hulsenbeck et al. 2001) at default Species concepts in birds
settings under the F81 and HKY + I + G models of A wide variety of species concepts is available for
evolution and demonstrated that both return the biologists to choose (e.g. Claridge et al. 1997). For
same tree identical to the one produced by ML and birds, making the right selection is particularly tricky,
other methods of analysis. since many pairs of taxa become morphologically
or behaviourally distinct long before they lose the
Taxonomic approach capacity to interbreed. Historically, the traditional
We advocate a ‘total evidence’ approach favourites have been morphological, biological and
which amounts to diagnosis of taxa as reliably phylogenetic (diagnostic) type concepts, together
differentiated as based on all available published with their modern counterparts, the General Lineage
information. The sections below outline our and Multidimensional Species Concepts (Helbig et al.
position on the use of molecular distance data and 2002). These authors combine diagnostic characters
phylogenetic separation in relation to systematic with evolutionary trajectory as a practical approach
and nomenclatural considerations. to resolving the dilemma of selecting an appropriate
concept. Their advice has been followed by ACAP
DNA barcodes (2006) and by OSNZ (2010).
This is a relatively new initiative. An international It is clear that once 2 taxa permanently cease to
consortium is building an open access ‘Tree of Life’ exchange genes they are on an inevitable pathway
based on biological classification using short DNA leading to genetic, morphological and behavioural
sequences from a standardized region of the genome divergence leading eventually to reproductive
(e.g. Stoeckle 2003, Edwards 2005). The preferred isolation (Randler 2006). But how can an observer tell
target is a 648-bp region of the mitochondrial where particular taxa are actually positioned on this
cytochrome c oxidase subunit I gene (COI). These pathway at any given time? Assortative mating in
sequences are used to create a public library of sympatry can be a good clue, as occurs in the black-
‘DNA barcodes’ linked to reliably identified and browed albatrosses on Campbell Is (see Discussion).
fully described specimens. Allopatric populations provide an extra challenge.
The leading assertion underlying this For instance, even strong natal philopatry has not
programme is that there is general concordance lead to divergence of the 2 southern populations
between mtDNA gene trees and species trees. This of T. b. bulleri on The Snares and Solander Is (van
is based partly on empirical observations and partly Bekkum et al. 2004). However, the microsatellite
on assumptions based on extrapolation. Hence, data do show that they are genetically separable
the organisers note that genetic variation forms a from the more distant northern populations of T. b.
discontinuous hierarchy paralleled by taxonomic platei (van Bekkum 2004), despite the similarity of
levels. It is then often taken to be the case that their cytochrome b sequences reported here.
86 Chambers et al.

The problem of avian subspecies is even more browed albatross nesting on the Falklands (Burg
difficult (Zink 2004). How can one decide what is & Croxall 2001) and the isolated population of P.
and what is not a subspecies when it is so difficult nigripes described by Edwards et al. (2001). We note
to say what species are in the first place? For birds, that the latter probably only represents an isolated
such fine distinctions can still be of value, but inbred population characterised by possession of
their reliability tends to depend on the group(s) an otherwise rare mtDNA haplotype. Consequently
of taxa under examination. It is well known that it is, therefore, unlikely to be recognised even at
particularly well-studied groups may be over subspecific level.
divided (Phillimore & Owens 2006). The nomenclature system used here follows
ACAP practice. OSNZ (2010) has a similar list but
Albatross nomenclature lump T. steadi and T. cauta as subspecies arguing
The scientific and common names applied to that the former is a Tasmanian population of the
albatross taxa have been in a constant state of flux as latter, the data of Abbott & Double (2003a,b) not
each succeeding wave of taxonomic revision breaks withstanding. Birdlife International also follow
upon the field. Luckily, there are many excellent ACAP but go further and split antipodensis from
technical guides to this history (Brooke 2004; ACAP gibsoni based on plumage features as mentioned in
2005, 2006, 2009, Onley & Scofield 2007, Christidis the text.
& Boles 2007, OSNZ 2010). Below we review some
tricky problems that have recently been resolved. RESULTS
In the wandering albatrosses (‘exulans complex’), The 2 new sequences passed our quality assurance
the taxon formerly known as D. chionoptera has now tests and were almost full length, lacking only 21 nt
been established as D. exulans as first described by from the 5’ end. These checks showed that our T.
Linnaeus (Medway 2006). A new type specimen platei sequence was most similar to that of T. bulleri,
has been described Schodde et al. (2009), and now and similarly, T. steadi was closest to T. cauta. These
generally adopted under the common name of observations are consistent with the Robertson &
wandering albatross, although some authors (Onley Nunn (1998) model. All tree-estimation methods
& Scofield 2007) still prefer to call it snowy albatross produced the same best-fit topology with the
after the common name originally attached to D. same relative branch lengths. Figure 1 shows this
chionoptera. Also in this group, the Antipodean tree optimised under the K81 + I + G maximum
albatrosses (taxon labels antipodensis and gibsoni) likelihood model. The results under alternative ML
have been lumped as subspecies. However, the models are indistinguishable to the eye. Internal
authors above prefer the term New Zealand nodes on this figure are numbered sequentially.
Albatross for the pair. Almost all internal branches have high bootstrap
In the genus Thalassarche, the 2 black-browed support across all methods; bootstrap scores
albatrosses include the widespread form T. are given in Table 2. In the Bayesian analyses, all
melanophris. For many years this has also appeared posterior probablites under the F81 model were
under the correct spelling melanophrys (Brooke ≥ 98% and under the more general HKY + I + G
2004) but the Rules of Zoological Nomenclature model they generally reflected the same order
require retention of the original incorrect spelling as the bootstrap values in the ML analyses. For
(ACAP 2009). Next, the taxon label platei, used example, the least well supported node 17 has
to describe the northern populations of Buller’s bootstrap <50% and a posterior probability of 62%,
albatross, must be dropped as the type specimen is with all other values >90% except for node 6 (81%)
reported (Robertson & Nunn 1998) to be a juvenile and node 12 (88%; full figures are available from
of the southern form, T. b. bulleri. These authors the authors on application). This demonstrates the
suggest T. sp. nov. as a provisional name until an now conventional 4-genus arrangement of taxa and
new type specimen can be described. This would matches those reported earlier (Nunn & Stanley
be appropriate if this taxon held up as a full species, 1998; Robertson & Nunn 1998; Penhallurick & Wink
but since most do not support this view the term T. 2004) with the addition of the 2 extra albatross
ssp. nov. might be more correct (Onley & Scofield taxa and the duplicate outgroups. The matrix of
2006). As Christidis & Boles (2007) and OSNZ uncorrected ‘p’ distances between pairs of taxa is
(2010) point out, the original observation was not given in Table 3.
supported by published evidence hence one must
resort to T. b. platei as we have done here. Finally, the DISCUSSION
term ‘mollymawk’ which was often used to describe We have produced the first complete mitochondrial
the smaller albatrosses that make up this genus has cytochrome b gene tree for all 24 named albatross taxa
now fallen into disuse. based on single accessions (i.e., 1 DNA sequence/
Two other taxa still remain to be named. These are taxon). The tree is robust to estimation methods with
the distinct lineage of the dark-eyed form of black- high boot strap support for most branches. It can
 Albatross molecular systematics 87

Fig. 1 Best-fit maximum

likelihood tree estimated by
PAUP and GARLI. The branch
lengths are optimised here
using the K81 + I + G maximum
likelihoodmodelrecommended
by MODELTEST but all other
models give the same tree
with virtually identical branch
lengths. The ML score (log
likelihood) under this model
is -4534. Internal nodes are
numbered as in the table of
bootstrap scores (Table 2)
and nodes accepted as valid
monophyletic groups on the
gene tree are shown by thick
lines.

thus be taken as indicating that there is strong signal related or otherwise the 2 pairs of outgroup taxa
in the data that may reflect historical relationships might be in taxonomic terms.
between those female lineages which carried these The 2 additional taxa are placed next to their
mitochondrial genomes. We acknowledge that closest relatives in accord with expectations. The
gene trees based on mtDNA characters may not T. b. platei sequence shows minimal divergence
correspond exactly to species trees and that gene (0.09%) from T. b. bulleri. The T. steadi sequence
trees are necessarily dichotomous due to the mode is placed in the ‘shy albatross complex’ together
of inheritance of this particular genome. We also with T. cauta, T. salvini and T. eremita, but closest
acknowledge that we have no information on intra– to T. cauta. This finding is in accord with evidence
taxon variation (see our comments on barcoding discussed at length in ACAP (2006) for the cauta/
in the Methods). Furthermore, as the majority of steadi pair and in ACAP (2009) for salvini/eremita.
internal branches are short in our tree, caution must Hence, we think that it is permissible to make the
guide our interpretations. The ‘p’ distances of 0.35- provisional claim that this particular arrangement
0.88 in the 2-sample outgoups illustrate this point of the 24 taxa enjoys general support. Further, we
well. However, we must be equally careful with believe that the comprehensive set of phylogenetic
respect to the values themselves, since we have analyses that we have carried out are sufficient to
no background information to show how closely satisfy the observations of Rheindt & Austin (2005)
88 Chambers et al.

Table 2. Bootstrap scores by node under a range of tree-estimation methods. Nodes judged to have sufficient evidential
support in these data to be accepted as delineating a monophyletic group on the gene tree are marked * here and are shown
in bold on Figure 1.

Bootstrap score % Gene-tree

nodes with
Node id #
Unweighted ML- PAUP ML -GARLI bootstrap
As in Fig. 1 NJ with ‘p’
parsimony (K81 + I + G) (GTR+I+G) support
distances
Branch & Bound ‘fast’ bootstrap full heuristic
1 100 100 100 100 *
2 100 100 85 86 *
3 100 100 100 100 *
4 100 100 91 100 *
5 100 99 98 99 *
6 <50 58 73 71
7 97 92 92 92 *
8 100 100 96 100 *
9 100 100 99 100 *
10 100 96 88 87 *
11 52 88 75 80
12 <50 60 55 66
13 80 77 82 74
14 100 100 95 97 *
15 100 100 97 98 *
16 100 100 97 99 *
17 55 69 <50 66
18 87 81 94 93 *
19 100 99 99 100 *
20 96 92 64 93 *
21 100 100 100 100 *
22 87 67 53 73
23 98 99 95 100 *
24 95 97 72 95 *

on deficiencies in prior analyses of the restricted 22 microsatellite loci as reported by van Bekkum 2004)
taxon dataset (Robertson & Nunn 1998, Penhallurick become available for all 24 taxa, rather than for
& Wink 2004). The fact that this mtDNA gene tree limited subsets of them. We do note the conventional
is robust may help explain why our present results reservation that although data from further mtDNA
match those reported earlier. But even if this set of targets have the potential to add to the existing
relationships may be said to be established beyond phylogenetic signal in the present dataset, the
reasonable doubt, this is not to say that these will mitochondrial genome should still be regarded as
remain unchanged when more extensive data from a single locus. Equally, if it should eventually prove
other mitochondrial targets (e.g. for Control Region possible to increase the number of cytochrome b
as reported by Burg & Croxall, 2001, 2004, Abbot & sequences per taxon, then the extended haplotype
Double 2003a,b, or for COI as observed by Cassidy, distributions may blur taxon boundaries. Where this
Steel & Chambers, unpubl. data) or nuclear loci (e.g. has been achieved for some taxa using other targets
for flanking regions and central repeat units of such as the mtDNA control region (see below) the
Table 3. The ‘p’ distance matrix

1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 M.giganteus -
2 M.giganteus 0.00875 -
3 P.cinerea 0.11374 0.10936 -
4 P.cinerea 0.11374 0.10936 0.00350 -
5 P.albatrus 0.14261 0.14611 0.15486 0.15311 -
6 P.irrorata 0.15048 0.15223 0.15836 0.15661 0.04374 -
7 P.immutabilis 0.14348 0.14698 0.15398 0.15223 0.03675 0.04724 -
8 P.nigripes 0.14348 0.14698 0.14611 0.14436 0.03500 0.04374 0.01750 -
9 D.epomorphora 0.14961 0.15311 0.15923 0.15748 0.07087 0.06912 0.06737 0.06737 -
10 D.sandfordi 0.14873 0.15223 0.15836 0.15661 0.06999 0.06824 0.06649 0.06649 0.00087 -
11 D.dabbenena 0.14086 0.14436 0.14786 0.14611 0.06387 0.07087 0.06737 0.06212 0.03062 0.03150 -
12 D.antip/gibsoni 0.13823 0.14173 0.14698 0.14523 0.06387 0.06912 0.06562 0.06037 0.03325 0.03412 0.00700 -
13 D.amsterdamensis 0.13998 0.14348 0.14873 0.14698 0.06737 0.07174 0.06737 0.06212 0.03500 0.03587 0.00875 0.00525 -
14 D.exulans 0.13998 0.14348 0.14698 0.14523 0.06212 0.06912 0.06387 0.05862 0.03500 0.03587 0.00875 0.00525 0.00525 -
15 P.fusca 0.12598 0.13473 0.13823 0.13648 0.10149 0.10761 0.09886 0.09274 0.09886 0.09799 0.09624 0.09711 0.09711 0.09361
16 P.palpebra 0.12598 0.13473 0.13473 0.13298 0.09974 0.10411 0.09624 0.09099 0.09624 0.09711 0.08749 0.08836 0.08836 0.08661
17 T.carteri 0.13211 0.13561 0.14436 0.14436 0.11286 0.11374 0.11286 0.10236 0.10586 0.10499 0.09974 0.09711 0.09886 0.09886
18 T.chlororhynchos 0.13298 0.13648 0.14261 0.14261 0.11111 0.11199 0.10936 0.09886 0.10586 0.10499 0.10149 0.09886 0.10061 0.10061
19 T.chrysostoma 0.13736 0.14086 0.14698 0.14523 0.10849 0.10936 0.10324 0.09449 0.10324 0.10236 0.10324 0.09974 0.10149 0.09974
20 T.impavida 0.13998 0.14348 0.14698 0.14523 0.11111 0.10761 0.10761 0.09886 0.10411 0.10324 0.10236 0.10061 0.10236 0.10236
21 T.melanophris 0.13736 0.14086 0.14086 0.13911 0.10849 0.11024 0.10499 0.09624 0.10324 0.10236 0.09974 0.09799 0.09974 0.09974
22 T. b. bulleri 0.13998 0.14173 0.14436 0.14261 0.11024 0.10936 0.10411 0.09799 0.10149 0.10061 0.10236 0.09974 0.10149 0.10149
23 T. b. platei 0.14261 0.14438 0.14698 0.14519 0.11313 0.11231 0.10695 0.10071 0.10430 0.10341 0.10520 0.10253 0.10429 0.10428
24 T.eremita 0.14261 0.14523 0.14611 0.14436 0.10674 0.11111 0.10586 0.09974 0.10061 0.09974 0.10061 0.09799 0.09974 0.09974
25 T.salvini 0.14173 0.14436 0.14523 0.14348 0.10586 0.11024 0.10499 0.09886 0.09974 0.09886 0.09974 0.09711 0.09886 0.09886
Albatross molecular systematics

26 T.cauta 0.13823 0.14086 0.14261 0.14086 0.10586 0.10849 0.10324 0.09536 0.09974 0.09886 0.09799 0.09536 0.09711 0.09711
27 T.steadi 0.13794 0.14062 0.14409 0.14231 0.10590 0.10863 0.10327 0.09526 0.09971 0.09883 0.09793 0.09527 0.09703 0.09702
89
90 Chambers et al.

Table 3. Continued. resultant data have supported the proposed splits

and even finer subdivisions, particularly between
allopatric populations of general morphotypes
such as the various black-browed albatrosses (Burg
& Croxall 2001).
27

-
We now turn to the rather more vexed question
of how best to label the terminal taxon on each

0.00176
branch: how many of them really merit species
rank? Final decisions in this regard go beyond
26

-
a simple molecular approach. However, an

0.00962
0.00976
important feature in any such debate is selection of
an appropriate species concept to fit the biological
25

situation (see Methods). For relatively simple data,

-
0.00262 such as those considered here, one tends to be
0.01050
0.01064
limited to the diagnostic version of the phylogenetic
species concept (Cracraft 1983, Mayden, 1997) which
24

emerges as a version of the barcoding paradigm

under MDSC (Penhallurick & Wink 2004). The
0.02135
0.02047
0.01782
0.01608

PSC has the well-known potential to inflate species

counts as in extremis each different DNA sequence
23

might be taken to represent a separate species.

0.00088
0.02012
0.01925
0.01662
0.01513

Thus, prudent investigators should properly collect

multiple DNA sequences to represent each taxon.
22

This must be done to show that the finite variation

-

within the taxon is bounded and well removed from

0.03150
0.03302
0.02712
0.02625
0.02800
0.02672

its nearest neighbour in phylogenetic space. Indeed,

the same requirement strictly applies equally to
21

barcoding studies in a general sense, but is rarely

practical.
0.00787
0.02887
0.03032
0.02800
0.02712
0.02712
0.02581

The efficient, but not robust, alternative is to

establish some cut off value for sequence divergence
20

above which taxa are considered species. The cut off

value should be selected with an eye to population
0.02012
0.01925
0.02625
0.02764
0.02362
0.02275
0.02362
0.02225

sizes, the rate of evolution of the molecular target

19

and the rate of speciation among the taxa under

-

study. In practice, it more often comes down to

0.02625
0.02712
0.02800
0.02975
0.03115
0.02887
0.02800
0.02712
0.02576

asking how far diverged well established or ‘good’

species pairs might be. This looks like a reasonable
18

and objective procedure, but is actually full of all

-

pitfalls. Not least of these is the near truism that

0.00350
0.02975
0.03062
0.03150
0.03325
0.03472
0.03062
0.02975
0.02887
0.02755

all the difficult cases will almost certainly be more

closely related than the exemplars. The greater
17

danger is that simple decision rules like a single cut

off number have a natural authority to persuade
0.07962
0.07787
0.07699
0.08136
0.07699
0.07699
0.07930
0.08049
0.07962
0.07699
0.07651

well beyond all reasonable expectations of their

16

actual performance. Nonetheless, we certainly agree

-

that adopting a 1% ‘p’ distance cut off provides

0.02100
0.08311
0.08136
0.07874
0.08486
0.08049
0.07699
0.07923
0.08311
0.08049
0.07612
0.07735

evidence for around 13 organism-level splits. To

go beyond this based on cytochrome b data alone
15

would be a step too far in our opinion. Rather, it

-

is the case that further splits must be argued on

the balance of total evidence and made despite
18 T.chlororhynchos

the contrary impression created by the small ‘p’

19 T.chrysostoma

21 T.melanophris

distance values. The tree itself is entirely in accord

20 T.impavida
16 P.palpebra

24 T.eremita

with the proposed relationships between taxa that

25 T.salvini
22 T.bulleri
17 T.carteri

27 T.steadi
23 T.platei

26 T.cauta
15 P.fusca

have emerged based on other evidence.

In their survey work for ACAP, the TWG follows
Helbig et al. (2002), who use a relaxed version of the
 Albatross molecular systematics 91

General Lineage Species Concept and starting (as species level by TWG in 2007 (ACAP 2009) based
we have here) from the species rich arrangement largely on breeding distributions and qualitative
of Robertson & Nunn (1998) in which pairs were morphological characters (Marchant & Higgins
lumped again if there is insufficient published 1990, Robertson 2002). This pair has also diverged
evidence of all types to support the split. We can quite recently and their cytochrome b distance is
now proceed to examine their recommendations only 0.4%. Thus, although the cytochrome b tree
(ACAP 2006, 2009) in the light of our cytochrome does not contradict their separation from the other
b sequence evidence for divergence between pairs 9 taxa in this genus, it is not sufficient in isolation to
of closely related birds. It will be clear from what justify splitting the yellow-nosed taxon pair.
follows that the general application of a cut off The second subclade, the grey-headed albatross
value of 1% divergence may be inappropriate for (T. chrysostoma), is a singleton and is accepted by all
such a recent radiation. This view is established authors at species level. In our tree, it groups with
from consideration of many cases and based on a the 2 black-browed albatrosses (T. impavida and T.
wide variety of molecular and other evidence. melanophris), but is well separated (‘p’ distances:
At the outset we observe that the branches 1.8 and 2.0%, respectively). Hence, this particular
within the genera Phoebetria (the 2 sooty albatrosses) species hypothesis is supported by the cytochrome
and Phobastria (4 northern hemisphere albatrosses) b data.
lead to well resolved singletons and are regarded The third subclade comprises the 2 members
as good species by all recent commentators. Our of the impavida/melanophris pair. The cytochrome b
cytochrome b trees give no grounds to question this sequences differ by 0.8% between these 2 species
position. but they have been decisively separated by Burg &
The genus Diomedea is divided in our gene tree Croxall (2001) on the basis of independent genetic
into clades containing the 2 royal albatrosses and the evidence and can be recognised by the colour of the
5 wandering albatrosses. The former are recognised iris; red in black-browed albatross (T. melanophris)
by TWG in 2007 as D. epomorpha and D. sandfordi and pale (straw) in the Campbell albatross (T.
on the basis of well-established morphological and impavida). They have recently been reported as
plumage differences. The cytochrome b distance is breeding in sympatry on Campbell Is, but have
only 0.08% and when considered alongside their distinct calls and mate assortatively (Moore et al.
potential to interbreed must leave this decision open 1997), although they are capable of hybridising
to review. Indeed, ACAP (2006) lumped the isolated when the sex ratio of one form is skewed (Moore et
population breeding on Adams Is in the subantarctic al. 1997, 2001). This view was supported by TWG in
as a subspecies of those breeding on the Antipodes their 2008 report (ACAP 2009).
Is; D. a. gibsoni and D. a. antipodensis, respectively. The fourth subclade involves the 2 forms of
Their decision was made in face of characteristic Buller’s albatross. No recognised authority presently
retention of darker juvenile plumage by adult grants them full species status (e.g. Brooke 2004,
female Antipodes birds, their distinct foraging Onley & Scofield 2007, ACAP 2009) despite the
patterns (Walker & Elliot 2006) and lack of gene large geographic separation of their breeding sites
flow between members of these 2 geographically (>900 km) and observations of Robertson & Nunn
distinct populations (Burg & Croxall 2004). These (1998) that they breed at different times. Our new
taxa have identical cytochrome b sequences (Nunn data confirm the close relationship between them
& Stanley 1998) and control region sequences, but with a minimal ‘p’ distance of only 0.09%, a value
do show some differentiation at microsatellite loci otherwise strongly indicative that they are con-
(Burg & Croxall 2004). The 3 remaining members of specific. The nomenclature suggested for this pair
Diomedea are provisionally recognised as species by is the Buller’s albatross (T. b. bulleri), breeding on
ACAP (2006) and other recent publications (Brooke Snares and Solander Is, and northern (or Pacific)
2004, Onley & Scofield 2007) and reviewed by TWG Buller’s albatross (T. b. platei), breeding 2 months
in 2008 (ACAP 2009). However, their cytochrome b earlier at Three Kings Is and at 2 sites in the Chatham
distances range from 0.5% between D. exulans and Is. Onley & Scofield (2007) suggest T. b. ssp nov is
D. amsterdamensis and 0.9% between D. dabbenea and preferable to T. b. platei, which would be more
each of the other two. Taken in isolation such values correct if Robertson & Nunn (1998) had provided
would not support there being more than one good published evidence to support their view regarding
species in this group. Burg & Croxall (2004) support the type specimen (see Methods). Microsatellite
the exulans/dabbenea split based on their distinct data and assignment tests (van Bekkum 2004)
mtDNA control region lineages. show that these breeding stocks are genetically
The genus Thalassarche is the most complex clade differentiated and probably no longer interbreed to
and consists of 5 subclades. First, the Indian and any significant extent. However, their separation is
Atlantic forms of the yellow-nosed albatross (T. carteri clearly recent and it has been argued that there are
and T. chlororhyncus, respectively) are recognised at few real grounds even for granting them subspecific
92 Chambers et al.

status (ACAP 2006). Our cytochrome b data support making such decisions. Recent technical advances
this latter position. by Bakström et al. (2008) and Kimball et al. (2009)
Finally, the shy albatross complex consists of 4 have opened up many new genetic targets for
taxa which deserve most careful attention, since they analysis in avian molecular systematics and have
are very closely related and were formerly classified already been applied across a wide range of bird
as members of a single species (e.g. Marchant & orders with considerable success (Hackett et al. 2008)
Higgins 1990). Our new data cover this group as albeit associated with obviously high experimental
a whole for the first time. The findings presented costs. However, in the case of the albatrosses, the
here justify the TWG approach of treating them as 2 rarity of nucleotide variation data reported for
pairs of 2 taxa: T. cauta + T. steadi (ACAP 2006) and chromosomal regions flanking microsatellite loci by
T. salvini + T. eremita (ACAP 2009), as shown in Fig. van Bekkum (2004), suggests that very large nuclear
1. The ‘p’ distances between pairs reflect this view targets will need to be examined to gain much in
being around 1% between the pairs but only 0.2% the way of resolving power.
between T. cauta and T. steadi and 0.3% between We conclude by noting that our study shows
T. salvini and T. eremita. Hence, they should not be comparatively short DNA sequences can give robust
grouped into a single species even under the strict gene trees displaying hypothetical evolutionary
Penhallurick & Wink (2004) decision framework. relationships between taxa. Such trees may be of
Further, ACAP (2006) presents extensive total assistance in recognising conservation priorities,
evidence arguments for the recognition of T. cauta but our experience also shows clearly the dangers
and T. steadi at specific level and similarly ACAP inherent in using just one type of information or
(2009) for T. salvini and T. eremita. overly simple and/or unsupported decision rules to
Taking all of the above together it is clear that determine what is or what is not a good species.
taxa separated by ‘p’ distances of 1% or greater are
widely recognised at species level. At the other end ACKNOWLEDGMENTS
of the scale there are few reliable sources that split Experimental work in the Chambers Laboratory at Victoria
pairs of taxa with ‘p’ distances equal to or less than university of Wellington was supported by various Science
0.1%. Those pairs of taxa with intermediate values Faculty Grants including a Summer Research Assistantship
must be viewed holistically. Here we see examples for Ryan Steel and an Awhina Scholarship for Cassidy
Moeke. The phylogenetic analyses were performed by
of pairs of taxa with ‘p’ distances as low as 0.2 or John Trueman with support from internal resources at the
0.3% for which there is other good evidence to show Australian National University, Canberra. The authors are
that they satisfy the Helbig et al. (2002) criteria for grateful to 2 anonymous reviewers whose comments helped
recognition as good species. Newton (2003) collated to improve the clarity and presentation of our manuscript.
reports that show many avian species are separated
by as much as 6 – 10% cytochrome b divergence.
LITERATURE CITED
However, there are several instances where this ACAP. 2004. Report on the First Meeting of the Parties
value may be as low as 0.1%. This author concludes (MoP1) held 10 – 12 November, 2004 in Hobart,
that “….. there can be no fixed degree of genetic Australia.
divergence that defines a speciation event.” ACAP. 2006. Report on the Second Meeting of the Parties
In summary, a mitochondrial gene tree has (MoP2) held 13 – 17 November, 2006 in Christchurch,
been prepared for 24 albatross taxa. It includes 22 New Zealand.
nominate species whose relationships have been ACAP. 2009. Report on the third Meeting of the Parties
recognised by other authorities with abundant (MoP3) held 27 April – 1 May November, 2006 in
Bergen, Norway.
support from other data sources. The list (after
Abbott, C.L.; Double, M.C. 2003a. Genetic structure,
ACAP 2009) is provisionally identical with Onley & conservation genetics, and evidence of speciation by
Scofield (2007) and differs only from Brooke (2004) range expansion in shy and white-capped albatrosses.
and OSNZ (2010) by splitting cauta and steadi. These Molecular Ecology 12: 2953-2962.
totals are subject to change as new data emerge. Abbott, C.L.; Double, M.C. 2003b. Phylogeography of
There are 2 remaining enigmatic taxa that we have shy and white-capped albatrosses inferred from
not had the opportunity to examine here. The first mitochondrial DNA sequences: implications for
is the unique Falklands Is lineage of black-browed population history and taxonomy. Molecular Ecology
albatross recorded by Burg & Croxall (2001). The 12: 2747-2758.
Alexander, W.; Falla, R.A.; Jouanin, C.; Murphy, R.C.;
second is the isolated Japanese population of P.
Salmonsen, F.; Voous, K.H.; Watson, G.E.; Bourne,
nigripes described by Edwards et al. (2001). At first W.R.P.; Kuroda, N.H.; Rowan, M.K.; Sevrenty, D.L.;
sight the available evidence would seem to support Tickell, W.L.N.; Warham, J.; Winterbottom, J.M. 1965.
species status for the former, but not the latter. Of The families and genera of petrels and their names.
course, it would be desirable to have more extensive Ibis 107: 401-405.
molecular data, especially from rapidly evolving Backström, N.; Fagerberg, S.; Ellegren, H. 2008. Genomics
nuclear loci, for these and indeed all taxa to aid in of natural bird populations: a gene-based set of
 Albatross molecular systematics 93

reference markers spread evenly across the avian Lindsey, T. 2008. Abatrosses. CSIRO Publishing:
genome. Molecular Ecology 17: 964-980. Melbourne.
Brooke, M. 2004. Albatrosses and petrels across the world. Marchant, S.; Higgins, P.J. 1990. Handbook of Australia, New
Oxford University Press: Oxford. Zealand and Antarctic birds. Oxford University Press:
Burg, T.M.; Croxall, J.P. 2001. Global relationships amongst Melbourne.
black-browed and grey-headed albatrosses: analysis Mayden, R.L. 1997. A hierarchy of species concepts: the
of population structure using mitochondrial DNA denoument in the saga of the species problem. pp.
and microsatellites. Molecular Ecology 10: 2647-2660. 381-424 in Species the units of biodiversity. Eds., M.F.
Burg, T.M.; Croxall, J.P. 2004. Global population structure Claridge; H.A. Dawah; M.R. Wilson. Chapman &
and taxonomy of the wandering albatross species Hall: London.
complex. Molecular Ecology 13: 2345-2355. Mayr E. 1996. What is a species and what is not? Philosophy
Christidis, L.; Boles, W.E. 1994. The Taxonomy and species of of Science 63: 262-277.
birds of Australia and its Territories. RAOU Monograph Medway, D. 2006. The identity of the chocolate albatrosss
No. 2, Royal Australasian Ornithologists Union: Diomedea spadaciea of Gmelin, 1789 and the wandering
Melbourne. albatross Diomedea exulans of Linnaeus, 1758. Notornis
Christidis, L.; Boles, W.E. 2007. Systematics and taxonomy of 40: 145-162.
Australian birds. CSIRO Publishing: Collingwood. Mills, M.S.L.; Ryan P.G. 2005. Modelling impacts of
Claridge, M.F.; Dawah, H.A.; Wilson M.R. 1997. Species the long-line fishing: what are the effects of pair bond
units of biodiversity. Chapman and Hall: London. disruption and sex-biased mortality on albatross
Cracraft, J. 1983. Species concepts and speciation analysis. fecundity? Animal Conservation 8: 359-367.
Current Ornithology 1: 159-187. Moore, P.J.; Taylor, G.A.; Amey, J.M. 1997. Interbreeding of
Croxall, J.P.; Gales, R. 1998. An assessment of the black-browed albatrosss Diomedea m. melanophris and
conservation status of albatrosses. In: Albatross Biology New Zealand black-browed albatross T. m. impavida
and Conservation. (Eds., G. Robertson; R. Gales.) pp. 46- on Campbell Island. Emu 97: 322-324.
65. Surrey Beatty & Sons: Chipping Norton. Moore, P.J.; Burg, T.M.; Taylor, G.A.; Millar, C.D. 2001.
Edwards, J.L. 2005. Consortium for the barcode of life: An Provenance and sex ratio of black-browed albatross,
introduction. Genes and Genetic Systematics 80: 443. Thalassarche melanophris, breeding on Campbell Island,
Edwards, S.V.; Silva, M.C.; Burg, T.; Friesen, V.; Warheit, New Zealand. Emu 101: 329-334.
K.I. 2001. Molecular genetics markers in the analysis of Moritz, C.; Cicero, C. 2004. DNA barcoding: Promise
seabird bycatch populations. pp. 115-140 in Proceedings and pitfalls. PloS Biol 2, e312 doi:10.1371/journal.
– seabird bycatch: trends, roadblocks and solutions. pbio.0020312
University of Alaska: Sea Grant AK-SG-01-01. Newton, I. 2003. Speciation and biogeography of birds.
Hackett, S.J.; Kimball, R.T.; Reddy, S.; Bowie, R.C.K.; Braun, Academic Press: Amsterdam.
E.L.; Braun, M.J.; Chojnovski, J.L.; Cox, W.A.; Han, Nunn, G.B.; Cooper, J.; Jouventin, P.; Robertson, C.J.R.;
K.-L.; Harshman, J.; Huddleston, C.J.; Marks, B.D.; Robertson, G.G. 1996. Evolutionary relationships
Miglia, K.J.; Moore, W.S.; Sheldon, F.H.; Steadman, among extant albatrosses (Procellariiformes:
D.W.; Witt, C.C.; Yuri, T. 2008. A phylogenetic study Diomedeidae) established from complete cytochrome-b
of birds reveals their evolutionary history. Science 320: gene sequences. Auk 113: 784-801.
1763-1768. Nunn, G.B.; Stanley, S.E. 1998. Body size effects and rates
Hebert, P.D.N.; Stoeckle, M.Y.; Zemlak, T.S.; Francis, C.M. of cytochrome b evolution in tube-nosed seabirds.
2004. Identification of birds through DNA barcodes. Molecular Biology and Evolution 15: 1360-1371.
PLoS Biology 2: 1657-1663. Onley D.; Scofield P. 2007. Albatrossses, petrels and
Helbig, A.J.; Knox, A.K.; Parkin, D.T.; Sangster, G.; shearwaters of the world. Christopher Helm: London.
Collinson, M. 2002. Guidelines for assigning species OSNZ Checklist Committee – E.G. Turbott, Convener.
rank. Ibis 144: 518-525. 1990. Checklist of the birds of New Zealand and the Ross
Hickerson, M.J.; Meyer, C.P.; Moritz, C. 2006. DNA Dependency, Antarctica, 3rd Ed. Ornithological Society
barcoding will often fail to discover new animal of New Zealand and Random Century: Auckland.
species over broad parameter space. Systematic Biology OSNZ Checklist Committee – B.J. Gill, Convener. 2010.
55: 729-739. Checklist of the birds of New Zealand, Norfolk and
Hulsenbeck, J.P.; Ronquist, F.; Posada, D.; Crandall, K.A. Macquarie Islands, and the Ross Dependency, Antarctica,
2001. MRBAYES: Bayesian inference of phylogeny. 4th Ed. Te Papa Press: Wellington– in press.
Bioinformatics 17: 754-755. Penhallurick, J.; Wink, M. 2004. Analysis of the taxonomy
Jouanin, C.; Mougin, J.-L. 1979. Order Procellariformes pp. and nomenclature of the Procellariiformes based on
48-121 in Checklist of the birds of the world. Vol I, 2nd ed. complete nucleotide sequences of the mitochondrial
Eds., E. Mayr; G.W. Cotrell. Museum of Comparative cytochrome b gene. Emu 104: 125-147.
Zoology: Cambridge, MA. Phillimore A.B.; Owens I.P.F. 2006. Are subspecies
Kimball, R.T.; Braun, E.L.; Barker, F.K.; Bowie, R.C.K.; useful in evolutionary and conservation biology?
Braun, M.J.; Chojnovski, J.L.; Hackett, S.J.; Han, K.-L.; Proceedings of the Royal Society of London Series B 273:
Harshman, J.; Heimer-Torres, V.; Holznagel, W.; 1049-1053.
Huddleston, C.J.; Marks, B.D.; Miglia, K.J.; Moore, Posada, D.; Crandall, K.A. 1998. MODELTEST: testing
W.S.; Reddy, S.; Sheldon, F.H.; Smith, J.V.; Witt, C.C.; the model of DNA substitution. Bioinformatics 14: 817-
Yuri, T. 2008. A well-tested set of primers to amplify 818.
regions spread across the avian genome. Molecular Randler, C. 2006. Behavioural and ecological correlates of
Phylogenetics and Evolution 50: 654-660. natural hybridisation in birds. Ibis 148: 459-467.
94 Chambers et al.

Rheindt, F.E.; Austin, J.J. 2005. Major analytical and Stoeckle, M. 2003. Taxonomy, DNA and the bar code of
conceptual shortcomings in a recent taxonomic revision life Bioscience 53: 796-797.
of the Procellariiformes - a reply to Penhallurick and Swofford. 2002. PAUP* v4.0b10 www.paup.csit.fsu.edu
Wink (2004). Emu 105: 181-186. Tickell, W.L.N. 2000. Albatrosses. Pica Press: Robertsbridge.
Robertson, C.J. 2002. The scientific name of the Indian Walker, K.; Elliot, G. 2006. At-sea distribution of Gibson’s and
yellow-nosed albatrosss Diomedia carteri. Marine Antipodean wandering albatrosses and relationships
Ornithology 28: 48-49. with longline fisheries. Notornis 53: 265-290.
Robertson, C.J.; Nunn, G.B. 1998. Towards a new taxonomy Van Bekkum, M. 2004. Microsatellite DNA analysis of
for albatrosses. In: Albatross Biology and Conservation. breeding behaviour and evolution in Buller’s albatross
Eds., G. Robertson; R. Gales. pp. 13-19. Surrey Beatty (Thalassarche bulleri). MSc Thesis, Victoria University:
and Sons: Chipping Norton. Wellington.
Schodde, R.; Gill, B.J.; Scofield, P.; Steinheimer, F.D.; Zink, R.M. 2004. The role of subspecies in obscuring avian
Groth J.G.; Lai J. 2009. Settling application of the name biological diversity and misleading conservation
Diomedea exulans Linnaeus, 1758, for the wandering policy. Proceedings of the Royal Society of London Series
albatross by neotypification. Unpublished manuscript. B 271: 561-564.
Sibley, C.G.; Monroe, B., Jr. 1990. Distribution and taxonomy Zwickl, D. 2006. GARLI v0.951. Computer program
of the birds of the world. Yale University Press: New distributed by the author. www.bio.utexas.edu/
Haven. faculty/antisense/garli/Garli.html.

View publication stats