Recognize Myrmeciza
palliata as a separate species from
Myrmeciza laemosticta
Proposal
(475) to South American Classification Committee
Three species of “Myrmeciza” antbirds are currently recognized in the laemosticta complex, namely M.
laemosticta, M. nigricauda, and M. berlepschi. These species have had an
unstable taxonomy, with sexes of single species treated in separate taxa, or
members classified in different and often distantly related genera (see Robbins
& Ridgely 1991, Zimmer & Isler 2003). Currently, only M. laemosticta is
polytypic, with two subspecies: the nominate subspecies from lower Central
America (NW Costa Rica to Darién in E Panama) and palliata from the Colombian and Venezuelan foothills of
inter-Andean valleys (Wetmore 1972, Zimmer & Isler 2003). The taxon palliata includes synonymized forms bolivari and venezuelae (Robbins & Ridgely 1991) and has never been
considered a separate species.
Observations
and recordings by B. Whitney and A. Cuervo (published in the Colombian Andes
Sound Guide; Alvarez et al. 2007) of the then-unknown loudsong of M. l. palliata from the Magdalena valley
were indicative of their distinctiveness and degree of differentiation with
respect to the loudsongs of M. l.
laemosticta. Chaves et al. (2010) recently evaluated species limits in the laemosticta complex and investigated
whether M. l. palliata merited
elevation to species rank. Specifically, they conducted a quantitative analysis
of vocal variation in male songs and a qualitative assessment of variation in
calls and female songs and calls, coupled with a phylogenetic hypothesis for
the group based on sequences of a mtDNA gene.
Fourteen
vocal traits in a total of 42 individuals of the complex plus M. griseiceps were analyzed
statistically. M. l. palliata is
indeed vocally diagnosable from any other species, and different from M. l. laemosticta in three vocal
characters. A multivariate analysis of songs showed that each taxon differs
significantly from the others: M. l.
palliata is vocally divergent from M.
l. laemosticta and closer to, but distinct from, M. nigricauda. Vocal results were consistent with the ND2 gene
tree, which showed a highly supported clade for the laemosticta complex with the four taxa forming a polytomy (i.e. no
support for a sister relationship between laemosticta
and palliata), and branches
leading to each taxon were relatively long, thus indicating a comparable time
of isolation. Thus, pairwise genetic distances were nearly the same among the
four taxa.
Chaves
et al. concluded that M. l. laemosticta and
M. l palliata should be treated as
distinct species under criteria of diagnosability and (inferred) potential
reproductive isolation. Vocal variation indicates that M. l. palliata merits elevation to species rank following the
standards for allopatric populations in antbirds (Isler et al. 1998), and
genetic variation indicated an independent history of evolutionary isolation
roughly congruent with the divergence among currently recognized species in the
group. I encourage all to read the paper for further details on methods and
results and to examine vocalizations in xeno-canto.
The
English name Magdalena Antbird is recommended for M. palliata (Chaves et al. 2010).
This name was previously used by Cory and Hellmayr (1924) for Formicivora grisea hondae. Meyer de
Schauensee (1950) didn’t follow Cory and Hellmayr’s name for F. g. hondae and used “Honda Ant-Bird”
instead. In both publications, Pale Antcatcher was used for M. laemosticta palliata. The English
names proposed for these subspecies have no tradition of usage (i.e. only used
in these publications) and taxonomic nomenclature rules do not apply for
English names.
Recommendation: I recommend a YES vote to recognize “Myrmeciza” palliata as a separate
species from “Myrmeciza” laemosticta
and to adopt the English name Magdalena Antbird for M. palliata.
Literature cited
Álvarez, M., V. Caro, O. Laverde & A. M. Cuervo.
2007. Guía Sonora de las Aves de los Andes Colombianos. Instituto Alexander von
Humboldt & Cornell Laboratory of Ornithology.
Chaves, J. C., A. M. Cuervo, M. J. Miller, & C.
D. Cadena. 2010. Revising species limits in a group of Myrmeciza antbirds reveals a cryptic species within M. laemosticta (Thamnophilidae). The
Condor 112: 718-730. http://www.museum.lsu.edu/cuervo/pubs_files/Chaves_etal.Condor.2010.pdf
Meyer de Schauensee, R. 1950. The birds of the
Republic of Colombia. Caldasia 5 (24): 645-871.
Other references in the SACC
reference list or in the Chaves et al. paper.
Andrés M. Cuervo, December 2010
Comments from Zimmer: “YES. As far as I can tell from listening to vocal samples of palliata (which I don’t know in life) versus those of laemosticta, which I have recorded extensively in Costa Rica and Panama, the two taxa are diagnosably different (vocally) to an extent consistent with species-level recognition in Thamnophilidae. The vocal and genetic analyses of Chaves et al. (2010) would seem to support that conclusion. “Magdalena Antbird” seems appropriate as an English name for palliata.”
Comments from Thomas
Donegan:
“Summary: Chaves et al. (2010) demonstrate clearly that palliata should not be treated as part
of the same species as the Dull-mantled Antbird M. laemosticta.
However, they do not deal adequately with the elephant in the room. Esmeraldas Antbird M. nigricauda (of the West Andes foothills) and palliata (Central
and East Andes foothills) are vocally almost identical, have similar habitat
requirements and elevational ranges on adjacent mountain ranges and were
historically treated as conspecifics.
Andrés Cuervo’s proposal does not discuss
whether palliata is a good species
with respect to nigricauda nor does
it note that the vocal differences fall below that typically used as a
benchmark for species rank in antbirds.
Chaves et al. (2010) briefly discuss the point, but their discussion is
inconsistent with the data presented and unconvincing.
“In January 2003, I made what I understand to be the first known
Colombian recordings of palliata in
Cerro de la Paz, Santander, Colombia (http://www.xeno-canto.org/recording.php?XC=24335)
and then compared sonograms of this with those of M. nigricauda and M.
laemosticta. At one point, I
prepared a short note with a view to publication, but this was not developed
further when I heard that Chaves, Cuervo and Cadena were conducting their own
more detailed study, a few years ago.
Chaves et al. (2010) have spent a lot of time and effort investigating
the situation further and speak to their own history of study in this group. They should be congratulated for
dealing with the laemosticta / palliata issue in a diligent way. The molecular data is also very
interesting. However, the authors
are not persuasive in concluding that palliata
is a separate species from nigricauda
based on vocal differences.
The reasons for this are set out below.
“Vocal issues
“Isler et al. in their various papers have generally adopted a
“three diagnosable differences” test for loudsongs in assessing species rank
for antbirds. Chaves et al. (2010)
do well to consider whether this test is appropriate for this particular group
of antbirds. It has recently been
shown that sympatric species of suboscine groups may show lower levels of
diagnosable differences in loudsong than the traditional benchmark of three
differences. Chaves et al. (2010)
call for a “2 diagnosable differences” test to be applied for these Myrmeciza based on the differences
between nigricauda and berlepschi. Some preliminary results with other Myrmeciza I am currently studying vocally would also support this
more liberal approach, as do studies of sympatric members of the
Warbling-Antbird group. However,
despite what is said in Chaves et al. (2010), it is a big stretch to conclude
that palliata and nigricauda meet even this “two
differences” test.
“The authors do not appear to have actually elucidated two differences
in loudsong between palliata and nigricauda. Their Table 1 (p.720) asserts the two differences to be in
“Note shape” and “Note structure”.
Those ‘two’ variables appear on their face to be the same thing. The authors list all vocal variables that
were studied in Appendix 2 (p. 729).
The variable “Note structure” is described but “Note shape” is not
mentioned. In Appendix 3 (p. 730)
the authors go on to describe the nature of the differences shown, and cite
differences in both “Note shape” (the same term not defined elsewhere) and
“Change in note structure” (a variable not mentioned in Table 1 for this
species pair, but which is defined in Appendix 2). The discussion of the differences between nigricauda and palliata in Appendix 3 refers to supposedly diagnosable differences
referable to “Change in note structure” (although more on that below) but does
not mention overall diagnosable differences in note structure.
“As can be seen from their Figure 2 (p. 721) and other recordings
on xeno-canto, both palliata and nigricauda give
inverted chevron shaped notes and also more “rounded” inverted-chevron-shaped
notes. Some of these note shapes
are given in a different order within songs in the sonograms presented in
Figure 2 and some of them are skewed differently. However, note shape should only be treated as a diagnosable
difference if none of the notes of the two populations are similar. It is clear that both nigricauda and palliata are capable of giving notes of the same shape: they just
do so at different points in time in the sonograms presented. The only difference that the authors
describe is the order in which the notes of particular structure are delivered
– i.e. a difference in “change in note structure”, or, more accurately
“The note structure of certain notes at a particular point in time or segment
within the song”, not actually “Note structure”. This single vocal
difference should not be double-counted.
“Even then, Chaves et al. (2010)’s description of changes in note
structure does not concord with available recordings. The sonograms of these taxa on xeno-canto
show considerable variation in note shape of both M. nigricauda …
http://www.xeno-canto.org/browse.php?query=myrmeciza
nigricauda&pagenumber=&order=taxonomy&view=3
… and “M. palliata”
http://www.xeno-canto.org/browse.php?query=laemosticta
cnt:colombia&pagenumber=&order=taxonomy&view=3
“Appendix 3 accurately discusses the subjective differences between
the sonograms shown in Figure 2, but it does not take account of individual
variation in these birds. The
authors describe palliata songs as
constituting three segments.
Supposedly, in the first segment, “notes of M. nigricauda are rounded, those of M. l. palliata are slurred up–down”. However, XC3870 of M. nigricauda is as up-down in the first segment as various palliata recordings. In the third segment, the authors claim
that “both taxa emit rounded notes (partially modulated in M. nigricauda).
However, several recordings of palliata
do not show any such “third segment”: examples are XC10725, XC16311 and
XC18154. One of these recordings
was cited in the paper as being part of the sample studied. The authors also conclude that in
recordings of nigricauda, note shapes “change from rounded to partially
modulated to rounded again”.
However, sonograms of nigricauda on
xeno-canto (none of which the authors studied)
include more variety in note shape, with examples where the third segment
involves a sharp up-down stroke (XC18884), more rounded inverted chevron
(XC30289), or down stroke with a small rising initial tail (XC58961). The methodology (p. 720) asserts that
the authors “examined qualitative characters through a blind inspection and
grouping of printed sonograms followed by an assessment of whether the
groupings matched the populations under study”. But their definition of the
differences in “change in note structure” between these taxa do not facilitate
the blind allocation of available recordings to particular species. More detailed and convincing analysis
on this issue is warranted. Given
that there is a larger sample available on xeno-canto,
it would be better to include these additional recordings in analyses.
“The authors also mis-describe the
results of their multivariate analysis as regards nigricauda and palliata. They state as follows: “Discriminant
function analysis of songs of the four taxa in the M. laemosticta complex revealed that each taxon’s
group centroid differs significantly from the others (Wilks’
L = 0.003, df = 36, P < 0.0001). The analysis classified correctly 100% of
individuals to their respective population designation (Table 3; Fig. 4),
demonstrating that M. l. laemosticta,
M. l. palliata, M. nigricauda, and M. berlepschi are all vocally diagnosable
with respect to each other in multivariate space.” (p. 721). This statement may be correct as
regards diagnosis as between most of the species pairs studied. However, as regards nigricauda and palliata it is highly misleading. The two taxa do not appear to be diagnosable based on
quantitative measures. The authors
cite Table 3 (p. 724) and Figure 4 (p. 724) for the proposition that “100% of
individuals” (including of palliata vs.
nigricauda) could be classified correctly. However, in Figure 3 (p. 723), nigricauda and palliata
overlap for all vocal variables.
Table 3 (p. 724) shows the weight given to the variables in the
analysis; and in Figure 4 (p. 724), palliata
and nigricauda cluster together and
would not appear diagnosable in the (x=10-11, y=-12 to -10) part of the
range. The authors cite a Wilks lamda test, but that compares the statistical
significance of differences between centroids, not dealing with
diagnosability. To consider
diagnosability using multivariate analysis, one should demonstrate that
recorded values for the two groups fall either side of a curve or line, depict
95% or higher ellipsoids on the diagram which should not overlap or carry out
other statistical tests. Figure 4
shows palliata and nigricauda clustering together, so it is
doubtful that those sorts of analyses would show diagnosability.
“Assuming that the authors were inadvertent in asserting that palliata and nigricauda are fully diagnosable in multivariate space, and given
that there are no diagnosable differences in the quantitative variables, we are
considering here only subjective differences in note shape. When comparing those to differences
between sympatric taxa, it should not be only the existence of a difference that is relevant, but the degree and nature of the
difference. The authors consider
the vocal differences between palliata and
nigricauda to be similar to those
between berlepschi and nigricauda, which are sympatric.
However, recordings of berlepschi show
more fundamental differences in “note structure”. M. berlepschi recordings
involve down strokes, with only a tiny rising “tail” in some instances and no
significant “change in note structure”.
See Figure 2 (p. 721) and also other recordings available on xeno-canto:
http://www.xeno-canto.org/browse.php?query=myrmeciza
berlepschi&pagenumber=&order=taxonomy&view=3
“In contrast, both palliata and
nigricauda songs consist of a series
of up-down strokes or “chevrons” of varying shape. As discussed above, perhaps only the “second segment” shows
diagnosable differences in note shape.
The degree of those differences are not very impressive compared with
those between nigricauda and berlepschi and are not accurately
described by Chaves et al. (2010).
“Finally, the authors refer to differences in call. Although this is not analysed in detail, various examples of sonograms of calls
are presented in Figure 5 (p. 724).
The authors’ selection of published sonograms is quite surprising. Similarly structured calls do not appear
to have been presented. Chaves et
al. (2010) show the palliata call
being a flat rasp, with that of nigricauda
being a very different up-down stroke. However, another of my recordings, XC31829 (and also XC10724
by Nick Athanas) are of calls of palliata and they are up-down strokes, with much more similar note
shape to the sonogram of the call presented by Chaves et al. (2010) of nigricauda. The calls of nigricauda
and palliata sound rather
similar, with the call of nigricauda perhaps
being a little raspier.
“A conservative approach under a comparative BSC approach of the
nature adopted in Remsen (2005), Helbig et al. (2002) and similar papers would
therefore be to lump palliata with nigricauda on the basis of the vocal
data presented.
“Sampling gaps
“The authors have overlooked to study some very interesting and
important specimens. It has never
been suggested that laemosticta, palliata or nigricauda exist in sympatry and Chaves et al. (2010) do not
suggest this either. However, the
three forms occur in very close proximity in the central to northern section of
the West Andes and surrounding region.
Some of them may indeed be sympatric or parapatric. The authors do not discuss whether or
not the taxa intergrade in this region.
“The closest materials studied by Chaves et al. (2010) for laemosticta are from Panama, palliata from
the northern Central Andes and nigricauda
from the southern-central part of the West Andes. M. nigricauda is
generally not considered to extend further north in range than the locality
sampled by the authors in major texts.
However, the literature may not be correct in this respect. Project Biomap (www.biomap.net) data include a specimen of nigricauda from the Baudó mountains in
Chocó department (ANSP 147226) and another from Remedios, Antioquia in the
northern foothills of the Central
Andes (Colegio San José, Medellín, 0021B). I have not seen these specimens or any photograph of them,
but per information on Biomap, the Remedios skin’s identification was confirmed
by Tomás Cuadros who noted
that it represented a significant range extension.
“Based on these specimens, M.
nigricauda may range further north and east than is generally thought and
could be sympatric or parapatric with M.
(x.) palliata. Or it may not
if these are database or identification errors. The Baudó specimen, whatever it
is, falls in an important sampling gap.
There are now several lower elevation localities in the northern West
Andes foothills and Baudó mountains that are safe to
study for sound recording or other field studies. There is a specimen of palliata
which one of the authors (Andrés Cuervo), Paul Salaman and myself collected
in 1999 from the northern end of the Central Andes below Anorí
(close to Remedios) which is close to these other localities, and this one was
sampled by the authors. Chaves et
al. (2010) should have studied the other specimens mentioned above before
concluding that palliata and nigricauda are two species. If palliata
and nigricauda are sympatric and
do not intergrade then they must be split, but
the specimens mentioned above must first be studied to confirm whether or not
this is the case.
“Plumage
“Plumage differences are notable in the context of Thamnophilidae,
with differences in the mantle colour and throat
pattern of males and wing bars of females. However, these are not so great such that the two forms were
historically treated as conspecifics, including as recently as by Hilty &
Brown (1986). Certainly, palliata is closer in plumage to laemosticta than it is to nigricauda. It would be nice to see some data on biometrics before
concluding that phenotype supports the split of palliata from nigricauda and
the authors do not justify this split with regard to plumages.
“Molecular data
“One could also treat palliata
as a species because nigricauda would
otherwise be paraphyletic based on the molecular data and applying phylogenetic
species concepts. SACC has
accepted some splits in the past where there is paraphyly but little vocal
support (e.g. Ecuadorian Thrush, although the disjunct distributions were also
relevant there). Genetic
differences (7%) are pretty impressive and more or less equal as between laemosticta, nigricauda and palliata. On the other hand, the laemosticta
to palliata node is not very
strongly supported (0.73 / 61) and there are “good” species that show greater
intraspecific mtDNA variation than these.
“Conclusion
“Chaves et al. (2010)’s vocal data strongly support moving palliata from laemosticta to nigricauda
but do not support splitting palliata from
nigricauda. At least, it would be
sensible to take the following steps before making this split: (i) studies of voice, the specimens referred to above, and
molecular samples from the “gap” in the northern West Andes and Baudó; (ii) a
more convincing study and accurate description of the differences (if any) in
what I would term “The note structure of certain notes at a particular point in
time or segment within the song”; (iii) a study of differences in the calls
involving comparison of more appropriate examples, a larger sample and
statistical analysis; and (iv) more information on the strength of molecular
support for palliata plus nigricauda not being monophyletic. Splitting palliata from nigricauda may
be warranted (particularly if they are sympatric) but it may not be. I just don’t know. In the absence of the information
mentioned above, I would suggest a prudent course of action would be to reject
this proposal for the time being and treat palliata
as a subspecies of nigricauda. M.
nigricauda was described before palliata
(1892 vs. 1917) so the former name has priority, resulting in no change
being necessary to the SACC list.
“These criticisms are presented with a heavy heart – not
least because the last thing that Colombian ornithology needs at the moment is
another argument. However, these
are genuinely held concerns about the methods, discussion and new taxonomy in
Chaves et al. (2010)’s paper. The
SACC ought to be made aware of these concerns. I would though warmly welcome any observations that the
authors or others may have in response to these comments.”
Comments from Bret Whitney: “I was a formal reviewer of the Chaves et al paper, and accepted it with minor revision, which objected mainly to their too-strict interpretation of the “three-character yardstick” as it was presented in Isler et al. (1998), and it was toned-down for publication. I think Thomas Donegan has also adopted an overly rigid interpretation of the “three-character yardstick”: we clearly stated that, for some groups, especially those for which other data sets provided support for significant divergences, less than three characters might be appropriate. I greatly appreciate Donegan’s criticisms, based in part on appropriate data not used by Chaves et al.; they are generally well-formulated and reasonable, and I find the bit about the geographic proximity of the three forms to be a sticking point, worth a focused trip to record and collect some specimens in an apparent, or at least potential, contact zone. That said, the Islers have formulated an objective and convincing test for the diagnosability of male loudsongs (albeit from a somewhat different sample than that employed by Chaves et al.), and on that basis I am willing to accept that, until further field collecting can be conducted, and a larger sample of vocalizations analyzed, the best course forward is the split of palliata and nigricauda. At present, I consider that decision an advancement in our understanding of the taxonomy of this poorly known complex, but it is always subject to revision should further collecting and analysis point to a different arrangement. To Donegan’s “heavy heart”, I suggest that ornithologists knowingly sharing a focused interest in the same groups of birds, and especially those gathering complementary data, make an effort to collaborate to produce the best publication, or set of publications possible. Then it’s left to the rest of us, who know a lot less about most of these individual situations, to try to make the best of the best.”
Additional
comments from Andrés Cuervo, Daniel Cadena, Juan Camilo
Chaves and Matt Miller: “We welcome the comments offered by Thomas Donegan on our
Myrmeciza laemosticta paper in which
he questioned our recommendation of elevating M. l. palliata to species rank (a hypothesis people started to
think about ever since Bret Whitney obtained the first recordings of songs of palliata in Colombia in 1992) by
pointing out that this taxon could be rather treated as a subspecies of M. nigricauda based on song similarity.
“To reach the conclusion that the
divergent taxa nigricauda and palliata are nothing else but
populations of the same biological species, one would have to ignore complementary evidence in current
existence and propel vocal similarity as the sole basis for that
interpretation. As Donegan points out, multiple lines of evidence should be
considered including better sampling in areas of potential geographic
contact and examining more vocal data than what we had at hand by mid 2008. We
believe this would be very worthwhile. Potential intergradation, which would
eventually support a lump between nigricauda
and palliata, when based on
inexistent data is mere speculation at this point. Thus, we
cannot comment on this more other than to say that complementary studies and
new specimens and recordings would be desirable. For the purpose of evaluating
this proposal we are left with the available vocal and molecular data. We
assume that the important baseline taxonomic and distributional work by Robbins
and Ridgely (1991) is being taken into consideration as well.
“Regarding Donegan’s specific
criticisms, we accept up front that the qualitative differentiation between the
samples of palliata and nigricauda studied by Chaves et al.
could be subtler than we surmised and that we could have done a better job at
characterizing it verbally. However, an independent assessment by Mort and
Phyllis Isler led to the conclusion that songs of these taxa can indeed be
diagnosed qualitatively. The Islers went a step further to suggest that female
songs are also likely diagnostic. We did not emphasize the latter finding in
the paper much owing to what we thought was a reduced sample size.
“The multivariate statistical
analysis (and the genetic analysis, for that matter) should be taken as a
complement to the “yardstick” approach based on individual diagnostic features.
This analysis specifically seeks to derive composite variables that summarize
vocal variation and tests whether predefined groups can be discriminated by
variation in the data. These results are summarized in the paper’s Fig. 4 and
do reflect that quantitative vocal differentiation exists between palliata and nigricauda, although there is marginal overlap. However, the
decision of whether taxa are vocally diagnosable should not be based solely on
a cursory examination of this graph (which does not show other dimensions of
the discriminant function). More telling is the fact that nearly all recordings
were correctly classified by our discriminant function to their respective
taxon. Note we say nearly all because following Donegan’s criticism we have
reanalyzed our data and have to admit we found a mistake: the truth is that all
nigricauda recordings but one were
assigned to nigricauda (the other was
assigned to palliata), and all palliata recordings but one were assigned to palliata (the other was assigned to nigricauda). We thank Donegan for
leading us to realize a mistake crept in to our paper and we stand corrected.
Regardless, the results of the DFA suggesting differentiation are complementary
to the gold-standard methodology to establish species limits in antbirds (M.
Isler, P. Isler, B. Whitney), which is based on diagnosability in individual
traits (the issue discussed in the paragraph above). In any event, we expect
that as more recordings become available and more vocal types are compared, we
will get a more robust idea of the magnitude of vocal differentiation and its
biological implications and we do encourage further work on this system.
“Donegan states that we should have
studied a number of museum study skins listed in the Project Biomap database before concluding that palliata merited species rank, but we note that: (1)
most of the relevant specimens (e.g. from the Baudó mountains) were indeed
examined in a careful review of the complex by Robbins and Ridgely (1991)
– except one male from Mutatá, NW Antioquia (J.
Haffer specimen at ICN) that is inseparable from palliata specimens, (2) the Biomap’s record from Remedios, NE
Dept. Antioquia, turned out to be a misidentified specimen (we have determined
it is not even a Myrmeciza of this
group but a male Cercomacra tyrannina),
and (3) independent of the inclusion or not of an analysis using specimens in
our study, the results from the vocal and genetic analysis are verifiable and
we think that the conclusions are sound.
“In sum, we suggest that even if
every point raised by Donegan was correct, then the information available taken
as a whole indicates that palliata
and nigricauda are differentiated
lineages that have been evolving independently to the extent that the inferred
potential reproductive isolation is likely. Direct tests of such a
hypothesis obviously require additional fieldwork, but the current data are
consistent with reproductive isolation of the two taxa, not the reverse. A separate
proposal (or better, a new, more comprehensive study) based on current or new
information for merging palliata with
nigricauda could be presented
elsewhere or to the consideration of this committee following formal analyses.
Consequently, we think that the most conservative hypothesis of species limits
in this group at this point is to maintain palliata
as a separate species from any other taxa of this complex.”
Additional comments
solicited from Mort and Phyllis Isler: “In response to Thomas Donegan’s thoughtful comments and the
Acting Chair’s request to consider his comments, we undertook an independent
review of the vocal characteristics of Myrmeciza
laemosticta palliata (henceforth palliata)
and M. nigricauda (henceforth nigricauda).
“Procedure: Mort examined spectrograms and
extracted a typical loudsong or multiple loudsongs if there was obvious
variation, including what appeared to be male and female songs, from every
available recording of palliata (n
=10) and every available recording of nigricauda
from Colombia and a random sample of nigricauda
recordings from Ecuador (n = 11, locations available on request). Recordings were allocated to taxa on
the basis of recordists’ identifications and geography. 39 loudsongs in all were
extracted. Each was printed on a
page with no identifying data that were hand written on the backside of the
page.
“The
unidentified loudsongs were given to Phyllis who had not yet read the Chaves et
al. 2010 paper and had no idea of the results. She was asked to sort them into groups on the basis of
qualitative characteristics and to identify the distinctive characters that she
used.
“Phyllis
sorted the 39 loudsongs into two primary groups and five subsets. After she finished, we turned the pages
over to check the taxon. The two primary groups corresponded to palliata and nigricauda, i.e., every loudsong was sorted correctly to taxon (100
percent). The loudsongs of nigricauda were further sorted into two
subsets, the smaller of which (n = 3) was later identified as a female
loudsong, identified as such in one recording. The loudsongs of palliata
were sorted into three subsets.
Again a small (n = 5) subset was identified as the female loudsong. The other two subsets of palliata differed only in the presence
or absence of a high-pitched terminal note.
“Phyllis
stated that she used two characters to distinguish the two primary groups: (1)
differences in note shape and (2) differences in the sequence of change of note
shape. With regard to the latter,
she pointed out that in one group the note change was gradual and subtle
whereas in the other group the note shape change was abrupt and
substantial. The sequential
placement of different types of notes is independent of their presence or
absence, and in our 1988 and sequential papers, we have treated them as
such. Although Phyllis’
description of change in note shape differed somewhat from that of the paper,
the independent “blind” test supported the finding of the paper of a difference
in two distinct qualitative characters between palliata and nigricauda.
“Moreover,
based on our brief review, it appears that other vocal distinctions between palliata and nigricauda will be found in future studies. Principally, despite the small samples,
it seems certain that female loudsongs differ diagnostically from male
loudsongs and that female loudsongs differ between the two populations at least
as much as males, and probably more.
The authors describe differences in female loudsongs, but conclude that sample
sizes are too small to include them as characters in the diagnosis. We understand this conservative
position, but we would have given them more weight in the analysis as a second
type of vocalization because of the apparent consistency of the female
loudsongs that we see in our (slightly larger) samples. Unfortunately, female loudsongs
illustrated in the paper (Fig. 6 which does not identify sources) do not seem
to typify (represent central tendencies) female loudsongs in our inventory that
appear to differ somewhat more dramatically between the two taxa. Because female loudsongs were not
included in the analysis, in our opinion the statement in the paper that
evidence is provided that two characters may be sufficient to support species
status is not supported by this study.
(By the way, we treat the three diagnostic vocal character yardstick as
a “point of reference, not a requirement” [Isler et al. 1998], and in principle
we would not object to a strong case for two character finding.)
“Finally,
beyond note shape characteristics, we observed visually that pallida and nigricauda loudsongs appear to differ in
note frequency, in particular the frequency of the highest point in the note
(peak frequency). Consequently, we
measured peak frequencies of the initial note and the middle (central in time)
note of samples of palliata (n = 10)
and nigricauda (n = 11). Peaks of palliata notes started lower and increased, whereas peaks of nigricauda notes started higher and
typically decreased in the central notes of the vocalization. The peak frequency of initial palliata notes was 4157–4958
Hz (mean 4157 ± 235) and of nigricauda 5041–5809 Hz (mean 5372 ± 271), although there is a
recording in our collection identified as Myrmotherula
schisticolor by the recordist but which appears to be nigricauda in which the peak of the first note is 4507 Hz. We also
computed the difference in peaks between the initial and middle notes. Peaks of palliata notes increased from 184–635 Hz (mean +442 ± 140), whereas those of nigricauda decreased from 33–317
Hz (mean -202 ± 116) excluding an
example in which the first note was lower pitched than the second note and
therefore slightly lower pitched (33 Hz) than the middle note. We do not present these preliminary
data as a formal diagnosis, but they provide support to the conclusion that
male loudsongs of the two populations are distinct.
“Thomas
Donegan also raises issues regarding the multivariate analysis, differences in
calls, geographic distribution, plumage distinctions, and the molecular
analysis presented in the paper.
The authors have responded to most of these comments, and all agree that
additional studies of the complex would be useful.
“In
summary, we recommend that the committee accept the authors’ recommendation
that Myrmeciza palliata be considered
specifically distinct on the basis of vocal differences between it and the
other three species in the complex.
We conclude that two distinct qualitative characters distinguish male
loudsongs and may be supported by differences in quantitative frequency measurements.
Furthermore, differences in female loudsongs, unevaluated in the paper, appear
certain to provide additional evidence that palliata
and nigricauda have evolved to
species status under the Biological Species Concept.”
Additional comments from Thomas Donegan: “The
Islers have done considerable work on my point (ii) above: (study of
qualitative vocal differences).
There are apparently diagnosable differences in the second section of
the song as per my comments above; and the Islers have now also come up with a
further “change in note structure” difference that Chaves et al. (2010) did not
elucidate. Based on that study,
the nigricauda / palliata split is
supported. Bret Whitney nicely
summarises the reasons for doing this.
“It is pleasing
that the authors have looked into some of the records at range
extremities. When I heard of the
Isler study last week, I also decided to look further into these issues,
obtaining a photo of the “nigricauda”
specimen at Colegio San Jose supposedly identified by Cuádros (new catalogue
number CSJ-a 2219). It is indeed
of a male Cercomacra parkeri/tyrannina,
probably the latter based on elevation.
The curator is now correcting their database and labels. As noted in my earlier comments, the
specimens needed to be checked before coming to any conclusions! Separately, Reserva Natural de Aves
“Las Tangaras” in the northern West Andes may be an ideal candidate locality
for future studies, as it lies between known localities for nigricauda and palliata in the northern West Andes. There are no records of these birds there to date in the few
studies that have taken place, only M.
berlepschi in lower elevation forest below the reserve. That locality deserves further study
for these birds.
“One response
to Bret Whitney’s comments: the discussion set out above was not based on a
rigid interpretation of the “three characters” yardstick, but on the authors
only describing a single difference and not describing it in a manner
consistent with the sample. The
Islers have now come up with two better-defined diagnosable qualitative
differences in loudsong, which are consistent with available recordings. The Chaves et al. “two differences”
yardstick is consistent with studies of Hypocnemis
(Isler et al. 2007) - assuming the observed differences in call
stand up to analysis.
Moreover, it may be supported once a forthcoming publication on some
other Myrmeciza including Myrmeciza goeldii / melanoceps is
out. In a study of these and
related species, including analysis of over 140 loudsongs for each of goeldii and melanoceps and tens of other vocalisations, there would appear to
be only a single diagnosable difference in loudsong – also in “the note
structure of certain notes at a particular point in time or segment”. This pair is also apparently allopatric
lowland Myrmeciza species with
broadly similar calls. As a
disclaimer, the paper is still being finalised, there are strong bare-skin as
well as plumage differences and they are in a different section of the
paraphyletic Myrmeciza tree. But this would otherwise seem a very
analogous situation to M. nigricauda /
palliata in a more closely related group than Hypocnemis and one that will give support to Chaves et al.’
proposals in light of the Isler study.
“I would thank
the authors and others for considering my concerns in such a detailed, sensible
and collegiate manner, and would now support this proposal, if not entirely for
the reasons set out in the Chaves et al., paper, then for the reasons set out
in this discussion. The authors
and Islers should also be encouraged to produce a short note for Condor making a few corrections and
discussing some of the additional points in this exchange of
communications. This proposal may
have taken up a lot of space on the SACC website, but hopefully it makes for
interesting reading and everyone involved in this discussion now seems to be in
agreement.
“[Thanks to
Danny Zurc (Museo de Ciencias Naturales de La Salle, un proyecto cultural del
Instituto Tecnologico Metropolitano) for the photograph of the CSJ specimens;
and Alonso Quevedo and Trevor Ellery for information on birds at Las
Tangaras.]”
Comments from Stiles: “YES,
especially given the additional information provided by the Islers that
resolved Donegan’s (reasonable) doubts regarding vocalizations. Perhaps unfortunate that Chaves et al.
did not discuss plumage characters, because males of palliata and nigricauda
differ strongly, females more subtly but consistently, adding another set of
characters favoring species status.”
Comments from Pacheco: “YES. Após
e a análise adicional dos
Islers, a partir das construtivas
colocações de Donegan.”
Comments from Robbins:
“YES. The process that has occurred during the evaluation
of this proposal is precisely how we want the Committee to function. All
parties should be congratulated not only in the new insights that were brought
to bear on this perplexing problem, but also in the manner in which it was
presented.”
Comments from Nores: “YES. Song differences (Xeno-canto) are noticeable. Moreover, the molecular analysis by Chavez et al. (2010) clearly shows a highly supported clade for the laemosticta complex with the four taxa forming a polytomy.”
Comments from Pérez-Emán: “YES.
As indicated by Mark, the evaluation of this proposal is a clear example of a
constructive peer-reviewed process leading to a stronger documentation of the
evidence supporting species status in this group. As Donegan pointed out, a
short note including some of the issues raised here and providing the new
evidence seems to be warranted.”