Proposal (582) to South American Classification Committee


Split Pterodroma heraldica and P. atrata from P. arminjoniana


Effect on South American CL: This proposal would separate Pterodroma arminjoniana into three species; the two Pacific taxa are not yet known to occur in South American waters.


Background: The genus Pterodroma holds some vexing issues with regards to their taxonomy. This particular case is one of the most problematic, as it is one of the few in which recent data confirm a certain level of hybridization in a situation of secondary contact, whereas in the rest of the distribution, members of the complex appear not to hybridize even though sympatric! It is also problematic in that a relatively large number of taxa are involved in this complex of tropical and wide ranging Pterodroma. The taxa are: the Kermadec Petrel (Pterodroma neglecta); the Herald Petrel (Pterodroma arminjoniana heraldica) of the Pacific; the “Trindade Petrel” (Pterodroma arminjoniana arminjoniana) of the Atlantic; the “Henderson Petrel” (P. a. “atrata”), which historically has been considered the dark morph of the Herald Petrel; and finally the Phoenix Petrel (P. alba), which is not usually treated as a taxon with any problematic issues, but is closely related and sympatric with up to three of the taxa in the complex. The group has been considered a subgenus within Pterodroma, known as Hallstroma (see Imber 1985). Murphy and Pennoyer (1952) in their review of the group lumped heraldica with arminjoniana; Imber (1985) separated them as species, and many authorities follow this treatment.


Sympatry and Natural History details:


1)   These petrels are surface nesters, they do not nest in burrows, and colony visits occur mainly in the late afternoon into the evening. They are not strictly nocturnal as most Pterodroma.

2)   In the Pacific, Herald Petrel (P.a. heraldica) and Kermadec Petrel (P. neglecta) are widely sympatric. There are no records of hybridization. The two are considered biological species by all modern authorities, including SACC.

3)   Similarly, Phoenix Petrel (P. alba) is widely sympatric with Herald and Kermadec petrels, and there are no records of hybridization. It too is widely considered a separate biological species, although recent information suggests it is quite closely related to the Herald Petrel.

4)   Vocalizations of Kermadec and Herald petrels are well known, and are quite different. Spectrograms available in Brown et al. (2010). Voices of all other members of the complex are similar to each other and to Herald Petrel.

5)   Kermadec Petrel is the largest, most robust species of the ones treated here. It is also the only one that shows white primary shafts on dorsal side of wing, a feature easily visible in the field and in specimens.

6)   Kermadec (P. neglecta) is polymorphic and highly variable with dark, pale, intermediate and white-headed (pale extreme) plumages.

7)   Herald (P. a. heraldica) is polymorphic (dark and pale); but if “Henderson Petrel” – P. atrata is separated from it, heraldica becomes  monomorphic based on current knowledge.

8)   Trindade (P. a. arminjoniana) is polymorphic with pale, dark and intermediate plumages.

9)   Trindade Petrel was named by Giglioli and Salvadori (1869) from waters near Trindade Island (Brazil). The type was a white-bellied pale morph. Note that other names have been associated with this species, including “trinitatis” for the dark morph birds, and “wilsoni” for intermediate morph birds; see Murphy and Pennoyer (1952) for more details.

10) The taxon heraldica was named by Salvin in 1888, from a bird caught at sea near Chesterfield Island, NW of New Caledonia.

11) Nominate arminjoniana show sexual dimorphism in size (males larger), whereas heraldica does not (Murphy and Pennoyer 1952). 

12) The form heraldica is significantly smaller in measurements than arminjoniana; see Murphy and Pennoyer (1952), Brooke and Rowe (1996).

13) In all plumages arminjoniana is separable in the hand and field from heraldica based on overall coloration, extent of white on underwing, and color of lores (dark vs. light), as well as different structure and size and bill proportions (Howell 2012).

14) Imber (1985) separated heraldica from arminjoniana based on size differences, distribution in separate ocean basins, and because they host different species of Halipeurus feather lice.


Recent information - Henderson Petrel: Murphy and Pennoyer (1952) were the first to note that the dark version of the heraldica is not evenly distributed through the Pacific, but is clustered in Henderson Island, whereas as few as 200 miles away at Ducie Atoll there were almost no dark birds, at Henderson an estimate of 40 dark to 1 light is given. No other nesting area showed a preponderance of dark birds, although the distribution of morphs in Kermadec (neglecta) is much more evenly distributed. Brooke and Rowe (1996) studied nesting petrels on Henderson Island, and discovered that:


a)   Individuals were either dark or light, with no intermediates. 

b)   Dark and pale birds did not differ significantly in linear dimensions.

c)   19 pairs for which both males and females were captured mated assortatively by plumage darkness; the probability of this happening at random was low P < 0.0001).

d)   Observations of displaying birds over the island included 63 dark-dark; 21 light-light, and 2 mixed.

e)   Three dark pairs studied raised dark offspring.

f)     Voices are structurally similar, a series of “kyek” notes given quickly; however, dark birds call significantly faster, showing limited overlap. Pitch of notes is slightly lower in dark birds, and differs significantly although with much overlap.

g)   No dark birds observed on other islands visited other than Henderson, including in the very large Herald Petrel colony (10 – 100K individuals) on Ducie.

h)   13 haplotypes (307 base pair sequence, cyt b) were found in 76 individuals; 5 haplotypes found only in light heraldica, and three only in the dark Henderson Island birds. These were mutually exclusive between them.

i)     Of the 13 haplotypes, 3 were shared between arminjoniana, heraldica and alba; two were exclusive to heraldica; and five were exclusive to arminjoniana. Note that arminjoniana samples are from Round Island NOT the type locality of Trindade Island (see below for the significance of this).

j)     Jaramillo et al. (2008) first reported Henderson Petrels on Rapa Nui, Chile, where heraldica is common. Ongoing monitoring has found that 3-4 nests of dark birds occur there, and nesting is always assortative (Pedro Lazo pers. comm.; Jaramillo unpublished data).

k)   Brooke and Rowe (1996) resurrected the name Pterodroma atrata (Mathews 1912), type from Henderson Island (AMNH 191641), based on the above data, clarifying species status for this population. They give it the name Henderson Petrel.


Recent information – heraldica compared to arminjoniana: In their study of Pacific Herald Petrels, Brooke and Rowe (1996) also summarized aspects of this pair of taxa. The found that:


a)   Statistical significance in size difference, confirming the general assertion made by Murphy and Pennoyer (1952). Pacific heraldica is smaller than arminjoniana.

b)    At least on Round Island (no data was available to them from Trindade Island), arminjoniana does not mate assortatively with respect to morph. Furthermore intermediate morphs occur in that form, which are known also from Trindade Island (previously given the name wilsoni).

c)   There are mtDNA haplotype frequency differences between arminjoniana (from Round Island) and heraldica. Of the 13 haplotypes (see above Brooke and Rowe 1996 data), 3 were shared between arminjoniana, heraldica, and alba; two were exclusive to heraldica; and five were exclusive to arminjoniana. Results suggest incomplete reproductive isolation, or polymorphism that predates the divergence of the two forms (three including alba!). But see below regarding the complexity of Round Island.

d)   Also - Voice data are few, but arminjoniana appears to call more quickly than heraldica, similar to the difference between atrata and heraldica (i.e. Xeno-canto, Brooke et al 2000). More data are needed, although voice is perhaps not as important in this group as in other Pterodroma, because these are diurnal visitors to the colony. Visual cues are likely of equal if not greater importance, such as dark coloration of atrata vs. sympatric nesting heraldica.

e)   Imber (1985) treated heraldica and arminjoniana as separate species based on their radically different intestinal structure and different feather lice species. Interestingly intestinal structure similarities as well as shared feather lice were found between arminjoniana and neglecta (Kermadec Petrel). Recall that the latter is widely sympatric with clear barriers to hybridization with heraldica throughout the Pacific.


Recent information – Round Island breeding Pterodroma: Round Island is in the Indian Ocean, and Pterodroma petrels appear to have colonized recently, because the first definitive records of nesting petrels there occurred in the 1940s (Brooke et al. 2000, Brown et al. 2010, 2011). Murphy and Pennoyer (1952) identified the Round Island petrels as arminjoniana based on size; they considered Round Island Petrels to be somewhat intermediate between heraldica and arminjoniana, and this is part of their basis for lumping the two taxa, although details of this intermediacy were not provided. Brooke et al. (2000) surprisingly determined that Kermadec Petrels are also present on Round Island, based on voice, morphology, and mtDNA; it is unclear when their arrival occurred, but they were not noted previous to the 1980s. In the 1990s a third phenotype was observed on Round Island, thought to be Pacific heraldica, and this eventually included a banded bird from Australia, which confirmed the assertion (Brown et al. 2011).

Brown et al. (2011) studied the interaction of these three taxa in secondary contact, the Atlantic arminjoniana with the two Pacific heraldica and neglecta species, on Round Island. They sequenced a 995 base pair fragment of the cyt b mitochondrial gene from 127 individuals, including Trindade Island arminjoniana (n=21), Round Island arminjoniana (n=26), intermediate birds (n=11), white-shafted (= neglecta type) birds from Round Island (n=8), as well as neglecta, heraldica, and atrata (Henderson Petrel) from Pacific breeding stations. This study results in the following:


a)   The 23 distinct haplotypes cluster out into four phylogroups, pertaining to arminjoniana, heraldica, neglecta and atrata. The genetic dataset confirms the contribution of arminjoniana, neglecta, and heraldica genes in the Round Island Petrel population.

b)   The haplotype network shows Trindade arminjoniana haplotypes are restricted to one region of the network, as are Pacific heraldica and Pacific Kermadec Petrels; atrata (Henderson) haplotypes are not shared with any other group and are centrally located in the network, suggesting they could be ancestral in the group.

c)   Trindade arminjoniana haplotypes are mutually exclusive from Pacific heraldica.

d)   Birds from Round Island of various morphologies (dark shafted, pale shafted, intermediate) share haplotypes with various populations, suggesting gene flow here through hybridization.

e)   Feather lice of Round Island petrels are Halipeurus heraldicus, as opposed to H. kermadecensis, which is found on Trindade arminjoniana. This suggests a host switch, most likely due to hybridization events there. Lice move from bird to bird during direct contact.

f)     Populations of Trindade arminjoniana and Pacific Kermadec petrels share no haplotypes; therefore, overlap of these haplotypes on Round Island is due to hybridization.

g)   The single heraldica type bird (based on morphology and genetics) sampled from Round Island was breeding with a dark morph dark-shafted individual (dark arminjoniana type).

h)   Of the 45 Round Island petrels sampled, only one was a heraldica; it is rare there.

i)  This is a rare example of a multi-species hybridization event in nature, in secondary contact, as this island was previously not used as a breeding station by these petrels. It appears that habitat changes on the island have allowed for petrels to colonize.

j)     Leakage of genes between species is occurring on this isolated island, but there is no evidence that gene flow is occurring in the main populations of these petrels.

k)   Given the differences in phenotype, anatomy, calls, and ectoparasites among these taxa, as well as the uneven distribution of haplotypes, it seems unlikely that they represent a single polytypic species complex.


Perhaps initial hybridization events occurred due to the rarity of finding a suitable mate by the few initial scouts of the second species to venture to this island?


Summary: This is a complex relationship muddied by a rare and unique situation, which is Round Island, where secondary contact and hybridization is ongoing, although no clear evidence of gene flow outside of this system has been found. That is to say that there is no introgression between Pacific and Atlantic populations of these taxa, and introgression is restricted to a single island in the Indian Ocean.


         What we do know is the following:


1)   In the Pacific, heraldica, neglecta, alba, and atrata are usually sympatric with one or two members (sometimes all as in Easter Island) of the complex, and are reproductively isolated. They behave as biological species.

2)   The mtDNA data suggests distinctiveness of neglecta, heraldica, arminjoniana, and atrata.

3)   These diurnal Pterodroma display and vocalize over nesting islands; the voice of neglecta is very distinct. The voices of the remaining taxa are similar. Voice of atrata and heraldica differ significantly in speed of delivery; these two similarly sounding forms are reproductively isolated and sympatric on at least two islands. The few recordings and spectrograms of arminjoniana suggest a speed difference between it and heraldica of the magnitude seen in atrata-heraldica.

4)   Plumage and structure may be important in mate selection, as is suggested by the plumage difference between reproductively isolated and sympatric heraldica and atrata. On the other hand arminjoniana is polymorphic (like neglecta), and in these forms coloration is likely not important in mate selection. In sympatry neglecta is larger and bulkier, and has white primary shafts and different voice that differentiates it to heraldica. Atlantic arminjoniana have no other member of the complex breeding in the area, and may not have as well developed barriers to avoid hybridization due to their isolation from related forms. All things being equal, heraldica may avoid breeding with dark arminjoniana based on coloration, and possibly due to vocal differences.

5)   This is a messy situation with similar and closely related taxa involved. But if heraldica, atrata, and alba are different species based on voice, coloration, lack of interbreeding in sympatry, and mtDNA haplotype distributions, then the allopatric arminjoniana should also be considered a separate species, not conspecific with heraldica. Otherwise it may be equally valid to lump arminjoniana with atrata, as opposed to heraldica! Or based on widespread hybridization with neglecta on Round Island, and their shared intestinal morphology, an argument could also be made for neglecta and arminjoniana as conspecific although other data refute this. Neither of these two latter options is satisfactory, and the most even-handed approach would be to separate arminjoniana from heraldica, which makes for a much more even and logical treatment of this group – particularly given its complexities.

6)   On Round Island most birds are arminjoniana and Kermadecs, and these are the ones that interbreed freely. Kermadec is widely sympatric with heraldica in the Pacific, and the two are not known to hybridize anywhere in the Pacific. On Round Island where there are few heraldica, only one hybridization event has been documented with arminjoniana, although genetic work clarifies that it has happened in previous years as well. It is unclear, but heraldica may be a vagrant or rare enough there that to consider that a population of it exists on Round Island may be a stretch. This is not detailed clearly in the papers.


Distribution in Neotropics: Kermadec Petrel (P. neglecta) breeds in our region with a population in the Juan Fernandez archipelago; this population is sometimes afforded subspecies status as P. n. juana.  Its at-sea range is not well known, but it is thought to be found offshore in Peru. Trindade Petrel (P. arminjoniana) also breeds in our region in the Atlantic islands of Trindade (Trindade) and Martim Vas off Brazil. It ranges offshore to the north, including the Gulf Stream off the US east coast. Savigny et al. (2005) reported a sight record from Argentina of Pterodroma arminjoniana. Further observations may determine that the Herald Petrel (P. heraldica) is a rare visitor to offshore waters of Galapagos or Peru, but this has not yet been confirmed.


English Names: The form atrata is the easy one: it should be named Henderson Petrel, a name used already and with little resistance. Although the composite species is named Herald Petrel, in this case leaving heraldica as the Herald Petrel makes sense for various reasons: (1) it matches the scientific name, (2) the period in which heraldica and arminjoniana were joined has not been long, and (3) heraldica as Herald has still been understood by many seabird researchers, and is well entrenched and used widely and for years now, often meant only for the Pacific population. For arminjoniana it is a bit trickier, because the English name that is widely used is often misspelled or misused. The name Trindade Petrel seems to have the most traction because that is the Portuguese name (= Trinity Island) of the main island where it breeds; it is sometimes erroneously listed as Trinidad Petrel (the island is sometimes known as South Trinidad Island), and an Anglicized version “Trinidade” has also been proposed. It seems to me that the Portuguese and official international name of the island is “Trindade,” so the petrel should bear this name. Arminjon’s Petrel is an old name, but it was originally given only to the pale morph, and currently has little to no use.


Recommendation: My recommendation is to split Pterodroma arminjoniana into three components, one of which is found in the Neotropics:


1)   Trindade Petrel Pterodroma arminjoniana – Atlantic breeder

2)   Herald Petrel Pterodroma heraldica – Pacific breeder.

3)   Henderson Petrel Pterodroma atrata – Pacific breeder.



Literature Cited:

Brooke, M. de L., Imber, M. J., and Rowe, G. 2000. Occurrence of two surface-breeding species of Pterodroma on Round Island, Indian Ocean. Ibis 142: 139-158.

Brooke, M. de L., and Rowe, G. 1996. Behavioural and molecular evidence for specific status of light and dark morphs of the Herald Petrel Pterodroma heraldica. Ibis 138: 420-432.

Brown R.M., Nichols R.A., Faulkes C.G., Jones C.G., Bugoni L., et al. (2010) Range expansion and hybridization in Round Island petrels (Pterodroma arminjoniana); evidence from microsatellite genotypes. Mol. Ecol. 19: 3157–3170.

Brown R.M., Jordan W.C., Faulkes C.G., Jones C.G., Bugoni L., et al. (2011) Phylogenetic relationships in Pterodroma petrels are obscured by recent secondary contact and hybridization. PLoS ONE 6(5): e20350. doi:10.1371/journal.pone.0020350

Howell, S. N. G. 2012. Petrels, Albatrosses & Storm – Petrels of North America. Princeton University Press, NJ.

Imber, M. J. 1985. Origins, phylogeny and taxonomy of the gadfly petrels Pterodroma spp. Ibis 127: 197-229.

Jaramillo, A., Johnson, M.T., Rothfelds, C. J., Johnson, R. A. 2008. The native and exotic avifauna of Easter Island: then and now. Boletín Chileno de Ornitología 14(1): 8-21.

Murphy, R. C. & Pennoyer, J.M. 1952. Larger petrels of the genus Pterodroma. Am. Mus. Novit. 1580: 1-43.

Savigny, C.; Caille, G.; González, R.; Harris, G. (2005) El petrel de Trinidae (Pterodroma arminjoniana) en el Golfo de San Matìas: una nueva especie para Argentina. Hornero 020 (02) : 183-186


Alvaro Jaramillo in collaboration with Peter Harrison, George Armistead, Ned Brinkley, Brian Patteson, and Angus Wilson.


July 2013




Comments from Stotz: “YES.  I’m hopeful that others will weigh in on this proposal, as I am uncertain.  My current vote is based on it seeming like the best response to a complicated story, but I could be talked into a different approach, especially given the evidence for hybridization occurring on Round Island in the complex.”


Comments from Stiles: “YES. This would be a pretty straightforward proposal were it not for the Round Island hodgepodge, which is really a fascinating situation!  Leaving Round Island in limbo for the moment, the combination of morphological and vocal differences, situations of sympatry or far distant allopatry or effective parapatry makes splitting arminjoniana logical. The situation on Round island is weird, and deserves monitoring - but the likelihood of the hybridizing species genetically "reinfecting" their ancestral populations seems remote, and should not be seen as an obstacle for clarifying the situation in the birds' natural populations. A neat natural experiment is underway.. reticulate speciation has occurred in a somewhat similar situation in Galapagos finches and might be one end result (?). In any case, the situation seems as near as one can get in the real world to the situation in captivity, where various hybrids have occurred between Crax species, for instance.”


Comments from Pacheco: “YES.  Great explanation of a complex situation! In view of the Alvaro´s explanatory and the data presented by Brown & Rowe (1996), I consider quite plausible to accept the two splits, for now, as the best treatment.”


Comments from Nores: “YES. Alvaro's proposal is quite convincing, but a little difficult to assess. The observations of Brooke and Rowe (1996) on Henderson Island of the light and dark morphs of Herald Petrels are good evidence that they are separate species. They found evidence of reproductive isolation between both morphs. The morphs bred and courted assortatively.  They also observed that they tended to breed in different parts of the island (dark birds nearer to the coast) and at slightly different seasons (dark birds do mostly in the winter, light birds more evenly throughout the year).”


Comments from Robbins: “YES.  Alvaro had done a good job of distilling this complicated issue.  Clearly, it is a matter of opinion on whether these forms continue to be treated as a single species or divide them into three.  Aside from the unusual situation on Round Island, it appears that these taxa are not hybridizing elsewhere and given the differences in morphology, genetics and intestinal flora in other areas of sympatry, I lean towards treating all three as species. However, I certainly do not have a strong opinion either way.  For now, I vote yes for recognizing all three.”


Comments from Pérez-Emán “YES.  One can go either way with this proposal. Information on differences and similarities are far from simple to assess the species status of these taxa. For example, intestinal structure and feather lice species are different in heraldica and arminjoniana, characters used to treat them as species. However, those characters are not useful to separate arminjoniana and neglecta (in fact, they are similar in these species). Vocal differences appear to be minor and Alvaro also suggested that voice might not be important in this group because of their diurnal habits. Size differences seem to be consistent but this is a character that could be geographically variable. Important data for this proposal is the lack of interbreeding in areas of sympatry and observations on assortative mating for some of these taxa color morphs. It is really intriguing why taxa, which do not hybridize in part of their area of distribution, interbreed in a recently colonized island (or better, what are the factors associated to reproductive isolation in most of their geographical range?). It is also interesting to learn that there are few places where one can find more than one of these Pterodroma species (see Fig. 1 from Brown et al. (2011)). Small genetic differentiation among these taxa could be associated to recent events of colony establishment in places unoccupied by these species, with posterior isolation through philopatry? At any rate, this study shows that reproductive isolation barriers could be broken at any time on the lifetime history of a species and the question is if there is potential for population/species homogenization in the future. It seems, from the data we have now, that it is a situation that could happen on a local basis but not necessarily throughout the geographical distribution of these taxa. Thus, I would incline for a YES. Also, a NO vote would require lumping more than one currently recognized species into one.”